From 2338648ace2642fe6c544406cf3714e3969762a6 Mon Sep 17 00:00:00 2001
From: pytorchbot <soumith+bot@pytorch.org>
Date: Tue, 29 Oct 2024 14:32:32 +0000
Subject: [PATCH] auto-generating sphinx docs

---
 main/_modules/functools.html                  |   2 +-
 main/_modules/index.html                      |   2 +-
 .../torchtune/config/_instantiate.html        |   2 +-
 main/_modules/torchtune/config/_parse.html    |   2 +-
 main/_modules/torchtune/config/_utils.html    |   2 +-
 main/_modules/torchtune/config/_validate.html |   2 +-
 main/_modules/torchtune/data/_collate.html    |   2 +-
 main/_modules/torchtune/data/_messages.html   |   2 +-
 .../torchtune/data/_prompt_templates.html     |   2 +-
 main/_modules/torchtune/data/_utils.html      |   2 +-
 main/_modules/torchtune/datasets/_alpaca.html |   2 +-
 main/_modules/torchtune/datasets/_chat.html   |   2 +-
 .../torchtune/datasets/_cnn_dailymail.html    |   2 +-
 main/_modules/torchtune/datasets/_concat.html |   2 +-
 .../_modules/torchtune/datasets/_grammar.html |   2 +-
 .../torchtune/datasets/_hh_rlhf_helpful.html  |   2 +-
 .../torchtune/datasets/_instruct.html         |   2 +-
 main/_modules/torchtune/datasets/_packed.html |   2 +-
 .../torchtune/datasets/_preference.html       |   2 +-
 main/_modules/torchtune/datasets/_samsum.html |   2 +-
 main/_modules/torchtune/datasets/_sft.html    |   2 +-
 .../torchtune/datasets/_slimorca.html         |   2 +-
 .../datasets/_stack_exchange_paired.html      |   2 +-
 .../torchtune/datasets/_text_completion.html  |   2 +-
 .../torchtune/datasets/_wikitext.html         |   2 +-
 .../datasets/multimodal/_llava_instruct.html  |   2 +-
 .../datasets/multimodal/_the_cauldron.html    |   2 +-
 .../torchtune/generation/_generation.html     |   2 +-
 .../models/clip/_component_builders.html      |   2 +-
 .../models/clip/_position_embeddings.html     |   2 +-
 .../models/code_llama2/_model_builders.html   |   2 +-
 .../models/gemma/_component_builders.html     |   2 +-
 .../models/gemma/_model_builders.html         |   2 +-
 .../models/llama2/_component_builders.html    |   2 +-
 .../models/llama2/_model_builders.html        |   2 +-
 .../models/llama2/_prompt_template.html       |   2 +-
 .../models/llama3/_component_builders.html    |   2 +-
 .../models/llama3/_model_builders.html        |   2 +-
 .../models/llama3_1/_component_builders.html  |   2 +-
 .../models/llama3_1/_model_builders.html      |   2 +-
 .../models/llama3_2/_model_builders.html      |   2 +-
 .../llama3_2_vision/_component_builders.html  |   2 +-
 .../models/llama3_2_vision/_encoder.html      |   2 +-
 .../llama3_2_vision/_model_builders.html      |   2 +-
 .../models/llama3_2_vision/_transform.html    |   2 +-
 .../models/mistral/_component_builders.html   |   2 +-
 .../models/mistral/_model_builders.html       |   2 +-
 .../models/mistral/_prompt_template.html      |   2 +-
 .../models/phi3/_component_builders.html      |   2 +-
 .../models/phi3/_model_builders.html          |   2 +-
 .../models/qwen2/_component_builders.html     |   2 +-
 .../models/qwen2/_model_builders.html         |   2 +-
 .../_modules/torchtune/modules/attention.html |   2 +-
 .../torchtune/modules/common_utils.html       |   2 +-
 .../torchtune/modules/feed_forward.html       |   2 +-
 main/_modules/torchtune/modules/kv_cache.html |   2 +-
 .../torchtune/modules/layer_norm.html         |   2 +-
 .../modules/loss/ce_chunked_output_loss.html  |   2 +-
 .../torchtune/modules/loss/kd_losses.html     |   2 +-
 .../modules/model_fusion/_fusion.html         |   2 +-
 .../modules/model_fusion/_fusion_utils.html   |   2 +-
 .../torchtune/modules/peft/_utils.html        |   2 +-
 .../_modules/torchtune/modules/peft/lora.html |   2 +-
 .../modules/position_embeddings.html          |   2 +-
 main/_modules/torchtune/modules/rms_norm.html |   2 +-
 .../_modules/torchtune/modules/tanh_gate.html |   2 +-
 .../torchtune/modules/tied_linear.html        |   2 +-
 .../modules/tokenizers/_sentencepiece.html    |   2 +-
 .../modules/tokenizers/_tiktoken.html         |   2 +-
 .../torchtune/modules/tokenizers/_utils.html  |   2 +-
 .../torchtune/modules/transformer.html        |   2 +-
 .../modules/transforms/_transforms.html       |   2 +-
 .../torchtune/modules/vision_transformer.html |   2 +-
 main/_modules/torchtune/rlhf/loss/dpo.html    |   2 +-
 main/_modules/torchtune/rlhf/loss/ppo.html    |   2 +-
 main/_modules/torchtune/rlhf/rewards.html     |   2 +-
 .../torchtune/rlhf/sequence_processing.html   |   2 +-
 .../torchtune/training/_distributed.html      |   2 +-
 .../torchtune/training/_profiler.html         |   2 +-
 .../torchtune/training/activations.html       |   2 +-
 .../training/checkpointing/_checkpointer.html |   2 +-
 .../training/checkpointing/_utils.html        |   2 +-
 .../torchtune/training/lr_schedulers.html     |   2 +-
 main/_modules/torchtune/training/memory.html  |   2 +-
 .../torchtune/training/metric_logging.html    |   2 +-
 main/_modules/torchtune/training/pooling.html |   2 +-
 .../torchtune/training/precision.html         |   2 +-
 .../torchtune/training/quantization.html      |   2 +-
 main/_modules/torchtune/training/seed.html    |   2 +-
 main/_modules/torchtune/utils/_device.html    |   2 +-
 main/_modules/torchtune/utils/_logging.html   |   2 +-
 main/_modules/torchtune/utils/_version.html   |   2 +-
 main/_modules/typing.html                     |   2 +-
 main/_sources/index.rst.txt                   |   2 +-
 .../tutorials/memory_optimizations.rst.txt    |  73 ++++++++++++++-
 main/api_ref_config.html                      |   2 +-
 main/api_ref_data.html                        |   2 +-
 main/api_ref_datasets.html                    |   2 +-
 main/api_ref_generation.html                  |   2 +-
 main/api_ref_models.html                      |   2 +-
 main/api_ref_modules.html                     |   2 +-
 main/api_ref_rlhf.html                        |   2 +-
 main/api_ref_training.html                    |   2 +-
 main/api_ref_utilities.html                   |   2 +-
 main/basics/chat_datasets.html                |   2 +-
 main/basics/custom_components.html            |   2 +-
 main/basics/datasets_overview.html            |   2 +-
 main/basics/instruct_datasets.html            |   2 +-
 main/basics/message_transforms.html           |   2 +-
 main/basics/messages.html                     |   2 +-
 main/basics/model_transforms.html             |   2 +-
 main/basics/multimodal_datasets.html          |   2 +-
 main/basics/packing.html                      |   2 +-
 main/basics/preference_datasets.html          |   2 +-
 main/basics/prompt_templates.html             |   2 +-
 main/basics/text_completion_datasets.html     |   2 +-
 main/basics/tokenizers.html                   |   2 +-
 main/deep_dives/checkpointer.html             |   6 +-
 main/deep_dives/comet_logging.html            |   2 +-
 main/deep_dives/configs.html                  |   2 +-
 main/deep_dives/recipe_deepdive.html          |   2 +-
 main/deep_dives/wandb_logging.html            |   2 +-
 .../torchtune.config.instantiate.html         |   2 +-
 .../torchtune.config.log_config.html          |   2 +-
 main/generated/torchtune.config.parse.html    |   2 +-
 main/generated/torchtune.config.validate.html |   2 +-
 .../torchtune.data.AlpacaToMessages.html      |   2 +-
 .../torchtune.data.ChatMLTemplate.html        |   2 +-
 ...rchtune.data.ChosenRejectedToMessages.html |   2 +-
 ...e.data.GrammarErrorCorrectionTemplate.html |   2 +-
 .../torchtune.data.InputOutputToMessages.html |   2 +-
 main/generated/torchtune.data.Message.html    |   2 +-
 .../torchtune.data.OpenAIToMessages.html      |   2 +-
 .../torchtune.data.PromptTemplate.html        |   2 +-
 ...orchtune.data.PromptTemplateInterface.html |   2 +-
 ...torchtune.data.QuestionAnswerTemplate.html |   2 +-
 main/generated/torchtune.data.Role.html       |   2 +-
 .../torchtune.data.ShareGPTToMessages.html    |   2 +-
 .../torchtune.data.SummarizeTemplate.html     |   2 +-
 ...htune.data.format_content_with_images.html |   2 +-
 .../torchtune.data.left_pad_sequence.html     |   2 +-
 main/generated/torchtune.data.load_image.html |   2 +-
 .../torchtune.data.padded_collate.html        |   2 +-
 .../torchtune.data.padded_collate_dpo.html    |   2 +-
 .../torchtune.data.padded_collate_sft.html    |   2 +-
 ....padded_collate_tiled_images_and_mask.html |   2 +-
 main/generated/torchtune.data.truncate.html   |   2 +-
 .../torchtune.data.validate_messages.html     |   2 +-
 .../torchtune.datasets.ConcatDataset.html     |   2 +-
 .../torchtune.datasets.PackedDataset.html     |   2 +-
 .../torchtune.datasets.PreferenceDataset.html |   2 +-
 .../torchtune.datasets.SFTDataset.html        |   2 +-
 ...chtune.datasets.TextCompletionDataset.html |   2 +-
 ...htune.datasets.alpaca_cleaned_dataset.html |   2 +-
 .../torchtune.datasets.alpaca_dataset.html    |   2 +-
 .../torchtune.datasets.chat_dataset.html      |   2 +-
 ...tasets.cnn_dailymail_articles_dataset.html |   2 +-
 .../torchtune.datasets.grammar_dataset.html   |   2 +-
 ...tune.datasets.hh_rlhf_helpful_dataset.html |   2 +-
 .../torchtune.datasets.instruct_dataset.html  |   2 +-
 ...ets.multimodal.llava_instruct_dataset.html |   2 +-
 ...asets.multimodal.the_cauldron_dataset.html |   2 +-
 ...torchtune.datasets.preference_dataset.html |   2 +-
 .../torchtune.datasets.samsum_dataset.html    |   2 +-
 .../torchtune.datasets.slimorca_dataset.html  |   2 +-
 ...atasets.stack_exchange_paired_dataset.html |   2 +-
 ...tune.datasets.text_completion_dataset.html |   2 +-
 .../torchtune.datasets.wikitext_dataset.html  |   2 +-
 .../torchtune.generation.generate.html        |   2 +-
 ...chtune.generation.generate_next_token.html |   2 +-
 ...ion.get_causal_mask_from_padding_mask.html |   2 +-
 ...on.get_position_ids_from_padding_mask.html |   2 +-
 .../torchtune.generation.sample.html          |   2 +-
 ...e.models.clip.TilePositionalEmbedding.html |   2 +-
 ...ls.clip.TiledTokenPositionalEmbedding.html |   2 +-
 ....models.clip.TokenPositionalEmbedding.html |   2 +-
 ...htune.models.clip.clip_vision_encoder.html |   2 +-
 ...ne.models.code_llama2.code_llama2_13b.html |   2 +-
 ...ne.models.code_llama2.code_llama2_70b.html |   2 +-
 ...une.models.code_llama2.code_llama2_7b.html |   2 +-
 ...dels.code_llama2.lora_code_llama2_13b.html |   2 +-
 ...dels.code_llama2.lora_code_llama2_70b.html |   2 +-
 ...odels.code_llama2.lora_code_llama2_7b.html |   2 +-
 ...els.code_llama2.qlora_code_llama2_13b.html |   2 +-
 ...els.code_llama2.qlora_code_llama2_70b.html |   2 +-
 ...dels.code_llama2.qlora_code_llama2_7b.html |   2 +-
 .../torchtune.models.gemma.gemma.html         |   2 +-
 .../torchtune.models.gemma.gemma_2b.html      |   2 +-
 .../torchtune.models.gemma.gemma_7b.html      |   2 +-
 ...orchtune.models.gemma.gemma_tokenizer.html |   2 +-
 .../torchtune.models.gemma.lora_gemma.html    |   2 +-
 .../torchtune.models.gemma.lora_gemma_2b.html |   2 +-
 .../torchtune.models.gemma.lora_gemma_7b.html |   2 +-
 ...torchtune.models.gemma.qlora_gemma_2b.html |   2 +-
 ...torchtune.models.gemma.qlora_gemma_7b.html |   2 +-
 ...tune.models.llama2.Llama2ChatTemplate.html |   2 +-
 .../torchtune.models.llama2.llama2.html       |   2 +-
 .../torchtune.models.llama2.llama2_13b.html   |   2 +-
 .../torchtune.models.llama2.llama2_70b.html   |   2 +-
 .../torchtune.models.llama2.llama2_7b.html    |   2 +-
 ...chtune.models.llama2.llama2_reward_7b.html |   2 +-
 ...chtune.models.llama2.llama2_tokenizer.html |   2 +-
 .../torchtune.models.llama2.lora_llama2.html  |   2 +-
 ...rchtune.models.llama2.lora_llama2_13b.html |   2 +-
 ...rchtune.models.llama2.lora_llama2_70b.html |   2 +-
 ...orchtune.models.llama2.lora_llama2_7b.html |   2 +-
 ...e.models.llama2.lora_llama2_reward_7b.html |   2 +-
 ...chtune.models.llama2.qlora_llama2_13b.html |   2 +-
 ...chtune.models.llama2.qlora_llama2_70b.html |   2 +-
 ...rchtune.models.llama2.qlora_llama2_7b.html |   2 +-
 ....models.llama2.qlora_llama2_reward_7b.html |   2 +-
 .../torchtune.models.llama3.llama3.html       |   2 +-
 .../torchtune.models.llama3.llama3_70b.html   |   2 +-
 .../torchtune.models.llama3.llama3_8b.html    |   2 +-
 ...chtune.models.llama3.llama3_tokenizer.html |   2 +-
 .../torchtune.models.llama3.lora_llama3.html  |   2 +-
 ...rchtune.models.llama3.lora_llama3_70b.html |   2 +-
 ...orchtune.models.llama3.lora_llama3_8b.html |   2 +-
 ...chtune.models.llama3.qlora_llama3_70b.html |   2 +-
 ...rchtune.models.llama3.qlora_llama3_8b.html |   2 +-
 .../torchtune.models.llama3_1.llama3_1.html   |   2 +-
 ...rchtune.models.llama3_1.llama3_1_405b.html |   2 +-
 ...orchtune.models.llama3_1.llama3_1_70b.html |   2 +-
 ...torchtune.models.llama3_1.llama3_1_8b.html |   2 +-
 ...rchtune.models.llama3_1.lora_llama3_1.html |   2 +-
 ...ne.models.llama3_1.lora_llama3_1_405b.html |   2 +-
 ...une.models.llama3_1.lora_llama3_1_70b.html |   2 +-
 ...tune.models.llama3_1.lora_llama3_1_8b.html |   2 +-
 ...e.models.llama3_1.qlora_llama3_1_405b.html |   2 +-
 ...ne.models.llama3_1.qlora_llama3_1_70b.html |   2 +-
 ...une.models.llama3_1.qlora_llama3_1_8b.html |   2 +-
 ...torchtune.models.llama3_2.llama3_2_1b.html |   2 +-
 ...torchtune.models.llama3_2.llama3_2_3b.html |   2 +-
 ...tune.models.llama3_2.lora_llama3_2_1b.html |   2 +-
 ...tune.models.llama3_2.lora_llama3_2_3b.html |   2 +-
 ...une.models.llama3_2.qlora_llama3_2_1b.html |   2 +-
 ...une.models.llama3_2.qlora_llama3_2_3b.html |   2 +-
 ...s.llama3_2_vision.Llama3VisionEncoder.html |   2 +-
 ...3_2_vision.Llama3VisionProjectionHead.html |   2 +-
 ...llama3_2_vision.Llama3VisionTransform.html |   2 +-
 ...s.llama3_2_vision.llama3_2_vision_11b.html |   2 +-
 ...ama3_2_vision.llama3_2_vision_decoder.html |   2 +-
 ...ama3_2_vision.llama3_2_vision_encoder.html |   2 +-
 ...a3_2_vision.llama3_2_vision_transform.html |   2 +-
 ...ma3_2_vision.lora_llama3_2_vision_11b.html |   2 +-
 ...2_vision.lora_llama3_2_vision_decoder.html |   2 +-
 ...2_vision.lora_llama3_2_vision_encoder.html |   2 +-
 ...a3_2_vision.qlora_llama3_2_vision_11b.html |   2 +-
 ...ne.models.mistral.MistralChatTemplate.html |   2 +-
 ...torchtune.models.mistral.lora_mistral.html |   2 +-
 ...chtune.models.mistral.lora_mistral_7b.html |   2 +-
 ...odels.mistral.lora_mistral_classifier.html |   2 +-
 ...models.mistral.lora_mistral_reward_7b.html |   2 +-
 .../torchtune.models.mistral.mistral.html     |   2 +-
 .../torchtune.models.mistral.mistral_7b.html  |   2 +-
 ...une.models.mistral.mistral_classifier.html |   2 +-
 ...tune.models.mistral.mistral_reward_7b.html |   2 +-
 ...tune.models.mistral.mistral_tokenizer.html |   2 +-
 ...htune.models.mistral.qlora_mistral_7b.html |   2 +-
 ...odels.mistral.qlora_mistral_reward_7b.html |   2 +-
 .../torchtune.models.phi3.lora_phi3.html      |   2 +-
 .../torchtune.models.phi3.lora_phi3_mini.html |   2 +-
 .../generated/torchtune.models.phi3.phi3.html |   2 +-
 .../torchtune.models.phi3.phi3_mini.html      |   2 +-
 ...htune.models.phi3.phi3_mini_tokenizer.html |   2 +-
 ...torchtune.models.phi3.qlora_phi3_mini.html |   2 +-
 .../torchtune.models.qwen2.lora_qwen2.html    |   2 +-
 ...orchtune.models.qwen2.lora_qwen2_0_5b.html |   2 +-
 ...orchtune.models.qwen2.lora_qwen2_1_5b.html |   2 +-
 .../torchtune.models.qwen2.lora_qwen2_7b.html |   2 +-
 .../torchtune.models.qwen2.qwen2.html         |   2 +-
 .../torchtune.models.qwen2.qwen2_0_5b.html    |   2 +-
 .../torchtune.models.qwen2.qwen2_1_5b.html    |   2 +-
 .../torchtune.models.qwen2.qwen2_7b.html      |   2 +-
 ...orchtune.models.qwen2.qwen2_tokenizer.html |   2 +-
 .../torchtune.modules.FeedForward.html        |   2 +-
 .../torchtune.modules.Fp32LayerNorm.html      |   2 +-
 main/generated/torchtune.modules.KVCache.html |   2 +-
 .../torchtune.modules.MultiHeadAttention.html |   2 +-
 main/generated/torchtune.modules.RMSNorm.html |   2 +-
 ...ne.modules.RotaryPositionalEmbeddings.html |   2 +-
 .../generated/torchtune.modules.TanhGate.html |   2 +-
 .../torchtune.modules.TiedLinear.html         |   2 +-
 ...odules.TransformerCrossAttentionLayer.html |   2 +-
 .../torchtune.modules.TransformerDecoder.html |   2 +-
 ...modules.TransformerSelfAttentionLayer.html |   2 +-
 .../torchtune.modules.VisionTransformer.html  |   2 +-
 ...modules.common_utils.delete_kv_caches.html |   2 +-
 ...modules.common_utils.disable_kv_cache.html |   2 +-
 ...e.modules.common_utils.local_kv_cache.html |   2 +-
 ...metrize_as_dtype_state_dict_post_hook.html |   2 +-
 ....modules.loss.CEWithChunkedOutputLoss.html |   2 +-
 .../torchtune.modules.loss.ForwardKLLoss.html |   2 +-
 ...s.loss.ForwardKLWithChunkedOutputLoss.html |   2 +-
 ....modules.model_fusion.DeepFusionModel.html |   2 +-
 ....modules.model_fusion.FusionEmbedding.html |   2 +-
 ...tune.modules.model_fusion.FusionLayer.html |   2 +-
 ...odules.model_fusion.get_fusion_params.html |   2 +-
 ...s.model_fusion.register_fusion_module.html |   2 +-
 .../torchtune.modules.peft.AdapterModule.html |   2 +-
 .../torchtune.modules.peft.LoRALinear.html    |   2 +-
 ...orchtune.modules.peft.disable_adapter.html |   2 +-
 ...htune.modules.peft.get_adapter_params.html |   2 +-
 ...une.modules.peft.set_trainable_params.html |   2 +-
 ...idate_missing_and_unexpected_for_lora.html |   2 +-
 ...les.peft.validate_state_dict_for_lora.html |   2 +-
 ...tune.modules.tokenizers.BaseTokenizer.html |   2 +-
 ...une.modules.tokenizers.ModelTokenizer.html |   2 +-
 ...tokenizers.SentencePieceBaseTokenizer.html |   2 +-
 ...ules.tokenizers.TikTokenBaseTokenizer.html |   2 +-
 ...es.tokenizers.parse_hf_tokenizer_json.html |   2 +-
 ...s.tokenize_messages_no_special_tokens.html |   2 +-
 ...orchtune.modules.transforms.Transform.html |   2 +-
 ...s.transforms.VisionCrossAttentionMask.html |   2 +-
 .../torchtune.rlhf.estimate_advantages.html   |   2 +-
 .../torchtune.rlhf.get_rewards_ppo.html       |   2 +-
 .../torchtune.rlhf.loss.DPOLoss.html          |   2 +-
 .../torchtune.rlhf.loss.PPOLoss.html          |   2 +-
 .../torchtune.rlhf.loss.RSOLoss.html          |   2 +-
 .../torchtune.rlhf.loss.SimPOLoss.html        |   2 +-
 ...truncate_sequence_at_first_stop_token.html |   2 +-
 .../torchtune.training.FSDPPolicyType.html    |   2 +-
 ...une.training.FormattedCheckpointFiles.html |   2 +-
 ...tune.training.FullModelHFCheckpointer.html |   2 +-
 ...ne.training.FullModelMetaCheckpointer.html |   2 +-
 ...aining.FullModelTorchTuneCheckpointer.html |   2 +-
 .../torchtune.training.ModelType.html         |   2 +-
 ...e.training.OptimizerInBackwardWrapper.html |   2 +-
 ...ly_selective_activation_checkpointing.html |   2 +-
 ....training.create_optim_in_bwd_wrapper.html |   2 +-
 ...ining.get_cosine_schedule_with_warmup.html |   2 +-
 .../torchtune.training.get_dtype.html         |   2 +-
 ...ng.get_full_finetune_fsdp_wrap_policy.html |   2 +-
 main/generated/torchtune.training.get_lr.html |   2 +-
 .../torchtune.training.get_memory_stats.html  |   2 +-
 ...torchtune.training.get_quantizer_mode.html |   2 +-
 ...raining.get_unmasked_sequence_lengths.html |   2 +-
 ...tune.training.get_world_size_and_rank.html |   2 +-
 .../torchtune.training.init_distributed.html  |   2 +-
 .../torchtune.training.is_distributed.html    |   2 +-
 .../torchtune.training.log_memory_stats.html  |   2 +-
 ...chtune.training.lora_fsdp_wrap_policy.html |   2 +-
 ...e.training.metric_logging.CometLogger.html |   2 +-
 ...ne.training.metric_logging.DiskLogger.html |   2 +-
 ....training.metric_logging.StdoutLogger.html |   2 +-
 ...ning.metric_logging.TensorBoardLogger.html |   2 +-
 ...e.training.metric_logging.WandBLogger.html |   2 +-
 ....training.register_optim_in_bwd_hooks.html |   2 +-
 ...training.set_activation_checkpointing.html |   2 +-
 .../torchtune.training.set_default_dtype.html |   2 +-
 .../torchtune.training.set_seed.html          |   2 +-
 ...rchtune.training.setup_torch_profiler.html |   2 +-
 ...ning.update_state_dict_for_classifier.html |   2 +-
 ...raining.validate_expected_param_dtype.html |   2 +-
 .../torchtune.utils.batch_to_device.html      |   2 +-
 .../generated/torchtune.utils.get_device.html |   2 +-
 .../generated/torchtune.utils.get_logger.html |   2 +-
 .../torchtune.utils.torch_version_ge.html     |   2 +-
 main/generated_examples/index.html            |   2 +-
 .../sg_execution_times.html                   |   2 +-
 main/genindex.html                            |   2 +-
 main/index.html                               |   2 +-
 main/install.html                             |   2 +-
 main/objects.inv                              | Bin 10917 -> 10945 bytes
 main/overview.html                            |   2 +-
 main/recipes/lora_finetune_single_device.html |   2 +-
 main/recipes/qat_distributed.html             |   2 +-
 main/recipes/recipes_overview.html            |   2 +-
 main/search.html                              |   2 +-
 main/searchindex.js                           |   2 +-
 main/sg_execution_times.html                  |   2 +-
 main/tune_cli.html                            |   2 +-
 main/tutorials/chat.html                      |   2 +-
 main/tutorials/e2e_flow.html                  |   6 +-
 main/tutorials/first_finetune_tutorial.html   |   2 +-
 main/tutorials/llama3.html                    |   2 +-
 main/tutorials/llama_kd_tutorial.html         |  10 +--
 main/tutorials/lora_finetune.html             |   2 +-
 main/tutorials/memory_optimizations.html      |  83 +++++++++++++++---
 main/tutorials/qat_finetune.html              |   2 +-
 main/tutorials/qlora_finetune.html            |   2 +-
 381 files changed, 526 insertions(+), 402 deletions(-)

diff --git a/main/_modules/functools.html b/main/_modules/functools.html
index ebd5302178..c2df1c0f9c 100644
--- a/main/_modules/functools.html
+++ b/main/_modules/functools.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/index.html b/main/_modules/index.html
index 65ea33fe3a..57592ce618 100644
--- a/main/_modules/index.html
+++ b/main/_modules/index.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/config/_instantiate.html b/main/_modules/torchtune/config/_instantiate.html
index 9dac7dcb7f..37683a48ab 100644
--- a/main/_modules/torchtune/config/_instantiate.html
+++ b/main/_modules/torchtune/config/_instantiate.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/config/_parse.html b/main/_modules/torchtune/config/_parse.html
index 7c0b90796d..91d22327a8 100644
--- a/main/_modules/torchtune/config/_parse.html
+++ b/main/_modules/torchtune/config/_parse.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/config/_utils.html b/main/_modules/torchtune/config/_utils.html
index f40bb91503..702fb37102 100644
--- a/main/_modules/torchtune/config/_utils.html
+++ b/main/_modules/torchtune/config/_utils.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/config/_validate.html b/main/_modules/torchtune/config/_validate.html
index cc31a2a761..a46f1092fd 100644
--- a/main/_modules/torchtune/config/_validate.html
+++ b/main/_modules/torchtune/config/_validate.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/data/_collate.html b/main/_modules/torchtune/data/_collate.html
index e5f4f7433f..b96d80240c 100644
--- a/main/_modules/torchtune/data/_collate.html
+++ b/main/_modules/torchtune/data/_collate.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/data/_messages.html b/main/_modules/torchtune/data/_messages.html
index 7c27c1bde1..f9b57e2270 100644
--- a/main/_modules/torchtune/data/_messages.html
+++ b/main/_modules/torchtune/data/_messages.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/data/_prompt_templates.html b/main/_modules/torchtune/data/_prompt_templates.html
index 1ce4082747..8701d712fb 100644
--- a/main/_modules/torchtune/data/_prompt_templates.html
+++ b/main/_modules/torchtune/data/_prompt_templates.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/data/_utils.html b/main/_modules/torchtune/data/_utils.html
index 368ce35201..56dbfe7a97 100644
--- a/main/_modules/torchtune/data/_utils.html
+++ b/main/_modules/torchtune/data/_utils.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/datasets/_alpaca.html b/main/_modules/torchtune/datasets/_alpaca.html
index 02c63ab6d8..b908da649f 100644
--- a/main/_modules/torchtune/datasets/_alpaca.html
+++ b/main/_modules/torchtune/datasets/_alpaca.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/datasets/_chat.html b/main/_modules/torchtune/datasets/_chat.html
index b8aae298f8..b0c624dedc 100644
--- a/main/_modules/torchtune/datasets/_chat.html
+++ b/main/_modules/torchtune/datasets/_chat.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/datasets/_cnn_dailymail.html b/main/_modules/torchtune/datasets/_cnn_dailymail.html
index cdbe42b54c..4821b47c4f 100644
--- a/main/_modules/torchtune/datasets/_cnn_dailymail.html
+++ b/main/_modules/torchtune/datasets/_cnn_dailymail.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/datasets/_concat.html b/main/_modules/torchtune/datasets/_concat.html
index 99b7d2c25c..5145be55ba 100644
--- a/main/_modules/torchtune/datasets/_concat.html
+++ b/main/_modules/torchtune/datasets/_concat.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/datasets/_grammar.html b/main/_modules/torchtune/datasets/_grammar.html
index 46a98ef47d..5634b39ba0 100644
--- a/main/_modules/torchtune/datasets/_grammar.html
+++ b/main/_modules/torchtune/datasets/_grammar.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/datasets/_hh_rlhf_helpful.html b/main/_modules/torchtune/datasets/_hh_rlhf_helpful.html
index 57238325e7..e9ce04f259 100644
--- a/main/_modules/torchtune/datasets/_hh_rlhf_helpful.html
+++ b/main/_modules/torchtune/datasets/_hh_rlhf_helpful.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/datasets/_instruct.html b/main/_modules/torchtune/datasets/_instruct.html
index fa1adc4547..fe4df1f146 100644
--- a/main/_modules/torchtune/datasets/_instruct.html
+++ b/main/_modules/torchtune/datasets/_instruct.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/datasets/_packed.html b/main/_modules/torchtune/datasets/_packed.html
index 3551a5abfa..c5667f5eb5 100644
--- a/main/_modules/torchtune/datasets/_packed.html
+++ b/main/_modules/torchtune/datasets/_packed.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/datasets/_preference.html b/main/_modules/torchtune/datasets/_preference.html
index 941fbc38f6..23e0a44171 100644
--- a/main/_modules/torchtune/datasets/_preference.html
+++ b/main/_modules/torchtune/datasets/_preference.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/datasets/_samsum.html b/main/_modules/torchtune/datasets/_samsum.html
index 0de850cd39..48cd49e156 100644
--- a/main/_modules/torchtune/datasets/_samsum.html
+++ b/main/_modules/torchtune/datasets/_samsum.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/datasets/_sft.html b/main/_modules/torchtune/datasets/_sft.html
index 60f1c91805..ca0093d2fe 100644
--- a/main/_modules/torchtune/datasets/_sft.html
+++ b/main/_modules/torchtune/datasets/_sft.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/datasets/_slimorca.html b/main/_modules/torchtune/datasets/_slimorca.html
index 809517e651..83d2ba23df 100644
--- a/main/_modules/torchtune/datasets/_slimorca.html
+++ b/main/_modules/torchtune/datasets/_slimorca.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/datasets/_stack_exchange_paired.html b/main/_modules/torchtune/datasets/_stack_exchange_paired.html
index 1454c42213..c460a4f329 100644
--- a/main/_modules/torchtune/datasets/_stack_exchange_paired.html
+++ b/main/_modules/torchtune/datasets/_stack_exchange_paired.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/datasets/_text_completion.html b/main/_modules/torchtune/datasets/_text_completion.html
index 8ded19c64e..927ec5f64c 100644
--- a/main/_modules/torchtune/datasets/_text_completion.html
+++ b/main/_modules/torchtune/datasets/_text_completion.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/datasets/_wikitext.html b/main/_modules/torchtune/datasets/_wikitext.html
index b611e53c12..e18f2b9337 100644
--- a/main/_modules/torchtune/datasets/_wikitext.html
+++ b/main/_modules/torchtune/datasets/_wikitext.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/datasets/multimodal/_llava_instruct.html b/main/_modules/torchtune/datasets/multimodal/_llava_instruct.html
index 59a476505a..a68d267cc8 100644
--- a/main/_modules/torchtune/datasets/multimodal/_llava_instruct.html
+++ b/main/_modules/torchtune/datasets/multimodal/_llava_instruct.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/datasets/multimodal/_the_cauldron.html b/main/_modules/torchtune/datasets/multimodal/_the_cauldron.html
index fd56aeeaca..ddd7c93a88 100644
--- a/main/_modules/torchtune/datasets/multimodal/_the_cauldron.html
+++ b/main/_modules/torchtune/datasets/multimodal/_the_cauldron.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/generation/_generation.html b/main/_modules/torchtune/generation/_generation.html
index ae7c1492f6..e6ed747e97 100644
--- a/main/_modules/torchtune/generation/_generation.html
+++ b/main/_modules/torchtune/generation/_generation.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/clip/_component_builders.html b/main/_modules/torchtune/models/clip/_component_builders.html
index 48c339d05d..993bc48e23 100644
--- a/main/_modules/torchtune/models/clip/_component_builders.html
+++ b/main/_modules/torchtune/models/clip/_component_builders.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/clip/_position_embeddings.html b/main/_modules/torchtune/models/clip/_position_embeddings.html
index b6106681d5..9c2f1e0905 100644
--- a/main/_modules/torchtune/models/clip/_position_embeddings.html
+++ b/main/_modules/torchtune/models/clip/_position_embeddings.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/code_llama2/_model_builders.html b/main/_modules/torchtune/models/code_llama2/_model_builders.html
index c98013cb1e..cbfaac1bfe 100644
--- a/main/_modules/torchtune/models/code_llama2/_model_builders.html
+++ b/main/_modules/torchtune/models/code_llama2/_model_builders.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/gemma/_component_builders.html b/main/_modules/torchtune/models/gemma/_component_builders.html
index bcd6d26bf0..388ab01a0f 100644
--- a/main/_modules/torchtune/models/gemma/_component_builders.html
+++ b/main/_modules/torchtune/models/gemma/_component_builders.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/gemma/_model_builders.html b/main/_modules/torchtune/models/gemma/_model_builders.html
index 37506c6575..3d6f4d237f 100644
--- a/main/_modules/torchtune/models/gemma/_model_builders.html
+++ b/main/_modules/torchtune/models/gemma/_model_builders.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/llama2/_component_builders.html b/main/_modules/torchtune/models/llama2/_component_builders.html
index 5ad269c9a8..322e1b0f72 100644
--- a/main/_modules/torchtune/models/llama2/_component_builders.html
+++ b/main/_modules/torchtune/models/llama2/_component_builders.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/llama2/_model_builders.html b/main/_modules/torchtune/models/llama2/_model_builders.html
index 26fbcac868..5b4a2fa70b 100644
--- a/main/_modules/torchtune/models/llama2/_model_builders.html
+++ b/main/_modules/torchtune/models/llama2/_model_builders.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/llama2/_prompt_template.html b/main/_modules/torchtune/models/llama2/_prompt_template.html
index c67ed861c1..6be995724e 100644
--- a/main/_modules/torchtune/models/llama2/_prompt_template.html
+++ b/main/_modules/torchtune/models/llama2/_prompt_template.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/llama3/_component_builders.html b/main/_modules/torchtune/models/llama3/_component_builders.html
index 577d8cd6f8..1999a6c247 100644
--- a/main/_modules/torchtune/models/llama3/_component_builders.html
+++ b/main/_modules/torchtune/models/llama3/_component_builders.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/llama3/_model_builders.html b/main/_modules/torchtune/models/llama3/_model_builders.html
index d863abd47b..a0fb5125fb 100644
--- a/main/_modules/torchtune/models/llama3/_model_builders.html
+++ b/main/_modules/torchtune/models/llama3/_model_builders.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/llama3_1/_component_builders.html b/main/_modules/torchtune/models/llama3_1/_component_builders.html
index ce267110c6..4f53045793 100644
--- a/main/_modules/torchtune/models/llama3_1/_component_builders.html
+++ b/main/_modules/torchtune/models/llama3_1/_component_builders.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/llama3_1/_model_builders.html b/main/_modules/torchtune/models/llama3_1/_model_builders.html
index b678f808e1..6d2fc3987d 100644
--- a/main/_modules/torchtune/models/llama3_1/_model_builders.html
+++ b/main/_modules/torchtune/models/llama3_1/_model_builders.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/llama3_2/_model_builders.html b/main/_modules/torchtune/models/llama3_2/_model_builders.html
index 920b6d431d..64eced78a4 100644
--- a/main/_modules/torchtune/models/llama3_2/_model_builders.html
+++ b/main/_modules/torchtune/models/llama3_2/_model_builders.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/llama3_2_vision/_component_builders.html b/main/_modules/torchtune/models/llama3_2_vision/_component_builders.html
index 35dbf813e1..e1c3df2ef6 100644
--- a/main/_modules/torchtune/models/llama3_2_vision/_component_builders.html
+++ b/main/_modules/torchtune/models/llama3_2_vision/_component_builders.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/llama3_2_vision/_encoder.html b/main/_modules/torchtune/models/llama3_2_vision/_encoder.html
index fa93ddf481..f2f9bdac13 100644
--- a/main/_modules/torchtune/models/llama3_2_vision/_encoder.html
+++ b/main/_modules/torchtune/models/llama3_2_vision/_encoder.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/llama3_2_vision/_model_builders.html b/main/_modules/torchtune/models/llama3_2_vision/_model_builders.html
index ac21aa445d..89a86b706d 100644
--- a/main/_modules/torchtune/models/llama3_2_vision/_model_builders.html
+++ b/main/_modules/torchtune/models/llama3_2_vision/_model_builders.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/llama3_2_vision/_transform.html b/main/_modules/torchtune/models/llama3_2_vision/_transform.html
index b06cdec814..5c46876a53 100644
--- a/main/_modules/torchtune/models/llama3_2_vision/_transform.html
+++ b/main/_modules/torchtune/models/llama3_2_vision/_transform.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/mistral/_component_builders.html b/main/_modules/torchtune/models/mistral/_component_builders.html
index 7f329d8671..bcf7b9e89b 100644
--- a/main/_modules/torchtune/models/mistral/_component_builders.html
+++ b/main/_modules/torchtune/models/mistral/_component_builders.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/mistral/_model_builders.html b/main/_modules/torchtune/models/mistral/_model_builders.html
index 5631427714..6ebe5c27ae 100644
--- a/main/_modules/torchtune/models/mistral/_model_builders.html
+++ b/main/_modules/torchtune/models/mistral/_model_builders.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/mistral/_prompt_template.html b/main/_modules/torchtune/models/mistral/_prompt_template.html
index c2bdeb0278..4bf294468f 100644
--- a/main/_modules/torchtune/models/mistral/_prompt_template.html
+++ b/main/_modules/torchtune/models/mistral/_prompt_template.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/phi3/_component_builders.html b/main/_modules/torchtune/models/phi3/_component_builders.html
index 4493d61213..a9b20abd81 100644
--- a/main/_modules/torchtune/models/phi3/_component_builders.html
+++ b/main/_modules/torchtune/models/phi3/_component_builders.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/phi3/_model_builders.html b/main/_modules/torchtune/models/phi3/_model_builders.html
index 94a04faf1a..5f256f6ba3 100644
--- a/main/_modules/torchtune/models/phi3/_model_builders.html
+++ b/main/_modules/torchtune/models/phi3/_model_builders.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/qwen2/_component_builders.html b/main/_modules/torchtune/models/qwen2/_component_builders.html
index 53713260c3..70cc35b67d 100644
--- a/main/_modules/torchtune/models/qwen2/_component_builders.html
+++ b/main/_modules/torchtune/models/qwen2/_component_builders.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/models/qwen2/_model_builders.html b/main/_modules/torchtune/models/qwen2/_model_builders.html
index dd162db90b..b8e319abff 100644
--- a/main/_modules/torchtune/models/qwen2/_model_builders.html
+++ b/main/_modules/torchtune/models/qwen2/_model_builders.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/attention.html b/main/_modules/torchtune/modules/attention.html
index 0bb908bd9e..54d98e139b 100644
--- a/main/_modules/torchtune/modules/attention.html
+++ b/main/_modules/torchtune/modules/attention.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/common_utils.html b/main/_modules/torchtune/modules/common_utils.html
index 0f487d41a8..13cf7a3cb1 100644
--- a/main/_modules/torchtune/modules/common_utils.html
+++ b/main/_modules/torchtune/modules/common_utils.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/feed_forward.html b/main/_modules/torchtune/modules/feed_forward.html
index ef998b161a..e344a55529 100644
--- a/main/_modules/torchtune/modules/feed_forward.html
+++ b/main/_modules/torchtune/modules/feed_forward.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/kv_cache.html b/main/_modules/torchtune/modules/kv_cache.html
index 7ea726f353..94d79e0724 100644
--- a/main/_modules/torchtune/modules/kv_cache.html
+++ b/main/_modules/torchtune/modules/kv_cache.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/layer_norm.html b/main/_modules/torchtune/modules/layer_norm.html
index 744828a632..713439c833 100644
--- a/main/_modules/torchtune/modules/layer_norm.html
+++ b/main/_modules/torchtune/modules/layer_norm.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/loss/ce_chunked_output_loss.html b/main/_modules/torchtune/modules/loss/ce_chunked_output_loss.html
index 51cc66be4d..5e8ef4f4c8 100644
--- a/main/_modules/torchtune/modules/loss/ce_chunked_output_loss.html
+++ b/main/_modules/torchtune/modules/loss/ce_chunked_output_loss.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/loss/kd_losses.html b/main/_modules/torchtune/modules/loss/kd_losses.html
index d17a9b5825..0d4c263b44 100644
--- a/main/_modules/torchtune/modules/loss/kd_losses.html
+++ b/main/_modules/torchtune/modules/loss/kd_losses.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/model_fusion/_fusion.html b/main/_modules/torchtune/modules/model_fusion/_fusion.html
index ab045c6aca..3dd3ec41df 100644
--- a/main/_modules/torchtune/modules/model_fusion/_fusion.html
+++ b/main/_modules/torchtune/modules/model_fusion/_fusion.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/model_fusion/_fusion_utils.html b/main/_modules/torchtune/modules/model_fusion/_fusion_utils.html
index 7231db2d9c..48510c7ae5 100644
--- a/main/_modules/torchtune/modules/model_fusion/_fusion_utils.html
+++ b/main/_modules/torchtune/modules/model_fusion/_fusion_utils.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/peft/_utils.html b/main/_modules/torchtune/modules/peft/_utils.html
index a2fc757c2a..01349e0971 100644
--- a/main/_modules/torchtune/modules/peft/_utils.html
+++ b/main/_modules/torchtune/modules/peft/_utils.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/peft/lora.html b/main/_modules/torchtune/modules/peft/lora.html
index c838f22414..a91ae2432b 100644
--- a/main/_modules/torchtune/modules/peft/lora.html
+++ b/main/_modules/torchtune/modules/peft/lora.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/position_embeddings.html b/main/_modules/torchtune/modules/position_embeddings.html
index 4723c96ece..8efa1cfe9e 100644
--- a/main/_modules/torchtune/modules/position_embeddings.html
+++ b/main/_modules/torchtune/modules/position_embeddings.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/rms_norm.html b/main/_modules/torchtune/modules/rms_norm.html
index 0121d00b34..a552866aff 100644
--- a/main/_modules/torchtune/modules/rms_norm.html
+++ b/main/_modules/torchtune/modules/rms_norm.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/tanh_gate.html b/main/_modules/torchtune/modules/tanh_gate.html
index 67f05b2930..c70ab86114 100644
--- a/main/_modules/torchtune/modules/tanh_gate.html
+++ b/main/_modules/torchtune/modules/tanh_gate.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/tied_linear.html b/main/_modules/torchtune/modules/tied_linear.html
index 96461e57d5..8b3dcaa420 100644
--- a/main/_modules/torchtune/modules/tied_linear.html
+++ b/main/_modules/torchtune/modules/tied_linear.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/tokenizers/_sentencepiece.html b/main/_modules/torchtune/modules/tokenizers/_sentencepiece.html
index 3f731b3d19..0bdc7fcb41 100644
--- a/main/_modules/torchtune/modules/tokenizers/_sentencepiece.html
+++ b/main/_modules/torchtune/modules/tokenizers/_sentencepiece.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/tokenizers/_tiktoken.html b/main/_modules/torchtune/modules/tokenizers/_tiktoken.html
index e72d4ec20f..80ff0d0c48 100644
--- a/main/_modules/torchtune/modules/tokenizers/_tiktoken.html
+++ b/main/_modules/torchtune/modules/tokenizers/_tiktoken.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/tokenizers/_utils.html b/main/_modules/torchtune/modules/tokenizers/_utils.html
index 3a84546e90..e8923d84e9 100644
--- a/main/_modules/torchtune/modules/tokenizers/_utils.html
+++ b/main/_modules/torchtune/modules/tokenizers/_utils.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/transformer.html b/main/_modules/torchtune/modules/transformer.html
index 50eabeda11..05b590dba8 100644
--- a/main/_modules/torchtune/modules/transformer.html
+++ b/main/_modules/torchtune/modules/transformer.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/transforms/_transforms.html b/main/_modules/torchtune/modules/transforms/_transforms.html
index 5a54b2ee58..0a63775c1d 100644
--- a/main/_modules/torchtune/modules/transforms/_transforms.html
+++ b/main/_modules/torchtune/modules/transforms/_transforms.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/modules/vision_transformer.html b/main/_modules/torchtune/modules/vision_transformer.html
index e50f323e0f..6eb444928b 100644
--- a/main/_modules/torchtune/modules/vision_transformer.html
+++ b/main/_modules/torchtune/modules/vision_transformer.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/rlhf/loss/dpo.html b/main/_modules/torchtune/rlhf/loss/dpo.html
index a6214d081c..3eb3d0a3a7 100644
--- a/main/_modules/torchtune/rlhf/loss/dpo.html
+++ b/main/_modules/torchtune/rlhf/loss/dpo.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/rlhf/loss/ppo.html b/main/_modules/torchtune/rlhf/loss/ppo.html
index 1b8448117f..09fc84625d 100644
--- a/main/_modules/torchtune/rlhf/loss/ppo.html
+++ b/main/_modules/torchtune/rlhf/loss/ppo.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/rlhf/rewards.html b/main/_modules/torchtune/rlhf/rewards.html
index 1f6f318200..53af220420 100644
--- a/main/_modules/torchtune/rlhf/rewards.html
+++ b/main/_modules/torchtune/rlhf/rewards.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/rlhf/sequence_processing.html b/main/_modules/torchtune/rlhf/sequence_processing.html
index b6bfc6ccfd..c05f89fb6b 100644
--- a/main/_modules/torchtune/rlhf/sequence_processing.html
+++ b/main/_modules/torchtune/rlhf/sequence_processing.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/training/_distributed.html b/main/_modules/torchtune/training/_distributed.html
index 059ae92bf8..65010951cb 100644
--- a/main/_modules/torchtune/training/_distributed.html
+++ b/main/_modules/torchtune/training/_distributed.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/training/_profiler.html b/main/_modules/torchtune/training/_profiler.html
index e9029bad89..4e9cbea672 100644
--- a/main/_modules/torchtune/training/_profiler.html
+++ b/main/_modules/torchtune/training/_profiler.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/training/activations.html b/main/_modules/torchtune/training/activations.html
index 653869589e..8e9c25b07a 100644
--- a/main/_modules/torchtune/training/activations.html
+++ b/main/_modules/torchtune/training/activations.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/training/checkpointing/_checkpointer.html b/main/_modules/torchtune/training/checkpointing/_checkpointer.html
index e71d3fc636..7d45bc6719 100644
--- a/main/_modules/torchtune/training/checkpointing/_checkpointer.html
+++ b/main/_modules/torchtune/training/checkpointing/_checkpointer.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/training/checkpointing/_utils.html b/main/_modules/torchtune/training/checkpointing/_utils.html
index 8a7f0acd20..b8f36f26c8 100644
--- a/main/_modules/torchtune/training/checkpointing/_utils.html
+++ b/main/_modules/torchtune/training/checkpointing/_utils.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/training/lr_schedulers.html b/main/_modules/torchtune/training/lr_schedulers.html
index 3824c83fd8..7a0959c8a0 100644
--- a/main/_modules/torchtune/training/lr_schedulers.html
+++ b/main/_modules/torchtune/training/lr_schedulers.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/training/memory.html b/main/_modules/torchtune/training/memory.html
index 3df1843f6c..ddf1ec7e2c 100644
--- a/main/_modules/torchtune/training/memory.html
+++ b/main/_modules/torchtune/training/memory.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/training/metric_logging.html b/main/_modules/torchtune/training/metric_logging.html
index 81eba2b86e..776666d721 100644
--- a/main/_modules/torchtune/training/metric_logging.html
+++ b/main/_modules/torchtune/training/metric_logging.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/training/pooling.html b/main/_modules/torchtune/training/pooling.html
index 1a62d3d9f4..f85d08f1fe 100644
--- a/main/_modules/torchtune/training/pooling.html
+++ b/main/_modules/torchtune/training/pooling.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/training/precision.html b/main/_modules/torchtune/training/precision.html
index da44ca5662..1b63801bde 100644
--- a/main/_modules/torchtune/training/precision.html
+++ b/main/_modules/torchtune/training/precision.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/training/quantization.html b/main/_modules/torchtune/training/quantization.html
index 57c91035b1..fe3071c416 100644
--- a/main/_modules/torchtune/training/quantization.html
+++ b/main/_modules/torchtune/training/quantization.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/training/seed.html b/main/_modules/torchtune/training/seed.html
index ae1a752099..72c7666c92 100644
--- a/main/_modules/torchtune/training/seed.html
+++ b/main/_modules/torchtune/training/seed.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/utils/_device.html b/main/_modules/torchtune/utils/_device.html
index 7171526213..c4c35a2f31 100644
--- a/main/_modules/torchtune/utils/_device.html
+++ b/main/_modules/torchtune/utils/_device.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/utils/_logging.html b/main/_modules/torchtune/utils/_logging.html
index 9aca2088db..39e543cdde 100644
--- a/main/_modules/torchtune/utils/_logging.html
+++ b/main/_modules/torchtune/utils/_logging.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/torchtune/utils/_version.html b/main/_modules/torchtune/utils/_version.html
index 4a5622d923..91d4c5f13e 100644
--- a/main/_modules/torchtune/utils/_version.html
+++ b/main/_modules/torchtune/utils/_version.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_modules/typing.html b/main/_modules/typing.html
index e7716c15d9..def5c06578 100644
--- a/main/_modules/typing.html
+++ b/main/_modules/typing.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/_sources/index.rst.txt b/main/_sources/index.rst.txt
index 5126dcb50e..318c82b3e2 100644
--- a/main/_sources/index.rst.txt
+++ b/main/_sources/index.rst.txt
@@ -146,8 +146,8 @@ torchtune tutorials.
    tutorials/qlora_finetune
    tutorials/qat_finetune
    tutorials/e2e_flow
-   tutorials/memory_optimizations
    tutorials/llama_kd_tutorial
+   tutorials/memory_optimizations
 
 .. toctree::
    :glob:
diff --git a/main/_sources/tutorials/memory_optimizations.rst.txt b/main/_sources/tutorials/memory_optimizations.rst.txt
index 04644093a9..c46f2237ce 100644
--- a/main/_sources/tutorials/memory_optimizations.rst.txt
+++ b/main/_sources/tutorials/memory_optimizations.rst.txt
@@ -16,9 +16,11 @@ To make things easy, we've summarized these components in the following table:
 
    ":ref:`glossary_precision`", "You'll usually want to leave this as its default ``bfloat16``. If you're struggling with training stability or accuracy due to precision, fp32 may help, but will significantly increase memory usage and decrease training speed."
    ":ref:`glossary_act_ckpt`", "Use when you're memory constrained and need to handle larger batch sizes or longer context lengths. Be aware that it may slow down training speed."
+   ":ref:`glossary_act_off`", "Similar to activation checkpointing, this can be used when memory constrained, but comes at the cost of training speed due to the overhead of moving tensors between GPU VRAM and CPU. This can also be used alongside activation checkpointing."
    ":ref:`glossary_grad_accm`", "Helpful when memory-constrained to simulate larger batch sizes. Often preferable to activation checkpointing for better training speed."
    ":ref:`glossary_low_precision_opt`", "When you need to further reduce memory usage beyond using ``bf16`` by reducing the precision in the optimizer states. Note that lower precision optimizers may reduce training stability/accuracy."
    ":ref:`glossary_opt_in_bwd`", "Helps reduce memory usage when using stateful optimizers, particularly when full-finetuning large models with high gradient memory usage. This is not compatible with ``gradient_accumulation_steps``, so training may slow down due to reduced model throughput."
+   ":ref:`glossary_cpu_offload`", "Offloads optimizer states and (optionally) gradients to CPU, and performs optimizer steps on CPU. This can be used to significantly reduce GPU memory usage at the cost of CPU RAM and training speed, as CPU optimizer steps can be slow and bottleneck training performance."
    ":ref:`glossary_lora`", "When you want to significantly reduce the number of trainable parameters, saving gradient and optimizer memory during training, and significantly speeding up training."
    ":ref:`glossary_qlora`", "When you need even more memory savings than LoRA, at the potential cost of some training speed. Useful for very large models or limited hardware."
 
@@ -95,7 +97,7 @@ efficiency technique that allows saving GPU VRAM by temporarily moving activatio
 them back when needed in the backward pass.
 
 See `PyTorch autograd hook tutorial <https://pytorch.org/tutorials/intermediate/autograd_saved_tensors_hooks_tutorial.html#saving-tensors-to-cpu>`_
-for more details about how this is implemented through saved_tensors_hooks.
+for more details about how this is implemented through :func:`torch.autograd.graph.saved_tensors_hooks`.
 
 This setting is especially helpful for larger batch sizes, or longer context lengths when you're memory constrained.
 While of course it takes runtime and resources to move Tensors from GPU to CPU and back, the implementation in
@@ -154,10 +156,13 @@ All of our finetuning recipes support simulating larger batch sizes by accumulat
 
   Gradient accumulation should always be set to 1 when :ref:`fusing the optimizer step into the backward pass <glossary_opt_in_bwd>`.
 
+Optimizers
+----------
+
 .. _glossary_low_precision_opt:
 
 Lower Precision Optimizers
---------------------------
+^^^^^^^^^^^^^^^^^^^^^^^^^^
 
 *What's going on here?*
 
@@ -186,7 +191,7 @@ or by directly :ref:`modifying a config file<config_tutorial_label>`:
 .. _glossary_opt_in_bwd:
 
 Fusing Optimizer Step into Backward Pass
-----------------------------------------
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
 *What's going on here?*
 
@@ -208,10 +213,70 @@ To understand how this works, we encourage you to read through the relevant PyTo
 .. todo ref full finetune recipe doc
 
 In torchtune, you can enable this feature using the ``optimizer_in_bwd`` flag, which is currently only supported in our
-single-device full finetune recipe. This feature works best when gradient memory is particularly large;
+single-device full finetune recipe. This feature works best when optimizer memory is particularly large;
 e.g. when using a stateful optimizer with a model with a lot of parameters, and when you don't need to use
 :ref:`gradient accumulation <glossary_grad_accm>`.
 
+.. _glossary_cpu_offload:
+
+Offloading Optimizer/Gradient states to CPU
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+*What's going on here?*
+
+We've mentioned above the concept of optimizer states - memory used by the stateful optimizers to maintain a state of gradient statistics, and
+model gradients - tensors used to store gradients when we perform model backwards passes. We support using CPU offloading in our single-device recipes
+through the `CPUOffloadOptimizer <https://github.com/pytorch/ao/tree/main/torchao/prototype/low_bit_optim#optimizer-cpu-offload>`_ from ``torchao``.
+
+This optimizer can wrap any base optimizer and works by keeping the optimizer states and performing the optimizer step on CPU, thus reducing
+GPU memory usage by the size of the optimizer states. Additionally, we can also offload gradients to the CPU by using `offload_gradients=True`.
+
+*Sounds great! How do I use it?*
+
+To use this optimizer in your recipes, set the ``optimizer`` key in your config to :class:`torchao.prototype.low_bit_optim.CPUOffloadOptimizer`, which
+will use the :class:`torch.optim.AdamW` optimizer with ``fused=True`` as the base optimizer. For example, to use this optimizer to offload
+both optimizer states and gradients to CPU:
+
+.. code-block:: bash
+
+  tune run <RECIPE> --config <CONFIG> \
+  optimizer=optimizer=torchao.prototype.low_bit_optim.CPUOffloadOptimizer \
+  optimizer.offload_gradients=True \
+  lr=4e-5
+
+
+or by directly :ref:`modifying a config file<config_tutorial_label>`:
+
+.. code-block:: yaml
+
+  optimizer:
+    _component_: torchao.prototype.low_bit_optim.CPUOffloadOptimizer
+    offload_gradients: True
+    # additional key-word arguments can be passed to torch.optim.AdamW
+    lr: 4e-5
+
+or using it directly in your code, which allows you to change the base optimizer:
+
+.. code-block:: python
+
+ from torchao.prototype.low_bit_optim import CPUOffloadOptimizer
+ from torch.optim import Adam
+
+ optimizer = CPUOffloadOptimizer(
+     model.parameters(), # your model here
+     Adam,
+     lr=1e-5,
+     fused=True
+ )
+
+Some helpful hints from the ``torchao`` `CPUOffloadOptimizer page <https://github.com/pytorch/ao/tree/main/torchao/prototype/low_bit_optim#optimizer-cpu-offload>`_:
+
+* The CPU optimizer step is often the bottleneck when optimizer CPU offload is used. To minimize the slowdown, it is recommended to (1) use full ``bf16`` training so that parameters, gradients, and optimizer states are in ``bf16``; and (2) give GPU more work per optimizer step (e.g. larger batch size with activation checkpointing, gradient accumulation).
+* Gradient accumulation should always be set to 1 when ``offload_gradients=True``, as gradients are cleared on GPU every backward pass.
+* This optimizer works by keeping a copy of parameters and pre-allocating gradient memory on CPU. Therefore, expect your RAM usage to increase by 4x model size.
+* This optimizer is only supported for single-device recipes. To use CPU-offloading in distributed recipes, use ``fsdp_cpu_offload=True`` in any distributed recipe. See :class:`torch.distributed.fsdp.FullyShardedDataParallel` for more details
+
+
 .. _glossary_peft:
 
 Parameter Efficient Fine-Tuning (PEFT)
diff --git a/main/api_ref_config.html b/main/api_ref_config.html
index fe498480d8..b3d95557ab 100644
--- a/main/api_ref_config.html
+++ b/main/api_ref_config.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/api_ref_data.html b/main/api_ref_data.html
index 0d3dbf4166..8ca4a61ee9 100644
--- a/main/api_ref_data.html
+++ b/main/api_ref_data.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/api_ref_datasets.html b/main/api_ref_datasets.html
index e435d10e08..49f1e9d15d 100644
--- a/main/api_ref_datasets.html
+++ b/main/api_ref_datasets.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/api_ref_generation.html b/main/api_ref_generation.html
index 38e55ca5fb..8d28ca0fce 100644
--- a/main/api_ref_generation.html
+++ b/main/api_ref_generation.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/api_ref_models.html b/main/api_ref_models.html
index 6eb99ba9e9..fe82ebbf21 100644
--- a/main/api_ref_models.html
+++ b/main/api_ref_models.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/api_ref_modules.html b/main/api_ref_modules.html
index bb0d048656..720d1a913b 100644
--- a/main/api_ref_modules.html
+++ b/main/api_ref_modules.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/api_ref_rlhf.html b/main/api_ref_rlhf.html
index ac282ab0b5..74d78350df 100644
--- a/main/api_ref_rlhf.html
+++ b/main/api_ref_rlhf.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/api_ref_training.html b/main/api_ref_training.html
index 61ca225d0e..f7d6eda740 100644
--- a/main/api_ref_training.html
+++ b/main/api_ref_training.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/api_ref_utilities.html b/main/api_ref_utilities.html
index 6080429949..4333c599e9 100644
--- a/main/api_ref_utilities.html
+++ b/main/api_ref_utilities.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/basics/chat_datasets.html b/main/basics/chat_datasets.html
index 27993300ef..1f8b8c7a91 100644
--- a/main/basics/chat_datasets.html
+++ b/main/basics/chat_datasets.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/basics/custom_components.html b/main/basics/custom_components.html
index 3b420aa8a4..7d83db6f14 100644
--- a/main/basics/custom_components.html
+++ b/main/basics/custom_components.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/basics/datasets_overview.html b/main/basics/datasets_overview.html
index 7e399ae34f..05d964b5f5 100644
--- a/main/basics/datasets_overview.html
+++ b/main/basics/datasets_overview.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/basics/instruct_datasets.html b/main/basics/instruct_datasets.html
index 9ce919ee70..31ed010a60 100644
--- a/main/basics/instruct_datasets.html
+++ b/main/basics/instruct_datasets.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/basics/message_transforms.html b/main/basics/message_transforms.html
index b3df9f89fa..4e36752be5 100644
--- a/main/basics/message_transforms.html
+++ b/main/basics/message_transforms.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/basics/messages.html b/main/basics/messages.html
index 7ae9dfdf3d..2d46552a99 100644
--- a/main/basics/messages.html
+++ b/main/basics/messages.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/basics/model_transforms.html b/main/basics/model_transforms.html
index cc9a3d8654..b5fefb565a 100644
--- a/main/basics/model_transforms.html
+++ b/main/basics/model_transforms.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/basics/multimodal_datasets.html b/main/basics/multimodal_datasets.html
index 227abc8212..cc5f54f5c1 100644
--- a/main/basics/multimodal_datasets.html
+++ b/main/basics/multimodal_datasets.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/basics/packing.html b/main/basics/packing.html
index afbb3fba1f..22b8e4de84 100644
--- a/main/basics/packing.html
+++ b/main/basics/packing.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/basics/preference_datasets.html b/main/basics/preference_datasets.html
index 3b00f986d1..04fea83ff6 100644
--- a/main/basics/preference_datasets.html
+++ b/main/basics/preference_datasets.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/basics/prompt_templates.html b/main/basics/prompt_templates.html
index 20a58e0c5b..1aa74f4fa0 100644
--- a/main/basics/prompt_templates.html
+++ b/main/basics/prompt_templates.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/basics/text_completion_datasets.html b/main/basics/text_completion_datasets.html
index 7f5fdd0cb6..4c864a403d 100644
--- a/main/basics/text_completion_datasets.html
+++ b/main/basics/text_completion_datasets.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/basics/tokenizers.html b/main/basics/tokenizers.html
index 19da36afe5..680fab53c3 100644
--- a/main/basics/tokenizers.html
+++ b/main/basics/tokenizers.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/deep_dives/checkpointer.html b/main/deep_dives/checkpointer.html
index 7c55940d0d..42539211a7 100644
--- a/main/deep_dives/checkpointer.html
+++ b/main/deep_dives/checkpointer.html
@@ -41,7 +41,7 @@
     <link rel="index" title="Index" href="../genindex.html" />
     <link rel="search" title="Search" href="../search.html" />
     <link rel="next" title="All About Configs" href="configs.html" />
-    <link rel="prev" title="Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation" href="../tutorials/llama_kd_tutorial.html" />
+    <link rel="prev" title="Memory Optimization Overview" href="../tutorials/memory_optimizations.html" />
   <!-- Google Tag Manager -->
     <script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':
     new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul class="current">
@@ -891,7 +891,7 @@ <h2>Putting this all together<a class="headerlink" href="#putting-this-all-toget
         <a href="configs.html" class="btn btn-neutral float-right" title="All About Configs" accesskey="n" rel="next">Next <img src="../_static/images/chevron-right-orange.svg" class="next-page"></a>
       
       
-        <a href="../tutorials/llama_kd_tutorial.html" class="btn btn-neutral" title="Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation" accesskey="p" rel="prev"><img src="../_static/images/chevron-right-orange.svg" class="previous-page"> Previous</a>
+        <a href="../tutorials/memory_optimizations.html" class="btn btn-neutral" title="Memory Optimization Overview" accesskey="p" rel="prev"><img src="../_static/images/chevron-right-orange.svg" class="previous-page"> Previous</a>
       
     </div>
   
diff --git a/main/deep_dives/comet_logging.html b/main/deep_dives/comet_logging.html
index 6cfbef09f0..7133a2bf65 100644
--- a/main/deep_dives/comet_logging.html
+++ b/main/deep_dives/comet_logging.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul class="current">
diff --git a/main/deep_dives/configs.html b/main/deep_dives/configs.html
index dcb14f0a1e..48b1eb8d0e 100644
--- a/main/deep_dives/configs.html
+++ b/main/deep_dives/configs.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul class="current">
diff --git a/main/deep_dives/recipe_deepdive.html b/main/deep_dives/recipe_deepdive.html
index bcc24c4c14..64d6d339c2 100644
--- a/main/deep_dives/recipe_deepdive.html
+++ b/main/deep_dives/recipe_deepdive.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul class="current">
diff --git a/main/deep_dives/wandb_logging.html b/main/deep_dives/wandb_logging.html
index 0e680505f7..8f02ee9326 100644
--- a/main/deep_dives/wandb_logging.html
+++ b/main/deep_dives/wandb_logging.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul class="current">
diff --git a/main/generated/torchtune.config.instantiate.html b/main/generated/torchtune.config.instantiate.html
index 42ed670929..e9b4a69b31 100644
--- a/main/generated/torchtune.config.instantiate.html
+++ b/main/generated/torchtune.config.instantiate.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.config.log_config.html b/main/generated/torchtune.config.log_config.html
index c50fbe375f..0613cb65bf 100644
--- a/main/generated/torchtune.config.log_config.html
+++ b/main/generated/torchtune.config.log_config.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.config.parse.html b/main/generated/torchtune.config.parse.html
index 315c84ef3b..a0a4ca253b 100644
--- a/main/generated/torchtune.config.parse.html
+++ b/main/generated/torchtune.config.parse.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.config.validate.html b/main/generated/torchtune.config.validate.html
index d064e7ae53..ac549fea98 100644
--- a/main/generated/torchtune.config.validate.html
+++ b/main/generated/torchtune.config.validate.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.AlpacaToMessages.html b/main/generated/torchtune.data.AlpacaToMessages.html
index 496c0d5206..cb7a34f416 100644
--- a/main/generated/torchtune.data.AlpacaToMessages.html
+++ b/main/generated/torchtune.data.AlpacaToMessages.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.ChatMLTemplate.html b/main/generated/torchtune.data.ChatMLTemplate.html
index 8a5ea8aa39..a52110a739 100644
--- a/main/generated/torchtune.data.ChatMLTemplate.html
+++ b/main/generated/torchtune.data.ChatMLTemplate.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.ChosenRejectedToMessages.html b/main/generated/torchtune.data.ChosenRejectedToMessages.html
index 1546f8ccd1..dae299fa56 100644
--- a/main/generated/torchtune.data.ChosenRejectedToMessages.html
+++ b/main/generated/torchtune.data.ChosenRejectedToMessages.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.GrammarErrorCorrectionTemplate.html b/main/generated/torchtune.data.GrammarErrorCorrectionTemplate.html
index 61b6c89e26..5851a299f4 100644
--- a/main/generated/torchtune.data.GrammarErrorCorrectionTemplate.html
+++ b/main/generated/torchtune.data.GrammarErrorCorrectionTemplate.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.InputOutputToMessages.html b/main/generated/torchtune.data.InputOutputToMessages.html
index 75c69085d4..dd0f322ba1 100644
--- a/main/generated/torchtune.data.InputOutputToMessages.html
+++ b/main/generated/torchtune.data.InputOutputToMessages.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.Message.html b/main/generated/torchtune.data.Message.html
index bfeb3de8b2..aa8cf5791e 100644
--- a/main/generated/torchtune.data.Message.html
+++ b/main/generated/torchtune.data.Message.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.OpenAIToMessages.html b/main/generated/torchtune.data.OpenAIToMessages.html
index a83206a7c0..c71cc9235c 100644
--- a/main/generated/torchtune.data.OpenAIToMessages.html
+++ b/main/generated/torchtune.data.OpenAIToMessages.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.PromptTemplate.html b/main/generated/torchtune.data.PromptTemplate.html
index 902eb3034f..ac5d9960f9 100644
--- a/main/generated/torchtune.data.PromptTemplate.html
+++ b/main/generated/torchtune.data.PromptTemplate.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.PromptTemplateInterface.html b/main/generated/torchtune.data.PromptTemplateInterface.html
index ff43fe678c..692173c3f4 100644
--- a/main/generated/torchtune.data.PromptTemplateInterface.html
+++ b/main/generated/torchtune.data.PromptTemplateInterface.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.QuestionAnswerTemplate.html b/main/generated/torchtune.data.QuestionAnswerTemplate.html
index 50376010e5..9f81cd9ea9 100644
--- a/main/generated/torchtune.data.QuestionAnswerTemplate.html
+++ b/main/generated/torchtune.data.QuestionAnswerTemplate.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.Role.html b/main/generated/torchtune.data.Role.html
index bf1d81b52d..3cf00ba49d 100644
--- a/main/generated/torchtune.data.Role.html
+++ b/main/generated/torchtune.data.Role.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.ShareGPTToMessages.html b/main/generated/torchtune.data.ShareGPTToMessages.html
index 9e78a1e4e9..404235c791 100644
--- a/main/generated/torchtune.data.ShareGPTToMessages.html
+++ b/main/generated/torchtune.data.ShareGPTToMessages.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.SummarizeTemplate.html b/main/generated/torchtune.data.SummarizeTemplate.html
index 55285f46fb..92dab781af 100644
--- a/main/generated/torchtune.data.SummarizeTemplate.html
+++ b/main/generated/torchtune.data.SummarizeTemplate.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.format_content_with_images.html b/main/generated/torchtune.data.format_content_with_images.html
index 92ffe7d8d3..3367b68929 100644
--- a/main/generated/torchtune.data.format_content_with_images.html
+++ b/main/generated/torchtune.data.format_content_with_images.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.left_pad_sequence.html b/main/generated/torchtune.data.left_pad_sequence.html
index 69aff34bb6..33cf74e174 100644
--- a/main/generated/torchtune.data.left_pad_sequence.html
+++ b/main/generated/torchtune.data.left_pad_sequence.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.load_image.html b/main/generated/torchtune.data.load_image.html
index f8bd7bdce2..fdc0df77ba 100644
--- a/main/generated/torchtune.data.load_image.html
+++ b/main/generated/torchtune.data.load_image.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.padded_collate.html b/main/generated/torchtune.data.padded_collate.html
index 915d26784b..ada4c1451c 100644
--- a/main/generated/torchtune.data.padded_collate.html
+++ b/main/generated/torchtune.data.padded_collate.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.padded_collate_dpo.html b/main/generated/torchtune.data.padded_collate_dpo.html
index 5d09a91d61..abd00d22b2 100644
--- a/main/generated/torchtune.data.padded_collate_dpo.html
+++ b/main/generated/torchtune.data.padded_collate_dpo.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.padded_collate_sft.html b/main/generated/torchtune.data.padded_collate_sft.html
index 8501d5a481..3404ee6edb 100644
--- a/main/generated/torchtune.data.padded_collate_sft.html
+++ b/main/generated/torchtune.data.padded_collate_sft.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.padded_collate_tiled_images_and_mask.html b/main/generated/torchtune.data.padded_collate_tiled_images_and_mask.html
index baa269a0d0..2600396099 100644
--- a/main/generated/torchtune.data.padded_collate_tiled_images_and_mask.html
+++ b/main/generated/torchtune.data.padded_collate_tiled_images_and_mask.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.truncate.html b/main/generated/torchtune.data.truncate.html
index 9053472f09..e10c14c11f 100644
--- a/main/generated/torchtune.data.truncate.html
+++ b/main/generated/torchtune.data.truncate.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.data.validate_messages.html b/main/generated/torchtune.data.validate_messages.html
index 14812035d5..ee075c78f0 100644
--- a/main/generated/torchtune.data.validate_messages.html
+++ b/main/generated/torchtune.data.validate_messages.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.datasets.ConcatDataset.html b/main/generated/torchtune.datasets.ConcatDataset.html
index 48ba0de9dc..69ae7091a7 100644
--- a/main/generated/torchtune.datasets.ConcatDataset.html
+++ b/main/generated/torchtune.datasets.ConcatDataset.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.datasets.PackedDataset.html b/main/generated/torchtune.datasets.PackedDataset.html
index ea42a55bc0..0c4508896f 100644
--- a/main/generated/torchtune.datasets.PackedDataset.html
+++ b/main/generated/torchtune.datasets.PackedDataset.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.datasets.PreferenceDataset.html b/main/generated/torchtune.datasets.PreferenceDataset.html
index 3677a14759..a6fee55877 100644
--- a/main/generated/torchtune.datasets.PreferenceDataset.html
+++ b/main/generated/torchtune.datasets.PreferenceDataset.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.datasets.SFTDataset.html b/main/generated/torchtune.datasets.SFTDataset.html
index d163ff7062..e57b0c85f9 100644
--- a/main/generated/torchtune.datasets.SFTDataset.html
+++ b/main/generated/torchtune.datasets.SFTDataset.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.datasets.TextCompletionDataset.html b/main/generated/torchtune.datasets.TextCompletionDataset.html
index e6a7427cdf..2da8d4978e 100644
--- a/main/generated/torchtune.datasets.TextCompletionDataset.html
+++ b/main/generated/torchtune.datasets.TextCompletionDataset.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.datasets.alpaca_cleaned_dataset.html b/main/generated/torchtune.datasets.alpaca_cleaned_dataset.html
index 40d9a31f97..edc8ee612e 100644
--- a/main/generated/torchtune.datasets.alpaca_cleaned_dataset.html
+++ b/main/generated/torchtune.datasets.alpaca_cleaned_dataset.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.datasets.alpaca_dataset.html b/main/generated/torchtune.datasets.alpaca_dataset.html
index 0c707bf219..27e498091b 100644
--- a/main/generated/torchtune.datasets.alpaca_dataset.html
+++ b/main/generated/torchtune.datasets.alpaca_dataset.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.datasets.chat_dataset.html b/main/generated/torchtune.datasets.chat_dataset.html
index df4ea09256..e3cbd9640d 100644
--- a/main/generated/torchtune.datasets.chat_dataset.html
+++ b/main/generated/torchtune.datasets.chat_dataset.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.datasets.cnn_dailymail_articles_dataset.html b/main/generated/torchtune.datasets.cnn_dailymail_articles_dataset.html
index 3734e93fd6..e2978c3098 100644
--- a/main/generated/torchtune.datasets.cnn_dailymail_articles_dataset.html
+++ b/main/generated/torchtune.datasets.cnn_dailymail_articles_dataset.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.datasets.grammar_dataset.html b/main/generated/torchtune.datasets.grammar_dataset.html
index 64047ec4d6..9fdb9b96a3 100644
--- a/main/generated/torchtune.datasets.grammar_dataset.html
+++ b/main/generated/torchtune.datasets.grammar_dataset.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.datasets.hh_rlhf_helpful_dataset.html b/main/generated/torchtune.datasets.hh_rlhf_helpful_dataset.html
index b350549d79..eb5524c390 100644
--- a/main/generated/torchtune.datasets.hh_rlhf_helpful_dataset.html
+++ b/main/generated/torchtune.datasets.hh_rlhf_helpful_dataset.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.datasets.instruct_dataset.html b/main/generated/torchtune.datasets.instruct_dataset.html
index 58f4e73a8f..1314f8ffef 100644
--- a/main/generated/torchtune.datasets.instruct_dataset.html
+++ b/main/generated/torchtune.datasets.instruct_dataset.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.datasets.multimodal.llava_instruct_dataset.html b/main/generated/torchtune.datasets.multimodal.llava_instruct_dataset.html
index fc9e5c1cbc..c537605941 100644
--- a/main/generated/torchtune.datasets.multimodal.llava_instruct_dataset.html
+++ b/main/generated/torchtune.datasets.multimodal.llava_instruct_dataset.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.datasets.multimodal.the_cauldron_dataset.html b/main/generated/torchtune.datasets.multimodal.the_cauldron_dataset.html
index cb0747b580..bc64c1c088 100644
--- a/main/generated/torchtune.datasets.multimodal.the_cauldron_dataset.html
+++ b/main/generated/torchtune.datasets.multimodal.the_cauldron_dataset.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.datasets.preference_dataset.html b/main/generated/torchtune.datasets.preference_dataset.html
index a267cdcba7..e0391598c7 100644
--- a/main/generated/torchtune.datasets.preference_dataset.html
+++ b/main/generated/torchtune.datasets.preference_dataset.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.datasets.samsum_dataset.html b/main/generated/torchtune.datasets.samsum_dataset.html
index 7c1f8afed8..9df5219fd1 100644
--- a/main/generated/torchtune.datasets.samsum_dataset.html
+++ b/main/generated/torchtune.datasets.samsum_dataset.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.datasets.slimorca_dataset.html b/main/generated/torchtune.datasets.slimorca_dataset.html
index 120417cc67..86f64962dd 100644
--- a/main/generated/torchtune.datasets.slimorca_dataset.html
+++ b/main/generated/torchtune.datasets.slimorca_dataset.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.datasets.stack_exchange_paired_dataset.html b/main/generated/torchtune.datasets.stack_exchange_paired_dataset.html
index ece2082357..d6e0b87962 100644
--- a/main/generated/torchtune.datasets.stack_exchange_paired_dataset.html
+++ b/main/generated/torchtune.datasets.stack_exchange_paired_dataset.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.datasets.text_completion_dataset.html b/main/generated/torchtune.datasets.text_completion_dataset.html
index 5bae71b219..6b6258af6e 100644
--- a/main/generated/torchtune.datasets.text_completion_dataset.html
+++ b/main/generated/torchtune.datasets.text_completion_dataset.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.datasets.wikitext_dataset.html b/main/generated/torchtune.datasets.wikitext_dataset.html
index 5c5276784e..8d99716929 100644
--- a/main/generated/torchtune.datasets.wikitext_dataset.html
+++ b/main/generated/torchtune.datasets.wikitext_dataset.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.generation.generate.html b/main/generated/torchtune.generation.generate.html
index 208a43f880..48a719f7db 100644
--- a/main/generated/torchtune.generation.generate.html
+++ b/main/generated/torchtune.generation.generate.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.generation.generate_next_token.html b/main/generated/torchtune.generation.generate_next_token.html
index c413cfda12..7d2fde7a8d 100644
--- a/main/generated/torchtune.generation.generate_next_token.html
+++ b/main/generated/torchtune.generation.generate_next_token.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.generation.get_causal_mask_from_padding_mask.html b/main/generated/torchtune.generation.get_causal_mask_from_padding_mask.html
index aae52e4571..87d9c9e13f 100644
--- a/main/generated/torchtune.generation.get_causal_mask_from_padding_mask.html
+++ b/main/generated/torchtune.generation.get_causal_mask_from_padding_mask.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.generation.get_position_ids_from_padding_mask.html b/main/generated/torchtune.generation.get_position_ids_from_padding_mask.html
index 1ce9de9eaf..3946919c0f 100644
--- a/main/generated/torchtune.generation.get_position_ids_from_padding_mask.html
+++ b/main/generated/torchtune.generation.get_position_ids_from_padding_mask.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.generation.sample.html b/main/generated/torchtune.generation.sample.html
index b52d493922..72d118605f 100644
--- a/main/generated/torchtune.generation.sample.html
+++ b/main/generated/torchtune.generation.sample.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.clip.TilePositionalEmbedding.html b/main/generated/torchtune.models.clip.TilePositionalEmbedding.html
index c6c06aa98d..f2d9e2f935 100644
--- a/main/generated/torchtune.models.clip.TilePositionalEmbedding.html
+++ b/main/generated/torchtune.models.clip.TilePositionalEmbedding.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.clip.TiledTokenPositionalEmbedding.html b/main/generated/torchtune.models.clip.TiledTokenPositionalEmbedding.html
index 9be0e9e01f..f663850439 100644
--- a/main/generated/torchtune.models.clip.TiledTokenPositionalEmbedding.html
+++ b/main/generated/torchtune.models.clip.TiledTokenPositionalEmbedding.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.clip.TokenPositionalEmbedding.html b/main/generated/torchtune.models.clip.TokenPositionalEmbedding.html
index 3a1395511a..33afce153f 100644
--- a/main/generated/torchtune.models.clip.TokenPositionalEmbedding.html
+++ b/main/generated/torchtune.models.clip.TokenPositionalEmbedding.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.clip.clip_vision_encoder.html b/main/generated/torchtune.models.clip.clip_vision_encoder.html
index c67d74cac7..522ca42196 100644
--- a/main/generated/torchtune.models.clip.clip_vision_encoder.html
+++ b/main/generated/torchtune.models.clip.clip_vision_encoder.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.code_llama2.code_llama2_13b.html b/main/generated/torchtune.models.code_llama2.code_llama2_13b.html
index 8e4c427220..f401ac2004 100644
--- a/main/generated/torchtune.models.code_llama2.code_llama2_13b.html
+++ b/main/generated/torchtune.models.code_llama2.code_llama2_13b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.code_llama2.code_llama2_70b.html b/main/generated/torchtune.models.code_llama2.code_llama2_70b.html
index 25ae8b7341..513c83944f 100644
--- a/main/generated/torchtune.models.code_llama2.code_llama2_70b.html
+++ b/main/generated/torchtune.models.code_llama2.code_llama2_70b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.code_llama2.code_llama2_7b.html b/main/generated/torchtune.models.code_llama2.code_llama2_7b.html
index b04b38df29..5531c37910 100644
--- a/main/generated/torchtune.models.code_llama2.code_llama2_7b.html
+++ b/main/generated/torchtune.models.code_llama2.code_llama2_7b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.code_llama2.lora_code_llama2_13b.html b/main/generated/torchtune.models.code_llama2.lora_code_llama2_13b.html
index 8e545f30c4..117e8740ca 100644
--- a/main/generated/torchtune.models.code_llama2.lora_code_llama2_13b.html
+++ b/main/generated/torchtune.models.code_llama2.lora_code_llama2_13b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.code_llama2.lora_code_llama2_70b.html b/main/generated/torchtune.models.code_llama2.lora_code_llama2_70b.html
index 969eda0e35..4387e92eda 100644
--- a/main/generated/torchtune.models.code_llama2.lora_code_llama2_70b.html
+++ b/main/generated/torchtune.models.code_llama2.lora_code_llama2_70b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.code_llama2.lora_code_llama2_7b.html b/main/generated/torchtune.models.code_llama2.lora_code_llama2_7b.html
index df246acd45..01c303c1a9 100644
--- a/main/generated/torchtune.models.code_llama2.lora_code_llama2_7b.html
+++ b/main/generated/torchtune.models.code_llama2.lora_code_llama2_7b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.code_llama2.qlora_code_llama2_13b.html b/main/generated/torchtune.models.code_llama2.qlora_code_llama2_13b.html
index 735722f724..846bae41ef 100644
--- a/main/generated/torchtune.models.code_llama2.qlora_code_llama2_13b.html
+++ b/main/generated/torchtune.models.code_llama2.qlora_code_llama2_13b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.code_llama2.qlora_code_llama2_70b.html b/main/generated/torchtune.models.code_llama2.qlora_code_llama2_70b.html
index 5fa0a50c9c..8a6f1f6422 100644
--- a/main/generated/torchtune.models.code_llama2.qlora_code_llama2_70b.html
+++ b/main/generated/torchtune.models.code_llama2.qlora_code_llama2_70b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.code_llama2.qlora_code_llama2_7b.html b/main/generated/torchtune.models.code_llama2.qlora_code_llama2_7b.html
index 5b6ebcf558..45480157d5 100644
--- a/main/generated/torchtune.models.code_llama2.qlora_code_llama2_7b.html
+++ b/main/generated/torchtune.models.code_llama2.qlora_code_llama2_7b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.gemma.gemma.html b/main/generated/torchtune.models.gemma.gemma.html
index bd3808242c..a74ca7fff1 100644
--- a/main/generated/torchtune.models.gemma.gemma.html
+++ b/main/generated/torchtune.models.gemma.gemma.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.gemma.gemma_2b.html b/main/generated/torchtune.models.gemma.gemma_2b.html
index dee79d54f5..1bd344a54a 100644
--- a/main/generated/torchtune.models.gemma.gemma_2b.html
+++ b/main/generated/torchtune.models.gemma.gemma_2b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.gemma.gemma_7b.html b/main/generated/torchtune.models.gemma.gemma_7b.html
index 78c0a868ca..31abea6301 100644
--- a/main/generated/torchtune.models.gemma.gemma_7b.html
+++ b/main/generated/torchtune.models.gemma.gemma_7b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.gemma.gemma_tokenizer.html b/main/generated/torchtune.models.gemma.gemma_tokenizer.html
index 8e2b875b43..be492b1702 100644
--- a/main/generated/torchtune.models.gemma.gemma_tokenizer.html
+++ b/main/generated/torchtune.models.gemma.gemma_tokenizer.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.gemma.lora_gemma.html b/main/generated/torchtune.models.gemma.lora_gemma.html
index 111d6b25eb..cc4d139c2b 100644
--- a/main/generated/torchtune.models.gemma.lora_gemma.html
+++ b/main/generated/torchtune.models.gemma.lora_gemma.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.gemma.lora_gemma_2b.html b/main/generated/torchtune.models.gemma.lora_gemma_2b.html
index 225c85a437..0c48a48524 100644
--- a/main/generated/torchtune.models.gemma.lora_gemma_2b.html
+++ b/main/generated/torchtune.models.gemma.lora_gemma_2b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.gemma.lora_gemma_7b.html b/main/generated/torchtune.models.gemma.lora_gemma_7b.html
index 0fdb4727e7..655e0b6858 100644
--- a/main/generated/torchtune.models.gemma.lora_gemma_7b.html
+++ b/main/generated/torchtune.models.gemma.lora_gemma_7b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.gemma.qlora_gemma_2b.html b/main/generated/torchtune.models.gemma.qlora_gemma_2b.html
index a798b4df96..a3d5d0673d 100644
--- a/main/generated/torchtune.models.gemma.qlora_gemma_2b.html
+++ b/main/generated/torchtune.models.gemma.qlora_gemma_2b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.gemma.qlora_gemma_7b.html b/main/generated/torchtune.models.gemma.qlora_gemma_7b.html
index ec6cb2fcd1..d6e79e5bbb 100644
--- a/main/generated/torchtune.models.gemma.qlora_gemma_7b.html
+++ b/main/generated/torchtune.models.gemma.qlora_gemma_7b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama2.Llama2ChatTemplate.html b/main/generated/torchtune.models.llama2.Llama2ChatTemplate.html
index 81911d1659..8ebbf57661 100644
--- a/main/generated/torchtune.models.llama2.Llama2ChatTemplate.html
+++ b/main/generated/torchtune.models.llama2.Llama2ChatTemplate.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama2.llama2.html b/main/generated/torchtune.models.llama2.llama2.html
index 62d0c03f1e..52ca9afa95 100644
--- a/main/generated/torchtune.models.llama2.llama2.html
+++ b/main/generated/torchtune.models.llama2.llama2.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama2.llama2_13b.html b/main/generated/torchtune.models.llama2.llama2_13b.html
index 053a0be7d0..ec0f75a5de 100644
--- a/main/generated/torchtune.models.llama2.llama2_13b.html
+++ b/main/generated/torchtune.models.llama2.llama2_13b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama2.llama2_70b.html b/main/generated/torchtune.models.llama2.llama2_70b.html
index 6cd4288ec5..bb76b0da93 100644
--- a/main/generated/torchtune.models.llama2.llama2_70b.html
+++ b/main/generated/torchtune.models.llama2.llama2_70b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama2.llama2_7b.html b/main/generated/torchtune.models.llama2.llama2_7b.html
index 88a0366cd3..304795969a 100644
--- a/main/generated/torchtune.models.llama2.llama2_7b.html
+++ b/main/generated/torchtune.models.llama2.llama2_7b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama2.llama2_reward_7b.html b/main/generated/torchtune.models.llama2.llama2_reward_7b.html
index 80c00054dc..825e3db00a 100644
--- a/main/generated/torchtune.models.llama2.llama2_reward_7b.html
+++ b/main/generated/torchtune.models.llama2.llama2_reward_7b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama2.llama2_tokenizer.html b/main/generated/torchtune.models.llama2.llama2_tokenizer.html
index 2ea170c714..f512af228c 100644
--- a/main/generated/torchtune.models.llama2.llama2_tokenizer.html
+++ b/main/generated/torchtune.models.llama2.llama2_tokenizer.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama2.lora_llama2.html b/main/generated/torchtune.models.llama2.lora_llama2.html
index 2b19ccc4b7..b13538e4a2 100644
--- a/main/generated/torchtune.models.llama2.lora_llama2.html
+++ b/main/generated/torchtune.models.llama2.lora_llama2.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama2.lora_llama2_13b.html b/main/generated/torchtune.models.llama2.lora_llama2_13b.html
index 4f5edbef18..1bb9dd07d9 100644
--- a/main/generated/torchtune.models.llama2.lora_llama2_13b.html
+++ b/main/generated/torchtune.models.llama2.lora_llama2_13b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama2.lora_llama2_70b.html b/main/generated/torchtune.models.llama2.lora_llama2_70b.html
index aaa6a48fb3..6f339db222 100644
--- a/main/generated/torchtune.models.llama2.lora_llama2_70b.html
+++ b/main/generated/torchtune.models.llama2.lora_llama2_70b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama2.lora_llama2_7b.html b/main/generated/torchtune.models.llama2.lora_llama2_7b.html
index 0cd3dca2a6..c796180c0f 100644
--- a/main/generated/torchtune.models.llama2.lora_llama2_7b.html
+++ b/main/generated/torchtune.models.llama2.lora_llama2_7b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama2.lora_llama2_reward_7b.html b/main/generated/torchtune.models.llama2.lora_llama2_reward_7b.html
index bdbf3cea30..501bbc4ab9 100644
--- a/main/generated/torchtune.models.llama2.lora_llama2_reward_7b.html
+++ b/main/generated/torchtune.models.llama2.lora_llama2_reward_7b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama2.qlora_llama2_13b.html b/main/generated/torchtune.models.llama2.qlora_llama2_13b.html
index c4a5ef4c3f..00257576ab 100644
--- a/main/generated/torchtune.models.llama2.qlora_llama2_13b.html
+++ b/main/generated/torchtune.models.llama2.qlora_llama2_13b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama2.qlora_llama2_70b.html b/main/generated/torchtune.models.llama2.qlora_llama2_70b.html
index 44b8c0b72c..06dd59b53a 100644
--- a/main/generated/torchtune.models.llama2.qlora_llama2_70b.html
+++ b/main/generated/torchtune.models.llama2.qlora_llama2_70b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama2.qlora_llama2_7b.html b/main/generated/torchtune.models.llama2.qlora_llama2_7b.html
index d114afba4d..cc9630d736 100644
--- a/main/generated/torchtune.models.llama2.qlora_llama2_7b.html
+++ b/main/generated/torchtune.models.llama2.qlora_llama2_7b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama2.qlora_llama2_reward_7b.html b/main/generated/torchtune.models.llama2.qlora_llama2_reward_7b.html
index 23a85154c3..fecf8d0e74 100644
--- a/main/generated/torchtune.models.llama2.qlora_llama2_reward_7b.html
+++ b/main/generated/torchtune.models.llama2.qlora_llama2_reward_7b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3.llama3.html b/main/generated/torchtune.models.llama3.llama3.html
index 43e408b7c8..234c6487d1 100644
--- a/main/generated/torchtune.models.llama3.llama3.html
+++ b/main/generated/torchtune.models.llama3.llama3.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3.llama3_70b.html b/main/generated/torchtune.models.llama3.llama3_70b.html
index b2cf53a85f..e570a9e69a 100644
--- a/main/generated/torchtune.models.llama3.llama3_70b.html
+++ b/main/generated/torchtune.models.llama3.llama3_70b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3.llama3_8b.html b/main/generated/torchtune.models.llama3.llama3_8b.html
index b8b7b0d257..84241ad9f1 100644
--- a/main/generated/torchtune.models.llama3.llama3_8b.html
+++ b/main/generated/torchtune.models.llama3.llama3_8b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3.llama3_tokenizer.html b/main/generated/torchtune.models.llama3.llama3_tokenizer.html
index c33e3a5bb4..a80863c7a0 100644
--- a/main/generated/torchtune.models.llama3.llama3_tokenizer.html
+++ b/main/generated/torchtune.models.llama3.llama3_tokenizer.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3.lora_llama3.html b/main/generated/torchtune.models.llama3.lora_llama3.html
index 5dc42066d6..4846b0ec7e 100644
--- a/main/generated/torchtune.models.llama3.lora_llama3.html
+++ b/main/generated/torchtune.models.llama3.lora_llama3.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3.lora_llama3_70b.html b/main/generated/torchtune.models.llama3.lora_llama3_70b.html
index 2dece09a7e..8b4a8d3425 100644
--- a/main/generated/torchtune.models.llama3.lora_llama3_70b.html
+++ b/main/generated/torchtune.models.llama3.lora_llama3_70b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3.lora_llama3_8b.html b/main/generated/torchtune.models.llama3.lora_llama3_8b.html
index 6fde89c5df..f4c250993b 100644
--- a/main/generated/torchtune.models.llama3.lora_llama3_8b.html
+++ b/main/generated/torchtune.models.llama3.lora_llama3_8b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3.qlora_llama3_70b.html b/main/generated/torchtune.models.llama3.qlora_llama3_70b.html
index 583385a5c9..c98ea2acaa 100644
--- a/main/generated/torchtune.models.llama3.qlora_llama3_70b.html
+++ b/main/generated/torchtune.models.llama3.qlora_llama3_70b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3.qlora_llama3_8b.html b/main/generated/torchtune.models.llama3.qlora_llama3_8b.html
index 8d03a9e43d..357d23aee4 100644
--- a/main/generated/torchtune.models.llama3.qlora_llama3_8b.html
+++ b/main/generated/torchtune.models.llama3.qlora_llama3_8b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_1.llama3_1.html b/main/generated/torchtune.models.llama3_1.llama3_1.html
index cf6f5300e7..8db7e0c282 100644
--- a/main/generated/torchtune.models.llama3_1.llama3_1.html
+++ b/main/generated/torchtune.models.llama3_1.llama3_1.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_1.llama3_1_405b.html b/main/generated/torchtune.models.llama3_1.llama3_1_405b.html
index 366a0f8dba..af2a4f2ee4 100644
--- a/main/generated/torchtune.models.llama3_1.llama3_1_405b.html
+++ b/main/generated/torchtune.models.llama3_1.llama3_1_405b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_1.llama3_1_70b.html b/main/generated/torchtune.models.llama3_1.llama3_1_70b.html
index 6ccb033540..4e868b0126 100644
--- a/main/generated/torchtune.models.llama3_1.llama3_1_70b.html
+++ b/main/generated/torchtune.models.llama3_1.llama3_1_70b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_1.llama3_1_8b.html b/main/generated/torchtune.models.llama3_1.llama3_1_8b.html
index bd7af5cabf..b1d3d15108 100644
--- a/main/generated/torchtune.models.llama3_1.llama3_1_8b.html
+++ b/main/generated/torchtune.models.llama3_1.llama3_1_8b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_1.lora_llama3_1.html b/main/generated/torchtune.models.llama3_1.lora_llama3_1.html
index 79b7f5d3f6..9da6740aa9 100644
--- a/main/generated/torchtune.models.llama3_1.lora_llama3_1.html
+++ b/main/generated/torchtune.models.llama3_1.lora_llama3_1.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_1.lora_llama3_1_405b.html b/main/generated/torchtune.models.llama3_1.lora_llama3_1_405b.html
index e0f9631c96..4d63cd3f88 100644
--- a/main/generated/torchtune.models.llama3_1.lora_llama3_1_405b.html
+++ b/main/generated/torchtune.models.llama3_1.lora_llama3_1_405b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_1.lora_llama3_1_70b.html b/main/generated/torchtune.models.llama3_1.lora_llama3_1_70b.html
index d1c93a203a..0ebe74e37d 100644
--- a/main/generated/torchtune.models.llama3_1.lora_llama3_1_70b.html
+++ b/main/generated/torchtune.models.llama3_1.lora_llama3_1_70b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_1.lora_llama3_1_8b.html b/main/generated/torchtune.models.llama3_1.lora_llama3_1_8b.html
index 4ee4a69487..0d6f96ba95 100644
--- a/main/generated/torchtune.models.llama3_1.lora_llama3_1_8b.html
+++ b/main/generated/torchtune.models.llama3_1.lora_llama3_1_8b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_1.qlora_llama3_1_405b.html b/main/generated/torchtune.models.llama3_1.qlora_llama3_1_405b.html
index a159249be0..fb523901b3 100644
--- a/main/generated/torchtune.models.llama3_1.qlora_llama3_1_405b.html
+++ b/main/generated/torchtune.models.llama3_1.qlora_llama3_1_405b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_1.qlora_llama3_1_70b.html b/main/generated/torchtune.models.llama3_1.qlora_llama3_1_70b.html
index 775098f683..ac05015dee 100644
--- a/main/generated/torchtune.models.llama3_1.qlora_llama3_1_70b.html
+++ b/main/generated/torchtune.models.llama3_1.qlora_llama3_1_70b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_1.qlora_llama3_1_8b.html b/main/generated/torchtune.models.llama3_1.qlora_llama3_1_8b.html
index 283c302d56..6e03363ee7 100644
--- a/main/generated/torchtune.models.llama3_1.qlora_llama3_1_8b.html
+++ b/main/generated/torchtune.models.llama3_1.qlora_llama3_1_8b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_2.llama3_2_1b.html b/main/generated/torchtune.models.llama3_2.llama3_2_1b.html
index 3b34234a45..f0e00ce0e9 100644
--- a/main/generated/torchtune.models.llama3_2.llama3_2_1b.html
+++ b/main/generated/torchtune.models.llama3_2.llama3_2_1b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_2.llama3_2_3b.html b/main/generated/torchtune.models.llama3_2.llama3_2_3b.html
index 3e94181e35..b903fca270 100644
--- a/main/generated/torchtune.models.llama3_2.llama3_2_3b.html
+++ b/main/generated/torchtune.models.llama3_2.llama3_2_3b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_2.lora_llama3_2_1b.html b/main/generated/torchtune.models.llama3_2.lora_llama3_2_1b.html
index 49664fabcf..6d15930ace 100644
--- a/main/generated/torchtune.models.llama3_2.lora_llama3_2_1b.html
+++ b/main/generated/torchtune.models.llama3_2.lora_llama3_2_1b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_2.lora_llama3_2_3b.html b/main/generated/torchtune.models.llama3_2.lora_llama3_2_3b.html
index 1ca8479f28..2e14530c64 100644
--- a/main/generated/torchtune.models.llama3_2.lora_llama3_2_3b.html
+++ b/main/generated/torchtune.models.llama3_2.lora_llama3_2_3b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_2.qlora_llama3_2_1b.html b/main/generated/torchtune.models.llama3_2.qlora_llama3_2_1b.html
index 02e1c89a52..40968838ee 100644
--- a/main/generated/torchtune.models.llama3_2.qlora_llama3_2_1b.html
+++ b/main/generated/torchtune.models.llama3_2.qlora_llama3_2_1b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_2.qlora_llama3_2_3b.html b/main/generated/torchtune.models.llama3_2.qlora_llama3_2_3b.html
index df76e3fb17..08865324be 100644
--- a/main/generated/torchtune.models.llama3_2.qlora_llama3_2_3b.html
+++ b/main/generated/torchtune.models.llama3_2.qlora_llama3_2_3b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_2_vision.Llama3VisionEncoder.html b/main/generated/torchtune.models.llama3_2_vision.Llama3VisionEncoder.html
index f86f111edc..3872db482a 100644
--- a/main/generated/torchtune.models.llama3_2_vision.Llama3VisionEncoder.html
+++ b/main/generated/torchtune.models.llama3_2_vision.Llama3VisionEncoder.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_2_vision.Llama3VisionProjectionHead.html b/main/generated/torchtune.models.llama3_2_vision.Llama3VisionProjectionHead.html
index 11ff0ac440..fee2636ea2 100644
--- a/main/generated/torchtune.models.llama3_2_vision.Llama3VisionProjectionHead.html
+++ b/main/generated/torchtune.models.llama3_2_vision.Llama3VisionProjectionHead.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_2_vision.Llama3VisionTransform.html b/main/generated/torchtune.models.llama3_2_vision.Llama3VisionTransform.html
index 9d095e9ec8..8132a64531 100644
--- a/main/generated/torchtune.models.llama3_2_vision.Llama3VisionTransform.html
+++ b/main/generated/torchtune.models.llama3_2_vision.Llama3VisionTransform.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_2_vision.llama3_2_vision_11b.html b/main/generated/torchtune.models.llama3_2_vision.llama3_2_vision_11b.html
index 6f18cee71b..67d4d4e0ef 100644
--- a/main/generated/torchtune.models.llama3_2_vision.llama3_2_vision_11b.html
+++ b/main/generated/torchtune.models.llama3_2_vision.llama3_2_vision_11b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_2_vision.llama3_2_vision_decoder.html b/main/generated/torchtune.models.llama3_2_vision.llama3_2_vision_decoder.html
index e8a7ed0907..dfa2001f63 100644
--- a/main/generated/torchtune.models.llama3_2_vision.llama3_2_vision_decoder.html
+++ b/main/generated/torchtune.models.llama3_2_vision.llama3_2_vision_decoder.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_2_vision.llama3_2_vision_encoder.html b/main/generated/torchtune.models.llama3_2_vision.llama3_2_vision_encoder.html
index b37f4bac11..0120a4dec6 100644
--- a/main/generated/torchtune.models.llama3_2_vision.llama3_2_vision_encoder.html
+++ b/main/generated/torchtune.models.llama3_2_vision.llama3_2_vision_encoder.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_2_vision.llama3_2_vision_transform.html b/main/generated/torchtune.models.llama3_2_vision.llama3_2_vision_transform.html
index 8c624d5230..60ebad86a0 100644
--- a/main/generated/torchtune.models.llama3_2_vision.llama3_2_vision_transform.html
+++ b/main/generated/torchtune.models.llama3_2_vision.llama3_2_vision_transform.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_11b.html b/main/generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_11b.html
index a05725a06e..fff8ff29a1 100644
--- a/main/generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_11b.html
+++ b/main/generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_11b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_decoder.html b/main/generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_decoder.html
index 653833e715..d4ef87b665 100644
--- a/main/generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_decoder.html
+++ b/main/generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_decoder.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_encoder.html b/main/generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_encoder.html
index 95cd1f924c..2c05a10bc5 100644
--- a/main/generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_encoder.html
+++ b/main/generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_encoder.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.llama3_2_vision.qlora_llama3_2_vision_11b.html b/main/generated/torchtune.models.llama3_2_vision.qlora_llama3_2_vision_11b.html
index 61111376fd..4142e4168b 100644
--- a/main/generated/torchtune.models.llama3_2_vision.qlora_llama3_2_vision_11b.html
+++ b/main/generated/torchtune.models.llama3_2_vision.qlora_llama3_2_vision_11b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.mistral.MistralChatTemplate.html b/main/generated/torchtune.models.mistral.MistralChatTemplate.html
index c2df9b4963..afdff6a024 100644
--- a/main/generated/torchtune.models.mistral.MistralChatTemplate.html
+++ b/main/generated/torchtune.models.mistral.MistralChatTemplate.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.mistral.lora_mistral.html b/main/generated/torchtune.models.mistral.lora_mistral.html
index 1fd800f4dd..0aa0d0e3cf 100644
--- a/main/generated/torchtune.models.mistral.lora_mistral.html
+++ b/main/generated/torchtune.models.mistral.lora_mistral.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.mistral.lora_mistral_7b.html b/main/generated/torchtune.models.mistral.lora_mistral_7b.html
index 39cde03b6b..14e9a41fa1 100644
--- a/main/generated/torchtune.models.mistral.lora_mistral_7b.html
+++ b/main/generated/torchtune.models.mistral.lora_mistral_7b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.mistral.lora_mistral_classifier.html b/main/generated/torchtune.models.mistral.lora_mistral_classifier.html
index c4b58530d6..e2a9cfc296 100644
--- a/main/generated/torchtune.models.mistral.lora_mistral_classifier.html
+++ b/main/generated/torchtune.models.mistral.lora_mistral_classifier.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.mistral.lora_mistral_reward_7b.html b/main/generated/torchtune.models.mistral.lora_mistral_reward_7b.html
index 379baaac45..789e55d5bd 100644
--- a/main/generated/torchtune.models.mistral.lora_mistral_reward_7b.html
+++ b/main/generated/torchtune.models.mistral.lora_mistral_reward_7b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.mistral.mistral.html b/main/generated/torchtune.models.mistral.mistral.html
index 3a8520a71b..6db7a7406c 100644
--- a/main/generated/torchtune.models.mistral.mistral.html
+++ b/main/generated/torchtune.models.mistral.mistral.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.mistral.mistral_7b.html b/main/generated/torchtune.models.mistral.mistral_7b.html
index b7944d0c78..a07e1b3f32 100644
--- a/main/generated/torchtune.models.mistral.mistral_7b.html
+++ b/main/generated/torchtune.models.mistral.mistral_7b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.mistral.mistral_classifier.html b/main/generated/torchtune.models.mistral.mistral_classifier.html
index 4130f9f847..f491c616ac 100644
--- a/main/generated/torchtune.models.mistral.mistral_classifier.html
+++ b/main/generated/torchtune.models.mistral.mistral_classifier.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.mistral.mistral_reward_7b.html b/main/generated/torchtune.models.mistral.mistral_reward_7b.html
index fd37ea7860..791859e9e6 100644
--- a/main/generated/torchtune.models.mistral.mistral_reward_7b.html
+++ b/main/generated/torchtune.models.mistral.mistral_reward_7b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.mistral.mistral_tokenizer.html b/main/generated/torchtune.models.mistral.mistral_tokenizer.html
index e8bc0d526a..db650284e3 100644
--- a/main/generated/torchtune.models.mistral.mistral_tokenizer.html
+++ b/main/generated/torchtune.models.mistral.mistral_tokenizer.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.mistral.qlora_mistral_7b.html b/main/generated/torchtune.models.mistral.qlora_mistral_7b.html
index 3ba639f6be..18c56233e9 100644
--- a/main/generated/torchtune.models.mistral.qlora_mistral_7b.html
+++ b/main/generated/torchtune.models.mistral.qlora_mistral_7b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.mistral.qlora_mistral_reward_7b.html b/main/generated/torchtune.models.mistral.qlora_mistral_reward_7b.html
index 3f5f496f36..f5e5dae74c 100644
--- a/main/generated/torchtune.models.mistral.qlora_mistral_reward_7b.html
+++ b/main/generated/torchtune.models.mistral.qlora_mistral_reward_7b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.phi3.lora_phi3.html b/main/generated/torchtune.models.phi3.lora_phi3.html
index ba33da635d..b4fd14a411 100644
--- a/main/generated/torchtune.models.phi3.lora_phi3.html
+++ b/main/generated/torchtune.models.phi3.lora_phi3.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.phi3.lora_phi3_mini.html b/main/generated/torchtune.models.phi3.lora_phi3_mini.html
index e2f7ff015e..2459e5b931 100644
--- a/main/generated/torchtune.models.phi3.lora_phi3_mini.html
+++ b/main/generated/torchtune.models.phi3.lora_phi3_mini.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.phi3.phi3.html b/main/generated/torchtune.models.phi3.phi3.html
index 8f125d7d11..196a265034 100644
--- a/main/generated/torchtune.models.phi3.phi3.html
+++ b/main/generated/torchtune.models.phi3.phi3.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.phi3.phi3_mini.html b/main/generated/torchtune.models.phi3.phi3_mini.html
index 110d02d9a9..e76a1fcc47 100644
--- a/main/generated/torchtune.models.phi3.phi3_mini.html
+++ b/main/generated/torchtune.models.phi3.phi3_mini.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.phi3.phi3_mini_tokenizer.html b/main/generated/torchtune.models.phi3.phi3_mini_tokenizer.html
index 3a9cb17419..bba4d2765f 100644
--- a/main/generated/torchtune.models.phi3.phi3_mini_tokenizer.html
+++ b/main/generated/torchtune.models.phi3.phi3_mini_tokenizer.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.phi3.qlora_phi3_mini.html b/main/generated/torchtune.models.phi3.qlora_phi3_mini.html
index 4cfcb357e2..3c51fbc97c 100644
--- a/main/generated/torchtune.models.phi3.qlora_phi3_mini.html
+++ b/main/generated/torchtune.models.phi3.qlora_phi3_mini.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.qwen2.lora_qwen2.html b/main/generated/torchtune.models.qwen2.lora_qwen2.html
index 0ff5fa6784..6bd363b07e 100644
--- a/main/generated/torchtune.models.qwen2.lora_qwen2.html
+++ b/main/generated/torchtune.models.qwen2.lora_qwen2.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.qwen2.lora_qwen2_0_5b.html b/main/generated/torchtune.models.qwen2.lora_qwen2_0_5b.html
index 745d73e5a9..7283e125a3 100644
--- a/main/generated/torchtune.models.qwen2.lora_qwen2_0_5b.html
+++ b/main/generated/torchtune.models.qwen2.lora_qwen2_0_5b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.qwen2.lora_qwen2_1_5b.html b/main/generated/torchtune.models.qwen2.lora_qwen2_1_5b.html
index a03123ec63..33622b9ba2 100644
--- a/main/generated/torchtune.models.qwen2.lora_qwen2_1_5b.html
+++ b/main/generated/torchtune.models.qwen2.lora_qwen2_1_5b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.qwen2.lora_qwen2_7b.html b/main/generated/torchtune.models.qwen2.lora_qwen2_7b.html
index c32a8b3b96..6989996968 100644
--- a/main/generated/torchtune.models.qwen2.lora_qwen2_7b.html
+++ b/main/generated/torchtune.models.qwen2.lora_qwen2_7b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.qwen2.qwen2.html b/main/generated/torchtune.models.qwen2.qwen2.html
index d00838ba49..5b8cd05c6a 100644
--- a/main/generated/torchtune.models.qwen2.qwen2.html
+++ b/main/generated/torchtune.models.qwen2.qwen2.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.qwen2.qwen2_0_5b.html b/main/generated/torchtune.models.qwen2.qwen2_0_5b.html
index 172f65eba8..28acce55ea 100644
--- a/main/generated/torchtune.models.qwen2.qwen2_0_5b.html
+++ b/main/generated/torchtune.models.qwen2.qwen2_0_5b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.qwen2.qwen2_1_5b.html b/main/generated/torchtune.models.qwen2.qwen2_1_5b.html
index 43b6d0d4b3..807c3e1b9a 100644
--- a/main/generated/torchtune.models.qwen2.qwen2_1_5b.html
+++ b/main/generated/torchtune.models.qwen2.qwen2_1_5b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.qwen2.qwen2_7b.html b/main/generated/torchtune.models.qwen2.qwen2_7b.html
index 8fee93eeaf..8fd3477a23 100644
--- a/main/generated/torchtune.models.qwen2.qwen2_7b.html
+++ b/main/generated/torchtune.models.qwen2.qwen2_7b.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.models.qwen2.qwen2_tokenizer.html b/main/generated/torchtune.models.qwen2.qwen2_tokenizer.html
index c11558e02c..b23ecea4b6 100644
--- a/main/generated/torchtune.models.qwen2.qwen2_tokenizer.html
+++ b/main/generated/torchtune.models.qwen2.qwen2_tokenizer.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.FeedForward.html b/main/generated/torchtune.modules.FeedForward.html
index c353c62e4e..fd37948d05 100644
--- a/main/generated/torchtune.modules.FeedForward.html
+++ b/main/generated/torchtune.modules.FeedForward.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.Fp32LayerNorm.html b/main/generated/torchtune.modules.Fp32LayerNorm.html
index 166e6a8958..5455998b75 100644
--- a/main/generated/torchtune.modules.Fp32LayerNorm.html
+++ b/main/generated/torchtune.modules.Fp32LayerNorm.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.KVCache.html b/main/generated/torchtune.modules.KVCache.html
index b08bb4b125..d4e57eccee 100644
--- a/main/generated/torchtune.modules.KVCache.html
+++ b/main/generated/torchtune.modules.KVCache.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.MultiHeadAttention.html b/main/generated/torchtune.modules.MultiHeadAttention.html
index 8762fdff5a..225a73176e 100644
--- a/main/generated/torchtune.modules.MultiHeadAttention.html
+++ b/main/generated/torchtune.modules.MultiHeadAttention.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.RMSNorm.html b/main/generated/torchtune.modules.RMSNorm.html
index d6973ccc08..bc7a5076d6 100644
--- a/main/generated/torchtune.modules.RMSNorm.html
+++ b/main/generated/torchtune.modules.RMSNorm.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.RotaryPositionalEmbeddings.html b/main/generated/torchtune.modules.RotaryPositionalEmbeddings.html
index 9407d74897..4268b8ffad 100644
--- a/main/generated/torchtune.modules.RotaryPositionalEmbeddings.html
+++ b/main/generated/torchtune.modules.RotaryPositionalEmbeddings.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.TanhGate.html b/main/generated/torchtune.modules.TanhGate.html
index 872cbfb9b8..6402eae4a7 100644
--- a/main/generated/torchtune.modules.TanhGate.html
+++ b/main/generated/torchtune.modules.TanhGate.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.TiedLinear.html b/main/generated/torchtune.modules.TiedLinear.html
index c4194e93cc..70670ff246 100644
--- a/main/generated/torchtune.modules.TiedLinear.html
+++ b/main/generated/torchtune.modules.TiedLinear.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.TransformerCrossAttentionLayer.html b/main/generated/torchtune.modules.TransformerCrossAttentionLayer.html
index b27d0afba6..7ca4a26a1f 100644
--- a/main/generated/torchtune.modules.TransformerCrossAttentionLayer.html
+++ b/main/generated/torchtune.modules.TransformerCrossAttentionLayer.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.TransformerDecoder.html b/main/generated/torchtune.modules.TransformerDecoder.html
index dd836e710c..d26f6058ca 100644
--- a/main/generated/torchtune.modules.TransformerDecoder.html
+++ b/main/generated/torchtune.modules.TransformerDecoder.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.TransformerSelfAttentionLayer.html b/main/generated/torchtune.modules.TransformerSelfAttentionLayer.html
index 4145fd0f0a..3072f0a713 100644
--- a/main/generated/torchtune.modules.TransformerSelfAttentionLayer.html
+++ b/main/generated/torchtune.modules.TransformerSelfAttentionLayer.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.VisionTransformer.html b/main/generated/torchtune.modules.VisionTransformer.html
index be1a4fd2cd..9ade001c45 100644
--- a/main/generated/torchtune.modules.VisionTransformer.html
+++ b/main/generated/torchtune.modules.VisionTransformer.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.common_utils.delete_kv_caches.html b/main/generated/torchtune.modules.common_utils.delete_kv_caches.html
index 6bb6339be7..df6914de68 100644
--- a/main/generated/torchtune.modules.common_utils.delete_kv_caches.html
+++ b/main/generated/torchtune.modules.common_utils.delete_kv_caches.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.common_utils.disable_kv_cache.html b/main/generated/torchtune.modules.common_utils.disable_kv_cache.html
index 4adfc7cca9..aa0efeaa93 100644
--- a/main/generated/torchtune.modules.common_utils.disable_kv_cache.html
+++ b/main/generated/torchtune.modules.common_utils.disable_kv_cache.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.common_utils.local_kv_cache.html b/main/generated/torchtune.modules.common_utils.local_kv_cache.html
index 95dd8c27cd..d6e975eea8 100644
--- a/main/generated/torchtune.modules.common_utils.local_kv_cache.html
+++ b/main/generated/torchtune.modules.common_utils.local_kv_cache.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.common_utils.reparametrize_as_dtype_state_dict_post_hook.html b/main/generated/torchtune.modules.common_utils.reparametrize_as_dtype_state_dict_post_hook.html
index 4adc34998e..15fa498121 100644
--- a/main/generated/torchtune.modules.common_utils.reparametrize_as_dtype_state_dict_post_hook.html
+++ b/main/generated/torchtune.modules.common_utils.reparametrize_as_dtype_state_dict_post_hook.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.loss.CEWithChunkedOutputLoss.html b/main/generated/torchtune.modules.loss.CEWithChunkedOutputLoss.html
index 6857511d95..3b7041e3b1 100644
--- a/main/generated/torchtune.modules.loss.CEWithChunkedOutputLoss.html
+++ b/main/generated/torchtune.modules.loss.CEWithChunkedOutputLoss.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.loss.ForwardKLLoss.html b/main/generated/torchtune.modules.loss.ForwardKLLoss.html
index be95334b0e..9051694ae6 100644
--- a/main/generated/torchtune.modules.loss.ForwardKLLoss.html
+++ b/main/generated/torchtune.modules.loss.ForwardKLLoss.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.loss.ForwardKLWithChunkedOutputLoss.html b/main/generated/torchtune.modules.loss.ForwardKLWithChunkedOutputLoss.html
index 9c4fecfc1a..372575946e 100644
--- a/main/generated/torchtune.modules.loss.ForwardKLWithChunkedOutputLoss.html
+++ b/main/generated/torchtune.modules.loss.ForwardKLWithChunkedOutputLoss.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.model_fusion.DeepFusionModel.html b/main/generated/torchtune.modules.model_fusion.DeepFusionModel.html
index 9aea3ef370..8fa8247611 100644
--- a/main/generated/torchtune.modules.model_fusion.DeepFusionModel.html
+++ b/main/generated/torchtune.modules.model_fusion.DeepFusionModel.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.model_fusion.FusionEmbedding.html b/main/generated/torchtune.modules.model_fusion.FusionEmbedding.html
index bc6fd3e41b..7466c456a9 100644
--- a/main/generated/torchtune.modules.model_fusion.FusionEmbedding.html
+++ b/main/generated/torchtune.modules.model_fusion.FusionEmbedding.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.model_fusion.FusionLayer.html b/main/generated/torchtune.modules.model_fusion.FusionLayer.html
index abca634188..ce59bb736c 100644
--- a/main/generated/torchtune.modules.model_fusion.FusionLayer.html
+++ b/main/generated/torchtune.modules.model_fusion.FusionLayer.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.model_fusion.get_fusion_params.html b/main/generated/torchtune.modules.model_fusion.get_fusion_params.html
index 0ababa5016..d8c3a6ec8b 100644
--- a/main/generated/torchtune.modules.model_fusion.get_fusion_params.html
+++ b/main/generated/torchtune.modules.model_fusion.get_fusion_params.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.model_fusion.register_fusion_module.html b/main/generated/torchtune.modules.model_fusion.register_fusion_module.html
index f961c99ae7..74432e4baf 100644
--- a/main/generated/torchtune.modules.model_fusion.register_fusion_module.html
+++ b/main/generated/torchtune.modules.model_fusion.register_fusion_module.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.peft.AdapterModule.html b/main/generated/torchtune.modules.peft.AdapterModule.html
index 447a5bd069..249cbed25f 100644
--- a/main/generated/torchtune.modules.peft.AdapterModule.html
+++ b/main/generated/torchtune.modules.peft.AdapterModule.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.peft.LoRALinear.html b/main/generated/torchtune.modules.peft.LoRALinear.html
index 26ac3d6292..efc9b93fef 100644
--- a/main/generated/torchtune.modules.peft.LoRALinear.html
+++ b/main/generated/torchtune.modules.peft.LoRALinear.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.peft.disable_adapter.html b/main/generated/torchtune.modules.peft.disable_adapter.html
index 59297aa739..652e67b005 100644
--- a/main/generated/torchtune.modules.peft.disable_adapter.html
+++ b/main/generated/torchtune.modules.peft.disable_adapter.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.peft.get_adapter_params.html b/main/generated/torchtune.modules.peft.get_adapter_params.html
index 4434049393..7ee8252bfe 100644
--- a/main/generated/torchtune.modules.peft.get_adapter_params.html
+++ b/main/generated/torchtune.modules.peft.get_adapter_params.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.peft.set_trainable_params.html b/main/generated/torchtune.modules.peft.set_trainable_params.html
index ac53136a07..908057ab07 100644
--- a/main/generated/torchtune.modules.peft.set_trainable_params.html
+++ b/main/generated/torchtune.modules.peft.set_trainable_params.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.peft.validate_missing_and_unexpected_for_lora.html b/main/generated/torchtune.modules.peft.validate_missing_and_unexpected_for_lora.html
index f98e329adf..5fbd951f3a 100644
--- a/main/generated/torchtune.modules.peft.validate_missing_and_unexpected_for_lora.html
+++ b/main/generated/torchtune.modules.peft.validate_missing_and_unexpected_for_lora.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.peft.validate_state_dict_for_lora.html b/main/generated/torchtune.modules.peft.validate_state_dict_for_lora.html
index a047c5c7c7..e904749b40 100644
--- a/main/generated/torchtune.modules.peft.validate_state_dict_for_lora.html
+++ b/main/generated/torchtune.modules.peft.validate_state_dict_for_lora.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.tokenizers.BaseTokenizer.html b/main/generated/torchtune.modules.tokenizers.BaseTokenizer.html
index 4e1d735ac8..a0ba30ee7d 100644
--- a/main/generated/torchtune.modules.tokenizers.BaseTokenizer.html
+++ b/main/generated/torchtune.modules.tokenizers.BaseTokenizer.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.tokenizers.ModelTokenizer.html b/main/generated/torchtune.modules.tokenizers.ModelTokenizer.html
index 9d7091ac92..28ebb39e40 100644
--- a/main/generated/torchtune.modules.tokenizers.ModelTokenizer.html
+++ b/main/generated/torchtune.modules.tokenizers.ModelTokenizer.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.tokenizers.SentencePieceBaseTokenizer.html b/main/generated/torchtune.modules.tokenizers.SentencePieceBaseTokenizer.html
index 7d55d6064e..85e7f74f23 100644
--- a/main/generated/torchtune.modules.tokenizers.SentencePieceBaseTokenizer.html
+++ b/main/generated/torchtune.modules.tokenizers.SentencePieceBaseTokenizer.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.tokenizers.TikTokenBaseTokenizer.html b/main/generated/torchtune.modules.tokenizers.TikTokenBaseTokenizer.html
index 3c960d08e3..f421ae192e 100644
--- a/main/generated/torchtune.modules.tokenizers.TikTokenBaseTokenizer.html
+++ b/main/generated/torchtune.modules.tokenizers.TikTokenBaseTokenizer.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.tokenizers.parse_hf_tokenizer_json.html b/main/generated/torchtune.modules.tokenizers.parse_hf_tokenizer_json.html
index 4fa23bc26b..482a958b40 100644
--- a/main/generated/torchtune.modules.tokenizers.parse_hf_tokenizer_json.html
+++ b/main/generated/torchtune.modules.tokenizers.parse_hf_tokenizer_json.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.tokenizers.tokenize_messages_no_special_tokens.html b/main/generated/torchtune.modules.tokenizers.tokenize_messages_no_special_tokens.html
index ba396206b1..06b628548e 100644
--- a/main/generated/torchtune.modules.tokenizers.tokenize_messages_no_special_tokens.html
+++ b/main/generated/torchtune.modules.tokenizers.tokenize_messages_no_special_tokens.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.transforms.Transform.html b/main/generated/torchtune.modules.transforms.Transform.html
index 009b48497a..d0d2d7c206 100644
--- a/main/generated/torchtune.modules.transforms.Transform.html
+++ b/main/generated/torchtune.modules.transforms.Transform.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.modules.transforms.VisionCrossAttentionMask.html b/main/generated/torchtune.modules.transforms.VisionCrossAttentionMask.html
index aae1698949..ea17292843 100644
--- a/main/generated/torchtune.modules.transforms.VisionCrossAttentionMask.html
+++ b/main/generated/torchtune.modules.transforms.VisionCrossAttentionMask.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.rlhf.estimate_advantages.html b/main/generated/torchtune.rlhf.estimate_advantages.html
index 6ac40025b0..b12902950d 100644
--- a/main/generated/torchtune.rlhf.estimate_advantages.html
+++ b/main/generated/torchtune.rlhf.estimate_advantages.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.rlhf.get_rewards_ppo.html b/main/generated/torchtune.rlhf.get_rewards_ppo.html
index b474e15147..b76b03b526 100644
--- a/main/generated/torchtune.rlhf.get_rewards_ppo.html
+++ b/main/generated/torchtune.rlhf.get_rewards_ppo.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.rlhf.loss.DPOLoss.html b/main/generated/torchtune.rlhf.loss.DPOLoss.html
index 07a2e4e0e2..b8fd04672b 100644
--- a/main/generated/torchtune.rlhf.loss.DPOLoss.html
+++ b/main/generated/torchtune.rlhf.loss.DPOLoss.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.rlhf.loss.PPOLoss.html b/main/generated/torchtune.rlhf.loss.PPOLoss.html
index 3a3451b8c1..4b4cae9ce0 100644
--- a/main/generated/torchtune.rlhf.loss.PPOLoss.html
+++ b/main/generated/torchtune.rlhf.loss.PPOLoss.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.rlhf.loss.RSOLoss.html b/main/generated/torchtune.rlhf.loss.RSOLoss.html
index 27d0d9cde4..3b4200c39f 100644
--- a/main/generated/torchtune.rlhf.loss.RSOLoss.html
+++ b/main/generated/torchtune.rlhf.loss.RSOLoss.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.rlhf.loss.SimPOLoss.html b/main/generated/torchtune.rlhf.loss.SimPOLoss.html
index 914fabcd23..05e89274ba 100644
--- a/main/generated/torchtune.rlhf.loss.SimPOLoss.html
+++ b/main/generated/torchtune.rlhf.loss.SimPOLoss.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.rlhf.truncate_sequence_at_first_stop_token.html b/main/generated/torchtune.rlhf.truncate_sequence_at_first_stop_token.html
index 79add233c6..5b13beabe1 100644
--- a/main/generated/torchtune.rlhf.truncate_sequence_at_first_stop_token.html
+++ b/main/generated/torchtune.rlhf.truncate_sequence_at_first_stop_token.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.FSDPPolicyType.html b/main/generated/torchtune.training.FSDPPolicyType.html
index b28f35f334..2d709dd94d 100644
--- a/main/generated/torchtune.training.FSDPPolicyType.html
+++ b/main/generated/torchtune.training.FSDPPolicyType.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.FormattedCheckpointFiles.html b/main/generated/torchtune.training.FormattedCheckpointFiles.html
index b34311f7e4..acbf58d28a 100644
--- a/main/generated/torchtune.training.FormattedCheckpointFiles.html
+++ b/main/generated/torchtune.training.FormattedCheckpointFiles.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.FullModelHFCheckpointer.html b/main/generated/torchtune.training.FullModelHFCheckpointer.html
index 5dcc7ae74c..e660f503a2 100644
--- a/main/generated/torchtune.training.FullModelHFCheckpointer.html
+++ b/main/generated/torchtune.training.FullModelHFCheckpointer.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.FullModelMetaCheckpointer.html b/main/generated/torchtune.training.FullModelMetaCheckpointer.html
index 71dde505b4..3cf14a82cb 100644
--- a/main/generated/torchtune.training.FullModelMetaCheckpointer.html
+++ b/main/generated/torchtune.training.FullModelMetaCheckpointer.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.FullModelTorchTuneCheckpointer.html b/main/generated/torchtune.training.FullModelTorchTuneCheckpointer.html
index b830d098e6..00c9797a98 100644
--- a/main/generated/torchtune.training.FullModelTorchTuneCheckpointer.html
+++ b/main/generated/torchtune.training.FullModelTorchTuneCheckpointer.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.ModelType.html b/main/generated/torchtune.training.ModelType.html
index 67e9b7603c..c4b793c830 100644
--- a/main/generated/torchtune.training.ModelType.html
+++ b/main/generated/torchtune.training.ModelType.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.OptimizerInBackwardWrapper.html b/main/generated/torchtune.training.OptimizerInBackwardWrapper.html
index fba2aa006e..0cd9f6b12a 100644
--- a/main/generated/torchtune.training.OptimizerInBackwardWrapper.html
+++ b/main/generated/torchtune.training.OptimizerInBackwardWrapper.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.apply_selective_activation_checkpointing.html b/main/generated/torchtune.training.apply_selective_activation_checkpointing.html
index b112075a3c..597fc6422f 100644
--- a/main/generated/torchtune.training.apply_selective_activation_checkpointing.html
+++ b/main/generated/torchtune.training.apply_selective_activation_checkpointing.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.create_optim_in_bwd_wrapper.html b/main/generated/torchtune.training.create_optim_in_bwd_wrapper.html
index c7aeb7bcac..72465bdf53 100644
--- a/main/generated/torchtune.training.create_optim_in_bwd_wrapper.html
+++ b/main/generated/torchtune.training.create_optim_in_bwd_wrapper.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.get_cosine_schedule_with_warmup.html b/main/generated/torchtune.training.get_cosine_schedule_with_warmup.html
index f687c56dc7..3a0ffbf8d2 100644
--- a/main/generated/torchtune.training.get_cosine_schedule_with_warmup.html
+++ b/main/generated/torchtune.training.get_cosine_schedule_with_warmup.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.get_dtype.html b/main/generated/torchtune.training.get_dtype.html
index e65718fe17..17d2ea7506 100644
--- a/main/generated/torchtune.training.get_dtype.html
+++ b/main/generated/torchtune.training.get_dtype.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.get_full_finetune_fsdp_wrap_policy.html b/main/generated/torchtune.training.get_full_finetune_fsdp_wrap_policy.html
index 40d60fbb24..934afa360d 100644
--- a/main/generated/torchtune.training.get_full_finetune_fsdp_wrap_policy.html
+++ b/main/generated/torchtune.training.get_full_finetune_fsdp_wrap_policy.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.get_lr.html b/main/generated/torchtune.training.get_lr.html
index 130f923e55..761629eb04 100644
--- a/main/generated/torchtune.training.get_lr.html
+++ b/main/generated/torchtune.training.get_lr.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.get_memory_stats.html b/main/generated/torchtune.training.get_memory_stats.html
index e7248f1ad4..28d3dca80d 100644
--- a/main/generated/torchtune.training.get_memory_stats.html
+++ b/main/generated/torchtune.training.get_memory_stats.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.get_quantizer_mode.html b/main/generated/torchtune.training.get_quantizer_mode.html
index 585bd72cf1..a50c65940f 100644
--- a/main/generated/torchtune.training.get_quantizer_mode.html
+++ b/main/generated/torchtune.training.get_quantizer_mode.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.get_unmasked_sequence_lengths.html b/main/generated/torchtune.training.get_unmasked_sequence_lengths.html
index 59351af8b8..6a7fd30516 100644
--- a/main/generated/torchtune.training.get_unmasked_sequence_lengths.html
+++ b/main/generated/torchtune.training.get_unmasked_sequence_lengths.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.get_world_size_and_rank.html b/main/generated/torchtune.training.get_world_size_and_rank.html
index 87ef3a9aeb..a974564735 100644
--- a/main/generated/torchtune.training.get_world_size_and_rank.html
+++ b/main/generated/torchtune.training.get_world_size_and_rank.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.init_distributed.html b/main/generated/torchtune.training.init_distributed.html
index bd70fca2e1..bc52c51af5 100644
--- a/main/generated/torchtune.training.init_distributed.html
+++ b/main/generated/torchtune.training.init_distributed.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.is_distributed.html b/main/generated/torchtune.training.is_distributed.html
index 12aee95ee2..9b85a8f938 100644
--- a/main/generated/torchtune.training.is_distributed.html
+++ b/main/generated/torchtune.training.is_distributed.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.log_memory_stats.html b/main/generated/torchtune.training.log_memory_stats.html
index ad70ab1c22..0d5923ba20 100644
--- a/main/generated/torchtune.training.log_memory_stats.html
+++ b/main/generated/torchtune.training.log_memory_stats.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.lora_fsdp_wrap_policy.html b/main/generated/torchtune.training.lora_fsdp_wrap_policy.html
index f7b46e3c03..e2e53bd058 100644
--- a/main/generated/torchtune.training.lora_fsdp_wrap_policy.html
+++ b/main/generated/torchtune.training.lora_fsdp_wrap_policy.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.metric_logging.CometLogger.html b/main/generated/torchtune.training.metric_logging.CometLogger.html
index 9b4dba7306..e7b333e16c 100644
--- a/main/generated/torchtune.training.metric_logging.CometLogger.html
+++ b/main/generated/torchtune.training.metric_logging.CometLogger.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.metric_logging.DiskLogger.html b/main/generated/torchtune.training.metric_logging.DiskLogger.html
index ed89c2e8c2..5d6863920b 100644
--- a/main/generated/torchtune.training.metric_logging.DiskLogger.html
+++ b/main/generated/torchtune.training.metric_logging.DiskLogger.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.metric_logging.StdoutLogger.html b/main/generated/torchtune.training.metric_logging.StdoutLogger.html
index 6a5d91ba98..b106560592 100644
--- a/main/generated/torchtune.training.metric_logging.StdoutLogger.html
+++ b/main/generated/torchtune.training.metric_logging.StdoutLogger.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.metric_logging.TensorBoardLogger.html b/main/generated/torchtune.training.metric_logging.TensorBoardLogger.html
index e62dea4fe3..dd6e70491f 100644
--- a/main/generated/torchtune.training.metric_logging.TensorBoardLogger.html
+++ b/main/generated/torchtune.training.metric_logging.TensorBoardLogger.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.metric_logging.WandBLogger.html b/main/generated/torchtune.training.metric_logging.WandBLogger.html
index c2d285a7f3..96be20a882 100644
--- a/main/generated/torchtune.training.metric_logging.WandBLogger.html
+++ b/main/generated/torchtune.training.metric_logging.WandBLogger.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.register_optim_in_bwd_hooks.html b/main/generated/torchtune.training.register_optim_in_bwd_hooks.html
index b0a549fc9b..2e2fe840a1 100644
--- a/main/generated/torchtune.training.register_optim_in_bwd_hooks.html
+++ b/main/generated/torchtune.training.register_optim_in_bwd_hooks.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.set_activation_checkpointing.html b/main/generated/torchtune.training.set_activation_checkpointing.html
index d916e7eb9d..9dc2ea3506 100644
--- a/main/generated/torchtune.training.set_activation_checkpointing.html
+++ b/main/generated/torchtune.training.set_activation_checkpointing.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.set_default_dtype.html b/main/generated/torchtune.training.set_default_dtype.html
index b82ac8f0a2..c296aff9b3 100644
--- a/main/generated/torchtune.training.set_default_dtype.html
+++ b/main/generated/torchtune.training.set_default_dtype.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.set_seed.html b/main/generated/torchtune.training.set_seed.html
index ecbb20075b..2b9deb4475 100644
--- a/main/generated/torchtune.training.set_seed.html
+++ b/main/generated/torchtune.training.set_seed.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.setup_torch_profiler.html b/main/generated/torchtune.training.setup_torch_profiler.html
index a984cc818b..98d510db84 100644
--- a/main/generated/torchtune.training.setup_torch_profiler.html
+++ b/main/generated/torchtune.training.setup_torch_profiler.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.update_state_dict_for_classifier.html b/main/generated/torchtune.training.update_state_dict_for_classifier.html
index 58c75ba903..042f68a6f1 100644
--- a/main/generated/torchtune.training.update_state_dict_for_classifier.html
+++ b/main/generated/torchtune.training.update_state_dict_for_classifier.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.training.validate_expected_param_dtype.html b/main/generated/torchtune.training.validate_expected_param_dtype.html
index 2c2e8b6ec1..a95820cea6 100644
--- a/main/generated/torchtune.training.validate_expected_param_dtype.html
+++ b/main/generated/torchtune.training.validate_expected_param_dtype.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.utils.batch_to_device.html b/main/generated/torchtune.utils.batch_to_device.html
index 725d7b165c..afa27e49db 100644
--- a/main/generated/torchtune.utils.batch_to_device.html
+++ b/main/generated/torchtune.utils.batch_to_device.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.utils.get_device.html b/main/generated/torchtune.utils.get_device.html
index 42863dccb3..4031314d68 100644
--- a/main/generated/torchtune.utils.get_device.html
+++ b/main/generated/torchtune.utils.get_device.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.utils.get_logger.html b/main/generated/torchtune.utils.get_logger.html
index 4f3c799faf..6137ce1657 100644
--- a/main/generated/torchtune.utils.get_logger.html
+++ b/main/generated/torchtune.utils.get_logger.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated/torchtune.utils.torch_version_ge.html b/main/generated/torchtune.utils.torch_version_ge.html
index 66d6e5496f..d841aab81f 100644
--- a/main/generated/torchtune.utils.torch_version_ge.html
+++ b/main/generated/torchtune.utils.torch_version_ge.html
@@ -332,8 +332,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated_examples/index.html b/main/generated_examples/index.html
index 2ecebdb8f3..b34ed8856c 100644
--- a/main/generated_examples/index.html
+++ b/main/generated_examples/index.html
@@ -331,8 +331,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/generated_examples/sg_execution_times.html b/main/generated_examples/sg_execution_times.html
index 95f71a66f0..f3dfb613ed 100644
--- a/main/generated_examples/sg_execution_times.html
+++ b/main/generated_examples/sg_execution_times.html
@@ -331,8 +331,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/genindex.html b/main/genindex.html
index a3ea408430..5a6b601df5 100644
--- a/main/genindex.html
+++ b/main/genindex.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/index.html b/main/index.html
index 4cb4cbb2c1..5d89d61c7c 100644
--- a/main/index.html
+++ b/main/index.html
@@ -332,8 +332,8 @@
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/install.html b/main/install.html
index 3cad6b83d4..c23f4511c3 100644
--- a/main/install.html
+++ b/main/install.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/objects.inv b/main/objects.inv
index 13405baf728725613d9cd850f29612ad5dcba7c8..7f6611ffae0a338ff21e8383a3809d10d08b828f 100644
GIT binary patch
delta 10183
zcmV;&Cpg%pRl!w|mw(;uXlbmrG}|#x462AyvBl!OBx^sN-#EWFUoyCoAPIn(0PUU!
zTSX9=YXO-^TnG|f*=+aSu^^}0ve;!0r&-Z<Nzr9VM>f^VUw6l1+ht|3Iomvtf;2Qa
z{i$;Zo$k9s{^~EA=wy~FFCSv@6RvrXaVV~mrbV#j*&XVhl7Bo)QRP<>CZ#kzIlaiM
zWSiWT*Q9Ne2huXt+dOGoi(aFURr}>W>8`Kt$f3$5?M2;V5r^ioCB+x=2ia03Nd7iq
zE7~tjayTT-ho&i;%d%+*uX?>)ESn3`t-%WXsi=<K=VM3zN0!kjX~mHlK$>OVC0)z~
zyM#8t_(0MuVSlQtrmRTQy{w^q*OZ4i&9)s=d?4MvOxIBRK)QkNdTKlJtz+S$BSp7X
zO?<9Maq$W17E>Xs_8VHiRflT2V-cX{Pqa>&U9$Co=w+=~zaL4<8rntCK9dI1V9k<N
zpf6<(Qp@sLmE7)=hWv7KhqUQj%#LtmQ7HQd!BoXe$bYJyHXTX8-mgUHj%BIKa?n`r
zCL-^C0@@=v=q-fz$S=Nw`Va@JS!WPIqm-prabVelScHyZOYzY`!`1R4kEXZWu;*kd
z9^2whHnP?_dWw+4C-<NXK*ZJ;M99v--Nm4Z7`l$gx9<S<Q8SLavN<Gu*B(D--9F9^
z%7GRJ>3<6vtYJ1MyDqMhG;T?KB*m6Mv1WpHh?3^wG{Cxh39C|yoRX9lHP2zQ7=sj0
zd7M_I9{`htg0#EN6NE}aLAop_vS!;jDbo0mw2z*!MqO4ST|*HNmQwkx3e|M954=##
zGeH|fOF9}jOGe|pxP%O$=2A}mH_7&qq*(nUDSx;AhU`eg+9w)gD(S?z{dk837(|?!
z@92Ezl7`TnFw0^=jM7d#7Cx!*Y?hDkreAa`GmnZJ+2$b;-bEJO&%&d31eZ5Oe(nTv
zSu-yd_e^dVA^pzn&1P&l30Ho#<fM_mmx5b_k!ZpvIPo?oNkOA^IuHYdN{0q#3z8!K
zV1E#hEggFLK<ILAfTE!Jvi#*h|BI8R%V-_6UXV^rP7)7dt<MK_BI<<M@8c%l@8Uhl
ztKBj8fwQP^!m=gx=D78;<5boObU5Z+b|}*%Kc$10rz8$RUQ`C9((OsSO^$imltmyB
zlPV{TA*lNx&!wGUZE|Rj2OprZq#Gosb$`+L#ht0F8>pk5W=y`(NK!ly+D&JTUubnI
za#EOdKR)0V6;9aa>@nkV9(Yq(6OdHEI#i_#0AY{<Dvt|Bs4E{y;fJFsScL3Y18tL>
z&y!;Q%z@3Qvf=?c3&W@b(5cEc<EGEjb`3smeG7zWL$7kAwS|mHF%60&ryXyeRez^<
zbe?)6sU`V`!#&}omSzwD6`sg!ndUiZ(#6L@D^W=An18NN1y)5tEWE43pUxWfu`;$_
zEqIa`nkrEu?Q%Pb0=*R1Jki_%ZXT`TSUxQcSDIA{;!_4JefD4cl(oD7#g^V_03m!q
zgUfsApO_B156Nr&cYJn!kHEDn@qd7R^YaSOREZDt8pu%Kylj$ql{z<R@+IM2`zB4^
zBv#`9rBJVDjXJ;_^?LTG0}N8HW|10ak_U1)B;tQKo+boS9lyT!(oW@o0B<}2pp_$L
zNZCIyy@OGF5sur>%OHa!0>r~KBVi2Gn-C~~rRtz!)wpuiK?SR!C6mqiihuuMn-!p&
z(3a$vF{e8YoFt+U7Kl>Y>54yaLyW|hYs06Pnu&s8P`%Jh6b|E$0taabn-&IO^$Ng{
z9?{yO007Bavaj<EvM3-wHyq330KFV?ERO^X#6!XbIH^g?GI$+o&KhmcI@F*w+M;!!
zNo%Z4>vfD;ds+2d{z1YiF@LCf58LUYV3@bwC~x5)zM*=q007w=)x$)CEDFfaSYvq{
z;M*`fi0$go0@mmf)`3N=k!6g}`X5*<O%wwY|J%?1ejk9v<xzkhmYay9;NAwI0d1yP
z*~@UcJQ~p3gfk_0XisBKb>M;DhU3C)Tdz{rwO?h|Yt(k_Ro(Rp^?zOCEBtl;BR<1J
zj-*jw=RsiSFfgq)e2X_xfrmX0hlQ24UZJ?PPkHMV3S4`XSU!&v3rOdG=NBJ_r+d(K
zGq<KN9@xj_-;dBuQ~n{|!2Xpa=?e8TR=5hQHEU}XdUs7ywCruA73!<k70Tr_CCm}l
zsHF8csri|DTjhT@yMOl2qSx;oaK@}?ofzR2IDHFYi!aftEjT3PK}y#3HeCn<(OPcf
z8IJ&knOw(bJ|hB$GKAYtr9}j5)}td822et83j!((pw#JZw+^Y!BW8_9W~(@pMPn$b
zaJO=34J8-uU=FPaa`pQ5tUU~R$Xe=*`RTRz>+#Urp#gw9dw=fjBoc)&L6zP?n%A?u
zfSb!C4X7bKJL>6*MFAF)*L=3iFl}Z*5tf3-qb0W<00D3F!YvzORFpl}Fr}9jS1@3J
z?R*}0SycdF>)@tZ<J)SzR!gbTO{Lmr@@W?P1xFmqfdMGDBJ)Mz!b?yd#ekH+mpqC$
z&*<avHVsRlb$^gyYOrPMXHv+OSpE|ZG?hev#6QQF=N(3I4BS~5+!+$?O$gYVWhgI0
z!EEGj9)xB5je@X;uPJa=@CgY<aSPlrC++D+Leh`JqVdA4VJZUDj9HdOK~(3juaXzi
z{0D4>c`gLjjxPcf=C{9JCfhxM(wEr*8c!P{U)=@~Mt>+iR_cqwYiYh_UpQckj29jI
zq8amZV13LbaX4th1_N@l#9L+b9|KAukJWbt{c?SaX@)X8KqJ~h5Y=UuG%toXAX_kh
zs|G(@d?~#M+$F{S7uXWfC_h9ex(FcU8A-3Qf+P)?axdZ3s}^!ebJ>(_+xs{V+F&fF
z7JvLx-hWctIB7^sisU{gh+6#toOn3-^BJKX?t2mOiRs^sj27$LAIs)Ffo>kLNaAo-
zY9+zg-mSqCd4DV(Ng9{@`^-NMH=woo0cx$P{c)%avp61P^~Sume=K}Z>&2nol6)83
z_05%dqp#GYzY`<xhQ21BfM&QRjleE%O)8;H27kw8vXcbfVrGIU6tvw@a7jPKb4|q~
zzX>B~zHef+Ef0sXh>u;Cx2Gw|Nk`(xC(Y8!t76M~T-7ob%vu(ZdP#2r6Fp!8dAUvU
zVEC~XFnB|%q)Djfu~R+bq>a<=rJ^CR<KI8ew*0$rUA!;LM_*<L2p8=gp@SpYTFeWa
z5r1;|(Y<bJ;?P&`Y<Vw2ESA|<3r(0gi(GNRmOd;iqY-+|a%s`k$b(KFPA@-v&AR=i
z7^;esM6PH$G*-HFxNPr~V_(5MQdiot2@V}zbWK^kfaB?Rp@a?UK_sNZ{&t17#7Gi{
z<5xsaz;K^)s|sWiSQmbjwj>_U7-LuijekNt5g8e*1$az3V`6pS$j_lbtpvkrg$jk(
z+OH9q3c{Zn#=9fmg!rD2>LdSo&61(5#-+pMn@+iaS>X5PgPQ|>7dF@^@?A)HEzW!w
zCSeOU--{A{M!E_UA%pqU?V=7}>@YuFjRe1r`OKpIGpL|e)ok)Un5#w66?p_i&wo%B
zD|ll7_*!1*Pr+xwZdC%|LD7f7;_xO4jmpSz6_~L#lt-bl3)zPt8BYRU6k=dRgk5CG
z&Mug>L*P1<TM^h98}g8~9ch%f$mGkztj$HHBD?PNB26kPb=_k_I;OcK4(BuxSJ)8f
zfqPYcxxj51nTR0}f{Nvbi0#L)P=9hU*5W0RpSeJ;Xr-i9%w=UG+KWkiu?ATCp4%0J
zIM=WU%Jyj+T{rSD?{dqq@?e96u1PXJVDv1$NylOZ4Ci%bZOcxRVP}ofT;D2wIyPI;
zk69&Q$w2kO32F}{f=L;4If)pKk$`MwjKVe(qIOW4csGYuJeinn6O9m?IbZF&PED9y
zQtaagw!d(`>oLOYOmu9cG~#B|t8&pAP1_eE-2O%9o=79U+W6m|zDrtir@ux3=?!i2
zF2A(Oz`wP!KDKzb`$u8((S1l|j}>^K`hBQn6(YBU9Y($-H<Mrt6a&j1xs!JcBms((
zoD3@gN0Z15QUSG-8x1=FQ<GN>A13c_K0~&p%>+PN!OO!)^*313HwdjBq(SPx+#*R!
z0gzS?(jfJ>lbj7M2OBi-%@XX5lhF+(e}BAve{)mj+4kk`rQ$R4H0crxrbWsPK>Hx}
zLQHzOC)>xW%!=+~mXpN>xv0Pk_4NLj<!QX_Q)zAGq(}~c#l1jA$2{k=xnDmHV#`ZI
zOY%OL)I$@!4iF_@kwlw3ksnXunsiB6{ZBRc;jUnVm57Af!H>utyLU(Lg_*;se<pyz
z3WiyoQn-oml!Tb<6F?^EC7gOcS9AiwmU2E7?~?5!8##S#lBz=3-mbzEnNK+=GX1ZC
zSMK+rQL@|dBYF8ViSRZsJWc;Isj%%2c@wwP{n`BXPqokgk{anq^<%4q=Ez^@7&B)F
z7|>9Zus;dkIx$2gxvO$v*oxNae<U`}sR_~oEe^P%kZ}fLgy+xPK3Ivu`exJ!$)Ep1
z7ay(Ad*MeIkm88vJ<C&<C}^gDnJvLt8#?X?*K<2Z17;!(VUbbvU0D&|Kd15Y(5pjZ
zf<uin46uH4Teew2`k)v;XWc%g0q!7Pk$_TIP_;;dzh#dF<U)(8chs-xe{{7V>{h&M
z(~4J4T=9uG7SpY5kxbrT0rO=W9>}3=UiiSH#j4k37lG<yLfZ^`q<N705HL_<=|!+(
zaY$PB#&Ew#lau11+oOwURkTRqxomPO#=itnq;ZoJk63Vn3MVZ6pE@@Cj(T=~?8Mtr
zP`s(E8>sb$X~7_Q`QYuHe^c27pe6}w+eQUWNWl#>F^dAnm$><$g)KGMHl~&Oz%DOK
zns7xg5c5a7iiLWzif!|<CGcVR`UGg&pA1iX&0?4R=`3)QvvoMU&)UZ@7aTPB;RZTj
zFnoyntpOfpc0UHUU0NOkLyxY*7dz1Qt568B|5^mXe888*9Vyzfe|cBZZY#_cW{QGX
z1UkjaWJ26yeJWv&62#{#9Xh`Y^Ak;lCvu=~tUw^dAJ%~n9j&ayBFGb=9Qrq#N9!u=
z!+)r0j6;nRo-N?7v0&`dvq&H%yX2UAEYHk@7ol2$n}sTwCB;X!orB(BXGOA23k?!v
zChr6k$C6Cns6-u|e+Sm(B+|d!*}r4TUxCG2(y7QqK`f`<CmpS`uB2Y@lx-o1MmakG
zUjW4E9z@(~^S-A6tIe(PMC1`J^h6rIwe<ncc`7(SBs8QY_axu6UHVt?D>~aGRTej7
z7x(t#`IBuVa$NI8i_LXXBoE|3ita=S$jHu=rR7#vCGK!Vf181xNbSY>_9ufdOP`9e
zw2>ffX01(_-7_g*>AezJf2=YMYKFU0MZ=w5l{m%9q(+`^HBli_b#b<x*dl*a4>2~g
z4amI)wnNP%&Uda&WLetD)Z`}NkvnYW&AeB_ye{i9a|!yJQ#IUIPsFpyI_XS~*&S*2
zEXmuSYQEV{f2^QO_hf<*O_yGJ!LC_Je!sY*C>wEn`bJjuJ+=Mtc1=5mp*OQ_`;+$O
z3rAowktW05W7-VYCsJ&?r`${)X}T{efgSt1d9NH&<vX@rd5GD^Udw{EGNZ0(9;uHf
zZ!QN>I$qg)A=|7Xy?#X@E@@<V%E+@p|16Ktuc&T5f6HX0Gb`GzIc}ZhpXCucKgr}-
zJ;;}hv^NqAO6Qq`)@zZz>G$MM)jE`oRSbgs(=~23lPppZ`&c-=mE<Nm{Y_Wh8FI~Z
zPwTHrwvUq*J-#C}+$M*LeIP}o^}@8tl7>I<aL_X35=w9S8*^IYW4Fy}$61t4mby{L
zWcLa?fAX#4Eg>g-l<MGX%PIiw=-p_zJ7Ycx8xz&zps~3d;)CAs<0E%hLyWUViB~Ks
zl1UMlm!WPPKDeZcl<p|XvTC~34XO4cjjz$&k#`3&da}o{1Ki1IA^;yrjKumfXlW<g
zJTo&d2pzo2!&>PNDcpA=o6D<DT$shHCQHd!e?BFoqNCs^LW|Ca9_^s1m!MJeK*MFk
zzI?G!skB<l{CFthLsA*-WsYjEgjB?&zOcQ{tS&_w(|C}hKGi(UM73N>DNTx$;FbJK
z`#8Rcexi-SW`@wlVRQgE8;OZOPVceSA-|aV&_z!F-IvFXa^sm#_(lWfE~5ZEt<F&h
ze}1K8qF&}wrVP<}))l%Mm$BP?C6c2dS^1{3JC(|2J&>zh)4oAL%iGbQL}JSun?dfR
zA1%*}+>e(>Cgvk%6@K_${Mo?AIA(sOLQXI93mfrf+l{=jQ8!LTA$LLRUrb|h^O$5>
ziRMv>;@L*M@O+%2d#}DTNekqbENY~te=nhA#ZZV7Tbs4%p0(?FOP;qitFoab8rqze
zv*BJFvlNzK(KpE@^nPV=|Dqwv^Z%yneWp#~exb;tJhLKyCE3HC+UlP-?=osN#=6ZO
zinwaBg5uHXjj2UXT!4#2gNSUw?|C{MI;iv$w)+6yW3k*y)v#^hBrTr{zF#+If8C?I
zuNAKb^}Bx54XIx$int<$BokTvBx2%nLSB=2mzU2r6m?&yrB@Np{ekvpT^Z3oo3CZ_
z$kTXVKDaM=qVay8GzE<`?vhL}s9YrFB=IlDV)KC}AH{6?$i7y#`H+)iw<pcTr~m$c
z|G(w;H@^z86Y`A@K4UftfsTfxe@w(>K8TAAzmWNQt*UFr^xH~%FRKvxgel4}G&cu+
z%gDV)(&QDEmZZ>n{|WUtl|1q|HR2<i|0<6Sy`%~@S6A0352SE4mhqha70s=din4Y~
zC_+KXV|!v+p^^g<R;_Gj7yA~O7Cc}QB9v$eE+p*<1Po{p5IDoN3x<Efe*=iX^9tqs
zr%SI|N(HJ`*@bp>Wdz`(hKrm|0F%4&+Auo?3#o2soN5PU6N>9AHK7ARjzxx5lp)Op
zSGJ`1LjK@O(a5segcSwzT|1dynuJa_nh#A=HkW175Wbv?stdOY)4Jz^h@I`HqB?fs
zFb-tB870-qkKvY`gIdyof1t{|qlFy1=Fb%=E<Pa(HWd<B(^NaCSk6Ehd@Q1`iY!}j
zobMAG+3u394{k4OBGBKoGx`0<K4?)E7e)I_8dL?CMU9?r&T41%rOaU*SVkivyJ6RV
zONUD2mzz7Jm(9iGs@*Kz0exg`KKlnj)%8ruSm;?UJB=NG&w_QKf9xYJVuycr0M~?F
zToAM0w^Pc=u8XT=*xe0cGLw?ygK4q@m(3kGIT&D{U5!N4wH<;INl8kpBL`c2ETYCG
zdKwfiPOH)bl8KpMSvnZH-E}_DRLlrX)4|YmG)$ynTE1hSy*uDT#;C|h&zRFrSbe+#
zL`r3r4PRc_l@A}Pf5t?(nbdg5+9z9a`IPFiWaQ+!4+M8EB}a04k{k0T**=mKi#d{#
zU8tV4fZ0D-LE}y(r%!4?__rVLu)u?e?7I4EHxzpubJ;(Q!J-<ahvz)ATW&51pVP$K
zoX{z4s%G+FGG=kX#(WP%$CHK$9t5ke#(+U}ApxnUEC3ele;HktzZ~d)anf|zmMsUO
zDLB<gQqDYXnh#=Dh2}Mp6GSG7Zsh%*Z|#WpB(HYI9L>NYFPK<vx*1K^)xcDyg2UpD
zp6Ui47S2v-_xO~=-fR{%$&r}x9++geC-F8p=4mrLjP{{u(j$9q!yPxh`7?bmqq(RY
zv&_=n6x!s_e;yBL__3IP9;Yq9=CnE+kGMJ&l+csZ1@Mjz0%G!w2FK!oojH;<9vzKS
zi=^es;^t`CJ^^NDk(ZuozUgN4oIPe-5KCn$C^$@0jo{jjKy`Zw)*vE-y}W_~#RX$0
zzE_9G)TGDo+Kpu<><qqRo8)||5$~Gvz-GNEf3y#nfA3&)DV7V_Ae{SyXR@>nChb->
zuy*HLqTPz8L2zn9pbvbNWj#vnMcw9Eb$UnVkA_2VmOue2<XGE$6T~a!d#F}mXH_(C
zv}Rdmx&}!%KVNF>zgT_|Ssee+U|1Q%<yd)s!^wsI5zj4O>%Zf(^Lr>Ay9UW<oan>m
z&Cj81f2f8z8=C0FCYX$g3z!Q3qCvQ=Ols5;=KPK~IX}~x)L6(Nj5wEFhntlem|4A&
zQER(Z>kzZnE7`TS8@3LyY`u<YYpZQ*q;Vg};gE>`fjl$`HAta9ff(`Ydpryk(vavC
z6zq);*qad;xb}t@LOd|L(a|!SN#ZgCZgjj6e+MxwFaXUt<i*-_Bsm2JSy#*c#vA+1
z92;Wn+P~nnq2x8P=(T^@YeV5{aOq{_;d}XCv)wY#2f<77axdpB9Y99>!4Z&u)XE$E
z4F*E=bu&2~aH&HZ3&T6GncN6;p_iHndty!?5s2dC5MSV4dW7Gn!f9ZmxJ7zI0~E;h
ze?gm<Uk`#>T!VY~klVw}$HFswsKaX#c&{G|&p817UzZ4>U}AtJ;M*olD&}-Bt)mX|
zkJ{)UHQqz&AQ!2PK2qbIqz>_t+T<oR+E3~g9Hn-6O6}(==kgDRh?3EQ<~bM$>jt`<
z4hS-RFZ|XI^X(Y&EdtT2W135k@ZlZPf4LS^u(llpr!nXphwEYLvG5EZYIIHFZ2*P0
z1_mG>sDo;uHdaH8uZKFMB5G4j)Yz(s&-x!Icuj891kO;9@xT52@AtmATzCx5lW+{M
z)}P2u@IiLhRlM~_e>(+-+FNrGywP>xdFwmFGw}0)nFa$Fgg1=Dd|YCxz(fb^e;gyJ
zw}E8d8o2;{r4Bb$ZL?MFX{<WfTD8SowV%D}P=nQGi`5<`d)@zt&yYMuqNC8l5^&h(
z{@CYJ>~kQtR#{$!pUBT!4ZIpgLg#^0&W%)nwNi&xOKq-~+N)yf(3+{uRa5)ajeJ@z
zmQc^dals!@PvilbwAI-04PS;ge@*#^cth}4lB6pT9V?vk9pCq*eK-0NM;PPrA=?|{
zEey}eb$o_yB@RuKU^d(cVVIH-H??RpGnF?(*pgyRwxeAymYX4LyFgF1>N7L$ygYUk
z?!w*HqR!Oscsj62N1zuwR62<J__Ik*pi?_kI>>vtcfGQ&)NW^~eZ1wse?H@q<fqr-
zugB9+hX(n+`TPci5igCrKvK~~HIvE|EI52E+9Extu(14W)T4d)IDpOMlgAs&O*A{~
zvb|f0fg5PuUTO7nn~}#LoW<KurY?f>j$p^I`GlW4gnEL-2}10Te$rr-A9DdcY*4!D
zuvk(XgGucXP3lz|Z0%~Xe>J|zR{KmoSzy0_WX3{Tv|yKTG|IHdydb&c1ed2Rf@4cL
z6cAK4uVPI~gewfsiEe2&@d*Q1Q{&2Nf3wymyVh{S*3YDnOGo@C5PK@7`4hj!K*m4E
z;Pqjn>=c|Qq8OY7FgP<YIKwh{<Bk7jj{XLKjkmUDXrPTf%{#R$f5xbu_MDm$V?{3m
zj%|lD8#-heWp(JqSjCSl_((|lv20k^2mqdn2s|ypja8k$zDiz5^B=H9mAR0_)03Qt
zzy19(+3pDxOJ)}+^1`W~f8d?1hF)~+yGD~!DsYn8C*w!yP6A)9Z!wiAvkNGB;nrVv
zN%LZOy{!d9S~W-(f7bg>Vs}Zg{{^-zFv>4q+LKpqh>WCHSwWHp4AM)ecNf*Q8%VAE
zk>;`~+qO47cXBYeQ;Q_z%mXF)9=u#<k(`lv?_P_Ws<|ckF1V8BN@ld?S)2?fPafc`
zW`csrEZK>^Ef0sXkZ0c0E`oF<ete3zEQo<yt5b<!HBCj3f6ChAJ|9H=NKKSTQj=6T
zsl43M$zc$su?jVG)6?K|8p03ZVh7u?gT-5M8kc@lSXH(iKV{B~D?-002tvSF%AAIq
z)5{NEvu=O6KNgQ96(1M9qUq3*xOB)5gqd$HGn3<&->%SnMv`)Xq9-A+!z;(FYQhd*
zRpJPp-H|^4e|S$w^^yO)X30>VE*<h9qj4gfkr2YX<iDsxR@pca%Lvs)+2nO5%YD&E
z3;WuI_$iL~YrVM4v<#op7$w7b^(Sabd6CJBoLQ9$VWxXdnibh~rx$5bQL$?jtGT57
z@V1_`jD1ypxxfuRCt~U(2X(y+srt}FFh$``WXGMde@t9vz+6dYMmnlReDKh5`Z5)E
z91A~ff|tO9j7fp~_>1ArGNIv<n_0__Zs6Y&rMcAEo$LfrwqArH2~|4%>Zdz5qA-&q
zQ2h&FP&6TJ$l)Zaof*c?6ud*M&p71;tx*z`2k#C;h>R7;6^Bt8l*%n(Uvl1(n~ZEp
zX#Logf60%MTJJ;Y-er$mCX}gBkDBZoZbA9+6DH#QZdkR8|7hW>Nz0nFgDmDZp)A$U
z{<FVvQN}H8{h4|U*?KSrXjNY~U(Dz($#A?EW*4GWcCuao#TL{0@%qa(;4^0kRr>nD
z{H~EGKZKTjmH)uPWRgB5MF;szv`tP>Fx;@Cf3tpCJf&^pYWW-GCYhnb*AnfBe9rX#
z<}+lTH4~7N6>+hEaf8LU(J^j77+-FY7*as4!`?zLZnFcnB)y1?<?7Ooq&{Zef2%*{
z=L)lPliIGVU|(2-YZ`w+-I?{od7%2o+xIs&Wu9$c?$}Z63j{4<wsV{(*%p}a;}+=6
zf8k`jk6FIlm=_iFD+tEB3#5*D&IcR6e!R4t^5-M3=EcUFKt|W3OTsu!HB4AdZv%PV
zv3u<3uP|~(H4}!$JZ_po4&&;*gg*3K^ybeM?L^pE;S(KPZ68@&|Jo!~g{TU<3UxA&
z?Y^6uI2gSpIeRhUiN*}}hkx%g9-XHPlTRTP1^eeTenyjNAzlG5lgJ?}8$-9pRfbi|
zK+KwDnd5WW<W!D7Nh;EqPGlZYlMW(OHT$f645Kn=kdxH@8b#!`OUq+mt>`-B3{Xo!
z(YYf<TQ=`XIyejCG*hIG9-1Vh*nOoz=UtPRB7+LKPdZxjT}l1zDU&24B!4H;@SOn<
z!17F`v_(5Db%2fKdGhRJ`;>e$f=u!VmH($GH(l1{<o}wKwGZ?+*&cbgqaoR{z~>U^
zM=I5^<6(2dlTU_&OU#GMYX_U~)8+rNF&%os)C?u=m8ev8(lq^f^1v306FQEh5j>`G
zyBmviN2)(d^7bb=+$_b44S$DGC)`UF5mga!6&HPPu(=%f20B`~+XaF#4I0+x-Hx3C
zI|ABPM+U^w1Hh=3NNE{RhUKfDerae5DOl}M)3?+ro6DR39B9aiLI4G&AsUoEP*7Tv
z+e3g_rX%7y`XXFxx5vXVm!71+`aJ@EI9&XyeBOLvlTW@YM;Ps2*neJ`f7NlGX@5qx
z;>Gnc$RVk&sMi%SU6cWhYd0!wuwXkdp!wt=D*&3_i?SkOO4!~p-C5{JoK$q$N_|8v
zdTwr#R%5{SSOj2m@f)#yl)`=3?XoSm?8mGi(cO`CjhlbD`S9^htM|U1LV$Z6Vrybw
z&jU;y=tn~ScezNDntxsDci~gBpsz*0uO#2nc_M45-JT3lbo0KX-cKt_xIa6^Fv;_=
z5}u@jsJQ=0o_5F>E1_7Fa8y=Z_?5hf#9gZ@{S?Lc$ezHPG~Xv}wq>2zSfCxCSaKwj
zBd$k7W;ZwT*jEj<(?0`&#p;E70F~07$S+hF3NpBZ_L0?B;(xa!%pU8=&koTV9gHWm
z#*ZmLFUzHpG>GUil^}HTP6r)1Ybyfk^lbC?ooKf*;q~V1-A1(T--_}%Cn+7tNIBs@
zVueI4-nv=oStTQy6saUhOK-#Ckbw<bDd$@bCvdFwRYp{CXM!2Cg5*RWF-{l=F-e30
zZf?2A=Diq{et#4hSqIED+X*Zedycj1wM+gJ+cQN<uUZ|WGl7)Qp4*Ru$`J`|k<unv
z-8Etuz=k0r$0jM#4c{QUsYt^%izEfD7tKIT|A3=0aZ=;Feg-4RG*K@lv2WFNdt!*1
zaY={6PZ@c3m)jn1Wr6iOk<!^WLSEv}T9(acndL-o6@MDOuCy4qW)eLUo*wn-qtJCN
z?hd<w+BtN~YSdiK{IGR1OdByNrN5i$Z=od`m!mb&XG--IDY$0{xk$<l;gKe8u0-OA
z*swk9{4$c&HiiksOeoTXM2|isD7A0{kMoal%*O2x^BvijzldwD9Ql&&Y-%LZnW$k}
zRJ}y%Mt}c&ob5ap6Qo0hutp`Bf1z!2OGhWe0BAUv7~GZ5MO2;jH;mo6R&9NUuG+3X
zkWp!qtuI3(b_J1<@T*B^FzBl2NSAr`QlVu&p;lguN_Q&`0F}2+SeTXTU#Nz|<f2|q
z>b}upHjCsp>i)_6wC3fK%tWO5h8srn-SoEy^M6sIhiVv&#ZRqi|ED3svt#8IDpS#q
z$)cw=k}4vpLhhbrJ(|co4A(aQypfqVa>M6@Q46S!v4D|PJE;lI;;tVwYAto43E#YQ
zdoc|eT5_F?o)ZFwM-vG_nI=YB;&`C)UVFAECn=)yyr9!O-c4m<_VxP(oqRF<<SEJB
zy?=Dgcf{#~6Gnl_Nzf1U=P}`{`fM&!_md+Y_D*zpAEc!L=(!l*&=Q@f)A|lomM>u$
z23<TYlH$q<jvX&q)@|7As4E``EoAf8&F}i4ko3?EiTByVJ||Cp*bRvW`Tyi&dx1oU
z{cZ9@<Va8ArG!)|+XCcSRwb|8hh9eP`hTyRcgHMG2XuB&vKT_vh&hEF7&xoo;3u-B
zD6(TTmKw9<_NPfcfhXHXFZ`8;q|F6w3|~Ym``Za%Hlp`i(8<ff48_vOO-QCVfNbhY
zmHZo<I%HcOxl>rA4qfUrn=6tTRZ7#e(~<oQsU->R(8roZntD-$-@l=uo95;1)qngJ
z7i{h&RYbjtNUDfV+)MB9b@p~oZS*i*T$EiE>og~Q_Fp#tq(uJdq^<Vf;s-jFn=kdp
z7J~X3Jn-qjD+Uu$n5dVDluXn<M7>NTU<`#+JeNo<G62DK<XgvMe@?g}=s?LLCpoaY
z(Qwx|WV##Y8OBY+bY+>vWCAo?kAI<t2@nrE+kEtqF_ktQQ(s<vlKCrk+Z?qu(N{b+
zZ>xwVNaV}1df6nKd)hCsrDe4SQZ}1yS?schUca+P5L@f}s5xE%Q=oDlK}S@(7Z$>j
z5ky)(7wicLun;P{>f+=jRVs!jq8Dp{+VWN{3DNL7t&EnB+67l;AAJ-&wtv!Wd~6-o
zurJnfhtl+s*CgKM<+E9*x-0a4C{jAWj_9Aw*Rpx!VqEdCa<yn3GmguYeR0U<zsh4n
zFKOP*)z$UMba~aRTR-2*y4vbkQFUmYExfL?(rW2ZEJS9Vj(c6zy<}#_ur98-SI*3%
zaWlkhyzU&F+lGY}-H198qkqIBon=ukW1B#?)AuG!qs8s?t%z5HhG?W=^bmRMgWM<L
z>EseI1LX>$={!`9{HOEalYX}dD2S`vS``1gOd4Y9GcE4DTKTh9)+m?I&Xd&ApIYr>
z)0D4-PGCAVX@8f|ey}}xW&@h6jhVk7CxA>e%G!uc1cd+}zh^CTsR+NF{108X5)>z|
BzCr*1

delta 10178
zcmV;zCq3A~Ri#ysmw(mmXlbmrbhcxj7*r9ZVvEIlNv-`fzcIfzUvh9IK@tEn0lIr0
zY!yLdt_5TwaUn=_WwYIP$AX-0%VL*3o@PbcB}JDd9obZ`f88C6ZI_kB=4|sw3ewQz
z^oPzNbh_^j`J2CNqLW#!ynKwsPq^km#-X@Mnij#9XLqQ3PJi+&MU`Jkn3U4=<n$u1
zl5KKd-jKFU9!bkoZ}X&WEqaYYR_&Mjq`SGkCx<GRv=?=cMI4&TmK0yfZ)8iAAo<&b
zt!O_r$>ESRpPHs@F3YANyz2FGv1~3#w+1Wl=b}1xUydF9A6Z7Dq!mYI0BM$amvk`~
z>=N1l;{!>vgny~7nzABI_qvAmT~i+7G~0Ge@qu*vGF?OMBk2ad>#6O??;Q&l9VxoC
zYT`>pii^)kx0ni9wcpbEtvXcG9g6@pf2MWP?2@eyL@#T_`t?X!*3d4B_JuT<25XkI
z0(~uWkXn|{s^o5;G~}n-d!$Y0Vs?Zhi$d8y2&O7#LVs5EwCP9!_I@QocPvX?mV?G}
zHxYUF6VM*XL2n_vM}F}g)Q31&%{qe!8l^0~iUZ3Y#3FPQTZ)ek8m^WXc{IJ{hCL@!
z@z@r3vXQmc(NlyRKDh^F03x=&AVPKq?k)yR#L#s_zI_L<kD77ZmCYgPyY~1c>-KSW
zP!6;(NPl0@U=6c5*>!Q1q;X5?BPq57iZv6oLzFZZrvcX8OIVds<dmeesCf>X#TcZ3
z%Hy;u{Q#IG6r|mCo*+~b3esgcku}@KNs-2fq<!*)HR`ew=^BcFu$0PgRj8(;ec**^
zo(b9@TGG+LSuz?Q#U*42HJ5VgzfHDJB*p3<Nq@QZH)KZ|);`e~Q%NVz-PJu7U=VR?
zzNhn@OBzCR!YqpgF-kk}SooyIvspgEn|{%)%seV?WSfUXco$i8KMRlE5nSFB`MDFw
zWzD=;+%vgdg!DVNH=D8LBwYE`l9NXMUJ7mzMxqIy;KbXUBn6Gu=|Bt+Djgb}El7&^
zgMUFlwsh#}1EI^g0g8g=%ktL){Vz_ME~9nOdO<ohIY~T<wLTxziKr81zmJ=Izl--I
zuXe}W2hO6x3Cot$o8#8Yj#F7D(BYVO*`Z96{FDw}o|8BPc~KdZO1CHRHaX^LQx<_l
zOsbqThM?|)JePKYwaKA99(;htl5UWk)_+Cg7k8$zZlI2KnlbsEMv~%@&~7?w{6ec!
zk(0ut`|$y{sBpr*WKS8F^T3<Rnt-GN)}bn000@H=P<dQ1LS6Yp3O^i8!6IbG8fcs3
ze4Z5ZXAW#el@*WBSr|qgfKFAm88>~FwrlWl>sug18+w%^tu16sifK?JIqi7!tbaPa
zr}NZXNiE4g9UcfLwKRhOsPIH)%QP=ZlP*3MT8Tn>&-`<RDzGXFV&PpK{&d!;kCn0g
zYQdAl&{T;MX_wnc6zHY6=85JGaPw#t$MR`uxYDdr5T7z&>9hah=d9%gD7N%Y0|?;@
z8eHB>|HO3AeMsKwzvHv>2L!HNiGK(5yB}A8rb>LE*Fc5>=Vg<`tJJwklP?MH+Ba$P
zCb1d^D1~}GYt#YesMoVc9bk}pHH*|ZlRT2cArb$>@iZZr>iF%0mv$-#1bF8O0IeJ`
zL(2Yv=^c#Xi*Ve2UIrN?5g;C>83|*c-i1H`EL8^;tHzb94k}m;EtzcA*MIyE+pGZH
zgtjEVj5*zL;3N@+ut1dBPFMVa8)783TpK>U)Jzl%gX)E5qHq|06gWsj*t9SJt5*Pq
z^oZ6L1pr9al6{?TkVOIcx#3tI2k7OHV|gTCARZDfz)4M7mci>#bJl2k)}aQi(H5-(
zO<H4ZTCZc&+RLiv@(&VDiGM-Wd)Q7F1;f1eMtKhh@eS2;1pvt2s2(O7WKlqV#v04x
z0N;n<L2Ort7O+N_unsI@jVxn)*8jj_X`&dI_}_l~_lE#1E{_8Au-rr(1@}G(4QMmX
z%3g-k<<WrNCY&k3Lwg!?ssj)FJ{%Wj+j^C{uKg;zUZb{auj;N>sDJMoU*T{2AMqI;
zawLrcI}ZXohk<Fe;aj|k3Oww2I4rER^$NwUeac&}P~h65#PWHZSU@`eJHPleJl%t?
zo4GZG@xVSV|9*sSoANjD2KLV+Nmr<!vBFhYtyx>E(7SJvqGfL@tx#XRu23$gDPfMV
zMkTGsNzKpH+baLN*?+Zv7QKG&fHP)A>%<7J!0B5ETYQODZNVWS4^pzOx9LI{h}Lo&
z&v*nV%;Y*g^BEC1lp)-HDlH;dvmPC(Fn|(zTM$rT0HscMyLCu)9x-b?GF!!&EE+>e
zg}aqQYbd#J2XkmekgM0XXYFCoL)KDf%ujE`Uyq004h;a@*?)6yCy^+O399rC(!8GK
z1>9UFX+RC>*-=kdEDErYyymlAhG{bkim((s9xb``00?-S7jD@QqoVA&hAF+QxPk!#
zZ0Ga1%c=qZTL(AQ8sApywOUGzZYtG2lTWkQFF4{@4h%rK6`3yz7hZz$C<deizT{E7
zc}5?Pw`o`ct$%|IQ-dv2zmP(%#PXkTps6GRB>pkRJnt}yW8luh;Leb6??S-dEkk)3
z3T7jJ^B^qaZxn<*d`*F~f=@^=id*20IcZO?2uZJoMdO88!&C&Q8M7>nf~d~lUMH`l
z`489%^IQn59bW_}%rAexOtyOhr7yDsG@dp@zPb$}jDJvktkf5U*V25$zHq=68815a
zMKk8-!1|a=;&9N04F=?9iMPt=e+(#vJXYTo^y|$XrWwlY0F7u1K~$Gr(!3hpfNa71
zts4Ar@ul=4aGw<WpI}Qwqx=w^=pulWXC%GO3X(Kn%DseBuUg0@&1F-zZSUheXoInw
zTKw@(d4Ef7<D?-mDUyerAZqmw;KakppU(*GaNmoFPfY)AWVBe{{;_O666od;izE(b
zrB)J*?cEwYk@v^qiKKDKzt8-~;RdueKR~TjwSOEc!z_*mS-ml@?LQVisP*Dd??}E2
z?)v6RywO)`(%*@ZcSB#3Pe3zVlSW{dw<eX)CVzwDGTBK2Z!t4L6bjnzD7d7b;<={c
zk>7+7G~YL|+LnhyS;WUK%iGhG<fJ3<)3avj<yEm|J+5jQ3uY|~NWG*tfr%b4fxO%%
zc`*D~3mCj1RnjEX^Vq2#ani<V_gc}A*zxb5XIuVVxGvt8<&!Tn1cZzBj?lr8Y%S&m
z&VLBG{ODdcHF4;xcecD2Ar{N*tA!@aoJFp<U`rpCmC*>jX1TQJYUDwu52u%(zGdD1
zQVdnaNg~%Y9U3cLI$XAQ%CWCt9;qvB*#w6UFS@3zUcvG7yHLUg^&k?`VSl+sTVf=M
z!|^MkCt$eGxm5)+39Ji0N?Q_-XN)ndfqzD!o`{SL)&e}GoH4OFaOCGupjLumwL*nL
zZ0*+wOa<Xj4ddOBZ$kV?NOi@3-mqk7t8wXY`KD7YU>5kj`QYY&--Qh}ihLImUW+r|
zg-O_g&G({&pOLP@M95%1b-Sp;7dy;PS0ll%V?MKJ{|qXqRW+Nu59Vr7bVVKk(SI|P
z#R}dS0KS$N`cv>(uv?Wtcu@3VusFPlLZdQrTm@!q4dqd2>_YY-NXC<Z7ljxY5n&fu
zva<_j?GU(5<yHiC#)dp*ZATg<E;9M@Fl%#>smQK7y-1UaO5OC>kdA3CiNiTf#1%FK
zdf;A{UoUW5MkZnigrH*iA!7S6EPs?-jJ0@4<Yz9BD_SY36?0kHi1uO<U#tPvzUOwu
zAkH-`g0g+uM%Rrz%)8t&tUTBtp=*+i4;Vd*Z_=??0mFHnS=+MHWY}4wG}rGHKOLK`
z=*O&*uw<Zm;RLk@62YVlx|~D|$4EdnGe%*X2~j&JO}v{!E1pcuwuwfF&3~NsU8g3@
zE-Ci$Bimm%-}M+_b|yNuQ5tbG>Q%XDji&935pMsYb5EoZUv2zvPd_9rxz}GKfb@no
zd6!>WW#HdhSsz=x+x<sj^U-}sWsenjq56HOWfdZKgdIk{CAS&bl2E_1D+^{5;E~^x
zmA|hBv%1foxJancjCw*i1e1OY907@wn+z)fTa(8OQUQ&V8Vx%EbCXsLA0{7fzd*L7
z%>+PN!OO!)^|x5kw+O8sq(SPx-XTd#0gzS?(jfJBlba1L2U|4o?Go&*lh6$&e_!2w
zyuB^+Z2NluTJafqnskW;(<0>tpnVW~Att@tlkHPgW<__E<z%rzE-LUsJ$*Q4c^Yr~
zR9ag(DUt(VaW9b3G0*vI?&qsPY<X#DN!|yOdT6580ixt9l4z4>^2d|7Azczy|5FWq
zxGUITB_iQ=@FQ~1?%mUSVdgNZe+gi)f?-yt6mB9sB_SsJ1dvI338&sK6`eq^rJT>j
zhh+Q2Mo!<Fq^c0Mx2y0(=2H%eO#f@(mHT~Yl<ao=L|*?)BD@U@Pt*TQDs1~h-o!0+
ze>T7UQ|<G=q((YY{jpU-bL6jdjG40o3}`4y*q?-Nofx8$+*P?SY(?vIe-az#)C6gP
z76)8W$T$Nr!t-ZtAFM=SeKTr=<j;Shi;q_5z3?LpNO8pTp5>`a6f{%7%$DG+4IOuc
z>$#ny0W*<?u*fL-uB?b3Uefqw=+&Vy!J)<(23WtjE!(UheNc>FvTh&K0Cy0tNI)qp
zs9L1K-?GO7a-l`lJL=bTf4W)_b}QbsX~ioiuJ}Y8i|N+3NG5Nvfcdfw59ClbuYBOq
zV%6)ii$L`;p>2je(mcq02pFib^di`?I3z85W4K?W$w~3p?a@WFDq5uQQZ_ji<6nX(
z(zr>ACoH%@g%g(kPaT_mM?HHucH(U*DBe`o4b*zWv|y0DeDwCtf2r&OP?H3;ZKDDw
zq~Hdcm_>o(OWb_W!j>9r8`DaCWS5sEO}M5Pi20*k#X>z<#kP6b68JEDeF8M?Pll(x
zX0glubQZYD**Y9PX6;j$3l19ma04AM7(T@P)&LJPyB~wQE-jCNp-0!@iydhDRVakm
ze=P!GKH$sZo)m4_e|#uuw-x3JGetox0-a)IG9m7<K9w*>3F7mO4xK-Q`H7~&6FJZ~
zRv-}K59`2(j#gG;5#)(b4*i?WqjeSb;Xl+g#-YXu&ld34STJ_!StO8>U2@DlmS<+d
zi%>1W%|eyTlHw!V&OvXmvm)81g$4;SlXn7&V@alORHBZ~e*^1s66s&=?B6lvufXCh
z=~QH*AeK`fl8)9{S5hx{&bAOlqnsUpF970n4<c^0dEe83)#lcCBJv0qdL|9u+WH9R
zJQW-u5*pHy2a<2vF8!<c6`gI8DvKMki+lU={K+;FId1r(#pWg{l1FkNMR%eEWMpT`
z(sHY-5_h<wf6YKor1s)``;$SKrB6j!+DMQ#v(_fe?wJ&@^j?XqKUSFrHN)MhqTx=j
zN}OV4QX|i|ny3(|x;WcTY>_{zhZvjL2IO7?+o5I>=R4OXvMlXnYI2kC$Q?HGX5K4d
zUYB*5xdi>qsT%I9C*s*;opdJ0?2a^hmgMaZHQ#I}e^$_?dosa@rb{oqVArf9zh2x^
zl#Mt(eIu*-f!cm}yQUq((3{z|{Xu*4g(EPTNRwgjF>QwH6DhXcQ*I`YG~E}Kz>fXh
zd{hpp@*Uf*JjCo{uVq17nNim?kJQJLH<yDb9j|P@l5JLzUcaIcmozdwXXM48f0jq+
zS5!A&e`K=KnH6o<9JkK$&+-VJpJno_9^}hL+8YT5rSnWe>y1d?^n3EBY8}ePDh5IR
z=^8hiNfxPyeJq^bN^%pO{-&$$47p~yr}bAQ+ows39^Vlf?vg{rK9C~PdSTjRNy8s_
zIA|Gi38lCFjXACHvD;>~<19)iOWmkrvU`OcfBC)REg>g-l<MGX%PIiw>D_3!J7Ycx
z8xz&zps~3h;)CAs<0E%hLyWUVi8m}Ml1UMlm!WPPKDeZcl<p|XvTC~34XO5u#@Fcn
z$h!j>J=x>f0q$fp5rB^*Mq+&#w6v3Lo|%~!gbrTiVXgFs6z)5b&E@rHF3jRplci)V
zf1eUk(NXX-p+)CIk9N@1OVFr!py4uNU%uF=R9dZNemoTMA*qb^GDo#HLMmcXU)f$~
zR+l1;X*|eLpK6|FqFOGclqN+=@LK+*eH>pzKhZ{EGec<OFggI7jl{$sr}tRvkY7xF
z=pv{89?D}!x$(>=e4_z#mr($oR_7=Lf4@;OQ7>~TQ-<g~>k3_s%h+wc5y{bztbEhi
zol0f19>`U$Y2Tor<?U!tBC+L-%^-KukCtaf?#Ig`6Y~+X3O{@={%qi595cUBA*Yx5
ziH&%(?MB|%s2eAvkh`GuFQ&1$c}y~`MDwUb@ob}Bcs@?iy;t9vqy=(I7By1Sf0t0Q
zVkpFkt<BnW&)W68CC}TMRoT!I4Q)=#*>JCoSqe+A=$qscdcU!_f6)--`G3>(zR)J|
zuu$Yto>`H<k?e6#ZS~Ka4;i%@W8G$tMO-ymLGkGH#?+!GF2F^iK}5FTk35|Y9aQ=W
z+kF7<u~=@UYS^}Ll9n$8->;jqf9}!U*NRtz`dvTjhSV<=MO=|Wl8LN-5;1W(A#X{%
z%gYxVin=e<(yNH){zUt;u8ioP&9}07;%U4uAKaHb(|Es6nu10ecS$A~R4$TolK9tS
zvH3)kk771`WM3=We9FnO+mq(v^Z)$Uf4BVp=4T;xLVo9i&zOxuprauve-m+;PvT<3
zFJ!)6tLmCD{kGEH%PNFEVTv*g&CP+|GIH;cG<k)kB`LJte?mP@C67E#jrhpsKg(l7
zFR6mf_4UokBPm>sWjv>ULvyR8qO9E#icpa9*q)eHsN{fzRV&-s#lA(R1rL~n2qjvA
z3rTwd0RtKY1kP~ng5jU=e*hxzyh1tu>C&r~Qh};fcA;He83FjH;UcFKz~sKXG0cv^
zLaG}Yr`kc;gyQB}P3S<7W07GMWk_?ul`Sd0lHd4JG_tHVVMW1w*G}fACZW@f=2O#@
z&1Km%gfHi!>cXwUwC=eeVrTogsE(aDj00J3MoG2uW4LAKpq6wXf2cC=Xd%b0`AbEL
zi_ge{O@##3G}R6&mNQTWAB*U#BFh#W=lje?w!38OgWJoR2=q7YOnyDG4_cJPMbW;H
z22}xOQKP4uv)WmGEpr$LmeGjFZrJtT(V-Ig>GmG!WpgpPYBx)FKp$C~&;CJBbv=_Z
z7J8P;PGiU4vtV5)fBT4w*x{cYz%^kP7sTxM?UZt|>*6XIc6WoA%%tS_V4Cc}Wpf8k
z4hGm~S0fR1ZHJ&lQj*f@$iWsLi>Pslo(6@B)2j4<WMU>*mJWt)cbyM36*EH9bTBj>
z4HIdYmhaeS?+*BoF)A|BGv>4tRv)hbky4pu!<Scf<->=ne=!kmCN&<i_Q_UUKBc-W
z89BM`1Hqk3$&s9%<i@;BwofF*VveL_7pf;MVD=AI(702{>601|{@v9*7I+YmT~~kY
zhGLInF8il3SX86*@SJCM%grU>bDDUY6FQ|$)l42t#w;$_nD2q;c+xPzgJAX57%-?V
zBp~&a1;Bzmf1}It*8}}8PMR*;vgJTD1*aNG%9+Pa^HI#I(7Yycg2*J%jlAFUtsU{6
z<kjw&qZwG_1ry6nH>2sg8kovda9G^YQ{CXh!r3Y99-ougo6VvoITAD81C#9bB;F>+
zJZ*-D(LNMSdStI{xZ|cbf2I#+G#8a)mRY)+LYo}gf8zlSKNb_v<Fp0XoK|P!5m%>z
z5_*!l0N&9-Kumt8!LfK`XO3izM@Qq-B5Ap@xH($3Pk`B3<fW&YZ@L-1WKS6v#8R0G
z3J%j$Be=FBP~BdFHHgSyFR!3LalzP$@73WkHR&<Dc4L_dJA?1oCOMyK#Ji?Euvu@)
zAMFF?e>)gmiseE!2<QIbnJjICNxPK|tljyRXt&~N5S*G2=mTG6S&x!?QMY+ko!-;=
zqu~&oB~XA0Io3Ac1o2Aw9;y}CSrrW&tyz|tu0hhx&zBneFP2|K7RNs{7*+;xIaZ$E
zaB`u4#B<BH`tSJc`~gbGu0b*yC;G5?_hTp<f2v{5h9-Kk2_|FW0;a;hXb^5IlNz;z
zIltpg&d+ovH5PIRBhF>l;bx@<W>&9c)Y@*<I>fB?N_MU7hOI*^Td!l<+G^VxY1~J0
zI3(hKAP-GK4N~Y&AV&Q50S`lkG$eWj1$*ZM_HG0QuD#)f5D(05bhON7lDN!(8y#=N
ze?d$O3_x=Zd9gMfNlt-5*447V^TvKR$A%cY_AhvCD0z)6dhK8K+EDl!TzVOK_+I|E
zY_|;bLGY5i+{-yj2apkea0KKZwem)PgMkoz-AqmgT<Xxq!tf4kCN~0I=%psYo|qFz
z1fn=O#22`i9^v<?a2nVsZjm0*00nY=f6(UT*Mp!I*WeyL<o0m$vG5EZ>hPKb-s{K0
za}GfN*Cj$Im>6IQ__hg?ia8xj>!^eLqc%E7jrWi`$VF<SkJNZ4sYAS^Hn~ZS_LF)A
zN2wj2Qv12ex%`76qGYt7c@74`x`8gI1A<K73%~cnd_RVKk3jV5nC8+We0axnf35`;
ztZfIuX$(5Y;d+>QEIh-98eNlkA3))~fdR+|>Y!Svjnz=&>!A**h}u*WHMT0^v;GGP
zUXvR&fio0j{BJ-0`-3ko7aoK2Bpd^*^(V3ue30FB74QAg-%r7z_SRekZ**OF-ulk)
z4E%gxroq4k;SD1(AD5UaFwp@!f5%AbeIS|lMlL{Ksl!cG+iX>P8mkVrR&6m??Psq#
z)L^yQVzr0K-u6G@GbE3Z=qR+X1RVCcKlb?)`y7a^RhC!bC-U=F1FwdW(0L%0b0ZaC
zt<<5_Qk(0g_NtgVv}S5^)zm(9BcGOwCDe0qT<|B<6M29pZ8dg$!<XT0e^dS@-Vpqm
zB<Tu7#|r0s$M=0{-;ch;5yp6Y$o9r~3&V4A9iO3Fi9^#Qm<=~V7^Wn|O)c8YOy$iG
zwxn2-?P%AF<z@)mF3?l0`pk?wFOMCCyKuL)s5A9Do(^o%5$MGZl@8)Q{%q0{=+q9C
z4)PxEU9apbwcA;0A8$FZf6uri`RR@L>+v+yp+UZHKEDBB#7iSDkW_S0&7?8~3l3k4
zwn)z@EG$18^=KbH4q!9+<nhLG6U`30Z0}ZL;09W^S6cntX5=vlXYuxvsf*yeBiJ!)
zKH=vMp`KuIf)M+opEOwI$6P=U8<eg(ESA*9U{ZTTlX{f~Tf16pe~oXl)jpF?7T7N!
znX!--E!ZU-jWR7VFGwyq!R2X-;Mh_Q1q7ANt5}l~;R?fZqFb6xe8K?M)VQ+R->kLC
zt~K1S^$RKF(h>g&#GZ<2{=}~_knxW(czxI?I|b*7C<bQ%49-jp&ae#LdE>vEqrU@S
z<E^b38farr^G+>`e=(}3J*TF`SkcRXW7{Fkh7MUqSsi*YR`DYXt_Vr5%7%4~0N|;J
zz|#`kSk?L4>*SR*{{dT6nF~ogJ;{mq%ik}P?VdofWOjifFP!@M2j1Cg=talAYcx5f
z0w<|`GJcfqB=Gg-4pW&jyMU4xZvADKG_Qu&+gdQBRfA+<f4%P{cApgcpJ2-Zqx|xv
zJ$dDZ$Vhsf6(niEAiacocTrutfz-+$X)c?xZF|#mCkKN&wMatFJW!G!!OL|P$r+iC
z?zOn7nmdy3f-7mRWJYVA#mR8;<N?lVCMcN9lAY+=@^B~%dFDOsB1lK#r{{Rff*829
zI+X}k(^Lehf2>U&@<G&()I^CSHA#h&%F8XC90pMut58EXJq=E$A^Z?7cCZ~gSiBXd
zap^~eRb|`pQ|7$5BJ_)bAOxJH%xSnez5Mho>-LxXWAQ{%@o~{>nhq_AONabGnEB>1
zGdX_w<r>XrBq;|ddJ+OVymH*CChYK4C63V99r*)*e~*MzSN!J<ONR1v>5vB*jT7OF
zgb?N>|3w|L%EpOUMyM{zCa*JD?u$lR*w-$^PjSRw>&0cJW%!iFC>hSHKS5K<i%eeR
z%&JrfGu?C2tjMlAy-1Uairt`C%_ZfBxAmlD?CbLD1#a*;5mP5QsOx1&)rTg6DGGNY
zJMNTaf8sI&=1MX%(orqqgNKgOm#MJhSomoZyaXO(ObX=3UkrDa2@Rjz%vyGI1OJ{V
z&85!nWG9HS^&%8WsM6_IKi#<zg_#_I>R$kZq6uk34kuCV%rJJQ;5}k}#wjmojgp`|
zcy|~=WUN50IE>PuRPG4-lJl0_W@Jl3>&LE4e}0tIdLK&nK6~Oap-hc>)MVdq3(Ak5
zFcI%}!>V2UTMJ)JTGpf;WHG-9WvPDlpZ$%CGHz+>&(vec)`KxXtNOb6Vn%mKhU2|3
zyAZ9ill1~9wwTtB*I#Y`pE*OQ($^2>ca22(A++qP{09~$lk_<$I>=|DZE}Kw;f5WZ
zfA!PiDQz29%ikzB$qXI7mS{)hbEc2CUm)|WnSh+Eh>Hb`TP(({j&TdZ_<D!LkOFcY
z_6~w^mmRPr=|yBLSC@7q^)d7QTm3OVSD2lf)OKYB`@$kz)A$SO&a5ZS1J$qYKHlDz
zdA5DMXGgIw5VVNd&T*b(TVTSETc9_Gf0Oa9vV6HQFDmF)5R7*hNFDQ>4>o?jy0o0~
z=OeG?#m1XJMmMBO!Z=MeOju2C19{!Ed+g}1FmgsU6Nbh-Zkj?4<LbSHKJ;Al<}VfP
zMA%s2GaX!QpIBV~)+AMhs0zCZbuy6czMGmj7`-Jqdokjf#tim{fA2FMou>;XlTaZP
z1&5b3enFFIAzlHmlgJ?}8)LV}Rfbi|K+KwDnd3{@<W!D7Nh;EqPGp`?lMW(OHOH)d
z3ZpV;kdxH@8b#!;OUq+mt>`-B3{Xo!(YYr@TQ(m`IyejCG*hIG9-1Vh*nOix=R=d1
zB7+KfNIF{cT}l1zIg=zKB!6ep@SOpV!17F`v_(5Db%2fKdGhRJ`;`1{1exR!D*sPW
zZn~_?$^SJeYai)vvOV%}M?<n@fzKt-k5sB-$HV4^C!Y)lmzWQi*A6z}r_29kV><MN
zsToS#D^aQHq-px|<dH2FCv+T1BX~^Xb~hI3j#Ph^<n0e~xLJx78-EU?PPmsUBB~<d
zDlYooU~@U{4Ro|}w+jSg8Z@jgyB#|Pb_66x=m9`VLo_ITY?71~VS5Nr%XCCs$4~TE
zMhaGK+vDMwOA{)v{^7hI4i~>JUp8OatdDPp5w-O%Y?I2r>Nqd7BOzPy#&{Xz2~=0q
z>x!5z%76xxTa`Af{C~9r1Da2cZUUg`y(lXpri5+Uq9dIiiVlHEMdyoMxlyZ)o7<$-
z7_iL^0oYvpR_xoPaG!R&Y|AZsl@%nqKe9e)^DnobuI{yZ@9QZ9xYr@}1orhjz|?_$
zB;<dUt0Sq|rG6DYHH)@d*!f2CEuEmzxW&%l8^YJ-V@bWAR)3aoe|CIelILS3JV`|_
zasRVCW{@#fLa`{}sI0p18+j3lyH-{D@rCh`J^D6jen{GE%X*KoKs!LO<VYq*T#ts#
zZf@jxs~T*le}4gs)eH9kDy2P<pN23LWN-)V>!xqSZ%LRvPm$mIp*1=fPiT#wQh;8T
zOC@O#(Niix=zpx4_VzhzD+22DZ1etuXty%q?dI&mMzro<it;5VDIG#cIpIHIX+o`<
zx>@O2B_o;?sU%5DZ^Pp2{u{PL&G-6E;8^RcjHu$y1T$s@$%#Hxn=lY!k_ZFb+;Nf3
zM==__5*b;~$2HptEEju@wd=J@{uA3XMM|$)mZC%Olz-6v*^h(D5eaRP(k5BmH@uhM
zu>N1<*w7jKhVM+>RHWg@-m{bIH3K#MlZD2_Ne!_2se>TXM7@;6zEjuji6LsnB^`%8
zXXM3QZhO3y1=jCGN@w2)d5OPhSvFr}mJ_*0XZWnqV&Ixd^h|hq)TfU^*NwP4Y{O~i
z&@HP`bAL7SQ_jsWZN#LM{%)qfg_dYsj@Cq<Db<&P;GQAmA}Ke7N1C{~7KtZfr}VJp
z%1Bn*ktGx}p-2-FJ^GNK)WQus&OgR68@E42cVu7w2CTVq<lC{cIf+PTqK5fM^%|)g
z{ex+?<y*{54i&;0m1O>vw#_XaoeTq@;Y46?SARYiQFYeeFm~r!we=mkYP<SCMx{;m
zoD7ZF6+}kDZziF^psS)IUFO+Kg_ik*T6r-l-JLjjQ{FmZVOB0Yp&Aa8i+VY!`&NtD
zERx@<`zP~bl9x*|6OrZ{ZWzsX+ut6{M~NP)VKf#$x2BDkh6vBjb62QLML#Bsp4&*O
zh<~ICxows8Xd?44T-*HfMrPi~ot6_uEucEa0!CKtq$W6v`+m@<wbX?seEZt%#WZAS
z$#pV%KnEBeO(X<mniy$`<AKUY?V+BWq=?S*f==^zH<gLm*RL0J^2PL%=OlOc(ly_T
zrVmaS1tup!KhdAZgfFMFxlG+pj(FHR(SPNAkd_9Z=VE+AOLU@+$vae8zJz5Mbn&!E
ziYq5LcD!U+w_&rRu6!i4kj-B=zv_cR(nB{S-e-^doILwsHzXS5{~sUQ3nV)1Z<A*t
zM|u)3C8SE(-W$)dDtYBT^fF@Cf8BgIW_dcGvlEKN5VA(hDeS<&Sp^3_ku61$9e<;-
z)R-l=KTProJlQ^a;cqk~Z7yhI_$pf2-%bFt5xw7nPF@yfD3(TULNdhxWK&nF<lorT
zA=~oEExICg=u)TIT#?MEQktfnj_hYhElFsHKGrPK)EgT7{w)pNG%s(j=GUiSb1$hP
z>QzKiMRej`^M)^Qw|i=%$LZpt?0>3Qr#b1f|FZcfCGt-vZMFYBexy^m`5Jp{A*e5Y
z1DECBFqnwKM7>O;WTN&l>SZDUV<@EJxkPG_0SK;R2i3FSofEDII#9C6Ne=98G~6`~
zneNAVhH=v{U0G%^nE(yfW2j*Q#KX=uAAQ|RrA^1wm)D<V{)*i;M{P~?4S$c#+bW_7
z68W;MUN_0+f%Xe*X<4m-l+9*a7Q5`R*YE5R#Gdv(YK~XH6sVj>&=J+{m4&cm1d*06
z1$)2&EQHFgx;S}Bm5Sks=*3!~w!BqKLNxqNE2HJ3cEOd|M;`@`tuz}STZc95Q?cBk
zG=1bPiFbMVV%Dkd3ca6-lz$GeBl>6at!$pS7*{;3TrFD1jN>w8UmUXe&+^#NOPY6c
zeSLE>U0yZo)(@7luC_W>R2^Dp3$N>}v|4%;3z1o;<6c*FFPWJ!tcz>zl{52b+zc@r
zuR90lwqc<~H=>TjDDg;VS=7te?#u1;g$2`SaXWn-;mx2S8fh3ky=xu&Aoq!QI=Mv5
zK)HfwIuDg2|LHvVq~E3i3gRla7RCQAlZKf3LW}#LR{pG&HOeKl^CY$Or&jyeH02wi
w6PS)o+COBpA8b!v*nlQ$W9Bc&2_O@VvNmE9K_S4$?^(-S>Tf6i9|JAktnUQC1^@s6

diff --git a/main/overview.html b/main/overview.html
index aca2fa4ca1..3a1fbc9d34 100644
--- a/main/overview.html
+++ b/main/overview.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/recipes/lora_finetune_single_device.html b/main/recipes/lora_finetune_single_device.html
index 084acebc6b..088afd9abc 100644
--- a/main/recipes/lora_finetune_single_device.html
+++ b/main/recipes/lora_finetune_single_device.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/recipes/qat_distributed.html b/main/recipes/qat_distributed.html
index 46cc0e22ed..d87c3ae724 100644
--- a/main/recipes/qat_distributed.html
+++ b/main/recipes/qat_distributed.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/recipes/recipes_overview.html b/main/recipes/recipes_overview.html
index 4fce2e7229..9556d75b20 100644
--- a/main/recipes/recipes_overview.html
+++ b/main/recipes/recipes_overview.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="../tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/search.html b/main/search.html
index acf0d3d103..4fdd4a355d 100644
--- a/main/search.html
+++ b/main/search.html
@@ -330,8 +330,8 @@
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/searchindex.js b/main/searchindex.js
index 4622a0f112..5bdeeb5453 100644
--- a/main/searchindex.js
+++ b/main/searchindex.js
@@ -1 +1 @@
-Search.setIndex({"docnames": ["api_ref_config", "api_ref_data", "api_ref_datasets", "api_ref_generation", "api_ref_models", "api_ref_modules", "api_ref_rlhf", "api_ref_training", "api_ref_utilities", "basics/chat_datasets", "basics/custom_components", "basics/datasets_overview", "basics/instruct_datasets", "basics/message_transforms", "basics/messages", "basics/model_transforms", "basics/multimodal_datasets", "basics/packing", "basics/preference_datasets", "basics/prompt_templates", "basics/text_completion_datasets", "basics/tokenizers", "deep_dives/checkpointer", "deep_dives/comet_logging", "deep_dives/configs", "deep_dives/recipe_deepdive", "deep_dives/wandb_logging", "generated/torchtune.config.instantiate", "generated/torchtune.config.log_config", "generated/torchtune.config.parse", "generated/torchtune.config.validate", "generated/torchtune.data.AlpacaToMessages", "generated/torchtune.data.ChatMLTemplate", "generated/torchtune.data.ChosenRejectedToMessages", "generated/torchtune.data.GrammarErrorCorrectionTemplate", "generated/torchtune.data.InputOutputToMessages", "generated/torchtune.data.Message", "generated/torchtune.data.OpenAIToMessages", "generated/torchtune.data.PromptTemplate", "generated/torchtune.data.PromptTemplateInterface", "generated/torchtune.data.QuestionAnswerTemplate", "generated/torchtune.data.Role", "generated/torchtune.data.ShareGPTToMessages", "generated/torchtune.data.SummarizeTemplate", "generated/torchtune.data.format_content_with_images", "generated/torchtune.data.left_pad_sequence", "generated/torchtune.data.load_image", "generated/torchtune.data.padded_collate", "generated/torchtune.data.padded_collate_dpo", "generated/torchtune.data.padded_collate_sft", "generated/torchtune.data.padded_collate_tiled_images_and_mask", "generated/torchtune.data.truncate", "generated/torchtune.data.validate_messages", "generated/torchtune.datasets.ConcatDataset", "generated/torchtune.datasets.PackedDataset", "generated/torchtune.datasets.PreferenceDataset", "generated/torchtune.datasets.SFTDataset", "generated/torchtune.datasets.TextCompletionDataset", "generated/torchtune.datasets.alpaca_cleaned_dataset", "generated/torchtune.datasets.alpaca_dataset", "generated/torchtune.datasets.chat_dataset", "generated/torchtune.datasets.cnn_dailymail_articles_dataset", "generated/torchtune.datasets.grammar_dataset", "generated/torchtune.datasets.hh_rlhf_helpful_dataset", "generated/torchtune.datasets.instruct_dataset", "generated/torchtune.datasets.multimodal.llava_instruct_dataset", "generated/torchtune.datasets.multimodal.the_cauldron_dataset", "generated/torchtune.datasets.preference_dataset", "generated/torchtune.datasets.samsum_dataset", "generated/torchtune.datasets.slimorca_dataset", "generated/torchtune.datasets.stack_exchange_paired_dataset", "generated/torchtune.datasets.text_completion_dataset", "generated/torchtune.datasets.wikitext_dataset", "generated/torchtune.generation.generate", "generated/torchtune.generation.generate_next_token", "generated/torchtune.generation.get_causal_mask_from_padding_mask", "generated/torchtune.generation.get_position_ids_from_padding_mask", "generated/torchtune.generation.sample", "generated/torchtune.models.clip.TilePositionalEmbedding", "generated/torchtune.models.clip.TiledTokenPositionalEmbedding", "generated/torchtune.models.clip.TokenPositionalEmbedding", "generated/torchtune.models.clip.clip_vision_encoder", "generated/torchtune.models.code_llama2.code_llama2_13b", "generated/torchtune.models.code_llama2.code_llama2_70b", "generated/torchtune.models.code_llama2.code_llama2_7b", "generated/torchtune.models.code_llama2.lora_code_llama2_13b", "generated/torchtune.models.code_llama2.lora_code_llama2_70b", "generated/torchtune.models.code_llama2.lora_code_llama2_7b", "generated/torchtune.models.code_llama2.qlora_code_llama2_13b", "generated/torchtune.models.code_llama2.qlora_code_llama2_70b", "generated/torchtune.models.code_llama2.qlora_code_llama2_7b", "generated/torchtune.models.gemma.gemma", "generated/torchtune.models.gemma.gemma_2b", "generated/torchtune.models.gemma.gemma_7b", "generated/torchtune.models.gemma.gemma_tokenizer", "generated/torchtune.models.gemma.lora_gemma", "generated/torchtune.models.gemma.lora_gemma_2b", "generated/torchtune.models.gemma.lora_gemma_7b", "generated/torchtune.models.gemma.qlora_gemma_2b", "generated/torchtune.models.gemma.qlora_gemma_7b", "generated/torchtune.models.llama2.Llama2ChatTemplate", "generated/torchtune.models.llama2.llama2", "generated/torchtune.models.llama2.llama2_13b", "generated/torchtune.models.llama2.llama2_70b", "generated/torchtune.models.llama2.llama2_7b", "generated/torchtune.models.llama2.llama2_reward_7b", "generated/torchtune.models.llama2.llama2_tokenizer", "generated/torchtune.models.llama2.lora_llama2", "generated/torchtune.models.llama2.lora_llama2_13b", "generated/torchtune.models.llama2.lora_llama2_70b", "generated/torchtune.models.llama2.lora_llama2_7b", "generated/torchtune.models.llama2.lora_llama2_reward_7b", "generated/torchtune.models.llama2.qlora_llama2_13b", "generated/torchtune.models.llama2.qlora_llama2_70b", "generated/torchtune.models.llama2.qlora_llama2_7b", "generated/torchtune.models.llama2.qlora_llama2_reward_7b", "generated/torchtune.models.llama3.llama3", "generated/torchtune.models.llama3.llama3_70b", "generated/torchtune.models.llama3.llama3_8b", "generated/torchtune.models.llama3.llama3_tokenizer", "generated/torchtune.models.llama3.lora_llama3", "generated/torchtune.models.llama3.lora_llama3_70b", "generated/torchtune.models.llama3.lora_llama3_8b", "generated/torchtune.models.llama3.qlora_llama3_70b", "generated/torchtune.models.llama3.qlora_llama3_8b", "generated/torchtune.models.llama3_1.llama3_1", "generated/torchtune.models.llama3_1.llama3_1_405b", "generated/torchtune.models.llama3_1.llama3_1_70b", "generated/torchtune.models.llama3_1.llama3_1_8b", "generated/torchtune.models.llama3_1.lora_llama3_1", "generated/torchtune.models.llama3_1.lora_llama3_1_405b", "generated/torchtune.models.llama3_1.lora_llama3_1_70b", "generated/torchtune.models.llama3_1.lora_llama3_1_8b", "generated/torchtune.models.llama3_1.qlora_llama3_1_405b", "generated/torchtune.models.llama3_1.qlora_llama3_1_70b", "generated/torchtune.models.llama3_1.qlora_llama3_1_8b", "generated/torchtune.models.llama3_2.llama3_2_1b", "generated/torchtune.models.llama3_2.llama3_2_3b", "generated/torchtune.models.llama3_2.lora_llama3_2_1b", "generated/torchtune.models.llama3_2.lora_llama3_2_3b", "generated/torchtune.models.llama3_2.qlora_llama3_2_1b", "generated/torchtune.models.llama3_2.qlora_llama3_2_3b", "generated/torchtune.models.llama3_2_vision.Llama3VisionEncoder", "generated/torchtune.models.llama3_2_vision.Llama3VisionProjectionHead", "generated/torchtune.models.llama3_2_vision.Llama3VisionTransform", "generated/torchtune.models.llama3_2_vision.llama3_2_vision_11b", "generated/torchtune.models.llama3_2_vision.llama3_2_vision_decoder", "generated/torchtune.models.llama3_2_vision.llama3_2_vision_encoder", "generated/torchtune.models.llama3_2_vision.llama3_2_vision_transform", "generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_11b", "generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_decoder", "generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_encoder", "generated/torchtune.models.llama3_2_vision.qlora_llama3_2_vision_11b", "generated/torchtune.models.mistral.MistralChatTemplate", "generated/torchtune.models.mistral.lora_mistral", "generated/torchtune.models.mistral.lora_mistral_7b", "generated/torchtune.models.mistral.lora_mistral_classifier", "generated/torchtune.models.mistral.lora_mistral_reward_7b", "generated/torchtune.models.mistral.mistral", "generated/torchtune.models.mistral.mistral_7b", "generated/torchtune.models.mistral.mistral_classifier", "generated/torchtune.models.mistral.mistral_reward_7b", "generated/torchtune.models.mistral.mistral_tokenizer", "generated/torchtune.models.mistral.qlora_mistral_7b", "generated/torchtune.models.mistral.qlora_mistral_reward_7b", "generated/torchtune.models.phi3.lora_phi3", "generated/torchtune.models.phi3.lora_phi3_mini", "generated/torchtune.models.phi3.phi3", "generated/torchtune.models.phi3.phi3_mini", "generated/torchtune.models.phi3.phi3_mini_tokenizer", "generated/torchtune.models.phi3.qlora_phi3_mini", "generated/torchtune.models.qwen2.lora_qwen2", "generated/torchtune.models.qwen2.lora_qwen2_0_5b", "generated/torchtune.models.qwen2.lora_qwen2_1_5b", "generated/torchtune.models.qwen2.lora_qwen2_7b", "generated/torchtune.models.qwen2.qwen2", "generated/torchtune.models.qwen2.qwen2_0_5b", "generated/torchtune.models.qwen2.qwen2_1_5b", "generated/torchtune.models.qwen2.qwen2_7b", "generated/torchtune.models.qwen2.qwen2_tokenizer", "generated/torchtune.modules.FeedForward", "generated/torchtune.modules.Fp32LayerNorm", "generated/torchtune.modules.KVCache", "generated/torchtune.modules.MultiHeadAttention", "generated/torchtune.modules.RMSNorm", "generated/torchtune.modules.RotaryPositionalEmbeddings", "generated/torchtune.modules.TanhGate", "generated/torchtune.modules.TiedLinear", "generated/torchtune.modules.TransformerCrossAttentionLayer", "generated/torchtune.modules.TransformerDecoder", "generated/torchtune.modules.TransformerSelfAttentionLayer", "generated/torchtune.modules.VisionTransformer", "generated/torchtune.modules.common_utils.delete_kv_caches", "generated/torchtune.modules.common_utils.disable_kv_cache", "generated/torchtune.modules.common_utils.local_kv_cache", "generated/torchtune.modules.common_utils.reparametrize_as_dtype_state_dict_post_hook", "generated/torchtune.modules.loss.CEWithChunkedOutputLoss", "generated/torchtune.modules.loss.ForwardKLLoss", "generated/torchtune.modules.loss.ForwardKLWithChunkedOutputLoss", "generated/torchtune.modules.model_fusion.DeepFusionModel", "generated/torchtune.modules.model_fusion.FusionEmbedding", "generated/torchtune.modules.model_fusion.FusionLayer", "generated/torchtune.modules.model_fusion.get_fusion_params", "generated/torchtune.modules.model_fusion.register_fusion_module", "generated/torchtune.modules.peft.AdapterModule", "generated/torchtune.modules.peft.LoRALinear", "generated/torchtune.modules.peft.disable_adapter", "generated/torchtune.modules.peft.get_adapter_params", "generated/torchtune.modules.peft.set_trainable_params", "generated/torchtune.modules.peft.validate_missing_and_unexpected_for_lora", "generated/torchtune.modules.peft.validate_state_dict_for_lora", "generated/torchtune.modules.tokenizers.BaseTokenizer", "generated/torchtune.modules.tokenizers.ModelTokenizer", "generated/torchtune.modules.tokenizers.SentencePieceBaseTokenizer", "generated/torchtune.modules.tokenizers.TikTokenBaseTokenizer", "generated/torchtune.modules.tokenizers.parse_hf_tokenizer_json", "generated/torchtune.modules.tokenizers.tokenize_messages_no_special_tokens", "generated/torchtune.modules.transforms.Transform", "generated/torchtune.modules.transforms.VisionCrossAttentionMask", "generated/torchtune.rlhf.estimate_advantages", "generated/torchtune.rlhf.get_rewards_ppo", "generated/torchtune.rlhf.loss.DPOLoss", "generated/torchtune.rlhf.loss.PPOLoss", "generated/torchtune.rlhf.loss.RSOLoss", "generated/torchtune.rlhf.loss.SimPOLoss", "generated/torchtune.rlhf.truncate_sequence_at_first_stop_token", "generated/torchtune.training.FSDPPolicyType", "generated/torchtune.training.FormattedCheckpointFiles", "generated/torchtune.training.FullModelHFCheckpointer", "generated/torchtune.training.FullModelMetaCheckpointer", "generated/torchtune.training.FullModelTorchTuneCheckpointer", "generated/torchtune.training.ModelType", "generated/torchtune.training.OptimizerInBackwardWrapper", "generated/torchtune.training.apply_selective_activation_checkpointing", "generated/torchtune.training.create_optim_in_bwd_wrapper", "generated/torchtune.training.get_cosine_schedule_with_warmup", "generated/torchtune.training.get_dtype", "generated/torchtune.training.get_full_finetune_fsdp_wrap_policy", "generated/torchtune.training.get_lr", "generated/torchtune.training.get_memory_stats", "generated/torchtune.training.get_quantizer_mode", "generated/torchtune.training.get_unmasked_sequence_lengths", "generated/torchtune.training.get_world_size_and_rank", "generated/torchtune.training.init_distributed", "generated/torchtune.training.is_distributed", "generated/torchtune.training.log_memory_stats", "generated/torchtune.training.lora_fsdp_wrap_policy", "generated/torchtune.training.metric_logging.CometLogger", "generated/torchtune.training.metric_logging.DiskLogger", "generated/torchtune.training.metric_logging.StdoutLogger", "generated/torchtune.training.metric_logging.TensorBoardLogger", "generated/torchtune.training.metric_logging.WandBLogger", "generated/torchtune.training.register_optim_in_bwd_hooks", "generated/torchtune.training.set_activation_checkpointing", "generated/torchtune.training.set_default_dtype", "generated/torchtune.training.set_seed", "generated/torchtune.training.setup_torch_profiler", "generated/torchtune.training.update_state_dict_for_classifier", "generated/torchtune.training.validate_expected_param_dtype", "generated/torchtune.utils.batch_to_device", "generated/torchtune.utils.get_device", "generated/torchtune.utils.get_logger", "generated/torchtune.utils.torch_version_ge", "generated_examples/index", "generated_examples/sg_execution_times", "index", "install", "overview", "recipes/lora_finetune_single_device", "recipes/qat_distributed", "recipes/recipes_overview", "sg_execution_times", "tune_cli", "tutorials/chat", "tutorials/e2e_flow", "tutorials/first_finetune_tutorial", "tutorials/llama3", "tutorials/llama_kd_tutorial", "tutorials/lora_finetune", "tutorials/memory_optimizations", "tutorials/qat_finetune", "tutorials/qlora_finetune"], "filenames": ["api_ref_config.rst", "api_ref_data.rst", "api_ref_datasets.rst", "api_ref_generation.rst", "api_ref_models.rst", "api_ref_modules.rst", "api_ref_rlhf.rst", "api_ref_training.rst", "api_ref_utilities.rst", "basics/chat_datasets.rst", "basics/custom_components.rst", "basics/datasets_overview.rst", "basics/instruct_datasets.rst", "basics/message_transforms.rst", "basics/messages.rst", "basics/model_transforms.rst", "basics/multimodal_datasets.rst", "basics/packing.rst", "basics/preference_datasets.rst", "basics/prompt_templates.rst", "basics/text_completion_datasets.rst", "basics/tokenizers.rst", "deep_dives/checkpointer.rst", "deep_dives/comet_logging.rst", "deep_dives/configs.rst", "deep_dives/recipe_deepdive.rst", "deep_dives/wandb_logging.rst", "generated/torchtune.config.instantiate.rst", "generated/torchtune.config.log_config.rst", "generated/torchtune.config.parse.rst", "generated/torchtune.config.validate.rst", "generated/torchtune.data.AlpacaToMessages.rst", "generated/torchtune.data.ChatMLTemplate.rst", "generated/torchtune.data.ChosenRejectedToMessages.rst", "generated/torchtune.data.GrammarErrorCorrectionTemplate.rst", "generated/torchtune.data.InputOutputToMessages.rst", "generated/torchtune.data.Message.rst", "generated/torchtune.data.OpenAIToMessages.rst", "generated/torchtune.data.PromptTemplate.rst", "generated/torchtune.data.PromptTemplateInterface.rst", "generated/torchtune.data.QuestionAnswerTemplate.rst", "generated/torchtune.data.Role.rst", "generated/torchtune.data.ShareGPTToMessages.rst", "generated/torchtune.data.SummarizeTemplate.rst", "generated/torchtune.data.format_content_with_images.rst", "generated/torchtune.data.left_pad_sequence.rst", "generated/torchtune.data.load_image.rst", "generated/torchtune.data.padded_collate.rst", "generated/torchtune.data.padded_collate_dpo.rst", "generated/torchtune.data.padded_collate_sft.rst", "generated/torchtune.data.padded_collate_tiled_images_and_mask.rst", "generated/torchtune.data.truncate.rst", "generated/torchtune.data.validate_messages.rst", "generated/torchtune.datasets.ConcatDataset.rst", "generated/torchtune.datasets.PackedDataset.rst", "generated/torchtune.datasets.PreferenceDataset.rst", "generated/torchtune.datasets.SFTDataset.rst", "generated/torchtune.datasets.TextCompletionDataset.rst", "generated/torchtune.datasets.alpaca_cleaned_dataset.rst", "generated/torchtune.datasets.alpaca_dataset.rst", "generated/torchtune.datasets.chat_dataset.rst", "generated/torchtune.datasets.cnn_dailymail_articles_dataset.rst", "generated/torchtune.datasets.grammar_dataset.rst", "generated/torchtune.datasets.hh_rlhf_helpful_dataset.rst", "generated/torchtune.datasets.instruct_dataset.rst", "generated/torchtune.datasets.multimodal.llava_instruct_dataset.rst", "generated/torchtune.datasets.multimodal.the_cauldron_dataset.rst", "generated/torchtune.datasets.preference_dataset.rst", "generated/torchtune.datasets.samsum_dataset.rst", "generated/torchtune.datasets.slimorca_dataset.rst", "generated/torchtune.datasets.stack_exchange_paired_dataset.rst", "generated/torchtune.datasets.text_completion_dataset.rst", "generated/torchtune.datasets.wikitext_dataset.rst", "generated/torchtune.generation.generate.rst", "generated/torchtune.generation.generate_next_token.rst", "generated/torchtune.generation.get_causal_mask_from_padding_mask.rst", "generated/torchtune.generation.get_position_ids_from_padding_mask.rst", "generated/torchtune.generation.sample.rst", "generated/torchtune.models.clip.TilePositionalEmbedding.rst", "generated/torchtune.models.clip.TiledTokenPositionalEmbedding.rst", "generated/torchtune.models.clip.TokenPositionalEmbedding.rst", "generated/torchtune.models.clip.clip_vision_encoder.rst", "generated/torchtune.models.code_llama2.code_llama2_13b.rst", "generated/torchtune.models.code_llama2.code_llama2_70b.rst", "generated/torchtune.models.code_llama2.code_llama2_7b.rst", "generated/torchtune.models.code_llama2.lora_code_llama2_13b.rst", "generated/torchtune.models.code_llama2.lora_code_llama2_70b.rst", "generated/torchtune.models.code_llama2.lora_code_llama2_7b.rst", "generated/torchtune.models.code_llama2.qlora_code_llama2_13b.rst", "generated/torchtune.models.code_llama2.qlora_code_llama2_70b.rst", "generated/torchtune.models.code_llama2.qlora_code_llama2_7b.rst", "generated/torchtune.models.gemma.gemma.rst", "generated/torchtune.models.gemma.gemma_2b.rst", "generated/torchtune.models.gemma.gemma_7b.rst", "generated/torchtune.models.gemma.gemma_tokenizer.rst", "generated/torchtune.models.gemma.lora_gemma.rst", "generated/torchtune.models.gemma.lora_gemma_2b.rst", "generated/torchtune.models.gemma.lora_gemma_7b.rst", "generated/torchtune.models.gemma.qlora_gemma_2b.rst", "generated/torchtune.models.gemma.qlora_gemma_7b.rst", "generated/torchtune.models.llama2.Llama2ChatTemplate.rst", "generated/torchtune.models.llama2.llama2.rst", "generated/torchtune.models.llama2.llama2_13b.rst", "generated/torchtune.models.llama2.llama2_70b.rst", "generated/torchtune.models.llama2.llama2_7b.rst", "generated/torchtune.models.llama2.llama2_reward_7b.rst", "generated/torchtune.models.llama2.llama2_tokenizer.rst", "generated/torchtune.models.llama2.lora_llama2.rst", "generated/torchtune.models.llama2.lora_llama2_13b.rst", "generated/torchtune.models.llama2.lora_llama2_70b.rst", "generated/torchtune.models.llama2.lora_llama2_7b.rst", "generated/torchtune.models.llama2.lora_llama2_reward_7b.rst", "generated/torchtune.models.llama2.qlora_llama2_13b.rst", "generated/torchtune.models.llama2.qlora_llama2_70b.rst", "generated/torchtune.models.llama2.qlora_llama2_7b.rst", "generated/torchtune.models.llama2.qlora_llama2_reward_7b.rst", "generated/torchtune.models.llama3.llama3.rst", "generated/torchtune.models.llama3.llama3_70b.rst", "generated/torchtune.models.llama3.llama3_8b.rst", "generated/torchtune.models.llama3.llama3_tokenizer.rst", "generated/torchtune.models.llama3.lora_llama3.rst", "generated/torchtune.models.llama3.lora_llama3_70b.rst", "generated/torchtune.models.llama3.lora_llama3_8b.rst", "generated/torchtune.models.llama3.qlora_llama3_70b.rst", "generated/torchtune.models.llama3.qlora_llama3_8b.rst", "generated/torchtune.models.llama3_1.llama3_1.rst", "generated/torchtune.models.llama3_1.llama3_1_405b.rst", "generated/torchtune.models.llama3_1.llama3_1_70b.rst", "generated/torchtune.models.llama3_1.llama3_1_8b.rst", "generated/torchtune.models.llama3_1.lora_llama3_1.rst", "generated/torchtune.models.llama3_1.lora_llama3_1_405b.rst", "generated/torchtune.models.llama3_1.lora_llama3_1_70b.rst", "generated/torchtune.models.llama3_1.lora_llama3_1_8b.rst", "generated/torchtune.models.llama3_1.qlora_llama3_1_405b.rst", "generated/torchtune.models.llama3_1.qlora_llama3_1_70b.rst", "generated/torchtune.models.llama3_1.qlora_llama3_1_8b.rst", "generated/torchtune.models.llama3_2.llama3_2_1b.rst", "generated/torchtune.models.llama3_2.llama3_2_3b.rst", "generated/torchtune.models.llama3_2.lora_llama3_2_1b.rst", "generated/torchtune.models.llama3_2.lora_llama3_2_3b.rst", "generated/torchtune.models.llama3_2.qlora_llama3_2_1b.rst", "generated/torchtune.models.llama3_2.qlora_llama3_2_3b.rst", "generated/torchtune.models.llama3_2_vision.Llama3VisionEncoder.rst", "generated/torchtune.models.llama3_2_vision.Llama3VisionProjectionHead.rst", "generated/torchtune.models.llama3_2_vision.Llama3VisionTransform.rst", "generated/torchtune.models.llama3_2_vision.llama3_2_vision_11b.rst", "generated/torchtune.models.llama3_2_vision.llama3_2_vision_decoder.rst", "generated/torchtune.models.llama3_2_vision.llama3_2_vision_encoder.rst", "generated/torchtune.models.llama3_2_vision.llama3_2_vision_transform.rst", "generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_11b.rst", "generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_decoder.rst", "generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_encoder.rst", "generated/torchtune.models.llama3_2_vision.qlora_llama3_2_vision_11b.rst", "generated/torchtune.models.mistral.MistralChatTemplate.rst", "generated/torchtune.models.mistral.lora_mistral.rst", "generated/torchtune.models.mistral.lora_mistral_7b.rst", "generated/torchtune.models.mistral.lora_mistral_classifier.rst", "generated/torchtune.models.mistral.lora_mistral_reward_7b.rst", "generated/torchtune.models.mistral.mistral.rst", "generated/torchtune.models.mistral.mistral_7b.rst", "generated/torchtune.models.mistral.mistral_classifier.rst", "generated/torchtune.models.mistral.mistral_reward_7b.rst", "generated/torchtune.models.mistral.mistral_tokenizer.rst", "generated/torchtune.models.mistral.qlora_mistral_7b.rst", "generated/torchtune.models.mistral.qlora_mistral_reward_7b.rst", "generated/torchtune.models.phi3.lora_phi3.rst", "generated/torchtune.models.phi3.lora_phi3_mini.rst", "generated/torchtune.models.phi3.phi3.rst", "generated/torchtune.models.phi3.phi3_mini.rst", "generated/torchtune.models.phi3.phi3_mini_tokenizer.rst", "generated/torchtune.models.phi3.qlora_phi3_mini.rst", "generated/torchtune.models.qwen2.lora_qwen2.rst", "generated/torchtune.models.qwen2.lora_qwen2_0_5b.rst", "generated/torchtune.models.qwen2.lora_qwen2_1_5b.rst", "generated/torchtune.models.qwen2.lora_qwen2_7b.rst", "generated/torchtune.models.qwen2.qwen2.rst", "generated/torchtune.models.qwen2.qwen2_0_5b.rst", "generated/torchtune.models.qwen2.qwen2_1_5b.rst", "generated/torchtune.models.qwen2.qwen2_7b.rst", "generated/torchtune.models.qwen2.qwen2_tokenizer.rst", "generated/torchtune.modules.FeedForward.rst", "generated/torchtune.modules.Fp32LayerNorm.rst", "generated/torchtune.modules.KVCache.rst", "generated/torchtune.modules.MultiHeadAttention.rst", "generated/torchtune.modules.RMSNorm.rst", "generated/torchtune.modules.RotaryPositionalEmbeddings.rst", "generated/torchtune.modules.TanhGate.rst", "generated/torchtune.modules.TiedLinear.rst", "generated/torchtune.modules.TransformerCrossAttentionLayer.rst", "generated/torchtune.modules.TransformerDecoder.rst", "generated/torchtune.modules.TransformerSelfAttentionLayer.rst", "generated/torchtune.modules.VisionTransformer.rst", "generated/torchtune.modules.common_utils.delete_kv_caches.rst", "generated/torchtune.modules.common_utils.disable_kv_cache.rst", "generated/torchtune.modules.common_utils.local_kv_cache.rst", "generated/torchtune.modules.common_utils.reparametrize_as_dtype_state_dict_post_hook.rst", "generated/torchtune.modules.loss.CEWithChunkedOutputLoss.rst", "generated/torchtune.modules.loss.ForwardKLLoss.rst", "generated/torchtune.modules.loss.ForwardKLWithChunkedOutputLoss.rst", "generated/torchtune.modules.model_fusion.DeepFusionModel.rst", "generated/torchtune.modules.model_fusion.FusionEmbedding.rst", "generated/torchtune.modules.model_fusion.FusionLayer.rst", "generated/torchtune.modules.model_fusion.get_fusion_params.rst", "generated/torchtune.modules.model_fusion.register_fusion_module.rst", "generated/torchtune.modules.peft.AdapterModule.rst", "generated/torchtune.modules.peft.LoRALinear.rst", "generated/torchtune.modules.peft.disable_adapter.rst", "generated/torchtune.modules.peft.get_adapter_params.rst", "generated/torchtune.modules.peft.set_trainable_params.rst", "generated/torchtune.modules.peft.validate_missing_and_unexpected_for_lora.rst", "generated/torchtune.modules.peft.validate_state_dict_for_lora.rst", "generated/torchtune.modules.tokenizers.BaseTokenizer.rst", "generated/torchtune.modules.tokenizers.ModelTokenizer.rst", "generated/torchtune.modules.tokenizers.SentencePieceBaseTokenizer.rst", "generated/torchtune.modules.tokenizers.TikTokenBaseTokenizer.rst", "generated/torchtune.modules.tokenizers.parse_hf_tokenizer_json.rst", "generated/torchtune.modules.tokenizers.tokenize_messages_no_special_tokens.rst", "generated/torchtune.modules.transforms.Transform.rst", "generated/torchtune.modules.transforms.VisionCrossAttentionMask.rst", "generated/torchtune.rlhf.estimate_advantages.rst", "generated/torchtune.rlhf.get_rewards_ppo.rst", "generated/torchtune.rlhf.loss.DPOLoss.rst", "generated/torchtune.rlhf.loss.PPOLoss.rst", "generated/torchtune.rlhf.loss.RSOLoss.rst", "generated/torchtune.rlhf.loss.SimPOLoss.rst", "generated/torchtune.rlhf.truncate_sequence_at_first_stop_token.rst", "generated/torchtune.training.FSDPPolicyType.rst", "generated/torchtune.training.FormattedCheckpointFiles.rst", "generated/torchtune.training.FullModelHFCheckpointer.rst", "generated/torchtune.training.FullModelMetaCheckpointer.rst", "generated/torchtune.training.FullModelTorchTuneCheckpointer.rst", "generated/torchtune.training.ModelType.rst", "generated/torchtune.training.OptimizerInBackwardWrapper.rst", "generated/torchtune.training.apply_selective_activation_checkpointing.rst", "generated/torchtune.training.create_optim_in_bwd_wrapper.rst", "generated/torchtune.training.get_cosine_schedule_with_warmup.rst", "generated/torchtune.training.get_dtype.rst", "generated/torchtune.training.get_full_finetune_fsdp_wrap_policy.rst", "generated/torchtune.training.get_lr.rst", "generated/torchtune.training.get_memory_stats.rst", "generated/torchtune.training.get_quantizer_mode.rst", "generated/torchtune.training.get_unmasked_sequence_lengths.rst", "generated/torchtune.training.get_world_size_and_rank.rst", "generated/torchtune.training.init_distributed.rst", "generated/torchtune.training.is_distributed.rst", "generated/torchtune.training.log_memory_stats.rst", "generated/torchtune.training.lora_fsdp_wrap_policy.rst", "generated/torchtune.training.metric_logging.CometLogger.rst", "generated/torchtune.training.metric_logging.DiskLogger.rst", "generated/torchtune.training.metric_logging.StdoutLogger.rst", "generated/torchtune.training.metric_logging.TensorBoardLogger.rst", "generated/torchtune.training.metric_logging.WandBLogger.rst", "generated/torchtune.training.register_optim_in_bwd_hooks.rst", "generated/torchtune.training.set_activation_checkpointing.rst", "generated/torchtune.training.set_default_dtype.rst", "generated/torchtune.training.set_seed.rst", "generated/torchtune.training.setup_torch_profiler.rst", "generated/torchtune.training.update_state_dict_for_classifier.rst", "generated/torchtune.training.validate_expected_param_dtype.rst", "generated/torchtune.utils.batch_to_device.rst", "generated/torchtune.utils.get_device.rst", "generated/torchtune.utils.get_logger.rst", "generated/torchtune.utils.torch_version_ge.rst", "generated_examples/index.rst", "generated_examples/sg_execution_times.rst", "index.rst", "install.rst", "overview.rst", "recipes/lora_finetune_single_device.rst", "recipes/qat_distributed.rst", "recipes/recipes_overview.rst", "sg_execution_times.rst", "tune_cli.rst", "tutorials/chat.rst", "tutorials/e2e_flow.rst", "tutorials/first_finetune_tutorial.rst", "tutorials/llama3.rst", "tutorials/llama_kd_tutorial.rst", "tutorials/lora_finetune.rst", "tutorials/memory_optimizations.rst", "tutorials/qat_finetune.rst", "tutorials/qlora_finetune.rst"], "titles": ["torchtune.config", "torchtune.data", "torchtune.datasets", "torchtune.generation", "torchtune.models", "torchtune.modules", "torchtune.rlhf", "torchtune.training", "torchtune.utils", "Chat Datasets", "Custom Components and Recipes", "Datasets Overview", "Instruct Datasets", "Message Transforms", "Messages", "Multimodal Transforms", "Multimodal Datasets", "Sample packing", "Preference Datasets", "Prompt Templates", "Text-completion Datasets", "Tokenizers", "Checkpointing in torchtune", "Logging to Comet", "All About Configs", "What Are Recipes?", "Logging to Weights &amp; Biases", "instantiate", "log_config", "parse", "validate", "AlpacaToMessages", "ChatMLTemplate", "ChosenRejectedToMessages", "torchtune.data.GrammarErrorCorrectionTemplate", "InputOutputToMessages", "Message", "OpenAIToMessages", "PromptTemplate", "PromptTemplateInterface", "torchtune.data.QuestionAnswerTemplate", "torchtune.data.Role", "ShareGPTToMessages", "torchtune.data.SummarizeTemplate", "format_content_with_images", "left_pad_sequence", "load_image", "padded_collate", "padded_collate_dpo", "padded_collate_sft", "padded_collate_tiled_images_and_mask", "truncate", "validate_messages", "ConcatDataset", "PackedDataset", "PreferenceDataset", "SFTDataset", "TextCompletionDataset", "alpaca_cleaned_dataset", "alpaca_dataset", "chat_dataset", "cnn_dailymail_articles_dataset", "grammar_dataset", "hh_rlhf_helpful_dataset", "instruct_dataset", "llava_instruct_dataset", "the_cauldron_dataset", "preference_dataset", "samsum_dataset", "slimorca_dataset", "stack_exchange_paired_dataset", "text_completion_dataset", "wikitext_dataset", "generate", "generate_next_token", "get_causal_mask_from_padding_mask", "get_position_ids_from_padding_mask", "sample", "TilePositionalEmbedding", "TiledTokenPositionalEmbedding", "TokenPositionalEmbedding", "clip_vision_encoder", "code_llama2_13b", "code_llama2_70b", "code_llama2_7b", "lora_code_llama2_13b", "lora_code_llama2_70b", "lora_code_llama2_7b", "qlora_code_llama2_13b", "qlora_code_llama2_70b", "qlora_code_llama2_7b", "gemma", "gemma_2b", "gemma_7b", "gemma_tokenizer", "lora_gemma", "lora_gemma_2b", "lora_gemma_7b", "qlora_gemma_2b", "qlora_gemma_7b", "Llama2ChatTemplate", "llama2", "llama2_13b", "llama2_70b", "llama2_7b", "llama2_reward_7b", "llama2_tokenizer", "lora_llama2", "lora_llama2_13b", "lora_llama2_70b", "lora_llama2_7b", "lora_llama2_reward_7b", "qlora_llama2_13b", "qlora_llama2_70b", "qlora_llama2_7b", "qlora_llama2_reward_7b", "llama3", "llama3_70b", "llama3_8b", "llama3_tokenizer", "lora_llama3", "lora_llama3_70b", "lora_llama3_8b", "qlora_llama3_70b", "qlora_llama3_8b", "llama3_1", "llama3_1_405b", "llama3_1_70b", "llama3_1_8b", "lora_llama3_1", "lora_llama3_1_405b", "lora_llama3_1_70b", "lora_llama3_1_8b", "qlora_llama3_1_405b", "qlora_llama3_1_70b", "qlora_llama3_1_8b", "llama3_2_1b", "llama3_2_3b", "lora_llama3_2_1b", "lora_llama3_2_3b", "qlora_llama3_2_1b", "qlora_llama3_2_3b", "Llama3VisionEncoder", "Llama3VisionProjectionHead", "Llama3VisionTransform", "llama3_2_vision_11b", "llama3_2_vision_decoder", "llama3_2_vision_encoder", "llama3_2_vision_transform", "lora_llama3_2_vision_11b", "lora_llama3_2_vision_decoder", "lora_llama3_2_vision_encoder", "qlora_llama3_2_vision_11b", "MistralChatTemplate", "lora_mistral", "lora_mistral_7b", "lora_mistral_classifier", "lora_mistral_reward_7b", "mistral", "mistral_7b", "mistral_classifier", "mistral_reward_7b", "mistral_tokenizer", "qlora_mistral_7b", "qlora_mistral_reward_7b", "lora_phi3", "lora_phi3_mini", "phi3", "phi3_mini", "phi3_mini_tokenizer", "qlora_phi3_mini", "lora_qwen2", "lora_qwen2_0_5b", "lora_qwen2_1_5b", "lora_qwen2_7b", "qwen2", "qwen2_0_5b", "qwen2_1_5b", "qwen2_7b", "qwen2_tokenizer", "FeedForward", "Fp32LayerNorm", "KVCache", "MultiHeadAttention", "RMSNorm", "RotaryPositionalEmbeddings", "TanhGate", "TiedLinear", "TransformerCrossAttentionLayer", "TransformerDecoder", "TransformerSelfAttentionLayer", "VisionTransformer", "delete_kv_caches", "disable_kv_cache", "local_kv_cache", "reparametrize_as_dtype_state_dict_post_hook", "CEWithChunkedOutputLoss", "ForwardKLLoss", "ForwardKLWithChunkedOutputLoss", "DeepFusionModel", "FusionEmbedding", "FusionLayer", "get_fusion_params", "register_fusion_module", "AdapterModule", "LoRALinear", "disable_adapter", "get_adapter_params", "set_trainable_params", "validate_missing_and_unexpected_for_lora", "validate_state_dict_for_lora", "BaseTokenizer", "ModelTokenizer", "SentencePieceBaseTokenizer", "TikTokenBaseTokenizer", "parse_hf_tokenizer_json", "tokenize_messages_no_special_tokens", "Transform", "VisionCrossAttentionMask", "estimate_advantages", "get_rewards_ppo", "DPOLoss", "PPOLoss", "RSOLoss", "SimPOLoss", "truncate_sequence_at_first_stop_token", "torchtune.training.FSDPPolicyType", "FormattedCheckpointFiles", "FullModelHFCheckpointer", "FullModelMetaCheckpointer", "FullModelTorchTuneCheckpointer", "ModelType", "OptimizerInBackwardWrapper", "apply_selective_activation_checkpointing", "create_optim_in_bwd_wrapper", "get_cosine_schedule_with_warmup", "get_dtype", "get_full_finetune_fsdp_wrap_policy", "get_lr", "get_memory_stats", "get_quantizer_mode", "get_unmasked_sequence_lengths", "get_world_size_and_rank", "init_distributed", "is_distributed", "log_memory_stats", "lora_fsdp_wrap_policy", "CometLogger", "DiskLogger", "StdoutLogger", "TensorBoardLogger", "WandBLogger", "register_optim_in_bwd_hooks", "set_activation_checkpointing", "set_default_dtype", "set_seed", "setup_torch_profiler", "update_state_dict_for_classifier", "validate_expected_param_dtype", "batch_to_device", "get_device", "get_logger", "torch_version_ge", "&lt;no title&gt;", "Computation times", "Welcome to the torchtune Documentation", "Install Instructions", "torchtune Overview", "LoRA Single Device Finetuning", "Distributed Quantization-Aware Training (QAT)", "Recipes Overview", "Computation times", "torchtune CLI", "Fine-Tuning Llama3 with Chat Data", "End-to-End Workflow with torchtune", "Fine-Tune Your First LLM", "Meta Llama3 in torchtune", "Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation", "Fine-Tuning Llama2 with LoRA", "Memory Optimization Overview", "Fine-Tuning Llama3 with QAT", "Fine-Tuning Llama2 with QLoRA"], "terms": {"instruct": [1, 2, 4, 9, 10, 11, 13, 15, 16, 17, 18, 19, 21, 31, 32, 33, 35, 37, 42, 54, 56, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 71, 145, 148, 149, 153, 161, 167, 168, 169, 176, 177, 178, 265, 268, 269, 272, 273, 275, 277, 278, 280, 281], "prompt": [1, 9, 10, 11, 12, 13, 18, 31, 33, 34, 35, 36, 37, 38, 39, 40, 42, 43, 55, 56, 59, 60, 62, 63, 64, 67, 68, 69, 70, 73, 74, 94, 100, 106, 119, 144, 148, 153, 162, 169, 179, 189, 199, 216, 274, 276], "chat": [1, 2, 11, 13, 16, 18, 32, 37, 42, 56, 60, 100, 169, 268], "includ": [1, 9, 11, 12, 16, 18, 19, 21, 22, 24, 25, 38, 39, 56, 77, 81, 91, 101, 116, 125, 146, 147, 148, 150, 151, 158, 169, 175, 189, 205, 211, 228, 229, 267, 270, 272, 273, 274, 275, 276, 277, 278, 281], "some": [1, 17, 18, 20, 21, 22, 24, 32, 156, 200, 202, 207, 208, 265, 267, 268, 269, 272, 273, 274, 275, 277, 278, 279, 280, 281], "specif": [1, 5, 11, 12, 15, 19, 21, 24, 25, 27, 55, 56, 65, 66, 144, 212, 237, 269, 273, 274, 279, 280, 281], "format": [1, 2, 7, 11, 19, 21, 36, 45, 46, 55, 56, 59, 60, 63, 64, 67, 100, 144, 153, 212, 227, 228, 229, 230, 231, 272, 273, 274, 275, 276, 278, 279], "differ": [1, 9, 10, 17, 18, 19, 21, 24, 26, 48, 53, 60, 64, 78, 79, 80, 144, 191, 213, 221, 231, 258, 267, 268, 269, 272, 273, 274, 276, 277, 278, 279, 280, 281], "dataset": [1, 10, 13, 14, 15, 17, 19, 24, 31, 33, 35, 36, 37, 42, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 221, 267, 275, 276, 277, 280], "model": [1, 2, 9, 10, 11, 12, 13, 14, 16, 17, 18, 19, 20, 22, 23, 24, 25, 27, 31, 32, 33, 35, 36, 37, 42, 53, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 182, 183, 184, 185, 187, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 215, 216, 217, 219, 220, 221, 222, 223, 224, 228, 229, 230, 231, 233, 234, 237, 239, 246, 247, 252, 253, 257, 265, 267, 268, 269, 273, 281], "convert": [1, 9, 11, 14, 21, 22, 33, 35, 37, 42, 49, 55, 56, 60, 65, 66, 67, 75, 142, 228, 274, 280, 281], "from": [1, 2, 4, 10, 11, 13, 14, 15, 17, 19, 22, 23, 24, 25, 26, 27, 31, 33, 36, 37, 42, 45, 46, 47, 50, 53, 54, 55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 76, 77, 78, 79, 80, 81, 82, 83, 84, 92, 93, 100, 102, 103, 104, 105, 119, 143, 144, 148, 159, 161, 169, 176, 177, 178, 179, 180, 183, 188, 189, 190, 191, 192, 193, 194, 196, 197, 198, 201, 202, 203, 204, 207, 210, 213, 215, 218, 221, 223, 224, 227, 228, 229, 230, 232, 234, 235, 247, 250, 251, 252, 257, 264, 266, 269, 271, 272, 274, 275, 276, 277, 278, 279, 280], "common": [1, 2, 5, 9, 14, 15, 24, 216, 272, 273, 276, 278, 279, 280], "schema": [1, 9, 11, 12, 16], "convers": [1, 13, 16, 18, 19, 21, 22, 33, 42, 52, 55, 56, 60, 65, 67, 69, 228, 230, 231, 267, 273, 274, 278, 279, 281], "json": [1, 9, 12, 13, 16, 18, 21, 22, 37, 42, 55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 119, 148, 169, 179, 215, 228, 272, 273, 274, 280], "list": [1, 9, 11, 14, 15, 18, 19, 21, 22, 24, 33, 36, 38, 44, 45, 47, 48, 49, 50, 51, 52, 53, 55, 56, 60, 61, 65, 66, 67, 72, 73, 81, 85, 86, 87, 88, 89, 90, 94, 95, 96, 97, 98, 99, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 119, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 143, 144, 147, 148, 149, 150, 151, 152, 154, 155, 156, 157, 162, 163, 164, 165, 166, 169, 170, 171, 172, 173, 174, 189, 191, 196, 198, 199, 200, 201, 204, 205, 209, 210, 211, 212, 213, 214, 216, 218, 227, 228, 229, 230, 247, 261, 270, 273, 274, 275, 276, 279, 280], "us": [1, 2, 4, 5, 9, 10, 11, 12, 13, 14, 16, 17, 18, 20, 21, 22, 23, 26, 27, 29, 32, 35, 36, 38, 44, 47, 50, 53, 54, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 77, 79, 80, 81, 100, 101, 107, 116, 119, 120, 125, 129, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 165, 169, 171, 175, 179, 180, 182, 183, 184, 185, 187, 189, 190, 191, 192, 193, 194, 195, 196, 199, 200, 203, 206, 209, 213, 214, 218, 219, 220, 221, 222, 224, 226, 228, 229, 231, 232, 236, 237, 239, 246, 247, 248, 249, 250, 251, 255, 257, 259, 260, 265, 266, 267, 268, 269, 270, 272, 275, 276, 278, 279, 280], "collect": [1, 24, 275], "sampl": [1, 9, 11, 12, 13, 14, 15, 16, 19, 20, 21, 23, 26, 33, 35, 36, 37, 42, 44, 50, 54, 55, 56, 57, 62, 63, 65, 66, 67, 68, 69, 71, 73, 74, 183, 185, 189, 190, 191, 199, 217, 218, 223, 273, 274, 279], "batch": [1, 11, 17, 25, 47, 48, 49, 50, 54, 59, 62, 65, 66, 68, 79, 142, 143, 182, 183, 185, 188, 189, 190, 191, 194, 199, 201, 219, 220, 221, 223, 224, 241, 256, 259, 267, 275, 276, 278, 279], "handl": [1, 13, 16, 17, 24, 29, 31, 53, 56, 144, 213, 214, 273, 274, 278, 279, 281], "ani": [1, 5, 10, 11, 13, 14, 15, 16, 17, 21, 22, 24, 25, 27, 29, 30, 33, 36, 37, 38, 42, 44, 47, 50, 51, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 76, 80, 181, 189, 195, 199, 201, 202, 207, 208, 209, 210, 211, 212, 213, 216, 228, 229, 230, 232, 243, 246, 247, 255, 258, 272, 273, 275, 278, 279, 280], "pad": [1, 45, 47, 48, 49, 50, 54, 73, 75, 76, 189, 191, 218, 220, 222, 225, 241], "miscellan": 1, "modifi": [1, 10, 21, 24, 25, 26, 193, 195, 232, 267, 274, 276, 277, 278, 279, 280, 281], "For": [2, 7, 9, 10, 12, 14, 16, 18, 19, 20, 21, 22, 24, 25, 33, 35, 36, 37, 38, 42, 50, 53, 54, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 78, 79, 80, 81, 91, 95, 101, 107, 116, 120, 125, 129, 143, 146, 147, 150, 151, 154, 156, 158, 160, 165, 167, 171, 175, 183, 189, 191, 196, 199, 200, 203, 206, 217, 228, 234, 240, 247, 251, 253, 255, 266, 268, 269, 270, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "detail": [2, 9, 10, 12, 13, 16, 21, 22, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 78, 79, 80, 81, 160, 182, 191, 196, 222, 226, 237, 246, 255, 268, 269, 272, 274, 275, 276, 277, 278, 279, 280, 281], "usag": [2, 21, 195, 196, 198, 227, 231, 232, 256, 266, 272, 274, 275, 276, 279, 280, 281], "guid": [2, 23, 24, 26, 33, 35, 37, 42, 60, 62, 63, 64, 65, 66, 67, 68, 69, 224, 247, 267, 273, 275, 277, 278], "pleas": [2, 7, 34, 40, 43, 78, 79, 80, 81, 88, 89, 90, 98, 99, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 152, 163, 164, 170, 191, 196, 226, 237, 246, 253, 266, 269, 270, 274, 276, 281], "see": [2, 7, 9, 10, 11, 12, 13, 14, 16, 18, 19, 21, 22, 23, 26, 34, 40, 43, 47, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 74, 88, 89, 90, 98, 99, 100, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 152, 153, 160, 163, 164, 170, 182, 188, 190, 191, 201, 204, 211, 212, 217, 226, 231, 237, 246, 247, 251, 253, 255, 261, 266, 267, 268, 269, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "overview": [2, 7, 24, 26, 199, 265, 268, 269, 275, 277, 278, 281], "support": [2, 4, 10, 11, 15, 16, 17, 18, 21, 22, 23, 25, 26, 27, 36, 37, 54, 55, 56, 59, 60, 61, 62, 65, 66, 67, 68, 69, 72, 77, 95, 107, 120, 129, 142, 149, 150, 151, 153, 154, 156, 165, 168, 169, 171, 181, 183, 191, 200, 201, 205, 223, 229, 230, 232, 236, 239, 240, 267, 268, 269, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "sever": [2, 279], "wide": [2, 9, 183, 277], "onli": [2, 4, 10, 16, 18, 22, 23, 26, 35, 36, 42, 54, 55, 56, 61, 67, 73, 77, 81, 95, 107, 120, 129, 144, 149, 150, 151, 153, 154, 156, 165, 171, 183, 187, 189, 191, 196, 198, 202, 205, 207, 209, 213, 228, 229, 230, 232, 236, 237, 239, 240, 246, 272, 274, 275, 277, 278, 279, 280, 281], "help": [2, 11, 18, 19, 22, 63, 100, 189, 191, 199, 228, 247, 265, 266, 267, 272, 273, 274, 275, 277, 279, 280, 281], "quickli": [2, 11, 24, 38, 57, 268, 273, 279], "bootstrap": [2, 11], "your": [2, 7, 9, 11, 12, 13, 14, 16, 17, 18, 21, 23, 26, 27, 38, 57, 60, 64, 67, 79, 80, 81, 147, 151, 191, 200, 247, 250, 251, 257, 265, 266, 267, 268, 269, 272, 273, 276, 277, 278, 279, 280, 281], "fine": [2, 9, 10, 11, 12, 16, 18, 19, 20, 22, 23, 25, 26, 36, 54, 55, 56, 71, 257, 265, 267, 268, 269, 270, 274], "tune": [2, 4, 9, 10, 11, 12, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 29, 36, 54, 55, 56, 71, 257, 265, 266, 267, 268, 269, 270, 272, 274], "also": [2, 9, 10, 12, 14, 16, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 53, 60, 64, 67, 71, 73, 74, 91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 169, 171, 175, 183, 189, 192, 205, 224, 237, 239, 246, 247, 251, 257, 260, 266, 269, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "like": [2, 6, 12, 22, 23, 24, 25, 26, 169, 191, 196, 198, 200, 230, 266, 272, 273, 274, 275, 277, 278, 279, 280], "These": [2, 5, 10, 13, 15, 18, 19, 21, 22, 24, 25, 27, 54, 55, 67, 191, 218, 268, 270, 273, 274, 275, 276, 278, 279, 280, 281], "ar": [2, 5, 9, 10, 11, 12, 13, 14, 15, 16, 18, 19, 20, 21, 22, 23, 24, 26, 27, 31, 35, 38, 39, 42, 45, 47, 48, 52, 54, 55, 56, 59, 60, 64, 65, 66, 67, 73, 75, 76, 79, 85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 100, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 144, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 182, 188, 189, 190, 191, 193, 199, 200, 201, 205, 206, 209, 210, 218, 220, 226, 228, 229, 231, 232, 234, 236, 238, 239, 244, 246, 256, 257, 266, 267, 268, 270, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "especi": [2, 267, 272, 274, 279], "specifi": [2, 10, 12, 16, 18, 20, 22, 24, 25, 27, 31, 33, 35, 37, 42, 44, 60, 62, 63, 64, 65, 66, 67, 68, 69, 73, 75, 77, 94, 101, 106, 107, 116, 119, 120, 125, 129, 146, 148, 150, 162, 169, 171, 175, 179, 183, 189, 190, 197, 198, 199, 226, 237, 240, 246, 251, 253, 256, 269, 270, 272, 273, 274, 275, 276, 279, 280, 281], "yaml": [2, 10, 17, 18, 20, 24, 25, 27, 28, 29, 53, 60, 64, 67, 71, 251, 267, 270, 272, 273, 274, 275, 276, 278, 280, 281], "config": [2, 9, 12, 13, 16, 17, 18, 19, 20, 21, 22, 23, 26, 27, 28, 29, 30, 53, 60, 64, 67, 71, 183, 209, 228, 232, 247, 251, 256, 267, 268, 269, 270, 273, 274, 276, 277, 278, 279, 280, 281], "represent": [2, 227, 277, 278, 280, 281], "abov": [2, 4, 9, 16, 17, 18, 20, 22, 55, 195, 244, 266, 269, 274, 276, 278, 279, 280, 281], "text": [4, 5, 9, 11, 12, 15, 18, 19, 21, 35, 36, 37, 38, 39, 42, 44, 50, 54, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 144, 200, 201, 211, 213, 214, 216, 218, 273, 274, 280], "version": [4, 58, 73, 95, 107, 120, 129, 149, 154, 156, 165, 171, 183, 262, 266, 276, 279, 280, 281], "famili": [4, 22, 25, 59, 61, 65, 66, 69, 70, 72, 231, 267, 272, 276, 277], "import": [4, 9, 10, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 24, 27, 60, 64, 65, 66, 67, 71, 77, 191, 192, 193, 194, 221, 247, 250, 251, 273, 274, 275, 276, 277, 278, 280, 281], "you": [4, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 36, 38, 55, 56, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 100, 182, 189, 191, 194, 196, 198, 201, 203, 231, 247, 250, 251, 257, 265, 266, 267, 268, 269, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "need": [4, 9, 10, 12, 14, 16, 18, 19, 20, 22, 23, 24, 25, 26, 38, 54, 56, 183, 189, 191, 199, 200, 224, 246, 247, 250, 251, 252, 266, 268, 269, 270, 272, 273, 274, 275, 276, 278, 279, 281], "request": [4, 236, 274], "access": [4, 10, 22, 24, 25, 53, 228, 234, 268, 269, 272, 274, 275], "hug": [4, 11, 22, 32, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 119, 148, 169, 179, 215, 235, 267, 272, 275, 276], "face": [4, 11, 22, 32, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 119, 148, 169, 179, 215, 235, 267, 272, 275, 276], "befor": [4, 19, 22, 38, 52, 54, 65, 78, 79, 81, 91, 95, 147, 151, 183, 188, 189, 190, 191, 196, 198, 199, 201, 205, 214, 228, 247, 269, 272, 274, 279, 280], "download": [4, 10, 11, 16, 22, 65, 263, 266, 268, 269, 273, 276, 277, 278, 280, 281], "To": [4, 9, 12, 13, 14, 16, 17, 18, 19, 20, 22, 24, 25, 26, 54, 65, 189, 191, 201, 228, 257, 266, 267, 269, 270, 272, 274, 275, 276, 277, 278, 279, 280, 281], "1b": [4, 10, 17, 136, 138, 140, 265], "meta": [4, 10, 15, 16, 20, 21, 22, 100, 185, 228, 229, 268, 269, 272, 273, 274, 275, 277], "output": [4, 10, 12, 13, 14, 20, 21, 22, 31, 35, 45, 53, 55, 56, 59, 62, 64, 68, 69, 73, 81, 85, 86, 87, 91, 95, 101, 105, 107, 108, 109, 110, 111, 116, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 142, 143, 146, 147, 149, 150, 151, 154, 155, 156, 157, 158, 161, 165, 166, 171, 174, 175, 180, 181, 183, 185, 186, 188, 189, 190, 191, 196, 198, 199, 200, 201, 205, 208, 209, 210, 218, 230, 237, 249, 256, 257, 266, 268, 269, 272, 274, 275, 276, 277, 278, 279, 281], "dir": [4, 10, 21, 22, 251, 266, 268, 269, 272, 274, 275, 276, 277, 280], "tmp": [4, 9, 10, 12, 14, 15, 16, 17, 18, 19, 20, 21, 24, 232, 268, 269, 273, 275, 277], "ignor": [4, 9, 10, 12, 22, 42, 71, 187, 188, 190, 197, 198, 233, 257, 268, 269, 272, 277], "pattern": [4, 10, 19, 214, 268, 269, 272, 277], "origin": [4, 10, 15, 16, 17, 20, 21, 22, 58, 59, 63, 195, 200, 201, 205, 268, 269, 273, 274, 276, 277, 278, 279, 280, 281], "consolid": [4, 10, 22, 268, 269, 277], "00": [4, 10, 16, 22, 60, 64, 264, 268, 269, 271, 275, 277], "pth": [4, 10, 22, 227, 268, 269, 274, 277], "hf": [4, 9, 18, 20, 21, 22, 221, 223, 228, 272, 273, 274, 275, 276], "token": [4, 9, 10, 11, 12, 13, 15, 16, 17, 18, 19, 20, 22, 24, 25, 36, 42, 47, 49, 50, 51, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 91, 94, 95, 101, 106, 107, 116, 119, 120, 125, 129, 144, 146, 147, 148, 150, 151, 154, 156, 158, 160, 162, 165, 167, 169, 171, 175, 179, 183, 185, 188, 189, 190, 191, 196, 198, 199, 200, 201, 211, 212, 213, 214, 215, 216, 218, 220, 222, 225, 237, 241, 268, 272, 274, 275, 276, 277, 278, 279, 280, 281], "hf_token": [4, 21, 269, 277], "3b": [4, 137, 139, 141], "The": [4, 9, 11, 12, 15, 16, 18, 19, 20, 21, 22, 23, 24, 25, 26, 29, 30, 32, 36, 46, 47, 52, 53, 54, 55, 56, 60, 63, 64, 65, 66, 67, 70, 78, 79, 80, 81, 85, 86, 87, 95, 96, 97, 107, 108, 109, 110, 111, 120, 121, 122, 129, 130, 131, 132, 138, 139, 142, 144, 147, 149, 150, 151, 154, 156, 165, 166, 171, 172, 173, 174, 181, 184, 185, 186, 187, 191, 195, 196, 197, 198, 199, 200, 201, 206, 211, 212, 213, 214, 215, 216, 218, 219, 221, 222, 223, 224, 226, 228, 230, 232, 235, 236, 238, 240, 247, 251, 254, 256, 260, 261, 262, 266, 267, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "reus": [4, 267], "llama3_token": [4, 15, 17, 20, 21, 65, 66, 73, 273, 276], "class": [4, 10, 13, 14, 15, 21, 24, 26, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 42, 43, 53, 54, 55, 56, 57, 65, 66, 78, 79, 80, 81, 94, 100, 105, 106, 119, 142, 143, 144, 148, 153, 156, 160, 161, 162, 169, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 196, 197, 198, 199, 200, 201, 202, 204, 205, 207, 208, 211, 212, 213, 214, 217, 218, 221, 222, 223, 224, 227, 228, 229, 230, 231, 232, 247, 248, 249, 250, 251, 270, 273, 275, 277, 278, 279, 281], "languag": [4, 10, 16, 32, 73, 150, 200, 201, 205, 221, 257, 278, 279], "11b": [4, 145, 152], "8b": [4, 15, 16, 20, 21, 118, 122, 124, 128, 130, 132, 135, 166, 265, 268, 269, 272, 273, 280], "70b": [4, 83, 86, 89, 103, 109, 113, 117, 121, 123, 127, 131, 134, 276], "405b": [4, 126, 130, 133], "weight": [4, 21, 22, 25, 50, 85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 145, 148, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 187, 195, 204, 205, 209, 213, 221, 228, 229, 230, 231, 240, 251, 257, 265, 269, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "can": [4, 5, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 30, 33, 35, 36, 37, 38, 39, 42, 50, 53, 55, 56, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 78, 79, 81, 144, 147, 151, 184, 185, 187, 188, 189, 191, 196, 198, 199, 201, 203, 206, 213, 214, 226, 228, 231, 233, 237, 246, 247, 250, 251, 253, 256, 265, 266, 267, 268, 269, 270, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "instead": [4, 9, 12, 14, 16, 22, 25, 31, 45, 54, 55, 71, 81, 129, 150, 151, 182, 187, 191, 205, 224, 272, 276, 278, 279, 280], "builder": [4, 9, 10, 11, 12, 13, 15, 16, 17, 22, 58, 60, 61, 64, 67, 82, 83, 84, 85, 86, 87, 88, 89, 90, 92, 93, 96, 97, 98, 99, 102, 103, 104, 105, 108, 109, 110, 111, 112, 113, 114, 115, 117, 118, 121, 122, 123, 124, 126, 127, 128, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 152, 155, 157, 159, 161, 163, 164, 166, 168, 170, 172, 173, 174, 176, 177, 178, 273, 279, 281], "all": [4, 5, 10, 11, 14, 15, 19, 21, 25, 30, 35, 36, 38, 42, 45, 47, 50, 53, 54, 55, 56, 81, 119, 142, 148, 169, 179, 183, 187, 189, 191, 192, 193, 194, 195, 199, 200, 201, 203, 206, 217, 228, 232, 234, 238, 244, 252, 258, 259, 263, 265, 267, 268, 269, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280], "7b": [4, 9, 12, 14, 18, 19, 20, 21, 22, 61, 72, 84, 87, 90, 93, 97, 104, 105, 110, 111, 114, 115, 155, 157, 159, 161, 164, 174, 178, 228, 229, 273, 275, 276, 278, 281], "13b": [4, 22, 82, 85, 88, 102, 108, 112], "codellama": 4, "size": [4, 14, 15, 16, 22, 25, 27, 45, 50, 59, 62, 65, 66, 68, 79, 80, 81, 142, 143, 144, 145, 147, 148, 149, 151, 182, 183, 184, 185, 188, 189, 190, 191, 194, 196, 198, 199, 200, 201, 218, 219, 220, 241, 242, 244, 267, 269, 272, 274, 275, 276, 278, 279, 280], "0": [4, 9, 10, 12, 14, 15, 16, 18, 20, 22, 25, 45, 47, 48, 49, 50, 54, 60, 64, 67, 73, 74, 76, 77, 81, 85, 86, 87, 88, 89, 90, 91, 95, 96, 97, 98, 99, 101, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 120, 121, 122, 123, 124, 125, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 143, 144, 146, 147, 149, 150, 151, 152, 154, 155, 156, 157, 158, 160, 163, 164, 165, 166, 167, 170, 171, 172, 173, 174, 175, 176, 177, 182, 183, 189, 191, 192, 193, 194, 200, 205, 216, 221, 222, 223, 224, 225, 235, 241, 247, 250, 251, 255, 260, 262, 264, 269, 271, 273, 274, 275, 276, 278, 279, 280, 281], "5b": [4, 172, 173, 176, 177, 279], "qwen2": [4, 10, 171, 172, 173, 174, 176, 177, 178, 179, 231, 279], "exampl": [4, 10, 19, 21, 22, 23, 24, 25, 26, 27, 29, 33, 35, 37, 38, 42, 44, 45, 46, 47, 48, 49, 50, 53, 54, 56, 59, 60, 61, 62, 64, 65, 66, 67, 68, 69, 71, 72, 73, 75, 76, 77, 81, 143, 144, 147, 151, 182, 183, 191, 192, 193, 194, 196, 198, 199, 200, 201, 203, 204, 206, 211, 212, 213, 214, 216, 217, 221, 223, 224, 225, 226, 227, 228, 229, 231, 232, 240, 241, 247, 250, 251, 254, 257, 260, 261, 262, 263, 264, 266, 268, 269, 271, 272, 273, 274, 276, 277, 278, 279, 280, 281], "none": [4, 9, 16, 25, 26, 28, 30, 31, 33, 35, 37, 42, 50, 51, 52, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 77, 81, 94, 101, 106, 107, 116, 119, 120, 125, 129, 142, 143, 144, 146, 148, 150, 162, 169, 179, 180, 182, 183, 185, 188, 189, 190, 191, 192, 193, 194, 199, 201, 206, 208, 209, 210, 213, 216, 218, 219, 220, 222, 228, 229, 230, 231, 232, 233, 236, 240, 245, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 258, 259, 260, 261, 272, 274, 280], "mini": [4, 21, 166, 167, 168, 169, 170], "4k": [4, 21, 167, 168, 169], "microsoft": [4, 168, 169], "ai": [4, 10, 12, 14, 19, 55, 56, 159, 251, 273, 276], "thi": [4, 9, 10, 12, 13, 14, 16, 17, 18, 19, 20, 21, 23, 24, 25, 26, 27, 31, 33, 34, 35, 36, 37, 42, 43, 44, 45, 47, 48, 50, 53, 54, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 75, 79, 80, 81, 91, 95, 101, 107, 116, 120, 125, 129, 142, 144, 146, 147, 150, 151, 153, 154, 156, 158, 160, 165, 167, 168, 169, 171, 175, 180, 182, 183, 185, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 199, 200, 201, 203, 204, 206, 209, 210, 211, 212, 213, 214, 216, 217, 218, 220, 221, 222, 224, 226, 227, 228, 229, 230, 232, 235, 236, 239, 241, 244, 246, 247, 248, 250, 251, 252, 253, 255, 257, 259, 260, 265, 266, 267, 268, 269, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "v0": [4, 9, 14, 18, 19, 21, 153], "mistralai": [4, 21, 272], "2b": [4, 92, 96], "gemma2": 4, "googl": [4, 92, 93], "gguf": 4, "compon": [4, 6, 14, 21, 22, 25, 30, 48, 55, 56, 65, 66, 267, 270, 275, 277, 278, 281], "multimod": [4, 11, 14, 36, 42, 56, 65, 66, 199, 266], "encod": [4, 5, 15, 21, 50, 56, 73, 74, 81, 142, 143, 145, 146, 147, 149, 150, 151, 183, 188, 189, 190, 194, 199, 200, 201, 203, 211, 213, 214, 216, 218, 221, 224, 273], "perform": [5, 12, 13, 17, 19, 20, 21, 22, 54, 73, 191, 196, 206, 217, 224, 267, 268, 269, 273, 274, 276, 277, 280, 281], "direct": [5, 18, 25, 48, 85, 86, 95, 96, 97, 107, 108, 109, 110, 120, 121, 122, 131, 132, 138, 139, 154, 155, 156, 157, 165, 166, 221, 266, 270], "id": [5, 14, 17, 21, 22, 47, 48, 49, 50, 54, 61, 65, 66, 72, 73, 74, 76, 77, 144, 183, 185, 189, 190, 199, 211, 212, 213, 214, 215, 216, 218, 228, 230, 247, 273, 274], "decod": [5, 9, 12, 14, 15, 16, 18, 20, 21, 60, 64, 67, 73, 91, 95, 101, 107, 116, 120, 125, 129, 143, 144, 145, 146, 147, 149, 150, 151, 154, 156, 158, 160, 165, 167, 171, 175, 183, 188, 189, 190, 194, 199, 201, 203, 211, 213, 214, 273], "typic": [5, 9, 12, 20, 24, 33, 37, 42, 50, 54, 55, 56, 57, 71, 169, 203, 221, 224, 279, 280, 281], "byte": [5, 21, 214, 279, 281], "pair": [5, 10, 18, 21, 24, 48, 49, 63, 67, 70, 214], "underli": [5, 13, 18, 21, 213, 279, 281], "helper": 5, "method": [5, 13, 14, 15, 19, 21, 22, 24, 25, 26, 29, 46, 55, 57, 59, 60, 61, 62, 63, 64, 67, 68, 69, 70, 71, 72, 144, 189, 195, 196, 199, 202, 203, 204, 207, 209, 211, 212, 232, 240, 266, 267, 278, 281], "two": [5, 15, 18, 19, 22, 24, 35, 50, 52, 65, 66, 73, 74, 79, 191, 200, 203, 218, 225, 227, 267, 269, 274, 275, 276, 278, 279, 280, 281], "pre": [5, 9, 11, 12, 17, 18, 19, 20, 54, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 100, 145, 148, 149, 191, 199, 201, 203, 269, 273], "train": [5, 9, 10, 11, 12, 13, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 31, 33, 35, 50, 53, 54, 55, 56, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 100, 144, 145, 148, 149, 181, 183, 185, 189, 190, 195, 196, 198, 199, 200, 201, 203, 221, 224, 227, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 265, 267, 268, 270, 272, 273, 274, 276, 277, 278, 279, 280, 281], "function": [5, 10, 22, 24, 25, 27, 29, 45, 46, 47, 48, 60, 64, 67, 73, 79, 80, 81, 147, 151, 180, 183, 191, 192, 195, 206, 209, 210, 221, 222, 226, 228, 242, 255, 257, 259, 260, 267, 277, 281], "preprocess": [5, 54, 191], "imag": [5, 11, 15, 35, 36, 37, 42, 44, 46, 50, 56, 65, 66, 78, 79, 80, 81, 142, 143, 144, 145, 147, 148, 149, 151, 191, 200, 218, 278], "loss": [6, 9, 12, 14, 24, 25, 36, 38, 55, 56, 59, 60, 62, 64, 67, 68, 69, 196, 197, 198, 221, 222, 223, 224, 275, 277, 278, 281], "algorithm": [6, 21, 219, 224, 255], "ppo": [6, 219, 220, 221, 222, 270], "dpo": [6, 18, 48, 55, 206, 221, 223, 224, 270], "offer": 7, "allow": [7, 10, 53, 201, 209, 250, 269, 272, 279, 280, 281], "seamless": 7, "transit": 7, "between": [7, 9, 18, 19, 21, 22, 55, 60, 67, 146, 150, 188, 189, 193, 199, 220, 222, 224, 228, 231, 247, 274, 276, 277, 278, 280, 281], "interoper": [7, 22, 25, 267, 274, 281], "rest": [7, 273, 279, 281], "ecosystem": [7, 22, 25, 267, 274, 276, 281], "comprehens": [7, 279], "deep": [7, 22, 23, 24, 25, 26, 201, 203, 267, 270, 275, 276, 279], "dive": [7, 22, 23, 24, 25, 26, 267, 269, 270, 275, 276, 279], "util": [7, 14, 16, 22, 24, 25, 27, 45, 47, 50, 142, 233, 250, 252, 253, 259, 260, 261, 262, 267, 274, 275, 279, 281], "work": [7, 22, 25, 35, 42, 187, 200, 201, 267, 269, 272, 274, 276, 279, 281], "set": [7, 9, 12, 17, 18, 20, 22, 23, 24, 25, 26, 33, 36, 37, 42, 50, 54, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 71, 72, 101, 107, 116, 120, 125, 129, 144, 146, 149, 150, 154, 156, 158, 160, 165, 167, 171, 175, 183, 185, 188, 189, 192, 193, 194, 199, 206, 208, 226, 232, 237, 244, 246, 247, 253, 254, 255, 256, 259, 260, 267, 270, 272, 273, 274, 275, 276, 277, 278, 279, 280], "enabl": [7, 10, 11, 17, 21, 23, 24, 25, 26, 53, 85, 86, 87, 88, 89, 90, 96, 97, 98, 99, 108, 109, 110, 111, 112, 113, 114, 115, 121, 122, 123, 124, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 152, 155, 157, 163, 164, 166, 170, 172, 173, 174, 176, 177, 183, 188, 189, 190, 192, 193, 194, 199, 201, 205, 255, 256, 269, 276, 278, 279, 281], "consumpt": [7, 53, 75, 268, 279], "dure": [7, 10, 11, 22, 54, 59, 60, 62, 64, 67, 68, 69, 182, 183, 185, 189, 190, 191, 195, 199, 200, 218, 224, 239, 268, 269, 273, 274, 276, 278, 279, 280, 281], "control": [7, 13, 18, 21, 25, 36, 59, 60, 62, 64, 67, 68, 69, 193, 194, 201, 206, 247, 255, 269, 274, 279], "lr": [7, 24, 232, 235, 238, 277, 279], "process": [7, 11, 14, 15, 17, 25, 26, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 81, 147, 151, 191, 195, 242, 243, 255, 275, 280, 281], "variou": 7, "provid": [7, 10, 11, 12, 14, 22, 24, 25, 27, 32, 33, 35, 37, 42, 46, 47, 51, 53, 54, 73, 75, 81, 183, 187, 189, 191, 199, 206, 216, 221, 230, 237, 247, 251, 256, 260, 267, 268, 269, 272, 273, 274, 275, 276, 279], "debug": [7, 22, 24, 25, 247, 272], "finetun": [7, 10, 22, 24, 25, 85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 172, 173, 174, 199, 265, 267, 269, 275, 276, 279], "job": [7, 10, 26, 255, 275], "involv": [9, 12, 17, 20, 56, 280], "multi": [9, 18, 25, 183, 276], "turn": [9, 18, 25, 33, 36, 37, 42, 52, 55, 67, 273, 279], "multipl": [9, 16, 17, 18, 22, 24, 25, 33, 36, 37, 42, 48, 53, 56, 67, 142, 143, 183, 189, 190, 191, 199, 205, 247, 248, 249, 250, 251, 256, 275, 276, 277, 279], "back": [9, 21, 22, 52, 206, 228, 278, 279, 281], "forth": [9, 52], "user": [9, 12, 13, 14, 15, 16, 18, 19, 21, 25, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 46, 52, 55, 56, 60, 64, 67, 94, 101, 106, 107, 116, 119, 120, 125, 129, 146, 148, 150, 154, 156, 158, 160, 162, 165, 167, 169, 171, 175, 179, 183, 216, 270, 273, 275, 280], "assist": [9, 12, 13, 14, 15, 16, 18, 19, 21, 31, 32, 33, 35, 36, 37, 38, 39, 41, 42, 44, 52, 55, 56, 60, 67, 73, 94, 100, 106, 119, 148, 162, 169, 179, 216, 273], "role": [9, 13, 14, 15, 16, 18, 19, 21, 33, 36, 37, 38, 39, 42, 44, 55, 56, 60, 67, 94, 106, 119, 144, 148, 162, 169, 179, 216, 273], "content": [9, 13, 15, 16, 18, 19, 21, 22, 33, 36, 37, 38, 39, 42, 44, 55, 56, 60, 67, 216, 273], "what": [9, 14, 15, 16, 18, 22, 23, 24, 26, 36, 37, 55, 56, 60, 64, 67, 100, 153, 191, 265, 270, 273, 274, 275, 276, 279], "answer": [9, 15, 16, 19, 40, 64, 274, 276], "ultim": [9, 280], "question": [9, 15, 16, 19, 40, 64, 274, 276], "life": 9, "42": [9, 73, 191], "That": [9, 273], "s": [9, 10, 12, 13, 14, 16, 17, 18, 19, 20, 22, 24, 25, 26, 27, 29, 32, 37, 42, 52, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 85, 86, 87, 100, 107, 108, 109, 110, 111, 120, 121, 122, 129, 130, 131, 132, 138, 139, 142, 143, 144, 149, 150, 151, 153, 154, 155, 156, 157, 165, 166, 169, 171, 174, 175, 182, 183, 185, 189, 190, 191, 195, 199, 202, 203, 204, 207, 209, 210, 214, 221, 223, 224, 225, 226, 228, 229, 232, 237, 239, 241, 246, 247, 250, 253, 254, 257, 259, 260, 267, 272, 273, 275, 277, 278, 279, 280, 281], "ridicul": 9, "oh": 9, "i": [9, 12, 14, 18, 19, 20, 25, 36, 67, 73, 100, 142, 143, 153, 183, 188, 189, 190, 191, 195, 199, 208, 227, 232, 274, 276, 279, 280, 281], "know": [9, 273, 274, 277, 278], "more": [9, 10, 11, 12, 13, 14, 16, 18, 19, 21, 22, 24, 25, 38, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 182, 191, 196, 203, 209, 226, 227, 230, 247, 251, 253, 255, 259, 267, 268, 269, 270, 272, 274, 275, 276, 277, 278, 279, 280, 281], "structur": [9, 12, 13, 14, 19, 25, 37, 39, 42, 60, 119, 144, 148, 169, 179, 218, 273, 274, 280], "than": [9, 10, 12, 16, 18, 24, 50, 52, 73, 75, 182, 183, 191, 221, 226, 230, 231, 258, 259, 262, 273, 274, 275, 276, 277, 278, 279, 281], "freeform": [9, 12, 57, 71], "associ": [9, 10, 11, 12, 22, 24, 25, 73, 74, 81, 91, 101, 116, 125, 146, 150, 158, 175, 247, 274, 278], "where": [9, 10, 12, 14, 16, 18, 19, 20, 36, 38, 45, 48, 59, 73, 75, 76, 79, 105, 142, 143, 161, 180, 183, 189, 191, 193, 196, 198, 199, 205, 213, 218, 219, 221, 222, 225, 237, 241, 246, 277, 279], "thei": [9, 11, 12, 19, 21, 24, 25, 53, 65, 66, 81, 142, 147, 151, 189, 191, 201, 210, 237, 272, 273, 278, 279, 280], "learn": [9, 12, 25, 53, 200, 201, 203, 232, 235, 238, 267, 268, 269, 270, 273, 275, 276, 278, 279, 280, 281], "simpli": [9, 12, 13, 14, 16, 20, 22, 24, 54, 56, 221, 272, 273, 274, 276, 277, 279, 281], "predict": [9, 12, 73, 74, 77, 219, 220, 222, 268], "next": [9, 12, 22, 54, 71, 73, 74, 81, 191, 218, 268, 276, 281], "respond": 9, "accur": 9, "primari": [9, 12, 16, 18, 20, 22, 24, 25, 55, 56, 270, 275], "entri": [9, 12, 16, 18, 20, 24, 25, 47, 50, 270, 275, 279], "point": [9, 10, 12, 16, 18, 20, 21, 24, 25, 46, 60, 216, 270, 274, 275, 276, 278, 280, 281], "torchtun": [9, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 35, 36, 37, 38, 39, 42, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 227, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 262, 266, 268, 269, 270, 273, 275, 279], "chat_dataset": [9, 12, 13, 18, 273], "let": [9, 10, 11, 12, 16, 18, 22, 24, 26, 272, 273, 274, 275, 276, 277, 278, 279, 281], "follow": [9, 10, 11, 12, 15, 16, 19, 22, 25, 36, 37, 38, 42, 50, 54, 55, 56, 64, 67, 144, 183, 188, 218, 222, 230, 231, 232, 235, 244, 251, 256, 265, 266, 269, 270, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "data": [9, 10, 12, 13, 14, 15, 16, 19, 21, 23, 31, 32, 33, 35, 36, 37, 38, 39, 42, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 100, 148, 179, 191, 217, 221, 223, 239, 247, 248, 249, 250, 251, 259, 268, 269, 274, 279, 280, 281], "directli": [9, 10, 12, 13, 14, 16, 22, 24, 25, 27, 31, 55, 56, 60, 64, 65, 67, 71, 221, 226, 228, 272, 274, 275, 276, 278, 279, 280, 281], "llm": [9, 10, 11, 12, 21, 25, 199, 201, 265, 266, 267, 268, 270, 274, 276, 277, 278], "my_data": [9, 12, 13, 16, 273], "human": [9, 16, 18, 36, 42, 60, 100, 221, 222, 223, 273], "valu": [9, 16, 22, 24, 33, 35, 37, 42, 45, 47, 48, 50, 59, 60, 62, 63, 64, 67, 68, 69, 70, 73, 74, 76, 77, 82, 83, 84, 91, 92, 93, 95, 101, 102, 103, 104, 105, 107, 116, 117, 118, 120, 125, 126, 127, 128, 129, 136, 137, 144, 146, 150, 154, 156, 158, 159, 160, 161, 165, 167, 171, 175, 176, 177, 178, 182, 183, 184, 188, 189, 190, 197, 198, 199, 201, 209, 219, 220, 222, 225, 228, 231, 232, 235, 241, 247, 248, 249, 250, 251, 255, 269, 272, 273, 275, 276, 278, 279, 280], "gpt": [9, 16, 42, 60, 74, 273, 274], "mistral": [9, 14, 18, 19, 21, 144, 153, 154, 155, 156, 157, 159, 160, 161, 162, 163, 164, 231, 272, 273, 274, 275], "mistral_token": [9, 14, 18, 19, 21], "m_token": [9, 14, 18, 19, 20, 21], "path": [9, 10, 12, 14, 15, 16, 17, 18, 19, 20, 21, 24, 25, 26, 27, 35, 42, 46, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 94, 106, 119, 144, 148, 162, 169, 179, 213, 214, 215, 228, 229, 230, 256, 272, 273, 274, 276, 278], "1": [9, 14, 16, 18, 19, 20, 21, 22, 25, 35, 42, 45, 47, 48, 49, 50, 54, 69, 73, 74, 76, 77, 78, 79, 101, 107, 116, 120, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 143, 144, 146, 150, 153, 154, 156, 158, 160, 165, 167, 171, 172, 173, 175, 176, 177, 182, 183, 189, 191, 192, 193, 194, 196, 197, 198, 213, 214, 216, 221, 222, 223, 224, 229, 231, 235, 241, 244, 247, 250, 251, 254, 255, 267, 268, 272, 273, 274, 275, 278, 279, 280, 281], "prompt_templ": [9, 12, 14, 16, 18, 19, 94, 106, 119, 144, 148, 162, 169, 179], "mistralchattempl": [9, 14, 18, 19, 162, 273], "max_seq_len": [9, 10, 12, 14, 16, 17, 18, 20, 21, 24, 27, 47, 50, 51, 54, 59, 60, 61, 62, 64, 65, 66, 68, 69, 71, 72, 91, 94, 95, 101, 106, 107, 116, 119, 120, 125, 129, 144, 146, 148, 150, 154, 156, 158, 160, 162, 165, 167, 169, 171, 175, 179, 182, 183, 185, 189, 194, 280], "8192": [9, 12, 14, 16, 17, 18, 20, 21, 148, 278, 280], "ds": [9, 10, 12, 15, 16, 18, 20, 54, 69, 273], "sourc": [9, 10, 12, 13, 16, 18, 20, 22, 24, 27, 28, 29, 30, 31, 32, 33, 35, 36, 37, 38, 39, 42, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 91, 92, 93, 94, 95, 96, 97, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 116, 117, 118, 119, 120, 121, 122, 125, 126, 127, 128, 129, 130, 131, 132, 136, 137, 138, 139, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 165, 166, 167, 168, 169, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 227, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 262, 263, 273, 274, 280], "data_fil": [9, 12, 13, 16, 18, 20, 55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 273], "split": [9, 10, 12, 13, 14, 16, 18, 20, 22, 44, 53, 54, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 214, 273, 274, 280], "conversation_column": [9, 60, 273], "conversation_styl": [9, 60, 273], "By": [9, 12, 22, 269, 272, 277, 278, 279, 280, 281], "default": [9, 10, 12, 16, 22, 24, 31, 32, 33, 35, 36, 37, 42, 45, 48, 49, 50, 51, 54, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 77, 81, 82, 83, 84, 85, 86, 87, 91, 92, 93, 94, 95, 96, 97, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 116, 117, 118, 119, 120, 121, 122, 125, 126, 127, 128, 129, 130, 131, 132, 136, 137, 138, 139, 144, 145, 148, 149, 150, 151, 154, 155, 156, 157, 158, 159, 160, 161, 162, 165, 166, 167, 169, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 183, 184, 185, 188, 189, 190, 195, 197, 198, 199, 205, 209, 213, 214, 216, 218, 219, 220, 221, 224, 228, 229, 230, 232, 235, 236, 242, 246, 247, 248, 251, 254, 255, 256, 266, 269, 272, 273, 274, 276, 277, 278, 279, 280, 281], "true": [9, 10, 12, 13, 14, 15, 16, 17, 22, 24, 31, 36, 45, 53, 54, 57, 58, 59, 60, 62, 64, 65, 66, 67, 68, 69, 71, 72, 75, 76, 81, 88, 89, 90, 91, 95, 98, 99, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 144, 145, 152, 163, 164, 170, 183, 188, 189, 190, 192, 193, 194, 195, 196, 197, 199, 201, 206, 213, 214, 216, 218, 219, 222, 225, 226, 228, 229, 230, 237, 238, 239, 241, 243, 244, 247, 250, 256, 262, 268, 272, 273, 274, 276, 278, 279, 280, 281], "train_on_input": [9, 12, 13, 18, 24, 31, 33, 35, 37, 42, 53, 58, 59, 60, 62, 63, 64, 67, 68, 69, 70], "new_system_prompt": [9, 12, 13, 33, 35, 37, 42, 60, 62, 63, 64, 65, 66, 67, 68, 69], "tokenized_dict": [9, 12, 15, 16, 18, 20], "label": [9, 12, 20, 25, 47, 48, 49, 50, 54, 61, 69, 72, 196, 197, 198, 221, 224, 277], "print": [9, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 26, 44, 50, 53, 59, 62, 65, 66, 68, 69, 73, 144, 191, 192, 193, 194, 213, 214, 216, 262, 273, 275, 278, 280, 281], "inst": [9, 14, 19, 21, 100, 144, 153, 273], "733": [9, 14, 21], "16289": [9, 14, 21], "28793": [9, 14, 21], "1824": 9, "349": 9, "272": 9, "4372": 9, "In": [9, 10, 12, 13, 14, 16, 17, 18, 19, 20, 21, 22, 24, 25, 55, 79, 80, 81, 147, 151, 185, 189, 191, 205, 226, 246, 250, 251, 269, 273, 274, 276, 277, 278, 279, 280, 281], "_component_": [9, 10, 12, 13, 16, 17, 18, 19, 20, 21, 22, 23, 24, 26, 27, 53, 60, 64, 67, 71, 256, 269, 273, 274, 276, 277, 278, 279, 280], "null": [9, 22, 24, 280], "have": [9, 10, 13, 14, 18, 21, 22, 24, 27, 35, 36, 55, 60, 67, 75, 79, 80, 81, 142, 147, 151, 181, 182, 183, 184, 187, 189, 191, 192, 193, 194, 196, 198, 199, 204, 210, 218, 224, 227, 230, 232, 237, 238, 250, 258, 266, 273, 274, 275, 276, 277, 278, 279, 280, 281], "singl": [9, 10, 16, 17, 18, 19, 22, 24, 27, 33, 35, 37, 42, 47, 53, 54, 55, 56, 57, 60, 67, 71, 79, 80, 81, 94, 105, 106, 119, 142, 143, 144, 147, 148, 151, 161, 162, 169, 183, 189, 191, 199, 228, 229, 230, 231, 232, 234, 270, 272, 273, 274, 275, 276, 277, 278, 279, 281], "name": [9, 12, 13, 14, 16, 18, 20, 22, 23, 24, 26, 28, 31, 33, 35, 37, 42, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 204, 208, 210, 214, 228, 229, 230, 231, 232, 234, 247, 248, 249, 250, 251, 257, 258, 260, 272, 273, 274, 276, 279, 280], "messag": [9, 11, 12, 15, 16, 18, 19, 21, 31, 32, 33, 35, 37, 38, 39, 42, 44, 52, 55, 56, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 94, 106, 119, 144, 148, 162, 169, 212, 216, 266, 272, 273], "contain": [9, 11, 13, 14, 15, 16, 18, 20, 22, 33, 35, 36, 42, 47, 48, 49, 50, 54, 55, 56, 57, 60, 65, 71, 119, 144, 148, 169, 179, 182, 183, 185, 189, 190, 199, 202, 204, 207, 208, 209, 214, 216, 219, 225, 228, 229, 230, 232, 234, 239, 245, 250, 256, 257, 259, 273, 274, 276, 278], "topic": [9, 265], "per": [9, 16, 47, 88, 89, 90, 98, 99, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 143, 144, 152, 163, 164, 170, 182, 191, 195, 218, 220, 221, 272, 279, 280, 281], "could": [9, 18, 19, 238, 277, 278], "system": [9, 12, 13, 18, 19, 32, 33, 35, 36, 37, 38, 39, 41, 42, 44, 52, 55, 56, 60, 62, 63, 64, 65, 66, 67, 68, 69, 94, 100, 106, 119, 148, 153, 162, 169, 179, 216, 273], "tool": [9, 18, 19, 22, 36, 38, 56, 153, 247, 274, 275], "call": [9, 14, 18, 21, 22, 27, 36, 38, 56, 65, 66, 153, 183, 189, 191, 192, 193, 195, 199, 209, 247, 248, 249, 250, 251, 252, 256, 257, 273, 278, 281], "return": [9, 10, 13, 15, 18, 19, 21, 27, 29, 36, 38, 44, 45, 46, 47, 48, 49, 50, 51, 54, 55, 56, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 91, 92, 93, 94, 95, 96, 97, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 116, 117, 118, 119, 120, 121, 122, 125, 126, 127, 128, 129, 130, 131, 132, 136, 137, 138, 139, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 154, 155, 156, 157, 158, 159, 160, 161, 162, 165, 166, 167, 168, 169, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 188, 189, 190, 191, 193, 194, 196, 197, 198, 199, 200, 201, 202, 204, 205, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, 230, 232, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 254, 255, 256, 260, 261, 262, 277, 278, 281], "dai": [9, 20], "todai": 9, "It": [9, 10, 14, 16, 32, 36, 38, 55, 56, 60, 62, 64, 65, 66, 68, 70, 144, 147, 151, 153, 187, 189, 191, 199, 221, 224, 247, 272, 273, 277, 281], "tuesdai": 9, "about": [9, 10, 13, 14, 18, 22, 25, 65, 66, 191, 221, 224, 247, 251, 267, 268, 269, 270, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "tomorrow": 9, "wednesdai": 9, "As": [9, 12, 16, 22, 24, 25, 26, 205, 267, 274, 279, 281], "an": [9, 10, 12, 14, 16, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 46, 50, 52, 53, 57, 60, 62, 64, 65, 66, 67, 68, 71, 72, 78, 79, 80, 107, 120, 129, 144, 147, 149, 151, 154, 156, 160, 165, 171, 172, 173, 176, 177, 183, 187, 189, 191, 199, 200, 201, 203, 204, 206, 207, 208, 212, 217, 218, 221, 226, 227, 228, 229, 230, 232, 233, 237, 238, 247, 251, 256, 260, 267, 268, 269, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "slimorca": [9, 69], "pass": [9, 10, 11, 12, 14, 15, 16, 17, 18, 19, 20, 21, 24, 27, 36, 38, 53, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 91, 95, 101, 107, 116, 120, 125, 129, 149, 154, 156, 158, 160, 165, 167, 171, 175, 183, 189, 193, 194, 195, 199, 206, 210, 214, 222, 226, 230, 236, 237, 239, 243, 246, 247, 250, 251, 253, 256, 272, 273, 278, 280, 281], "repo": [9, 10, 12, 16, 18, 20, 22, 65, 228, 229, 231, 272, 274], "select": [9, 233], "one": [9, 10, 11, 12, 13, 16, 18, 22, 25, 33, 35, 37, 42, 47, 50, 52, 60, 66, 67, 191, 196, 198, 216, 230, 247, 274, 275, 276, 279, 281], "most": [9, 12, 13, 16, 18, 20, 22, 24, 36, 38, 273, 275, 278, 279, 281], "gemma": [9, 12, 18, 20, 92, 93, 94, 95, 96, 97, 98, 99, 187, 231, 279], "gemma_token": [9, 12, 18, 20], "g_token": [9, 12, 18, 20], "open": [9, 20, 46, 69, 92, 93, 274], "orca": [9, 69], "dedup": [9, 69], "recip": [9, 11, 12, 16, 18, 20, 22, 23, 24, 26, 27, 28, 29, 144, 189, 199, 228, 229, 230, 267, 268, 269, 273, 274, 276, 279, 281], "via": [9, 12, 14, 16, 17, 18, 20, 23, 24, 26, 55, 60, 64, 67, 71, 183, 189, 190, 205, 228, 278, 281], "http": [9, 12, 16, 27, 46, 57, 61, 63, 65, 71, 72, 74, 82, 83, 84, 85, 86, 87, 88, 89, 90, 92, 93, 95, 96, 97, 98, 99, 102, 103, 104, 105, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 152, 154, 155, 156, 157, 159, 161, 163, 164, 165, 166, 168, 169, 170, 172, 173, 174, 176, 177, 178, 183, 184, 185, 191, 196, 197, 218, 219, 221, 222, 223, 224, 226, 228, 229, 235, 244, 247, 250, 251, 253, 255, 261, 266, 274, 276, 277], "ha": [9, 18, 22, 64, 73, 143, 186, 188, 189, 191, 194, 196, 198, 199, 202, 204, 206, 207, 210, 225, 230, 232, 257, 258, 273, 274, 275, 276, 277, 278, 279, 281], "addition": [9, 22, 213, 214, 224, 255, 273, 278], "argument": [9, 10, 12, 16, 22, 24, 27, 34, 40, 43, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 88, 89, 90, 98, 99, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 152, 163, 164, 170, 226, 237, 243, 247, 248, 250, 251, 253, 272, 273, 278, 280], "load_dataset": [9, 12, 16, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 273], "document": [9, 12, 16, 17, 78, 79, 80, 81, 183, 189, 190, 226, 237, 246, 268, 270, 272, 279], "file": [9, 10, 11, 12, 16, 22, 23, 24, 25, 26, 27, 28, 29, 46, 55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 119, 144, 148, 169, 179, 213, 214, 215, 228, 229, 230, 248, 251, 256, 264, 267, 269, 271, 272, 273, 274, 275, 276, 278, 279, 280, 281], "raw": [9, 11, 13, 14, 16, 21, 44], "vari": [9, 50, 54, 189], "field": [9, 10, 14, 15, 27, 31, 35, 36, 42, 44, 54, 55, 56, 59, 65, 66, 245], "indic": [9, 14, 16, 18, 19, 50, 53, 54, 75, 76, 81, 147, 151, 183, 185, 189, 190, 191, 199, 200, 218, 219, 222, 225, 226, 241, 244, 273], "There": [9, 24, 52, 79, 273, 275, 276, 277, 278, 279], "few": [9, 10, 201, 276, 278, 281], "standard": [9, 12, 14, 15, 17, 19, 22, 34, 55, 56, 60, 63, 101, 107, 116, 120, 125, 129, 144, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 183, 249, 267, 273, 274, 276, 277], "across": [9, 22, 25, 50, 53, 228, 250, 255, 274, 276, 277, 280], "mani": [9, 14, 16, 19, 24, 54, 268, 269, 274, 277], "we": [9, 10, 11, 12, 18, 19, 20, 21, 22, 23, 24, 25, 26, 47, 50, 54, 55, 56, 60, 61, 67, 72, 73, 77, 182, 183, 185, 189, 190, 191, 193, 196, 198, 199, 205, 221, 224, 228, 229, 230, 236, 240, 246, 252, 257, 267, 268, 269, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "ipython": [9, 14, 19, 36, 38, 41, 55, 56, 94, 106, 119, 148, 162, 169, 179], "transform": [9, 10, 11, 16, 22, 25, 31, 33, 35, 55, 56, 59, 60, 62, 63, 65, 66, 67, 68, 69, 81, 85, 86, 87, 91, 95, 96, 97, 101, 107, 108, 109, 110, 111, 116, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 143, 144, 146, 147, 148, 149, 150, 151, 154, 155, 156, 157, 158, 160, 165, 166, 167, 171, 172, 173, 174, 175, 188, 189, 190, 191, 201, 218, 235, 253, 278, 279, 280], "sharegpttomessag": [9, 13, 60, 69], "expect": [9, 12, 13, 15, 16, 18, 19, 20, 22, 24, 27, 31, 33, 35, 36, 37, 42, 46, 50, 55, 56, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 143, 144, 185, 199, 210, 232, 247, 251, 258, 273, 278, 280], "code": [9, 10, 12, 13, 16, 19, 21, 22, 25, 82, 83, 84, 85, 86, 87, 88, 89, 90, 189, 247, 263, 267, 275], "openaitomessag": [9, 13, 60, 67], "If": [9, 10, 13, 14, 16, 17, 19, 21, 22, 24, 30, 33, 35, 36, 37, 42, 44, 46, 47, 50, 51, 52, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 71, 73, 75, 77, 81, 94, 101, 106, 107, 116, 119, 120, 125, 129, 142, 144, 146, 148, 150, 162, 169, 171, 175, 179, 182, 183, 185, 187, 189, 190, 191, 193, 194, 195, 196, 198, 199, 205, 210, 216, 228, 229, 230, 231, 232, 233, 236, 237, 238, 239, 240, 243, 247, 250, 251, 255, 256, 258, 260, 266, 272, 273, 274, 275, 276, 277, 278, 279, 280], "doe": [9, 17, 22, 44, 50, 54, 67, 71, 91, 153, 158, 168, 183, 187, 189, 190, 192, 193, 194, 197, 198, 199, 204, 216, 228, 230, 232, 257, 272, 274, 280], "fit": [9, 25, 54, 61, 71, 72, 191, 221, 273], "creat": [9, 10, 13, 16, 19, 22, 24, 27, 38, 54, 56, 60, 67, 75, 82, 83, 84, 85, 86, 87, 88, 89, 90, 92, 93, 96, 97, 98, 99, 102, 103, 104, 105, 108, 109, 110, 111, 112, 113, 114, 115, 117, 118, 121, 122, 123, 124, 126, 127, 128, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 152, 155, 157, 159, 161, 163, 164, 166, 168, 170, 172, 173, 174, 176, 177, 178, 182, 183, 189, 190, 191, 199, 226, 228, 229, 230, 234, 235, 247, 248, 250, 272, 274, 281], "custom": [9, 15, 16, 21, 24, 25, 31, 38, 55, 56, 60, 64, 65, 66, 67, 71, 94, 106, 119, 148, 162, 169, 179, 253, 267, 268, 269, 272, 275, 276, 278, 279], "dialogu": [9, 16, 43, 68, 273], "defin": [9, 10, 17, 22, 24, 25, 38, 55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 183, 188, 189, 199, 202, 204, 205, 207, 220, 275, 278], "same": [9, 10, 11, 15, 18, 22, 24, 38, 45, 78, 79, 85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 143, 166, 172, 173, 174, 181, 182, 184, 186, 187, 188, 190, 191, 194, 199, 201, 216, 222, 224, 225, 232, 237, 238, 251, 257, 259, 269, 272, 273, 274, 276, 277, 278, 279, 280, 281], "wai": [9, 14, 19, 22, 24, 55, 56, 209, 227, 272, 274, 275, 276, 277], "instruct_dataset": [9, 12, 13, 53], "info": [9, 261, 275], "slimorca_dataset": [9, 24], "command": [10, 12, 17, 21, 23, 25, 26, 266, 269, 270, 272, 273, 274, 275, 276, 277, 278, 280, 281], "line": [10, 17, 22, 23, 25, 270, 272, 275, 276], "both": [10, 14, 15, 21, 22, 37, 50, 53, 63, 67, 180, 199, 201, 203, 210, 272, 274, 277, 278, 280, 281], "built": [10, 11, 13, 23, 24, 26, 63, 67, 70, 266, 273, 275, 281], "done": [10, 17, 54, 189, 209, 236, 246, 257, 278, 280, 281], "run": [10, 17, 22, 23, 24, 26, 29, 91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 182, 183, 189, 195, 196, 228, 229, 230, 232, 233, 234, 244, 247, 250, 251, 252, 266, 267, 268, 269, 270, 273, 275, 276, 277, 278, 279, 280, 281], "cli": [10, 24, 26, 28, 29, 266, 268, 274, 275, 279], "which": [10, 11, 12, 14, 16, 17, 18, 19, 20, 21, 22, 24, 25, 46, 47, 53, 54, 57, 59, 60, 62, 64, 67, 68, 69, 71, 76, 77, 85, 86, 87, 94, 95, 96, 97, 106, 107, 108, 109, 110, 111, 119, 120, 121, 122, 129, 130, 131, 132, 138, 139, 144, 147, 148, 149, 150, 151, 153, 154, 155, 156, 157, 162, 165, 166, 169, 171, 172, 173, 174, 182, 183, 185, 189, 190, 191, 192, 193, 194, 199, 201, 209, 210, 213, 228, 229, 230, 232, 235, 236, 248, 251, 253, 257, 267, 268, 269, 270, 272, 273, 274, 275, 277, 278, 279, 280, 281], "folder": [10, 22], "first": [10, 17, 22, 24, 27, 42, 52, 54, 65, 76, 81, 147, 151, 189, 191, 192, 193, 199, 225, 228, 265, 267, 268, 273, 274, 276, 277, 278, 280, 281], "ensur": [10, 19, 21, 22, 24, 30, 52, 55, 56, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 183, 192, 228, 230, 236, 267, 275], "instal": [10, 23, 24, 26, 244, 247, 250, 251, 265, 272, 274, 275, 276, 277, 278, 279, 280, 281], "environ": [10, 25, 244, 247, 260, 266, 270, 272, 274, 275, 280], "so": [10, 13, 18, 19, 20, 22, 24, 54, 65, 187, 191, 228, 266, 267, 273, 274, 275, 276, 278, 279, 280, 281], "directori": [10, 22, 24, 35, 42, 65, 228, 229, 230, 248, 250, 251, 256, 272, 274, 275, 276], "new": [10, 14, 15, 16, 19, 21, 25, 37, 42, 59, 61, 62, 63, 65, 67, 68, 69, 159, 182, 200, 201, 231, 247, 248, 250, 273, 274, 275, 276, 277, 278, 281], "librari": [10, 221, 223, 236, 255, 261, 265, 266, 267, 272, 279, 281], "mkdir": 10, "my_project": [10, 247, 251], "cd": [10, 21, 266, 274], "llama": [10, 15, 16, 17, 20, 21, 22, 100, 142, 144, 145, 146, 147, 148, 150, 151, 184, 185, 228, 229, 268, 269, 272, 273, 274, 275, 276, 277, 278], "3": [10, 15, 16, 17, 20, 21, 22, 45, 47, 48, 49, 50, 54, 76, 77, 81, 142, 144, 145, 146, 147, 148, 150, 151, 153, 166, 168, 169, 191, 231, 241, 254, 261, 268, 269, 272, 273, 274, 275, 276, 277, 280, 281], "2": [10, 14, 15, 17, 21, 22, 26, 45, 47, 48, 49, 50, 52, 54, 69, 76, 77, 78, 79, 136, 137, 138, 139, 140, 141, 142, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 182, 183, 191, 213, 214, 216, 222, 224, 225, 228, 229, 231, 241, 254, 255, 256, 262, 269, 273, 274, 275, 276, 278, 279, 280], "lora": [10, 24, 85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 205, 206, 209, 210, 228, 246, 265, 267, 270, 273, 275, 276, 277], "devic": [10, 17, 24, 25, 194, 232, 236, 239, 259, 260, 270, 272, 273, 274, 275, 276, 278, 279, 281], "lora_finetune_single_devic": [10, 24, 268, 272, 273, 274, 275, 276, 277, 278, 279, 281], "llama3_2": [10, 17, 136, 137, 138, 139, 140, 141, 192, 193, 194, 231, 277], "1b_lora_single_devic": 10, "often": [10, 278, 279], "ll": [10, 18, 20, 22, 24, 25, 73, 240, 267, 269, 273, 274, 275, 276, 277, 279, 280, 281], "want": [10, 12, 19, 22, 24, 25, 26, 27, 50, 55, 56, 73, 203, 266, 272, 273, 274, 275, 276, 277, 278, 279], "start": [10, 23, 25, 26, 46, 76, 216, 231, 247, 266, 267, 273, 274, 275, 277, 279, 280], "our": [10, 12, 13, 20, 22, 25, 267, 268, 269, 270, 273, 274, 275, 277, 278, 279, 280, 281], "particular": [10, 11, 13, 19, 21, 24, 53, 144, 226, 278, 281], "adjust": [10, 268, 269, 277, 279, 280], "hyperparamet": [10, 23, 224, 232, 267, 275, 278, 281], "cp": [10, 24, 266, 272, 273, 274, 275, 276, 280], "copi": [10, 273, 274, 275, 276, 280, 281], "make": [10, 17, 19, 22, 23, 24, 25, 26, 145, 191, 267, 272, 274, 275, 276, 277, 278, 279, 280, 281], "modif": [10, 280], "show": [10, 144, 218, 266, 268, 269, 272, 273, 277, 278], "each": [10, 12, 15, 18, 19, 20, 22, 25, 38, 39, 42, 47, 48, 50, 53, 54, 55, 56, 78, 79, 80, 81, 85, 86, 87, 95, 96, 97, 107, 108, 109, 110, 111, 120, 121, 122, 129, 130, 131, 132, 138, 139, 143, 144, 147, 149, 150, 151, 154, 155, 156, 157, 165, 166, 171, 172, 173, 174, 183, 185, 189, 190, 191, 196, 198, 199, 201, 209, 210, 216, 218, 219, 220, 221, 223, 224, 241, 255, 256, 267, 269, 270, 272, 274, 275, 278, 279, 280], "ls": [10, 21, 266, 270, 272, 274, 275, 276], "full": [10, 11, 13, 16, 22, 24, 25, 34, 40, 43, 55, 72, 88, 89, 90, 98, 99, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 149, 152, 163, 164, 170, 199, 209, 210, 216, 233, 266, 267, 270, 272, 274, 276, 278, 279, 280], "5b_full_single_devic": 10, "qwen_config": 10, "now": [10, 19, 22, 182, 193, 232, 234, 269, 273, 274, 275, 276, 277, 278, 280, 281], "sure": [10, 17, 22, 24, 274, 275, 276, 277, 278, 279, 280, 281], "correct": [10, 12, 14, 19, 25, 34, 62, 184, 185, 189, 260, 267, 273], "ve": [10, 18, 21, 24, 182, 269, 272, 273, 274, 276, 277, 278, 279], "even": [10, 191, 257, 266, 272, 273, 276, 277, 278, 279, 281], "didn": 10, "t": [10, 13, 14, 18, 19, 20, 22, 24, 25, 45, 142, 143, 196, 201, 236, 251, 255, 272, 273, 274, 275, 277, 279, 281], "complet": [10, 11, 12, 18, 22, 25, 37, 54, 61, 71, 169, 273, 274, 275, 276, 279], "note": [10, 16, 21, 22, 24, 95, 199, 204, 232, 252, 255, 257, 269, 273, 274, 277, 278, 279, 280, 281], "must": [10, 13, 17, 27, 38, 53, 65, 66, 183, 193, 204, 227, 247, 281], "extens": [10, 25, 230, 267], "full_finetune_single_devic": [10, 17, 238, 272, 274, 275], "Or": [10, 199, 266], "rel": [10, 16, 17, 54, 183, 185, 189, 190, 199, 221, 239, 277, 278], "discuss": [10, 14, 19, 21, 24, 274, 275, 276, 278], "workflow": [10, 11, 265, 275, 278], "write": [10, 16, 22, 25, 228, 229, 230, 248, 275], "own": [10, 13, 18, 21, 22, 38, 246, 255, 272, 273, 274, 276, 277, 278], "loop": 10, "logic": [10, 15, 25, 31, 56, 212, 231, 267, 270, 275, 278], "case": [10, 14, 16, 22, 25, 26, 36, 38, 55, 79, 80, 81, 147, 151, 191, 193, 228, 232, 236, 240, 246, 248, 253, 267, 272, 273, 274, 276, 278, 279, 281], "similar": [10, 13, 16, 60, 61, 63, 65, 66, 67, 70, 71, 72, 209, 221, 274, 276, 277, 278, 279, 281], "scratch": 10, "local": [10, 11, 14, 46, 55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 119, 148, 169, 179, 247, 251, 255, 266, 272, 273, 274, 275], "single_devic": 10, "py": [10, 13, 24, 27, 74, 85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 172, 173, 174, 182, 184, 185, 197, 221, 222, 223, 224, 235, 272, 274, 276], "recommend": [10, 60, 61, 62, 67, 68, 70, 72, 153, 189, 196, 247, 250, 273, 274, 279, 281], "python": [10, 24, 247, 251, 255, 261, 263, 272, 274, 280], "convent": [10, 188], "main": [10, 27, 29, 169, 184, 185, 266, 269, 274, 276], "script": [10, 22, 26, 270, 272, 274, 275, 276], "decor": [10, 25, 29], "pars": [10, 24, 27, 28, 215, 270, 275], "omegaconf": [10, 27], "dictconfig": [10, 24, 25, 27, 28, 29, 30, 247, 251, 256], "def": [10, 13, 15, 19, 21, 24, 25, 26, 29, 65, 66, 226, 231, 277, 278, 281], "cfg": [10, 24, 25, 28, 29, 30], "add": [10, 12, 13, 14, 16, 19, 21, 23, 24, 26, 50, 54, 57, 71, 144, 153, 191, 203, 214, 216, 230, 231, 273, 274, 276, 278, 281], "here": [10, 12, 14, 15, 16, 18, 20, 21, 22, 23, 24, 26, 32, 62, 65, 66, 184, 185, 238, 268, 269, 270, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "attribut": [10, 187, 206, 216, 224, 234], "__name__": 10, "__main__": 10, "don": [10, 13, 14, 18, 19, 20, 22, 24, 25, 251, 255, 272, 273, 274, 275, 277, 279, 281], "experiment": [10, 21, 24], "optim": [10, 18, 19, 22, 24, 25, 48, 53, 55, 91, 158, 168, 221, 222, 223, 224, 230, 232, 234, 235, 238, 239, 252, 256, 268, 269, 270, 273, 274, 275, 276, 277, 278, 281], "them": [10, 12, 15, 18, 19, 22, 24, 53, 67, 191, 195, 201, 216, 259, 269, 272, 273, 274, 278, 279, 280, 281], "when": [10, 16, 17, 18, 20, 21, 22, 24, 25, 29, 53, 54, 55, 56, 57, 67, 71, 73, 75, 182, 183, 185, 189, 190, 191, 193, 194, 195, 196, 198, 199, 200, 206, 209, 220, 235, 237, 250, 252, 257, 268, 272, 274, 276, 277, 278, 279, 280, 281], "mean": [10, 24, 144, 183, 184, 188, 189, 190, 199, 219, 246, 272, 273, 275, 278, 280], "high": [10, 53, 55, 56, 267, 277, 278, 279], "level": [10, 25, 55, 56, 196, 198, 217, 234, 246, 261, 267, 277, 281], "paramet": [10, 13, 14, 15, 16, 25, 27, 28, 29, 30, 31, 33, 35, 36, 37, 38, 42, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 91, 92, 93, 94, 95, 96, 97, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 116, 117, 118, 119, 120, 121, 122, 125, 126, 127, 128, 129, 130, 131, 132, 136, 137, 138, 139, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 154, 155, 156, 157, 158, 159, 160, 161, 162, 165, 166, 167, 169, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, 229, 230, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 243, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 262, 265, 267, 268, 269, 270, 272, 273, 274, 275, 276, 277, 278, 280, 281], "easili": [10, 16, 22, 24, 267, 277, 278, 280, 281], "custom_decod": 10, "customtransformerdecod": 10, "nn": [10, 27, 45, 47, 50, 81, 142, 143, 180, 182, 183, 187, 188, 189, 190, 191, 192, 193, 194, 195, 199, 200, 201, 202, 203, 204, 206, 207, 208, 226, 233, 234, 246, 252, 253, 257, 258, 277, 278, 281], "modul": [10, 13, 15, 21, 24, 27, 65, 66, 78, 79, 80, 81, 142, 143, 144, 147, 151, 156, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 217, 218, 221, 222, 223, 226, 231, 233, 234, 237, 246, 252, 253, 255, 275, 277, 278, 281], "A": [10, 13, 15, 19, 25, 26, 33, 34, 37, 40, 42, 43, 47, 48, 49, 50, 53, 54, 67, 81, 179, 183, 187, 188, 189, 190, 191, 195, 199, 205, 209, 213, 214, 216, 218, 219, 220, 221, 222, 223, 224, 225, 226, 231, 232, 238, 239, 240, 245, 246, 264, 265, 271, 272, 273, 278, 279, 280, 281], "architectur": [10, 25, 100, 153, 189, 191, 199, 201, 231, 272], "present": [10, 33, 37, 42, 62, 63, 65, 66, 67, 68, 69, 214, 230, 257], "custom_model": 10, "num_lay": [10, 27, 81, 91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 189, 191, 199, 201], "int": [10, 15, 21, 24, 26, 47, 48, 49, 50, 51, 54, 61, 65, 66, 72, 73, 74, 75, 77, 78, 79, 80, 81, 85, 86, 87, 88, 89, 90, 91, 94, 95, 96, 97, 98, 99, 101, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 119, 120, 121, 122, 123, 124, 125, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 154, 155, 156, 157, 158, 160, 162, 163, 164, 165, 166, 167, 169, 170, 171, 172, 173, 174, 175, 179, 182, 183, 184, 185, 188, 189, 190, 191, 194, 196, 197, 198, 199, 200, 201, 205, 211, 212, 213, 214, 215, 216, 218, 225, 226, 228, 229, 230, 232, 233, 235, 237, 242, 246, 247, 248, 249, 250, 251, 253, 255, 256, 272, 277, 278, 279, 281], "classification_head": 10, "bool": [10, 15, 19, 21, 24, 31, 33, 35, 36, 37, 42, 45, 54, 57, 58, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 81, 85, 86, 87, 88, 89, 90, 91, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 144, 145, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 175, 183, 188, 189, 190, 195, 196, 197, 199, 201, 205, 209, 210, 212, 213, 214, 216, 219, 225, 226, 228, 229, 230, 237, 239, 243, 244, 246, 247, 250, 253, 256, 257, 262, 279, 281], "fals": [10, 13, 14, 15, 16, 18, 19, 22, 24, 33, 35, 36, 37, 42, 45, 53, 54, 58, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 75, 76, 81, 85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 145, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 175, 183, 189, 190, 192, 193, 194, 199, 200, 201, 205, 206, 209, 213, 225, 228, 229, 230, 241, 244, 256, 257, 272, 273, 274, 276, 278, 280, 281], "setup": [10, 22, 24, 25, 75, 182, 183, 188, 189, 190, 192, 193, 194, 199, 201, 233, 256, 272, 274, 278, 281], "expos": [10, 13, 24, 25, 230, 270, 275], "friendli": [10, 60, 64, 67, 71, 73, 273], "manner": [10, 20], "rather": [10, 221, 279], "everi": [10, 12, 22, 25, 62, 63, 67, 68, 69, 78, 79, 80, 146, 150, 191, 193, 250, 256, 266, 272, 279, 281], "construct": [10, 36, 63, 218, 270, 278], "care": [10, 22, 228, 230, 274, 276, 278], "how": [10, 13, 14, 18, 22, 23, 24, 25, 26, 191, 226, 247, 253, 265, 268, 269, 272, 273, 274, 275, 276, 279, 280, 281], "implement": [10, 19, 21, 22, 25, 55, 57, 59, 60, 61, 62, 63, 64, 67, 68, 69, 70, 71, 72, 180, 184, 185, 186, 191, 197, 204, 205, 211, 212, 217, 221, 222, 223, 224, 228, 235, 240, 250, 267, 269, 277, 278, 279, 280, 281], "llama3_2_vision_11b": 10, "custom_dataset": [10, 13], "sftdataset": [10, 13, 24, 55, 58, 59, 60, 62, 64, 65, 66, 68, 69], "packeddataset": [10, 17, 53, 58, 59, 60, 62, 64, 68, 69, 71, 72], "inputoutputtomessag": [10, 13, 14, 62, 68], "modeltoken": [10, 15, 21, 24, 36, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 67, 68, 69, 70, 71, 72, 216], "build": [10, 25, 71, 81, 91, 101, 116, 125, 146, 147, 150, 151, 158, 160, 175, 227, 267, 276, 278], "block": [10, 25, 54, 85, 86, 87, 91, 95, 96, 97, 101, 107, 108, 109, 110, 111, 116, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 146, 149, 150, 151, 154, 155, 156, 157, 158, 165, 166, 171, 172, 173, 174, 175, 183, 189, 190, 209, 210, 267], "tiny_cod": 10, "pack": [10, 54, 58, 59, 60, 62, 64, 65, 66, 68, 69, 71, 72, 183, 185, 189, 190, 199, 280], "subset": [10, 15, 16, 47, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 95, 107, 120, 129, 149, 154, 156, 165, 171, 202, 207], "nampdn": 10, "tini": 10, "respons": [10, 12, 13, 18, 19, 21, 32, 33, 35, 36, 37, 42, 55, 56, 60, 62, 63, 64, 65, 66, 67, 68, 69, 216, 219, 220, 221, 223, 224, 274, 275, 276], "model_transform": [10, 13, 15, 16, 55, 56, 62, 65, 66, 68, 69, 144], "message_transform": [10, 13, 55, 56], "column_map": [10, 12, 13, 16, 18, 31, 33, 35, 37, 42, 53, 58, 59, 62, 63, 64, 65, 66, 67, 68, 69, 70], "input": [10, 11, 12, 13, 14, 15, 20, 21, 22, 31, 35, 47, 48, 49, 50, 54, 55, 56, 59, 61, 62, 64, 65, 66, 68, 69, 72, 78, 79, 80, 81, 94, 106, 119, 142, 143, 144, 147, 148, 151, 162, 169, 171, 175, 180, 181, 183, 184, 185, 186, 187, 188, 189, 190, 191, 197, 198, 199, 200, 201, 205, 213, 214, 218, 228, 230, 238, 255, 258, 273, 278, 281], "filter_fn": [10, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72], "lambda": [10, 219], "x": [10, 22, 45, 73, 74, 75, 78, 79, 80, 142, 143, 180, 181, 183, 184, 185, 186, 188, 189, 190, 191, 199, 200, 201, 205, 241, 254, 277, 278, 280, 281], "split_across_pack": [10, 54, 71], "els": [10, 11, 12, 19, 25, 251, 267, 281], "posit": [10, 17, 24, 27, 54, 74, 76, 78, 79, 80, 81, 91, 95, 125, 129, 142, 147, 151, 154, 156, 158, 160, 165, 167, 182, 183, 185, 188, 189, 190, 191, 199, 200, 276], "automat": [10, 12, 16, 17, 19, 21, 23, 24, 26, 27, 59, 60, 272, 274, 281], "instanti": [10, 30, 38, 82, 83, 84, 85, 86, 87, 91, 92, 93, 94, 95, 96, 97, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 116, 117, 118, 119, 120, 121, 122, 125, 126, 127, 128, 129, 130, 131, 132, 136, 137, 138, 139, 145, 146, 147, 148, 149, 150, 151, 154, 155, 156, 157, 158, 159, 160, 161, 162, 165, 166, 167, 168, 169, 171, 172, 173, 174, 175, 176, 177, 178, 179, 232], "separ": [10, 55, 201, 216, 228, 273, 275, 276, 278, 281], "under": [10, 24, 256, 279, 281], "best": [10, 16, 18, 25, 269, 273, 277, 279], "root": [10, 184, 250, 251], "custom_finetun": 10, "32": [10, 27, 182, 191, 199, 201, 247, 276, 278, 279, 280, 281], "option": [10, 12, 18, 21, 22, 24, 25, 31, 33, 35, 37, 42, 50, 51, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 77, 80, 81, 85, 86, 87, 94, 95, 96, 97, 101, 106, 107, 108, 109, 110, 111, 116, 119, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 142, 143, 144, 146, 147, 148, 149, 150, 151, 154, 155, 156, 157, 162, 165, 166, 169, 171, 172, 173, 174, 175, 179, 180, 183, 185, 188, 189, 190, 191, 194, 195, 199, 209, 210, 211, 213, 216, 218, 219, 220, 222, 228, 229, 230, 232, 233, 236, 240, 247, 248, 251, 255, 256, 260, 261, 266, 267, 272, 273, 274], "param": [10, 22, 25, 85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 145, 149, 166, 172, 173, 174, 202, 203, 205, 207, 208, 210, 228, 278, 280, 281], "omit": [10, 277, 278, 279], "being": [10, 19, 22, 56, 193, 228, 229, 230, 234, 260, 279, 280, 281], "found": [10, 11, 22, 23, 24, 26, 184, 185, 228, 229, 230, 269, 272, 277, 278, 281], "correctli": [10, 21, 22, 25, 30, 209, 228, 266, 270, 273, 275, 281], "try": [10, 22, 24, 273, 274, 275, 276, 281], "after": [10, 19, 20, 23, 25, 38, 56, 65, 66, 94, 106, 119, 144, 148, 162, 169, 182, 183, 186, 187, 189, 190, 199, 201, 225, 246, 247, 248, 249, 250, 251, 269, 273, 274, 276, 280, 281], "pythonpath": 10, "pwd": 10, "vlm": [11, 16], "hub": [11, 22, 55, 56, 272, 275], "remot": [11, 14, 35, 42, 46, 55, 56], "url": [11, 16, 35, 37, 42, 46, 266], "project": [11, 23, 26, 81, 85, 86, 87, 91, 95, 101, 105, 107, 108, 109, 110, 111, 116, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 142, 143, 146, 147, 149, 150, 151, 154, 155, 156, 157, 158, 161, 165, 166, 171, 174, 175, 180, 183, 189, 191, 199, 203, 209, 210, 231, 237, 247, 251, 265, 278, 279, 281], "prefer": [11, 13, 25, 48, 55, 63, 67, 70, 221, 222, 223, 224, 267, 270, 272, 279], "align": [11, 65, 66, 221, 273, 277], "continu": [11, 20, 54, 191, 247], "pretrain": [11, 142, 143, 144, 199, 201, 203, 213, 214, 272, 273, 275, 278, 281], "beyond": [11, 274, 279, 281], "those": [11, 22, 231, 274, 276, 278], "customiz": 11, "task": [11, 12, 16, 18, 19, 34, 40, 43, 53, 61, 144, 268, 273, 274, 276, 277, 278, 279, 280, 281], "supervis": [11, 20, 56], "rlhf": [11, 55, 63, 219, 220, 221, 222, 223, 224, 225], "queri": [11, 91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 182, 183, 189, 190, 199, 276, 279], "time": [11, 16, 17, 22, 60, 64, 91, 158, 193, 196, 198, 216, 219, 248, 250, 256, 269, 272, 273, 274, 276, 281], "take": [11, 12, 13, 16, 18, 22, 24, 25, 27, 48, 55, 56, 65, 66, 67, 142, 182, 191, 195, 201, 228, 230, 259, 260, 269, 273, 274, 275, 276, 277, 278, 279, 281], "object": [11, 13, 14, 15, 19, 21, 24, 27, 28, 81, 183, 221, 224, 226, 240], "appli": [11, 12, 15, 19, 22, 25, 47, 55, 56, 59, 65, 66, 85, 86, 87, 88, 89, 90, 91, 95, 96, 97, 98, 99, 101, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 120, 121, 122, 123, 124, 125, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 146, 149, 150, 151, 152, 154, 155, 156, 157, 158, 163, 164, 165, 166, 170, 171, 172, 173, 174, 175, 183, 187, 188, 189, 190, 199, 209, 210, 253, 267, 268, 277, 279, 281], "templat": [11, 31, 32, 34, 38, 39, 40, 43, 55, 56, 59, 62, 68, 94, 100, 106, 119, 144, 148, 153, 162, 169, 179], "anyth": [11, 61, 259], "requir": [11, 15, 17, 19, 21, 22, 24, 47, 48, 53, 55, 56, 57, 65, 66, 67, 71, 144, 187, 189, 200, 228, 230, 232, 243, 244, 246, 247, 250, 251, 255, 256, 266, 269, 272, 273, 275, 279, 280, 281], "collat": [11, 47, 49, 50, 54], "packag": [11, 23, 26, 247, 250, 251, 266], "togeth": [11, 25, 54, 196, 251, 270, 275, 278, 279, 280], "form": [12, 18, 22, 24, 25, 31, 44, 52, 55, 56, 272], "along": [12, 22, 278], "describ": [12, 253], "hand": [12, 36], "grammar": [12, 19, 34, 62], "head": [12, 81, 91, 95, 101, 107, 116, 120, 125, 129, 142, 146, 147, 150, 151, 154, 156, 158, 160, 165, 167, 171, 175, 182, 183, 185, 189, 199, 203, 231, 276], "csv": [12, 55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72], "incorrect": [12, 19], "cat": [12, 16, 19, 218], "grammarerrorcorrectiontempl": [12, 19, 62], "prepend": [12, 14, 16, 19, 33, 35, 37, 38, 39, 42, 60, 62, 63, 64, 65, 66, 67, 68, 69, 94, 106, 119, 144, 148, 162, 169, 179, 213], "english": [12, 19, 34], "ncorrect": [12, 34], "mask": [12, 13, 14, 15, 17, 19, 21, 36, 38, 50, 54, 56, 59, 60, 62, 64, 65, 66, 67, 68, 69, 74, 75, 76, 144, 183, 188, 189, 190, 199, 212, 216, 218, 219, 222, 241, 273, 277], "out": [12, 15, 18, 20, 22, 24, 25, 59, 60, 62, 64, 67, 68, 69, 75, 76, 218, 228, 229, 241, 265, 267, 268, 269, 270, 272, 273, 274, 275, 276, 278, 279, 281], "100": [12, 18, 25, 48, 49, 50, 59, 60, 62, 64, 67, 68, 69, 73, 196, 197, 198, 200, 277, 278, 281], "27957": 12, "736": 12, "577": 12, "anoth": [12, 13, 16, 24, 56, 187, 247, 274, 279], "c4": [12, 71, 280], "200m": 12, "liweili": [12, 62], "c4_200m": [12, 62], "chang": [12, 13, 16, 21, 22, 23, 24, 26, 31, 33, 35, 64, 66, 70, 230, 266, 272, 274, 275, 276, 277, 278, 279, 280, 281], "remap": 12, "someth": [12, 22, 25, 26, 273, 274, 280], "hello": [12, 13, 14, 19, 21, 44, 213, 214, 261, 273, 274, 276], "world": [12, 13, 14, 19, 21, 44, 213, 214, 242, 244, 261, 274], "bye": [12, 13], "robot": [12, 15], "am": [12, 14, 16, 60, 64, 100, 153, 273, 274, 276], "prompttempl": [12, 31, 34, 40, 43, 144], "relev": [12, 14, 25, 188, 189, 190, 199, 272, 274, 278, 279], "inform": [12, 14, 22, 247, 251, 253, 267, 272, 274, 275], "mai": [12, 16, 17, 24, 26, 60, 73, 191, 194, 200, 237, 257, 268, 269, 273, 275, 277, 278, 279], "alpaca_dataset": [12, 17, 24, 58], "grammar_dataset": 12, "samsum_dataset": 12, "dictionari": [13, 14, 15, 36, 38, 44, 47, 48, 49, 54, 55, 56, 94, 106, 119, 148, 162, 169, 179, 239, 245, 247, 248, 249, 250, 251, 259, 274], "onc": [13, 21, 24, 38, 189, 199, 274, 275, 276, 278, 281], "repres": [13, 36, 48, 78, 79, 191, 227, 233, 273, 279, 280], "prepar": [13, 15, 273, 280], "ad": [13, 16, 19, 21, 25, 38, 50, 78, 79, 80, 146, 150, 160, 191, 199, 200, 203, 213, 216, 230, 231, 273, 278, 279, 280, 281], "column": [13, 16, 18, 20, 31, 33, 35, 37, 42, 55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 183, 189, 190, 199, 280], "worri": [13, 22, 273, 275], "itself": [13, 24], "do": [13, 15, 18, 21, 22, 23, 25, 36, 47, 65, 67, 193, 209, 216, 247, 251, 257, 272, 274, 275, 276, 278, 279, 280], "well": [13, 18, 22, 24, 25, 267, 272, 274, 276, 277, 279, 281], "flexibl": [13, 24, 53, 279], "inherit": [13, 14, 19, 25, 267], "__call__": [13, 15, 19, 65, 66, 144], "simpl": [13, 22, 25, 191, 224, 265, 275, 278, 280, 281], "contriv": [13, 19], "would": [13, 15, 19, 22, 24, 26, 38, 54, 189, 191, 199, 266, 273, 274, 278, 279, 281], "inde": [13, 236, 274], "quit": [13, 279, 281], "type": [13, 14, 15, 16, 21, 26, 27, 29, 36, 37, 44, 45, 46, 47, 48, 49, 50, 51, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 91, 92, 93, 94, 95, 96, 97, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 116, 117, 118, 119, 120, 121, 122, 125, 126, 127, 128, 129, 130, 131, 132, 136, 137, 138, 139, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 154, 155, 156, 157, 158, 159, 160, 161, 162, 165, 166, 167, 168, 169, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 188, 189, 190, 191, 194, 195, 196, 197, 198, 199, 200, 201, 202, 205, 207, 211, 212, 213, 214, 215, 216, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, 229, 230, 231, 232, 234, 236, 237, 238, 239, 240, 241, 242, 243, 244, 246, 253, 254, 255, 256, 258, 260, 261, 262, 269, 274, 278, 279, 280, 281], "map": [13, 15, 19, 21, 22, 31, 33, 35, 37, 38, 42, 47, 53, 54, 59, 62, 63, 64, 65, 66, 67, 68, 69, 70, 94, 106, 119, 144, 148, 162, 169, 179, 208, 214, 215, 228, 232, 234, 247, 248, 249, 250, 251, 252, 256, 274, 278], "messagetransform": 13, "self": [13, 15, 18, 19, 20, 21, 25, 26, 54, 65, 66, 85, 86, 87, 91, 95, 96, 97, 101, 107, 108, 109, 110, 111, 116, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 146, 149, 150, 151, 154, 155, 156, 157, 158, 160, 165, 166, 167, 171, 172, 173, 174, 175, 183, 188, 189, 190, 196, 198, 199, 201, 204, 209, 210, 228, 231, 232, 277, 278, 281], "str": [13, 15, 21, 24, 27, 28, 31, 33, 35, 36, 37, 38, 42, 44, 46, 47, 48, 49, 50, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 94, 106, 119, 144, 148, 149, 152, 162, 169, 179, 195, 200, 201, 202, 204, 205, 207, 208, 209, 210, 211, 212, 213, 214, 215, 227, 228, 229, 230, 231, 232, 233, 236, 239, 240, 243, 245, 247, 248, 249, 250, 251, 255, 256, 257, 258, 260, 261, 262, 279], "eot": [13, 14, 19, 36, 144], "_messag": 13, "0x7fb0a10094e0": 13, "0x7fb0a100a290": 13, "msg": [13, 14, 16, 19, 21, 273], "text_cont": [13, 14, 16, 19, 36, 273], "manipul": 13, "load_dataset_kwarg": [13, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72], "mymessagetransform": 13, "chosenrejectedtomessag": [13, 63, 67], "core": [14, 25, 55, 56, 267, 270, 275, 281], "govern": [14, 273], "serv": [14, 19, 24, 33, 35, 37, 42, 60, 62, 63, 64, 65, 66, 67, 68, 69, 216, 226, 278], "interfac": [14, 25, 38, 39, 53, 204, 217], "api": [14, 25, 26, 34, 40, 43, 55, 56, 59, 65, 66, 88, 89, 90, 98, 99, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 152, 163, 164, 170, 209, 247, 266, 270, 272, 273, 275, 276, 281], "oper": [14, 25, 191, 206, 217, 255, 280], "send": 14, "other": [14, 15, 18, 20, 22, 25, 27, 35, 38, 53, 230, 237, 256, 259, 268, 269, 273, 275, 276, 277, 278, 279, 280], "special": [14, 16, 19, 36, 42, 119, 144, 146, 148, 150, 169, 179, 191, 200, 211, 212, 214, 215, 216, 218, 232], "individu": [14, 36, 54, 199, 239, 251, 253, 273], "ref": [14, 55, 56, 59, 65, 66, 168, 169, 251], "constructor": [14, 21], "ident": [14, 18, 20, 45, 47, 54, 65, 67, 153, 189, 274, 280], "from_dict": [14, 36, 273], "becaus": [14, 21, 55, 56, 95, 182, 189, 191, 199, 230, 272, 273, 280], "correspond": [14, 18, 21, 36, 48, 74, 75, 76, 202, 204, 207, 219, 222, 236, 269, 275, 276, 279, 280], "begin": [14, 22, 54, 71, 191, 214, 216, 273, 276, 281], "pil": [14, 15, 16, 36, 37, 44, 46], "img_msg": 14, "place": [14, 16, 20, 257, 273, 279], "mode": [14, 15, 16, 194, 233, 240, 247, 274], "rgb": [14, 15, 16, 142], "4": [14, 15, 16, 22, 24, 45, 47, 48, 49, 50, 76, 81, 144, 147, 151, 182, 183, 191, 241, 262, 267, 269, 272, 274, 276, 277, 278, 279, 280, 281], "appropri": [14, 36, 53, 76, 100, 200, 228, 235, 281], "load_imag": [14, 16], "image_path": [14, 16], "jpg": [14, 16, 35, 42, 46], "tag": [14, 16, 19, 21, 38, 42, 44, 94, 100, 106, 119, 144, 148, 153, 162, 169, 179, 247, 248, 249, 250, 251, 273], "placehold": [14, 16, 42, 227], "should": [14, 15, 16, 18, 20, 22, 24, 25, 33, 35, 36, 37, 38, 42, 47, 54, 59, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 75, 76, 85, 86, 87, 95, 96, 97, 100, 101, 107, 108, 109, 110, 111, 116, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 142, 146, 149, 150, 151, 153, 154, 155, 156, 157, 158, 160, 165, 166, 167, 171, 172, 173, 174, 175, 182, 183, 189, 191, 194, 199, 204, 209, 210, 219, 222, 226, 227, 245, 247, 248, 249, 250, 251, 266, 267, 274, 275, 276, 277, 278, 279, 280, 281], "insert": [14, 201, 280], "format_content_with_imag": [14, 16], "image_tag": [14, 16, 42, 44], "conveni": [14, 24, 25, 46, 272], "prompttemplateinterfac": [14, 19, 94, 106, 119, 148, 162, 169, 179], "templated_msg": [14, 19], "contains_media": [14, 16, 36], "get_media": [14, 15, 16, 36], "4x4": 14, "0x7f8d27e72740": 14, "tokenize_messsag": 14, "hi": [14, 20, 73, 273], "tokenize_messag": [14, 15, 21, 36, 55, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 144, 212, 216, 273], "22557": 14, "1526": [14, 21], "28808": 14, "28705": [14, 21], "28748": [14, 21], "15359": 14, "28725": 14, "315": [14, 20], "837": 14, "396": 14, "16107": 14, "13892": 14, "28723": 14, "modal": [15, 16, 56, 144, 201], "current": [15, 16, 18, 22, 35, 42, 54, 67, 75, 91, 95, 107, 120, 129, 149, 150, 151, 154, 156, 158, 165, 168, 171, 182, 183, 185, 189, 190, 199, 222, 229, 230, 232, 237, 240, 242, 248, 250, 252, 255, 269, 270, 275, 276, 277, 279, 280], "intend": [15, 259, 273], "drop": [15, 144, 200, 277, 280], "replac": [15, 16, 42, 51, 59, 60, 62, 64, 67, 68, 69, 144, 195, 200, 257, 278], "llama3_2_vis": [15, 16, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152], "llama3visiontransform": [15, 16, 148], "__init__": [15, 24, 25, 65, 66, 277, 278, 281], "transform_imag": 15, "clipimagetransform": [15, 65, 66, 144, 191], "xattn_mask": 15, "visioncrossattentionmask": [15, 144, 217], "224": [15, 16, 144], "tile_s": [15, 79, 80, 81, 144, 147, 151, 191, 218], "patch_siz": [15, 79, 80, 81, 144, 147, 151, 191, 218], "14": [15, 48, 144, 191, 280, 281], "skip_special_token": [15, 16, 67, 144], "begin_of_text": [15, 16, 21, 273], "start_header_id": [15, 16, 273], "end_header_id": [15, 16, 273], "n": [15, 16, 18, 19, 21, 34, 38, 40, 43, 183, 191, 216, 264, 271, 272, 273, 280], "eot_id": [15, 16, 21, 273], "na": [15, 273], "encoder_input": [15, 16, 50, 188, 189, 199], "shape": [15, 16, 22, 47, 50, 73, 74, 75, 76, 78, 79, 80, 81, 142, 143, 144, 147, 151, 180, 181, 182, 183, 184, 185, 186, 188, 189, 190, 191, 196, 197, 198, 199, 200, 201, 205, 218, 219, 220, 221, 222, 223, 224, 225, 241, 256, 257, 277], "num_til": [15, 16, 142, 143, 191], "num_channel": [15, 16, 191], "tile_height": [15, 16], "tile_width": [15, 16], "torch": [15, 16, 22, 24, 45, 47, 48, 49, 50, 73, 74, 75, 76, 77, 78, 79, 80, 81, 142, 143, 144, 180, 181, 182, 183, 184, 185, 186, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 205, 219, 220, 221, 222, 223, 224, 225, 230, 232, 234, 235, 236, 238, 239, 241, 243, 244, 250, 252, 253, 254, 255, 256, 257, 258, 259, 260, 262, 266, 269, 274, 275, 276, 277, 278, 279, 281], "just": [15, 19, 22, 267, 269, 272, 273, 275, 276, 278, 279, 280], "the_cauldron_dataset": [15, 16], "ai2d": [15, 66], "respir": 15, "combust": 15, "give": [15, 21, 24, 227, 277, 278, 279], "choic": [15, 18], "oxygen": 15, "b": [15, 25, 45, 47, 142, 143, 182, 183, 185, 189, 190, 199, 205, 219, 220, 224, 241, 251, 278, 281], "carbon": 15, "dioxid": 15, "c": [15, 45, 47, 50, 65, 142, 273], "nitrogen": 15, "d": [15, 24, 36, 65, 142, 143, 182, 183, 189, 199, 272, 273, 277, 278, 280], "heat": 15, "letter": 15, "mymultimodaltransform": 15, "my_tokenizer_build": 15, "myimagetransform": 15, "add_eo": [15, 57, 71, 144, 213, 214, 273], "tupl": [15, 19, 21, 24, 27, 38, 48, 73, 74, 80, 94, 106, 119, 144, 148, 162, 169, 179, 182, 191, 195, 212, 216, 219, 220, 221, 222, 223, 224, 225, 226, 242, 256, 257, 258], "infer": [15, 19, 22, 50, 56, 91, 100, 158, 182, 183, 185, 189, 190, 199, 218, 260, 265, 269, 270, 273, 274, 275, 276, 280, 281], "vision": [15, 16, 56, 81, 142, 144, 145, 146, 147, 148, 149, 150, 151, 152, 200, 231], "aspect_ratio": [15, 50, 78, 79, 142, 191], "append": [15, 19, 38, 39, 94, 106, 119, 144, 148, 162, 169, 179, 189, 199, 213, 247, 266], "addit": [15, 21, 22, 24, 25, 27, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 144, 146, 147, 150, 151, 153, 193, 200, 201, 209, 221, 226, 228, 229, 230, 236, 237, 243, 246, 247, 248, 250, 251, 253, 267, 273, 275, 278, 279], "kei": [15, 21, 22, 24, 26, 33, 35, 37, 42, 47, 48, 55, 56, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 182, 183, 188, 189, 190, 199, 201, 208, 209, 210, 224, 228, 230, 232, 247, 256, 272, 274, 275, 278, 279, 281], "e": [16, 18, 19, 36, 46, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 78, 79, 80, 81, 142, 143, 144, 147, 151, 183, 191, 195, 199, 204, 208, 218, 227, 228, 232, 239, 256, 260, 266, 269, 274, 276, 278, 279, 280, 281], "g": [16, 18, 46, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 78, 79, 80, 81, 142, 143, 144, 147, 151, 183, 191, 199, 204, 218, 227, 228, 239, 256, 260, 269, 276, 278, 279, 280, 281], "base": [16, 18, 20, 22, 27, 36, 38, 85, 86, 87, 88, 89, 90, 91, 95, 96, 97, 98, 99, 101, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 125, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 144, 145, 148, 149, 150, 151, 152, 154, 155, 156, 157, 158, 160, 163, 164, 165, 166, 167, 170, 171, 172, 173, 174, 175, 185, 205, 206, 208, 209, 210, 220, 221, 223, 224, 228, 235, 237, 238, 246, 248, 257, 260, 265, 273, 274, 275, 276, 277, 278, 279, 281], "multimodal_chat_dataset": 16, "visual": [16, 201], "get": [16, 22, 23, 24, 25, 26, 50, 144, 232, 236, 239, 242, 247, 261, 266, 267, 268, 269, 273, 274, 275, 277, 278, 279, 280], "below": [16, 23, 26, 47, 226, 276, 277, 278, 281], "clock": 16, "10": [16, 45, 47, 48, 49, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 191, 200, 274, 276, 279, 280, 281], "llama3_2_vision_transform": 16, "questionanswertempl": [16, 19, 70], "image_s": [16, 145, 148, 149, 152, 191], "560": [16, 145, 148, 149, 152], "image_dir": [16, 35, 42, 65], "home": [16, 24, 35, 42, 46], "nquestion": 16, "nit": 16, "00am": 16, "sharegpt": [16, 42, 60, 273], "q1": [16, 33, 55, 60, 67], "a1": [16, 33, 55, 60], "sharegpt4v": 16, "lin": 16, "chen": 16, "renam": 16, "themselv": [16, 281], "pathlib": 16, "pil_imag": 16, "Then": [16, 20, 26, 206, 275, 277, 279], "relat": [16, 188, 189, 199, 278], "user_messag": [16, 34, 40, 43, 144, 273], "locat": [16, 21, 24, 35, 42, 272, 276, 278, 280, 281], "long": [16, 54, 214, 273, 278], "image_dog": 16, "image_cat": 16, "image_bird": 16, "dog": [16, 218], "bird": [16, 46], "pet": 16, "three": [16, 22, 25, 50, 144, 221, 223, 224, 270, 275], "referenc": 16, "huggingfac": [16, 57, 61, 63, 71, 72, 161, 168, 169, 176, 177, 178, 221, 223, 224, 228, 229, 235, 272, 274], "co": [16, 57, 61, 63, 71, 72, 161, 168, 169, 176, 177, 178, 228, 229, 274], "img": 16, "llava_instruct_dataset": 16, "concaten": [17, 21, 48, 53, 147, 151, 212, 216], "sequenc": [17, 45, 47, 48, 49, 50, 54, 57, 61, 65, 66, 71, 72, 75, 76, 91, 94, 95, 101, 106, 107, 116, 119, 120, 125, 129, 142, 143, 144, 146, 148, 150, 154, 156, 158, 160, 162, 165, 167, 169, 171, 175, 179, 182, 183, 185, 188, 189, 190, 191, 194, 199, 201, 214, 216, 218, 220, 224, 225, 241, 273], "upto": [17, 185], "maximum": [17, 24, 47, 50, 51, 54, 61, 72, 75, 78, 79, 81, 91, 94, 95, 101, 106, 107, 116, 119, 120, 125, 129, 144, 146, 147, 148, 150, 151, 154, 156, 158, 160, 162, 165, 167, 169, 171, 175, 182, 183, 185, 188, 189, 190, 194, 199, 201, 218, 227, 272], "length": [17, 45, 47, 49, 50, 51, 52, 53, 54, 61, 72, 75, 91, 94, 95, 101, 106, 107, 116, 119, 120, 125, 129, 142, 143, 144, 146, 148, 150, 154, 156, 158, 160, 162, 165, 167, 168, 169, 171, 175, 179, 182, 183, 185, 188, 189, 190, 194, 196, 198, 199, 201, 214, 218, 219, 220, 229, 241, 247, 279], "slow": [17, 279, 281], "down": [17, 191, 230, 278, 279, 281], "introduc": [17, 85, 86, 95, 96, 97, 107, 108, 109, 110, 120, 121, 122, 131, 132, 138, 139, 154, 155, 156, 157, 165, 166, 183, 184, 201, 205, 224, 269, 273, 277, 278, 279, 280, 281], "signific": [17, 279, 280], "speedup": [17, 274, 276], "depend": [17, 25, 26, 228, 256, 272, 274, 277, 278, 279, 281], "iter": [17, 256, 257, 258, 281], "through": [17, 18, 22, 23, 24, 25, 26, 55, 81, 147, 151, 180, 182, 191, 201, 206, 267, 268, 269, 270, 272, 273, 274, 275, 277, 279, 280, 281], "greedi": [17, 54], "upon": [17, 25, 53, 189, 193, 199, 276], "initi": [17, 22, 25, 29, 53, 54, 82, 83, 84, 92, 93, 102, 103, 104, 105, 117, 118, 126, 127, 128, 136, 137, 159, 161, 176, 177, 178, 221, 232, 243, 244, 257, 269, 275, 278, 281], "max": [17, 50, 54, 179, 189, 191, 199, 214, 227, 235, 272, 278], "llama3": [17, 20, 21, 24, 65, 66, 73, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 144, 146, 148, 149, 150, 152, 165, 196, 198, 231, 237, 265, 267, 268, 269, 272, 274, 279], "load": [17, 22, 25, 35, 42, 46, 53, 54, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 199, 209, 228, 229, 230, 232, 250, 257, 274, 276, 278], "isinst": [17, 226], "1b_full_single_devic": 17, "prevent": [17, 22, 54, 221, 272, 279], "irrelev": 17, "cross": [17, 50, 54, 146, 150, 188, 196, 198, 199, 201, 218, 277], "attend": [17, 54, 183, 188, 189, 190, 199, 218], "pytorch": [17, 24, 25, 74, 189, 195, 196, 226, 244, 250, 253, 255, 256, 265, 266, 267, 269, 274, 276, 278, 279, 280, 281], "flex": 17, "attent": [17, 50, 54, 74, 75, 76, 81, 85, 86, 87, 91, 95, 96, 97, 101, 107, 108, 109, 110, 111, 116, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 146, 147, 149, 150, 151, 154, 155, 156, 157, 158, 160, 165, 166, 167, 168, 171, 172, 173, 174, 175, 182, 183, 185, 188, 189, 190, 192, 199, 201, 209, 210, 218, 276, 278, 279, 281], "flash": 17, "non": [17, 197, 198, 210, 220, 277], "causal": [17, 54, 75, 183, 189, 190, 199], "hardwar": [17, 236, 267, 274, 278, 279], "cuda": [17, 24, 236, 239, 256, 260, 266, 274, 279, 281], "ture": 17, "sdpa": 17, "memori": [17, 21, 25, 53, 54, 57, 61, 71, 72, 187, 189, 195, 196, 198, 199, 209, 237, 239, 245, 246, 256, 265, 267, 268, 269, 274, 275, 276, 277, 280], "effici": [17, 209, 237, 265, 267, 268, 274, 275, 278, 280], "fallback": 17, "while": [17, 24, 25, 85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 172, 173, 174, 200, 267, 269, 274, 279, 280, 281], "retain": [17, 221, 279, 281], "reward": [18, 105, 111, 115, 157, 161, 164, 219, 220, 221, 223, 224, 231], "downstream": 18, "captur": 18, "ground": [18, 196, 197, 198, 279], "truth": [18, 24, 196, 197, 198, 274, 276], "usual": [18, 21, 22, 185, 189, 225, 228, 241, 251, 272, 274, 278, 279], "outcom": 18, "binari": 18, "comparison": [18, 25, 278, 281], "annot": 18, "accord": [18, 19, 65, 66, 76, 153, 273], "criterion": 18, "style": [18, 31, 54, 58, 59, 60, 69, 201, 281], "interact": [18, 25, 55, 67, 265, 270, 275], "free": [18, 224, 270, 278], "preference_dataset": 18, "my_preference_dataset": [18, 67], "chosen_convers": [18, 67], "hole": [18, 67], "my": [18, 19, 23, 67, 73, 272, 273, 274, 276], "trouser": [18, 67], "fix": [18, 20, 67, 280], "rejected_convers": [18, 67], "off": [18, 25, 38, 67, 268, 269, 274, 280], "chosen": [18, 33, 55, 63, 67, 70, 221, 223, 224, 256], "reject": [18, 33, 55, 63, 67, 70, 221, 223, 224], "rejected_input_id": [18, 48, 67], "nwhat": 18, "ntake": 18, "rejected_label": [18, 48], "128006": 18, "78191": 18, "128007": 18, "271": 18, "18293": 18, "1124": 18, "1022": 18, "13": [18, 20, 21, 48, 191, 216, 225, 281], "128009": [18, 273], "accomplish": [18, 20, 53, 60, 64, 67, 71], "shown": [18, 274, 280], "di": 18, "look": [18, 19, 22, 24, 25, 234, 250, 266, 273, 274, 275, 276, 277, 278, 280], "anthrop": [18, 63], "harmless": [18, 63], "granni": 18, "her": [18, 20], "mobil": [18, 274], "phone": [18, 274], "issu": [18, 270, 280], "grandmoth": 18, "manag": [18, 22, 53, 193, 194, 206, 247, 254, 273], "behavior": [18, 22, 246, 273], "thing": [18, 279, 281], "grandma": 18, "feel": [18, 270, 278], "box": [18, 267, 269, 281], "hh_rlhf_helpful_dataset": 18, "hendrydong": 18, "preference_700k": 18, "stack_exchange_paired_dataset": 18, "purpos": [19, 65, 66, 275, 276], "whenev": [19, 144, 196, 278], "llama2": [19, 22, 24, 25, 27, 61, 72, 82, 83, 84, 85, 86, 87, 88, 89, 90, 100, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 144, 180, 189, 190, 231, 265, 268, 272, 275, 276, 279, 280], "were": [19, 20, 21, 53, 142, 191, 206, 222, 275, 280], "gear": [19, 144], "summar": [19, 43, 68, 273, 279], "summarizetempl": [19, 68, 273], "commun": [19, 144, 274, 279], "chatmltempl": [19, 144, 179], "gec_templ": 19, "extend": [19, 21, 22, 25, 267], "customprompttempl": 19, "achiev": [19, 38, 252, 269, 274, 276, 277, 278, 280, 281], "prepend_tag": [19, 38], "append_tag": [19, 38], "thu": [19, 31, 38, 55, 56, 189, 279, 280], "empti": [19, 47, 50, 52, 77, 272], "standalon": [19, 182], "my_custom_templ": 19, "Is": 19, "overhyp": 19, "advanc": [19, 79, 80, 81, 147, 151, 191], "configur": [19, 21, 25, 55, 56, 59, 60, 61, 62, 64, 65, 66, 67, 68, 69, 70, 71, 72, 95, 107, 120, 129, 149, 154, 165, 171, 247, 267, 269, 270, 273, 275, 276, 277, 278, 279, 280, 281], "doesn": [19, 274], "neatli": 19, "fall": 19, "protocol": [19, 21, 204, 211, 212, 217], "arg": [19, 21, 24, 27, 39, 80, 181, 189, 195, 201, 204, 211, 212, 217, 249, 256, 269, 280], "whether": [19, 31, 33, 35, 36, 37, 42, 47, 50, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 85, 86, 87, 91, 95, 96, 97, 107, 108, 109, 110, 111, 120, 121, 122, 129, 130, 131, 132, 138, 139, 144, 145, 149, 150, 151, 154, 155, 156, 157, 165, 166, 171, 172, 173, 174, 175, 195, 197, 199, 205, 209, 210, 213, 214, 226, 236, 238, 239, 247, 257, 273, 277], "sai": [19, 272, 273, 275], "eureka": 19, "eurekatempl": 19, "formatted_dialogu": 19, "llama2chattempl": [19, 106, 153, 179, 273], "paradigm": [20, 25, 268, 279], "unstructur": [20, 57, 71, 72], "unlabel": 20, "text_complet": 20, "odyssei": 20, "clear": 20, "river": 20, "oceanu": 20, "had": 20, "got": [20, 50], "sea": 20, "went": 20, "till": 20, "reach": 20, "aeaean": 20, "island": 20, "dawn": 20, "sunris": 20, "drew": 20, "ship": 20, "sand": 20, "shore": 20, "sleep": 20, "wait": [20, 256], "break": [20, 144, 214], "child": 20, "morn": 20, "rosi": 20, "finger": 20, "appear": [20, 279], "sent": [20, 251], "men": 20, "circ": 20, "hous": 20, "fetch": [20, 278], "bodi": 20, "elpenor": 20, "cut": 20, "firewood": 20, "wood": 20, "headland": 20, "jut": 20, "wept": 20, "over": [20, 21, 25, 36, 56, 197, 198, 221, 235, 267, 269, 272, 274, 277, 278, 279, 281], "him": 20, "lament": 20, "funer": 20, "rite": 20, "armour": 20, "been": [20, 73, 75, 182, 189, 199, 225, 232, 237, 273, 280], "burn": 20, "ash": 20, "rais": [20, 22, 27, 30, 33, 35, 37, 42, 44, 46, 47, 50, 52, 53, 59, 60, 62, 64, 65, 66, 68, 69, 71, 75, 81, 171, 182, 183, 187, 188, 189, 191, 192, 193, 194, 209, 210, 216, 228, 229, 230, 232, 236, 238, 239, 243, 247, 251, 255, 257, 258, 259], "cairn": 20, "stone": 20, "top": [20, 74, 77, 147, 151, 234, 279, 281], "oar": 20, "he": 20, "row": [20, 55, 56, 183, 189, 190, 199], "text_completion_dataset": [20, 280], "128000": [20, 273, 280], "6153": 20, "584": 20, "1051": 20, "2867": 20, "279": 20, "15140": 20, "22302": 20, "355": 20, "11": [20, 22, 45, 47, 48, 191, 274, 280, 281], "323": 20, "1047": 20, "2751": 20, "704": 20, "1139": 20, "1825": 20, "9581": 20, "4024": 20, "389": 20, "12222": 20, "8813": 20, "362": 20, "12791": 20, "5420": 20, "13218": 20, "1405": 20, "1070": 20, "374": 20, "39493": 20, "64919": 20, "439": 20, "304": 20, "1023": 20, "7634": 20, "1226": 20, "1243": 20, "24465": 20, "1057": 20, "8448": 20, "311": 20, "70163": 20, "1077": 20, "31284": 20, "6212": 20, "30315": 20, "1938": 20, "1288": 20, "1464": 20, "128001": [20, 280], "similarli": [20, 119, 148, 169, 179, 280], "wikimedia": 20, "wikipedia": [20, 46, 72], "cnn_dailymail_articles_dataset": 20, "index": [21, 48, 49, 50, 53, 54, 183, 185, 190, 197, 199, 220, 235, 260, 266, 273, 274], "embed": [21, 22, 78, 79, 80, 81, 91, 95, 101, 107, 116, 120, 125, 129, 142, 143, 144, 146, 147, 150, 151, 154, 156, 158, 160, 165, 167, 171, 175, 182, 183, 184, 185, 188, 189, 191, 199, 200, 201, 203, 237, 273, 276, 279, 280], "vector": [21, 223, 273], "understood": 21, "plai": [21, 274, 279], "necessari": [21, 22, 55, 56, 247, 248, 249, 250, 251, 273, 278], "phi3": [21, 22, 165, 166, 168, 169, 170, 231, 272], "phi3_mini_token": 21, "p_token": 21, "phi": [21, 168, 169, 231], "32010": 21, "29871": 21, "1792": [21, 216], "9508": [21, 216], "32007": 21, "32001": 21, "4299": 21, "2933": [21, 216], "nuser": 21, "nmodel": 21, "sentencepiec": [21, 213, 276], "tiktoken": [21, 144, 214, 276], "host": [21, 266, 272, 275, 279], "distribut": [21, 77, 232, 243, 244, 253, 255, 260, 267, 270, 272, 275, 276, 277, 279], "alongsid": [21, 237], "alreadi": [21, 24, 33, 37, 42, 62, 63, 65, 66, 67, 68, 69, 182, 183, 193, 194, 199, 231, 243, 266, 272, 274, 277, 278], "_token": [21, 25], "mistraltoken": [21, 162, 273], "adher": [21, 37, 42], "arbitrarili": 21, "small": [21, 184, 274], "seq": [21, 189, 199], "len": [21, 22, 50, 53, 59, 62, 65, 66, 68, 189, 191, 199], "demonstr": [21, 279, 280], "7": [21, 22, 45, 47, 48, 49, 50, 182, 191, 218, 222], "6312": 21, "28709": 21, "assign": [21, 24, 55, 56], "uniqu": [21, 55, 56, 231], "abil": 21, "NOT": [21, 22, 91, 144, 158], "presenc": [21, 31], "certain": [21, 22, 24, 256, 273], "proper": [21, 266, 275], "end_of_text": 21, "special_token": [21, 144, 214, 273], "added_token": 21, "128257": 21, "128258": 21, "remain": [21, 37, 42, 235, 277, 278, 279], "special_tokens_path": [21, 119, 148, 169, 179], "basetoken": 21, "actual": [21, 23, 24, 26, 31, 33, 35, 55, 56, 59, 62, 63, 64, 66, 67, 68, 70, 144, 269, 273, 280], "string": [21, 22, 35, 36, 38, 44, 60, 61, 94, 106, 119, 144, 148, 162, 169, 179, 204, 211, 213, 214, 216, 227, 233, 236, 240, 247, 260, 272, 279], "kwarg": [21, 24, 27, 39, 179, 181, 188, 190, 195, 201, 204, 211, 212, 217, 243, 247, 248, 249, 250, 251, 253, 256], "dict": [21, 22, 24, 25, 26, 27, 31, 33, 35, 36, 37, 38, 42, 44, 47, 48, 49, 50, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 94, 106, 119, 144, 148, 162, 169, 179, 188, 190, 195, 199, 201, 202, 207, 208, 209, 210, 211, 212, 214, 215, 217, 228, 229, 230, 232, 234, 239, 243, 245, 247, 252, 257, 259], "given": [21, 25, 27, 44, 47, 52, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 76, 144, 193, 194, 205, 206, 211, 212, 220, 236, 240, 246, 252, 260, 262, 267, 278], "token_id": [21, 144, 211, 214], "its": [21, 54, 100, 153, 156, 183, 185, 189, 190, 199, 201, 232, 252, 255, 272, 273, 274, 276, 278, 279], "sentencepiecebasetoken": [21, 211], "bpe": 21, "sp_token": 21, "reason": [21, 25, 73, 274, 279, 280], "walk": [22, 25, 250, 267, 273, 274, 275, 280, 281], "design": [22, 25, 224], "cover": [22, 23, 24, 25, 26, 273, 274, 281], "scenario": [22, 53, 144], "compos": [22, 191], "plug": [22, 279], "evalu": [22, 25, 265, 267, 269, 270, 275, 277, 278, 281], "gener": [22, 25, 47, 54, 61, 71, 74, 75, 76, 77, 144, 193, 194, 206, 219, 238, 247, 254, 255, 256, 263, 265, 269, 273, 277, 278, 279, 280, 281], "easi": [22, 25, 267, 278, 279], "understand": [22, 24, 25, 201, 265, 267, 268, 273, 278, 279, 281], "concept": [22, 270, 274, 275, 279], "talk": 22, "close": [22, 25, 247, 248, 249, 250, 251, 278], "veri": [22, 53, 189, 199, 272, 274, 279], "dictat": 22, "state_dict": [22, 195, 200, 201, 209, 228, 229, 230, 231, 232, 257, 278, 281], "store": [22, 55, 56, 247, 248, 251, 278, 279, 281], "disk": [22, 57, 248], "identifi": [22, 247], "state": [22, 25, 143, 189, 191, 193, 195, 199, 202, 207, 208, 209, 210, 219, 221, 228, 229, 230, 232, 234, 257, 274, 276, 278, 279, 281], "match": [22, 44, 210, 247, 257, 266, 272, 274, 276, 278], "up": [22, 23, 25, 26, 50, 54, 61, 72, 144, 189, 193, 194, 199, 214, 218, 234, 247, 256, 268, 269, 270, 272, 273, 275, 276, 278, 279, 281], "exactli": [22, 210, 227, 280], "definit": [22, 278], "either": [22, 47, 55, 56, 73, 183, 189, 190, 210, 228, 247, 253, 266, 272, 278, 279, 280, 281], "explicit": 22, "error": [22, 24, 34, 52, 228, 255, 272], "except": [22, 36, 153, 216, 277], "wors": [22, 279], "silent": 22, "succe": 22, "popular": [22, 199, 267, 274], "offici": [22, 100, 273, 275, 276], "websit": 22, "inspect": [22, 274, 278, 281], "mmap": [22, 274], "weights_onli": [22, 230], "map_loc": [22, 274], "cpu": [22, 25, 194, 195, 236, 256, 260, 266, 272, 274, 279, 281], "tensor": [22, 45, 47, 48, 49, 50, 73, 74, 75, 76, 77, 78, 79, 80, 81, 142, 143, 180, 181, 182, 183, 184, 185, 186, 188, 189, 190, 191, 195, 196, 197, 198, 199, 200, 201, 205, 219, 220, 221, 222, 223, 224, 225, 228, 241, 247, 248, 249, 250, 251, 254, 257, 259, 277, 278, 279, 281], "item": 22, "f": [22, 26, 59, 62, 65, 66, 68, 227, 273, 274, 277, 278, 281], "tok_embed": [22, 189, 199, 200], "32000": [22, 27, 278], "4096": [22, 27, 61, 72, 183, 185, 278, 280], "292": 22, "tabl": [22, 200, 273, 274, 276, 277, 279, 281], "layer": [22, 25, 81, 85, 86, 87, 88, 89, 90, 91, 95, 96, 97, 98, 99, 101, 105, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 120, 121, 122, 123, 124, 125, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 143, 146, 147, 149, 150, 151, 152, 154, 155, 156, 157, 158, 160, 161, 163, 164, 165, 166, 167, 170, 171, 172, 173, 174, 175, 183, 186, 187, 188, 189, 190, 191, 192, 193, 194, 199, 201, 203, 205, 209, 210, 226, 233, 237, 267, 268, 276, 278, 279, 280, 281], "dim": [22, 50, 142, 143, 180, 183, 184, 185, 189, 196, 198, 199, 277], "within": [22, 24, 27, 54, 73, 77, 78, 95, 107, 120, 129, 149, 150, 151, 154, 156, 165, 171, 191, 193, 194, 250, 255, 256, 272, 278, 281], "big": 22, "bin": [22, 272, 274], "piec": 22, "pytorch_model": [22, 274], "00001": [22, 272, 277], "00002": [22, 272, 277], "embed_token": 22, "241": 22, "Not": 22, "fewer": [22, 183], "sinc": [22, 24, 27, 55, 56, 198, 228, 230, 273, 274, 276, 279, 280], "mismatch": 22, "caus": [22, 213], "re": [22, 24, 193, 201, 224, 230, 267, 268, 269, 273, 274, 275, 278, 279], "end": [22, 25, 36, 57, 71, 144, 214, 216, 265, 267, 273, 276, 278, 280], "number": [22, 25, 44, 50, 54, 61, 72, 73, 78, 79, 81, 91, 95, 101, 107, 116, 120, 125, 129, 142, 143, 144, 146, 147, 150, 151, 154, 156, 158, 160, 165, 167, 171, 175, 182, 183, 189, 191, 197, 198, 218, 228, 229, 230, 232, 233, 235, 242, 255, 256, 272, 275, 277, 278, 279], "save": [22, 25, 26, 189, 195, 196, 198, 199, 228, 229, 230, 232, 237, 246, 251, 265, 269, 272, 273, 274, 276, 278, 279, 280], "less": [22, 50, 73, 274, 275, 276, 279, 281], "prone": 22, "invari": 22, "accept": [22, 24, 226, 275, 279, 281], "explicitli": [22, 204, 267, 278], "produc": [22, 232, 269, 280, 281], "One": [22, 50, 280], "advantag": [22, 219, 222, 269, 278], "abl": [22, 25, 274, 275, 280], "post": [22, 191, 252, 256, 269, 274, 276, 280, 281], "quantiz": [22, 85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 205, 230, 240, 265, 266, 268, 270, 275, 281], "eval": [22, 265, 267, 280], "without": [22, 24, 26, 183, 187, 189, 193, 199, 209, 266, 267, 269, 273, 274, 278, 279, 280], "OR": 22, "surround": [22, 25, 267], "load_checkpoint": [22, 25, 228, 229, 230, 231], "save_checkpoint": [22, 25, 26, 228, 229, 230], "permut": 22, "behav": 22, "further": [22, 191, 224, 272, 277, 278, 279, 280, 281], "illustr": [22, 65, 66, 276], "whilst": [22, 268, 279], "read": [22, 228, 229, 230, 267, 279], "compat": [22, 228, 230, 279, 280], "framework": [22, 25, 267], "mention": [22, 274, 281], "assum": [22, 35, 42, 45, 47, 65, 94, 106, 119, 148, 162, 169, 179, 182, 183, 185, 190, 199, 200, 202, 207, 214, 232, 234, 235, 236, 238, 273, 274, 278], "checkpoint_dir": [22, 24, 228, 229, 230, 274, 276, 277, 280], "easiest": [22, 274, 275], "everyth": [22, 25, 267, 270, 275], "flow": [22, 54, 280, 281], "safetensor": [22, 227, 228, 272, 277], "output_dir": [22, 24, 228, 229, 230, 256, 274, 276, 278, 280, 281], "snippet": 22, "explain": [22, 279], "fullmodelhfcheckpoint": [22, 274, 277], "sort": [22, 228, 230], "order": [22, 23, 25, 228, 230, 250, 251, 275, 279], "matter": [22, 228, 230, 272, 278], "checkpoint_fil": [22, 24, 26, 228, 229, 230, 274, 276, 277, 278, 280, 281], "restart": [22, 272], "previou": [22, 54, 228, 229, 230, 277], "section": [22, 25, 239, 265, 274, 276, 279, 281], "recipe_checkpoint": [22, 228, 229, 230, 280], "model_typ": [22, 228, 229, 230, 274, 276, 280], "resume_from_checkpoint": [22, 228, 229, 230], "discrep": [22, 228], "github": [22, 27, 74, 85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 172, 173, 174, 184, 185, 196, 197, 221, 222, 223, 224, 235, 266, 274, 276, 277], "repositori": [22, 55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 100, 268, 269, 274, 275], "fullmodelmetacheckpoint": [22, 276, 280], "test": [22, 24, 25, 73, 267, 269, 273, 279], "written": [22, 24, 25, 228, 229, 247, 248, 249, 250, 251, 267], "partit": [22, 228, 281], "key_1": [22, 230], "weight_1": 22, "key_2": 22, "weight_2": 22, "mid": 22, "chekpoint": 22, "middl": [22, 201, 274, 279], "subsequ": [22, 25, 182, 189, 191, 218], "recipe_st": [22, 228, 229, 230], "pt": [22, 26, 228, 229, 230, 274, 276, 277, 280], "epoch": [22, 25, 26, 228, 229, 230, 232, 235, 272, 273, 274, 275, 276, 280], "etc": [22, 25, 143, 228, 239, 275], "flood": 22, "overwritten": 22, "updat": [22, 24, 25, 38, 182, 183, 189, 199, 217, 221, 222, 228, 232, 256, 259, 266, 273, 274, 275, 276, 278, 279, 280, 281], "hf_model_0001_0": [22, 274, 277], "hf_model_0002_0": [22, 274], "adapt": [22, 85, 86, 95, 96, 97, 107, 108, 109, 110, 120, 121, 122, 131, 132, 138, 139, 143, 154, 155, 156, 157, 165, 166, 199, 201, 202, 204, 205, 206, 207, 208, 228, 229, 230, 246, 268, 273, 274, 278, 281], "merg": [22, 27, 28, 179, 228, 274, 276, 281], "tutori": [22, 253, 267, 268, 269, 273, 274, 275, 276, 277, 278, 279, 280, 281], "save_adapter_weights_onli": 22, "choos": [22, 60, 278], "resum": [22, 25, 228, 229, 230, 235, 281], "frozen": [22, 143, 149, 152, 200, 221, 278, 279, 281], "learnt": [22, 273, 274], "refer": [22, 24, 25, 184, 185, 187, 191, 196, 206, 220, 221, 222, 223, 224, 247, 267, 278, 279, 280], "adapter_checkpoint": [22, 228, 229, 230], "adapter_0": [22, 274], "knowledg": [22, 265], "forward": [22, 25, 78, 79, 80, 142, 143, 180, 181, 183, 184, 185, 186, 188, 189, 190, 191, 193, 194, 196, 197, 198, 199, 200, 201, 205, 221, 222, 223, 224, 239, 256, 276, 277, 278, 279, 281], "modeltyp": [22, 228, 229, 230], "llama2_13b": [22, 108], "right": [22, 47, 50, 76, 189, 228, 274, 276, 278], "pytorch_fil": 22, "00003": [22, 227, 277], "torchtune_sd": 22, "load_state_dict": [22, 199, 200, 201, 209, 232, 257, 278], "successfulli": [22, 272, 275], "vocab": [22, 27, 179, 189, 199, 200, 276], "70": [22, 117], "randint": 22, "no_grad": 22, "6": [22, 45, 47, 48, 49, 50, 54, 91, 95, 184, 191, 241, 269, 280, 281], "3989": 22, "9": [22, 45, 47, 48, 50, 182, 191, 241, 274, 280, 281], "0531": 22, "2375": 22, "5": [22, 24, 45, 47, 48, 49, 50, 75, 191, 221, 224, 225, 235, 274, 275, 276, 277, 279], "2822": 22, "4872": 22, "7469": 22, "8": [22, 45, 47, 48, 50, 59, 62, 65, 66, 68, 85, 86, 87, 88, 89, 90, 96, 97, 98, 99, 108, 109, 110, 111, 112, 113, 114, 115, 121, 122, 123, 124, 125, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 155, 157, 163, 164, 166, 170, 172, 173, 174, 182, 191, 196, 198, 274, 277, 278, 279, 280, 281], "6737": 22, "0023": 22, "8235": 22, "6819": 22, "2424": 22, "0109": 22, "6915": 22, "3618": 22, "1628": 22, "8594": 22, "5857": 22, "1151": 22, "7808": 22, "2322": 22, "8850": 22, "9604": 22, "7624": 22, "6040": 22, "3159": 22, "5849": 22, "8039": 22, "9322": 22, "2010": [22, 191], "6824": 22, "8929": 22, "8465": 22, "3794": 22, "3500": 22, "6145": 22, "5931": 22, "find": [22, 23, 25, 26, 221, 272, 274, 275, 277, 278, 279], "hope": 22, "deeper": [22, 268, 269, 275, 279], "insight": [22, 274], "happi": [22, 274], "cometlogg": 23, "checkpoint": [23, 24, 25, 195, 199, 201, 214, 227, 228, 229, 230, 231, 232, 233, 251, 253, 257, 267, 269, 272, 276, 277, 278, 280, 281], "workspac": [23, 26, 247], "seen": [23, 26, 278, 281], "screenshot": [23, 26], "comet_ml": [23, 247], "featur": [23, 25, 26, 266, 267, 268, 269, 274, 275, 279], "pip": [23, 26, 247, 250, 251, 266, 274, 276, 279], "login": [23, 26, 247, 251, 272, 274], "metric_logg": [23, 24, 25, 26], "metric_log": [23, 24, 26, 247, 248, 249, 250, 251], "experiment_nam": [23, 247], "experi": [23, 24, 247, 251, 265, 267, 276, 277, 278], "grab": [23, 26, 276], "tab": [23, 26], "asset": 23, "artifact": [23, 26, 256], "click": [23, 26], "effect": [24, 224, 277, 279, 280], "prerequisit": [24, 273, 274, 275, 276, 277, 278, 280, 281], "Be": [24, 273, 274, 275, 276, 277, 278, 279, 280, 281], "familiar": [24, 273, 274, 275, 276, 277, 278, 280, 281], "fundament": [24, 280], "reproduc": [24, 247], "overridden": [24, 256], "quick": 24, "seed": [24, 25, 26, 255, 275, 280], "shuffl": [24, 54, 280], "dtype": [24, 25, 77, 182, 183, 188, 189, 190, 192, 193, 194, 195, 199, 201, 236, 254, 258, 274, 277, 279, 280, 281], "fp32": [24, 189, 196, 198, 279, 280, 281], "enable_fsdp": 24, "keyword": [24, 27, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 195, 273], "subfield": 24, "dotpath": [24, 94, 106, 119, 148, 162, 169, 179], "wish": [24, 182, 193, 257], "exact": [24, 27, 274], "normal": [24, 54, 144, 181, 183, 184, 188, 189, 190, 196, 197, 198, 213, 273, 278, 280, 281], "instanc": [24, 27, 53, 107, 120, 129, 149, 154, 156, 165, 171, 172, 173, 176, 177, 195, 202, 207, 208, 278], "preced": [24, 27, 272, 276, 278], "throw": 24, "notic": [24, 78, 79, 80, 191, 273, 278], "miss": [24, 209, 210, 256, 278], "llama2_token": [24, 273, 274], "llama2token": [24, 106], "512": [24, 81, 281], "overwrit": [24, 230, 257, 266, 272], "duplic": [24, 25, 267, 272], "sometim": 24, "resolv": [24, 28, 275], "alpaca": [24, 31, 53, 58, 59, 85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 172, 173, 174, 277], "disklogg": 24, "log_dir": [24, 248, 250, 251], "verifi": [24, 236, 237, 260, 273, 275, 278], "properli": [24, 209, 244, 272], "wa": [24, 35, 42, 50, 75, 79, 80, 81, 147, 151, 191, 209, 273, 278, 280, 281], "7b_lora_single_devic": [24, 274, 275, 278, 281], "my_config": [24, 272], "guidelin": 24, "tempt": 24, "put": [24, 25, 270, 275, 278, 280], "much": [24, 200, 224, 274, 276, 278, 280, 281], "switch": 24, "encourag": [24, 224, 278, 279], "clariti": 24, "significantli": [24, 221, 268, 269, 279], "easier": [24, 274, 275], "dont": 24, "privat": 24, "parent": [24, 272], "guarante": 24, "stabil": [24, 196, 198, 267, 269, 279, 280, 281], "underscor": 24, "_alpaca": 24, "k1": [24, 25], "v1": [24, 25, 72], "k2": [24, 25], "v2": [24, 25, 247], "my_model_checkpoint": 24, "file_1": 24, "file_2": 24, "my_tokenizer_path": 24, "nest": [24, 259], "dot": 24, "notat": [24, 50, 142, 143, 183, 185, 189, 199, 219, 220, 241], "flag": [24, 25, 36, 59, 60, 62, 64, 67, 68, 69, 226, 230, 237, 272, 279, 281], "bitsandbyt": [24, 279], "pagedadamw8bit": [24, 279], "delet": [24, 189, 192, 193, 194, 199], "foreach": 24, "8b_full": [24, 272], "adamw": [24, 278, 279], "2e": [24, 279], "fuse": [24, 146, 150, 199, 200, 201, 202, 252, 280], "nproc_per_nod": [24, 269, 276, 278, 280], "full_finetune_distribut": [24, 238, 272, 274, 275], "thought": [25, 267, 270, 275, 281], "target": [25, 75, 197, 198, 224, 267, 277], "pipelin": [25, 267, 269], "eg": [25, 189, 199, 228, 267], "meaning": [25, 267, 274], "fsdp": [25, 187, 226, 232, 237, 246, 275, 276], "activ": [25, 81, 180, 233, 239, 245, 253, 256, 267, 269, 280, 281], "gradient": [25, 197, 198, 246, 252, 256, 267, 269, 274, 276, 278, 281], "accumul": [25, 252, 256, 267, 269], "mix": [25, 181, 272, 274, 279], "precis": [25, 181, 195, 236, 267, 269, 275, 281], "complex": 25, "becom": [25, 191, 266], "harder": 25, "anticip": 25, "methodolog": 25, "possibl": [25, 54, 227, 272, 279], "trade": [25, 279], "vs": [25, 275], "qualiti": [25, 274, 278, 280], "believ": 25, "suit": [25, 275, 279], "solut": 25, "result": [25, 65, 81, 147, 151, 191, 198, 216, 218, 256, 269, 274, 276, 277, 278, 279, 280, 281], "meant": [25, 195, 232], "expertis": 25, "routin": 25, "yourself": [25, 272, 276, 278], "exist": [25, 194, 201, 232, 247, 266, 272, 274, 275, 276, 281], "ones": [25, 50, 182], "modular": [25, 267], "wandb": [25, 26, 251, 275], "log": [25, 28, 221, 222, 223, 224, 239, 245, 247, 248, 249, 250, 251, 261, 274, 275, 276, 277, 278, 279, 281], "fulli": [25, 53, 149], "nativ": [25, 265, 267, 278, 280, 281], "numer": [25, 66, 267, 269, 280], "pariti": [25, 267], "verif": [25, 184], "benchmark": [25, 255, 267, 274, 276, 278, 280], "limit": [25, 232, 277, 279, 280], "hidden": [25, 81, 143, 147, 151, 180, 189, 191], "behind": 25, "unnecessari": 25, "abstract": [25, 211, 212, 267, 275, 281], "No": [25, 230, 267], "go": [25, 81, 100, 147, 151, 153, 191, 216, 267, 274, 275, 277, 279, 281], "figur": [25, 278, 281], "spectrum": 25, "decid": 25, "avail": [25, 35, 42, 72, 199, 201, 236, 244, 260, 267, 272, 274, 276, 278, 279], "consist": [25, 33, 37, 42, 65, 66, 72, 270, 275], "overrid": [25, 28, 29, 33, 37, 42, 62, 63, 65, 66, 67, 68, 69, 257, 270, 272, 274, 275, 276, 277, 281], "valid": [25, 52, 76, 197, 209, 210, 220, 238, 257, 258, 266, 270, 274, 275], "closer": [25, 277, 278], "monolith": [25, 267], "trainer": [25, 221, 223, 224], "wrapper": [25, 181, 213, 214, 232, 234, 272, 278], "around": [25, 144, 181, 213, 214, 239, 272, 273, 274, 278, 279, 280, 281], "extern": 25, "primarili": [25, 53, 278], "eleutherai": [25, 72, 267, 277, 278, 280], "har": [25, 267, 277, 278, 280], "stage": [25, 191], "distil": [25, 265], "dataload": [25, 54, 59, 62, 65, 66, 68], "applic": [25, 228, 229, 251], "clean": [25, 26, 58, 277], "group": [25, 183, 242, 243, 247, 248, 249, 250, 251, 272, 276, 280], "init_process_group": [25, 243], "backend": [25, 272, 280], "gloo": 25, "nccl": 25, "fullfinetunerecipedistribut": 25, "cleanup": 25, "stuff": 25, "carri": [25, 56], "metric": [25, 275, 277, 279, 280], "logger": [25, 245, 247, 248, 249, 250, 251, 261, 275], "_devic": 25, "get_devic": 25, "_dtype": 25, "get_dtyp": 25, "ckpt_dict": 25, "wrap": [25, 201, 226, 233, 237, 246, 253, 273], "_model": [25, 232], "_setup_model": 25, "_setup_token": 25, "_optim": 25, "_setup_optim": 25, "_loss_fn": 25, "_setup_loss": 25, "_sampler": 25, "_dataload": 25, "_setup_data": 25, "backward": [25, 232, 234, 252, 256, 281], "zero_grad": 25, "curr_epoch": 25, "rang": [25, 200, 221, 222, 224, 255, 272, 276, 280], "epochs_run": [25, 26], "total_epoch": [25, 26], "idx": [25, 54], "enumer": 25, "_autocast": 25, "logit": [25, 73, 74, 77, 196, 197, 198, 241, 277], "global_step": 25, "_log_every_n_step": 25, "_metric_logg": 25, "log_dict": [25, 247, 248, 249, 250, 251], "step": [25, 54, 55, 56, 65, 66, 189, 199, 219, 232, 234, 235, 247, 248, 249, 250, 251, 252, 256, 265, 269, 274, 278, 280, 281], "recipe_main": [25, 29], "fullfinetunerecip": 25, "wandblogg": [26, 278, 281], "tip": 26, "straggler": 26, "background": 26, "crash": 26, "otherwis": [26, 45, 47, 50, 79, 80, 81, 147, 151, 187, 189, 191, 244, 247, 273, 280], "exit": [26, 193, 194, 206, 266, 272], "resourc": [26, 247, 248, 249, 250, 251, 279, 280], "kill": 26, "ps": 26, "aux": 26, "grep": 26, "awk": 26, "xarg": 26, "desir": [26, 55, 56, 254, 273, 279], "suggest": [26, 277], "approach": [26, 53, 277], "full_finetun": 26, "joinpath": 26, "_checkpoint": [26, 274], "_output_dir": [26, 228, 229, 230], "torchtune_model_": 26, "with_suffix": 26, "wandb_at": 26, "descript": [26, 272], "whatev": 26, "metadata": [26, 280], "seed_kei": 26, "epochs_kei": 26, "total_epochs_kei": 26, "max_steps_kei": 26, "max_steps_per_epoch": [26, 280], "add_fil": 26, "log_artifact": 26, "hydra": 27, "facebook": 27, "research": 27, "com": [27, 74, 85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 172, 173, 174, 184, 185, 196, 197, 221, 222, 223, 224, 235, 247, 266, 274, 276, 277], "facebookresearch": [27, 184], "blob": [27, 74, 85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 169, 172, 173, 174, 184, 185, 197, 221, 222, 223, 224, 235], "_intern": 27, "_instantiate2": 27, "l148": 27, "num_head": [27, 81, 91, 95, 101, 107, 116, 120, 125, 129, 146, 147, 150, 151, 154, 156, 158, 160, 165, 167, 171, 175, 182, 183, 185, 189], "num_kv_head": [27, 91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 182, 183], "vocab_s": [27, 73, 74, 91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 196, 197, 198, 200], "parsed_yaml": 27, "embed_dim": [27, 78, 79, 80, 81, 91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 183, 185, 188, 189, 190, 191, 200, 201, 257, 278], "valueerror": [27, 33, 35, 37, 42, 44, 46, 47, 50, 52, 53, 59, 60, 62, 64, 65, 66, 68, 69, 71, 171, 182, 183, 191, 192, 193, 194, 228, 229, 230, 236, 239, 255, 258], "recipe_nam": 28, "rank": [28, 85, 86, 87, 95, 96, 97, 107, 108, 109, 110, 111, 120, 121, 122, 129, 130, 131, 132, 138, 139, 149, 150, 151, 154, 155, 156, 157, 165, 166, 171, 172, 173, 174, 205, 242, 244, 255, 268, 275, 278, 281], "zero": [28, 50, 182, 184, 189, 199, 227, 274, 276, 280], "displai": 28, "callabl": [29, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 81, 189, 206, 226, 237, 240, 246, 253], "With": [29, 274, 277, 278, 280, 281], "my_recip": 29, "foo": 29, "bar": [29, 267, 275, 279], "configerror": 30, "cannot": [30, 46, 230, 276], "equival": [31, 35, 79, 223, 224], "condit": [31, 73, 244, 272], "dedic": 31, "due": [31, 213, 278, 279, 281], "keep": [31, 33, 35, 37, 42, 63, 64, 66, 67, 70, 200, 274, 278, 279], "openai": [32, 37, 60, 222], "markup": 32, "im_start": 32, "context": [32, 168, 193, 194, 206, 254, 256, 279], "im_end": 32, "goe": [32, 206], "a2": [33, 55], "functool": [34, 40, 43, 226], "partial": [34, 40, 43, 226], "_prompt_templ": [34, 40, 43], "assistant_messag": [34, 40, 43], "respect": [35, 53, 100, 182, 208, 256, 273], "final": [35, 42, 55, 56, 85, 86, 87, 91, 95, 101, 107, 108, 109, 110, 111, 116, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 146, 147, 149, 150, 151, 154, 155, 156, 157, 158, 165, 166, 171, 174, 175, 180, 189, 199, 209, 210, 274, 276, 277, 278, 279, 281], "leav": [35, 42, 279], "liter": [36, 38, 41, 85, 86, 87, 88, 89, 90, 94, 95, 96, 97, 98, 99, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 119, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 148, 149, 150, 151, 152, 154, 155, 156, 157, 162, 163, 164, 165, 166, 169, 170, 171, 172, 173, 174, 179, 209, 210], "union": [36, 46, 47, 58, 59, 60, 62, 64, 68, 69, 71, 72, 94, 106, 119, 148, 162, 169, 179, 189, 199, 210, 228, 233, 238, 247, 248, 249, 250, 251, 253, 255], "interleav": [36, 218], "attach": 36, "writer": 36, "calcul": [36, 38, 76, 142, 144, 183, 188, 190, 191, 219, 220, 222, 276], "consecut": [36, 52, 182, 218], "last": [36, 51, 54, 71, 189, 220, 232, 235], "properti": [36, 278, 279], "media": [36, 56], "classmethod": 36, "image_url": 37, "unmask": [37, 42, 197], "consid": [38, 53, 55, 56, 79, 80, 81, 147, 151, 191, 279], "come": [38, 52, 204, 278, 279], "nanswer": 40, "alia": [41, 226], "alwai": [42, 247, 257, 273, 279], "nsummari": [43, 273], "summari": [43, 53, 68, 191, 239], "batch_first": 45, "padding_valu": 45, "float": [45, 73, 74, 77, 85, 86, 87, 88, 89, 90, 91, 95, 96, 97, 98, 99, 101, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 120, 121, 122, 123, 124, 125, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 144, 149, 150, 151, 152, 154, 155, 156, 157, 158, 160, 163, 164, 165, 166, 167, 170, 171, 172, 173, 174, 175, 183, 184, 205, 219, 220, 221, 222, 223, 224, 232, 235, 238, 239, 245, 247, 248, 249, 250, 251, 278, 279, 280, 281], "rnn": [45, 47, 50], "pad_sequ": [45, 47, 50], "variabl": [45, 231, 244, 247, 279, 281], "left": [45, 47, 50, 144, 189, 278], "longest": [45, 49, 50], "trail": 45, "dimens": [45, 50, 91, 95, 101, 107, 116, 120, 125, 129, 142, 146, 147, 150, 151, 154, 156, 158, 160, 165, 167, 171, 175, 180, 182, 183, 185, 189, 191, 200, 205, 276, 278, 279, 281], "element": [45, 47, 50, 53, 197, 241, 274], "12": [45, 47, 48, 69, 191, 266, 280], "image_loc": 46, "www": [46, 247], "org": [46, 65, 82, 83, 84, 85, 86, 88, 89, 90, 95, 96, 97, 98, 99, 102, 103, 104, 105, 107, 108, 109, 110, 112, 113, 114, 115, 120, 121, 122, 123, 124, 131, 132, 133, 134, 135, 138, 139, 140, 141, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 183, 184, 185, 191, 218, 219, 221, 222, 223, 224, 226, 244, 250, 253, 255, 261, 266], "en": [46, 57, 61, 63, 71, 72, 280], "pad_direct": [47, 50], "keys_to_pad": 47, "padding_idx": [47, 48, 49, 50, 54], "left_pad_sequ": [47, 50], "integ": [47, 49, 200, 226, 227, 233, 255], "batch_siz": [47, 59, 62, 65, 66, 68, 182, 183, 188, 189, 190, 192, 193, 194, 196, 197, 198, 199, 200, 201, 221, 223, 225, 274, 279, 280], "ignore_idx": [48, 49, 50], "input_id": [48, 241], "chosen_input_id": [48, 67], "chosen_label": 48, "15": [48, 191, 237, 273, 274, 278, 281], "16": [48, 85, 86, 87, 88, 89, 90, 96, 97, 98, 99, 108, 109, 110, 111, 112, 113, 114, 115, 121, 122, 123, 124, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 155, 157, 163, 164, 166, 170, 172, 173, 174, 182, 191, 278, 281], "17": [48, 191, 278], "18": [48, 191, 276], "19": [48, 191, 281], "20": [48, 191, 225, 280], "token_pair": 49, "padded_col": 49, "pad_max_til": 50, "pad_max_imag": 50, "tile": [50, 78, 79, 80, 81, 142, 143, 144, 145, 147, 148, 149, 151, 191, 218], "aspect": [50, 78, 79, 267], "ratio": [50, 78, 79, 221, 222], "text_seq_len": [50, 218], "n_tile": [50, 78, 79, 191], "h": [50, 142, 182, 191, 196, 198, 266, 272], "w": [50, 82, 83, 84, 92, 93, 102, 103, 104, 105, 117, 118, 126, 127, 128, 136, 137, 142, 159, 161, 176, 177, 178, 191, 247, 250, 251, 273, 274, 278, 281], "h_ratio": 50, "w_ratio": 50, "encoder_mask": [50, 188, 189, 199], "image_seq_len": [50, 218], "channel": [50, 81, 142, 144, 147, 151, 191, 280], "height": [50, 142], "largest": 50, "bsz": [50, 73, 74, 75, 76, 78, 79, 191, 196, 198], "max_num_imag": 50, "max_num_til": [50, 78, 79, 81, 144, 147, 151, 191, 218], "tokens_per_til": 50, "image_id": 50, "four": [50, 278], "model_input": 50, "max_text_seq_len": 50, "40": [50, 79, 80, 81, 147, 151, 191, 218, 279, 281], "did": [50, 276, 281], "extra": [50, 144, 199, 266, 273, 278, 279, 280, 281], "second": [50, 183, 200, 274, 278, 279, 281], "eos_id": [51, 144, 214, 216], "shorter": [52, 189], "min": [52, 278], "invalid": 52, "sub": [53, 250], "unifi": [53, 161], "simplifi": [53, 221, 272, 277, 278], "simultan": 53, "intern": 53, "aggreg": 53, "transpar": 53, "howev": [53, 169, 266, 277, 279], "constitu": 53, "might": [53, 193, 200, 203, 272, 274, 279], "larg": [53, 196, 198, 205, 256, 272, 279, 281], "comput": [53, 55, 56, 101, 107, 116, 120, 125, 129, 142, 143, 146, 150, 171, 175, 183, 185, 189, 190, 196, 198, 199, 218, 221, 223, 224, 239, 255, 269, 274, 277, 279, 280, 281], "cumul": 53, "maintain": [53, 201, 268, 279, 281], "deleg": 53, "retriev": [53, 55, 56, 189, 237], "lead": [53, 213, 227, 269], "scale": [53, 73, 74, 77, 85, 86, 87, 95, 96, 97, 107, 108, 109, 110, 111, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 149, 150, 151, 154, 155, 156, 157, 165, 166, 171, 172, 173, 174, 184, 186, 188, 190, 205, 220, 224, 278, 279, 280, 281], "strategi": [53, 269], "stream": [53, 261, 279], "demand": 53, "deriv": [53, 180, 189, 190], "instans": 53, "dataset1": 53, "mycustomdataset": 53, "params1": 53, "dataset2": 53, "params2": 53, "concat_dataset": 53, "total": [53, 220, 222, 235, 242, 264, 271, 274, 276, 277, 278, 279], "data_point": 53, "1500": 53, "vicgal": 53, "gpt4": 53, "samsum": [53, 68], "focus": [53, 270, 275, 279], "enhanc": [53, 191, 224, 279, 281], "divers": 53, "machin": [53, 223, 260, 272, 274], "max_pack": 54, "outsid": [54, 255, 256, 278], "sampler": [54, 275], "part": [54, 200, 223, 273, 281], "buffer": [54, 189, 199, 279], "enough": [54, 273], "lower": [54, 269, 277, 278], "triangular": 54, "wise": 54, "made": [54, 60, 64, 67, 71, 144, 274], "smaller": [54, 200, 274, 276, 277, 278, 279, 280, 281], "jam": 54, "s1": [54, 213], "s2": [54, 213], "s3": 54, "s4": 54, "contamin": 54, "input_po": [54, 74, 183, 185, 189, 190, 199], "matrix": [54, 188, 189, 199], "increment": 54, "move": [54, 71, 189, 259, 279], "entir": [54, 71, 196, 203, 246, 273, 281], "avoid": [54, 71, 184, 191, 195, 255, 272, 280, 281], "truncat": [54, 61, 71, 72, 94, 106, 119, 144, 148, 162, 169, 179, 214, 225], "sentenc": [54, 71], "techniqu": [55, 267, 268, 269, 274, 275, 276, 277, 278, 279, 280], "repons": 55, "At": [55, 56, 189, 199], "extract": [55, 56, 61, 215], "against": [55, 56, 224, 262, 280, 281], "unit": [55, 56, 246, 267], "filepath": [55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72], "filter": [55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 280], "prior": [55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 257], "doc": [55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 226, 244, 247, 250, 251, 255, 261, 272, 274], "round": [56, 280], "incorpor": [56, 221], "happen": [56, 196, 198], "ti": [56, 95, 171, 175, 187, 279], "agnost": 56, "treat": [56, 191, 206, 273], "minimum": [56, 65, 66], "corpu": [57, 61, 71, 72], "package_refer": [57, 61, 63, 71, 72], "loading_method": [57, 61, 63, 71, 72], "tabular": [57, 71], "txt": [57, 71, 179, 248, 275], "eo": [57, 71, 169, 213, 216, 273], "yahma": 58, "variant": [58, 62, 68], "page": [58, 72, 266, 267, 272, 275, 276, 279], "tatsu": 59, "lab": [59, 74], "codebas": [59, 274], "independ": 59, "contribut": [59, 60, 62, 64, 67, 68, 69, 197, 198, 220, 222], "alpacatomessag": 59, "alpaca_d": 59, "altern": [60, 64, 67, 193, 275, 279], "toward": [60, 224], "my_dataset": [60, 64], "london": [60, 64], "ccdv": 61, "cnn_dailymail": 61, "textcompletiondataset": [61, 71, 72], "cnn": 61, "dailymail": 61, "articl": [61, 72], "highlight": [61, 281], "disabl": [61, 72, 189, 193, 199, 206, 255, 280], "highest": [61, 72], "conjunct": [62, 68, 70, 189, 279], "grammar_d": 62, "rlhflow": 63, "hh": 63, "preferencedataset": [63, 67, 70], "liuhaotian": 65, "llava": 65, "150k": 65, "coco": 65, "train2017": 65, "llava_instruct_150k": 65, "2017": 65, "visit": [65, 274], "cocodataset": 65, "wget": 65, "zip": [65, 263], "unzip": 65, "minim": [65, 66, 275, 277, 278, 280, 281], "clip": [65, 66, 78, 79, 80, 81, 142, 143, 144, 147, 151, 191, 222], "mymodeltransform": [65, 66], "tokenizer_path": [65, 66], "image_transform": [65, 66], "yet": [65, 66, 153, 273, 274], "llava_instruct_d": 65, "huggingfacem4": 66, "the_cauldron": 66, "cauldron": 66, "card": 66, "cauldron_d": 66, "compris": 67, "share": [67, 183, 187, 274], "c1": 67, "r1": 67, "chosen_messag": 67, "rejected_messag": 67, "samsung": 68, "samsum_d": 68, "351": 69, "82": 69, "391": 69, "221": 69, "220": 69, "193": 69, "471": 69, "lvwerra": 70, "stack": [70, 191, 256], "exchang": 70, "allenai": [71, 280], "data_dir": 71, "realnewslik": 71, "wikitext_document_level": 72, "wikitext": [72, 280], "103": [72, 274], "transformerdecod": [73, 74, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 95, 96, 97, 98, 99, 101, 102, 103, 104, 105, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 146, 150, 154, 155, 156, 157, 158, 159, 160, 161, 163, 164, 165, 166, 167, 168, 170, 171, 172, 173, 174, 175, 176, 177, 178, 188, 190, 199, 200, 201, 278], "max_generated_token": 73, "pad_id": [73, 225], "temperatur": [73, 74, 77, 221, 223, 224, 274], "top_k": [73, 74, 77, 274], "stop_token": [73, 225], "rng": 73, "custom_generate_next_token": 73, "seq_length": [73, 74, 75, 188, 190, 200, 201], "prune": [73, 77, 281], "probabl": [73, 77, 85, 86, 87, 95, 96, 97, 107, 109, 110, 111, 120, 121, 122, 129, 130, 131, 132, 138, 139, 149, 150, 151, 154, 155, 156, 157, 165, 166, 171, 172, 173, 174, 205, 221, 222, 223, 224, 274, 277], "stop": [73, 225], "random": [73, 191, 255, 275], "compil": [73, 196, 274, 276, 279, 281], "generate_next_token": 73, "llama3_8b": [73, 122, 130, 199, 276, 279, 280], "manual_se": 73, "tolist": 73, "jeremi": 73, "m": [73, 195, 273, 280], "seq_len": [73, 75, 76, 189], "num_generated_token": 73, "q": [74, 77, 183, 278], "randomli": [74, 77, 257], "softmax": [74, 77, 183, 189, 190, 199, 277], "trick": [74, 77], "fast": [74, 274], "32971d3129541c5bfb4f715abc33d1c5f408d204": 74, "l40": 74, "k": [74, 77, 79, 183, 278], "padding_mask": [75, 76, 222, 225], "target_seq_len": 75, "suitabl": 75, "scaled_dot_product_attent": [75, 91, 95, 101, 107, 116, 120, 125, 129, 154, 156, 158, 160, 165, 167, 171, 175, 183], "static": 75, "kv": [75, 182, 183, 189, 190, 192, 193, 194, 199, 280], "cach": [75, 182, 183, 185, 188, 189, 190, 192, 193, 194, 199, 201, 266, 272], "longer": [75, 182, 279], "boolean": [75, 76, 81, 183, 188, 189, 190, 199, 201, 226, 241], "assertionerror": [75, 81, 182, 188, 189, 209, 210, 257], "shift": [76, 189], "uniform_": 77, "int32": 77, "patch": [78, 79, 80, 81, 143, 144, 147, 151, 191, 218], "check": [78, 79, 80, 81, 188, 189, 190, 191, 199, 201, 209, 236, 244, 262, 265, 267, 268, 269, 270, 273, 274, 275, 278, 279], "vision_transform": [78, 79, 80, 81], "visiontransform": [78, 79, 80, 81], "divid": [78, 79, 80, 81, 144, 147, 151, 191, 197, 198, 218], "dimension": [78, 79, 80, 81, 147, 151, 191], "n_img": [78, 79, 191], "n_tokens_per_til": [78, 79, 80], "crop": [78, 79, 80, 81, 142, 147, 151, 191], "local_token_positional_embed": 79, "_position_embed": [79, 191], "tokenpositionalembed": [79, 191], "gate": [79, 186, 231, 268, 269, 272, 275], "global_token_positional_embed": 79, "400": [79, 80, 81, 147, 151, 191, 218], "10x10": [79, 80, 81, 147, 151, 191, 218], "grid": [79, 80, 81, 147, 151, 191, 218], "th": [79, 182], "silu": [81, 180], "cls_output_dim": [81, 191], "attn_bia": 81, "out_indic": [81, 191], "output_cls_project": 81, "in_channel": [81, 147, 151, 191], "intermediate_act": 81, "transformerencoderlay": 81, "cl": [81, 143, 191], "mlp": [81, 85, 86, 87, 91, 95, 96, 97, 101, 107, 108, 109, 110, 111, 116, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 146, 149, 150, 151, 154, 155, 156, 157, 158, 160, 165, 166, 167, 171, 172, 173, 174, 175, 188, 189, 190, 209, 210, 276, 278, 279], "bia": [81, 187, 204, 205, 257, 278, 280, 281], "intermedi": [81, 91, 95, 101, 107, 116, 120, 125, 129, 146, 147, 150, 151, 154, 156, 158, 160, 165, 167, 171, 175, 191, 230, 253, 276, 281], "fourth": [81, 147, 151, 191], "determin": [81, 147, 151, 210], "divis": [81, 184], "code_llama2": [82, 83, 84, 85, 86, 87, 88, 89, 90, 272], "arxiv": [82, 83, 84, 85, 86, 88, 89, 90, 95, 96, 97, 98, 99, 102, 103, 104, 105, 107, 108, 109, 110, 112, 113, 114, 115, 120, 121, 122, 123, 124, 131, 132, 133, 134, 135, 138, 139, 140, 141, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 183, 184, 185, 191, 218, 219, 221, 222, 223, 224], "pdf": [82, 83, 84, 218, 219], "2308": [82, 83, 84], "12950": [82, 83, 84], "lora_attn_modul": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 209, 210, 268, 278, 279, 281], "q_proj": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 183, 209, 210, 268, 278, 279, 280, 281], "k_proj": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 183, 209, 210, 268, 278, 279, 280, 281], "v_proj": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 183, 209, 210, 268, 278, 279, 280, 281], "output_proj": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 183, 209, 210, 278, 279, 280, 281], "apply_lora_to_mlp": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 209, 210, 268, 278, 279], "apply_lora_to_output": [85, 86, 87, 88, 89, 90, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 174, 209, 210, 278, 279], "lora_rank": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 268, 278, 279], "lora_alpha": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 268, 278, 279], "lora_dropout": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 279], "use_dora": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 131, 132, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174], "quantize_bas": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 205, 281], "code_llama2_13b": 85, "tloen": [85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 172, 173, 174], "8bb8579e403dc78e37fe81ffbb253c413007323f": [85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 172, 173, 174], "l41": [85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 172, 173, 174], "l43": [85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 172, 173, 174], "linear": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 143, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 187, 189, 204, 205, 209, 210, 278, 279, 280, 281], "low": [85, 86, 87, 95, 96, 97, 107, 108, 109, 110, 111, 120, 121, 122, 129, 130, 131, 132, 138, 139, 149, 150, 151, 154, 155, 156, 157, 165, 166, 171, 172, 173, 174, 205, 268, 274, 277, 278, 281], "approxim": [85, 86, 87, 95, 96, 97, 107, 108, 109, 110, 111, 120, 121, 122, 129, 130, 131, 132, 138, 139, 149, 150, 151, 154, 155, 156, 157, 165, 166, 171, 172, 173, 174, 205, 278], "factor": [85, 86, 87, 95, 96, 97, 107, 108, 109, 110, 111, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 149, 150, 151, 154, 155, 156, 157, 165, 166, 171, 172, 173, 174, 205, 219, 274], "dropout": [85, 86, 87, 91, 95, 96, 97, 101, 107, 109, 110, 111, 116, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 149, 150, 151, 154, 155, 156, 157, 158, 160, 165, 166, 167, 171, 172, 173, 174, 175, 183, 205, 278, 279, 281], "decompos": [85, 86, 95, 96, 97, 107, 108, 109, 110, 120, 121, 122, 131, 132, 138, 139, 154, 155, 156, 157, 165, 166], "magnitud": [85, 86, 95, 96, 97, 107, 108, 109, 110, 120, 121, 122, 131, 132, 138, 139, 154, 155, 156, 157, 165, 166, 279], "dora": [85, 86, 95, 96, 97, 107, 108, 109, 110, 120, 121, 122, 129, 131, 132, 138, 139, 150, 151, 154, 155, 156, 157, 165, 166], "ab": [85, 86, 88, 89, 90, 95, 96, 97, 98, 99, 102, 103, 104, 105, 107, 108, 109, 110, 112, 113, 114, 115, 120, 121, 122, 123, 124, 131, 132, 133, 134, 135, 138, 139, 140, 141, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 183, 184, 185, 191, 221, 222, 223, 224], "2402": [85, 86, 95, 96, 97, 107, 108, 109, 110, 120, 121, 122, 131, 132, 138, 139, 154, 155, 156, 157, 165, 166], "09353": [85, 86, 95, 96, 97, 107, 108, 109, 110, 120, 121, 122, 131, 132, 138, 139, 154, 155, 156, 157, 165, 166], "code_llama2_70b": 86, "code_llama2_7b": 87, "qlora": [88, 89, 90, 98, 99, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 152, 163, 164, 170, 195, 265, 267, 268, 276, 278], "paper": [88, 89, 90, 98, 99, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 152, 163, 164, 170, 218, 221, 223, 224, 277, 278, 281], "2305": [88, 89, 90, 98, 99, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 152, 163, 164, 170, 183, 221, 223], "14314": [88, 89, 90, 98, 99, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 152, 163, 164, 170], "lora_code_llama2_13b": 88, "lora_code_llama2_70b": 89, "lora_code_llama2_7b": 90, "head_dim": [91, 95, 182, 183, 189], "intermediate_dim": [91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175], "attn_dropout": [91, 95, 101, 107, 116, 120, 125, 129, 154, 156, 158, 160, 165, 167, 171, 175, 183, 189], "norm_ep": [91, 95, 101, 107, 116, 120, 125, 129, 154, 156, 158, 160, 165, 167, 171, 175], "1e": [91, 95, 101, 107, 116, 120, 125, 129, 154, 156, 158, 160, 165, 167, 171, 175, 184, 277], "06": [91, 95, 184, 278], "rope_bas": [91, 95, 101, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175], "10000": [91, 95, 101, 154, 156, 158, 160, 165, 167, 185], "norm_embed": [91, 95], "transformerselfattentionlay": [91, 101, 116, 125, 158, 175, 188, 189, 199, 201], "rm": [91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175], "norm": [91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 189], "space": [91, 101, 116, 125, 146, 150, 158, 175, 189, 203, 279], "slide": [91, 158, 168], "window": [91, 158, 168], "vocabulari": [91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 196, 198, 278, 279], "mha": [91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 183, 189], "onto": [91, 95, 101, 107, 116, 120, 125, 129, 154, 156, 158, 160, 165, 167, 171, 175, 183, 203], "epsilon": [91, 95, 101, 107, 116, 120, 125, 129, 154, 156, 158, 160, 165, 167, 171, 175, 222], "rotari": [91, 95, 101, 125, 129, 154, 156, 158, 160, 165, 167, 185, 276], "10_000": [91, 95, 154, 156, 158, 160, 167], "blog": [92, 93], "technolog": [92, 93], "develop": [92, 93, 266, 281], "gemmatoken": 94, "_templatetyp": [94, 106, 119, 148, 162, 169, 179], "gemma_2b": 96, "gemma_7b": 97, "lora_gemma_2b": 98, "lora_gemma_7b": 99, "taken": [100, 278, 281], "sy": [100, 273], "honest": [100, 273], "pari": [100, 153], "capit": [100, 153], "franc": [100, 153], "known": [100, 153, 240, 280], "stun": [100, 153], "05": [101, 107, 116, 120, 125, 129, 154, 156, 158, 160, 165, 167, 171, 175], "gqa": [101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 183], "mqa": [101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 183], "kvcach": [101, 107, 116, 120, 125, 129, 146, 150, 165, 171, 175, 183, 189, 192, 193, 194, 199], "scale_hidden_dim_for_mlp": [101, 107, 116, 120, 125, 129, 146, 150, 171, 175], "2307": [102, 103, 104, 105], "09288": [102, 103, 104, 105], "classif": [105, 156, 160, 161, 231], "llama2_70b": 109, "llama2_7b": [110, 278], "classifi": [111, 156, 160, 161, 257, 279], "llama2_reward_7b": [111, 231], "lora_llama2_13b": 112, "lora_llama2_70b": 113, "lora_llama2_7b": [114, 278], "lora_llama2_reward_7b": 115, "500000": [116, 120, 125, 129, 146, 150], "llama3token": [119, 144, 212], "regist": [119, 144, 148, 169, 179, 195, 252, 281], "canon": [119, 144, 148, 169, 179], "llama3_70b": 121, "lora_llama3_70b": 123, "lora_llama3_8b": 124, "scale_factor": [125, 129], "500_000": [125, 129], "rope": [125, 129, 171, 175, 183, 185], "llama3_1": [126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 268, 277], "llama3_1_70b": 131, "llama3_1_8b": 132, "lora_llama3_1_405b": 133, "lora_llama3_1_70b": 134, "lora_llama3_1_8b": 135, "llama3_2_1b": [138, 192, 193, 194], "llama3_2_3b": 139, "lora_llama3_2_1b": 140, "lora_llama3_2_3b": 141, "projection_head": [142, 199, 203], "combin": [142, 144, 147, 151, 189, 199, 201, 203, 220, 277], "learnabl": [142, 186, 199, 201, 274], "fusion": [142, 145, 146, 147, 149, 150, 151, 199, 200, 201, 202, 203], "encoder_dim": [142, 143], "decoder_dim": [142, 143], "num_img": [142, 143], "num_emb": [142, 143], "broken": [142, 143, 191, 201], "width": [142, 280], "clip_embeds_per_til": 142, "emb": [142, 143, 183, 188, 189, 199], "num_hidden_input": 143, "sequenti": [143, 199, 203], "num_hidden": 143, "hidden_st": [143, 191], "image_mean": 144, "image_std": 144, "tranform": 144, "possible_resolut": 144, "448": [144, 145, 148, 149], "deviat": 144, "still": [144, 196, 198, 200, 201, 268, 278, 280, 281], "transformed_data": 144, "img1": [144, 218], "img2": [144, 218], "31587": [144, 213, 214], "29644": [144, 213, 214], "102": [144, 213, 214], "truncate_at_eo": [144, 214], "skip": [144, 183], "tokenize_head": 144, "tokenize_end": 144, "header": 144, "eom": 144, "wether": 144, "decoder_train": [145, 149, 152, 199], "encoder_train": [145, 149, 152, 199], "fusion_train": [145, 149, 152, 199], "deepfusionmodel": [145, 149, 152], "trainabl": [145, 149, 201, 205, 208, 246, 278, 279, 281], "resiz": [145, 148, 149], "fusion_interv": [146, 150], "num_special_token": [146, 150], "encoder_max_seq_len": [146, 150, 188, 189, 190, 194, 199, 201], "causalselfattent": [146, 150], "interv": [146, 150, 275], "clip_embed_dim": [147, 151], "clip_num_lay": [147, 151], "clip_hidden_st": [147, 151], "num_layers_project": [147, 151], "decoder_embed_dim": [147, 151], "llama3visionencod": [147, 151], "spatial": [147, 151], "backbon": [147, 151], "trainbl": 149, "decoder_lora": 150, "fusion_lora": [150, 151], "encoder_lora": 151, "lora_llama3_2_vision_11b": 152, "num_class": [156, 160, 257], "announc": 159, "ray2333": 161, "feedback": [161, 221], "lora_mistral_7b": 163, "lora_mistral_reward_7b": 164, "phi3_mini": [166, 231], "128k": 168, "nor": 168, "phi3minitoken": 169, "tokenizer_config": 169, "spm": 169, "lm": [169, 222, 277], "bo": [169, 213, 216, 273], "unk": 169, "augment": [169, 281], "endoftext": 169, "phi3minisentencepiecebasetoken": 169, "lora_phi3_mini": 170, "1000000": [171, 175], "tie_word_embed": [171, 172, 173, 175, 176, 177], "qwen2transformerdecod": 171, "period": [171, 175], "word": [171, 175, 280], "qwen2_0_5b": [172, 187], "qwen2_1_5b": [173, 187], "qwen2_7b": 174, "qwen": [176, 177, 178], "merges_fil": 179, "qwen2token": 179, "gate_proj": 180, "down_proj": 180, "up_proj": 180, "feed": [180, 188, 190], "network": [180, 206, 278, 281], "fed": [180, 273], "multipli": [180, 279], "in_dim": [180, 204, 205, 278, 279, 281], "out_dim": [180, 189, 204, 205, 278, 279, 281], "layernorm": 181, "past": 182, "expand": 182, "dpython": [182, 183, 188, 189, 190, 194, 195, 199, 201, 254, 258], "reset": [182, 183, 188, 189, 190, 199, 201, 239], "k_val": 182, "v_val": 182, "fill": 182, "bfloat16": [182, 195, 254, 274, 275, 276, 278, 279, 280], "greater": [182, 191, 262], "pos_embed": [183, 188, 278, 280], "q_norm": 183, "k_norm": 183, "kv_cach": [183, 192, 193, 194], "is_caus": 183, "13245v1": 183, "multihead": 183, "extrem": 183, "credit": 183, "litgpt": 183, "v": [183, 189, 199, 278], "n_kv_head": 183, "rotarypositionalembed": [183, 278, 280], "rmsnorm": 183, "vice": [183, 272], "versa": [183, 272], "y": 183, "s_x": 183, "s_y": 183, "_masktyp": [183, 189, 190], "score": [183, 189, 190, 220], "encoder_max_cache_seq_len": [183, 189, 190], "j": [183, 188, 189, 190, 199], "blockmask": [183, 189, 190], "create_block_mask": [183, 189, 190], "flex_attent": [183, 189, 190], "n_h": [183, 185], "num": [183, 185], "n_kv": 183, "h_d": [183, 185], "reset_cach": [183, 188, 189, 190, 199, 201], "setup_cach": [183, 188, 189, 190, 192, 193, 199, 201], "ep": 184, "squar": 184, "1910": 184, "07467": 184, "propos": [185, 279], "2104": 185, "09864": 185, "verfic": 185, "l80": 185, "init": [185, 239, 251, 281], "exceed": 185, "freq": 185, "recomput": [185, 279], "geometr": 185, "progress": [185, 270, 275, 279], "rotat": 185, "angl": 185, "basic": [186, 276], "tied_modul": 187, "lost": 187, "whose": [187, 206, 247, 252], "attributeerror": [187, 259], "attn": [188, 190, 192, 193, 194, 278, 280, 281], "multiheadattent": [188, 190, 278, 280], "ca_norm": 188, "mlp_norm": [188, 190], "ca_scal": 188, "mlp_scale": [188, 190], "ff": [188, 190], "caches_are_en": [188, 189, 190, 192, 193, 194, 199, 201], "func": [188, 190, 201], "caches_are_setup": [188, 189, 190, 192, 193, 194, 199, 201], "token_sequ": 188, "embed_sequ": 188, "decoder_max_seq_len": [188, 189, 190, 192, 193, 194, 199, 201], "modulelist": 189, "output_hidden_st": [189, 199], "belong": [189, 234], "reduc": [189, 221, 267, 268, 269, 277, 278, 279, 280, 281], "statement": 189, "improv": [189, 214, 223, 237, 269, 276, 277, 278, 279], "readabl": [189, 274], "behaviour": [189, 199, 257], "alter": [189, 199], "common_util": [189, 192, 193, 194, 195], "disable_kv_cach": [189, 199], "chunked_output": 189, "last_hidden_st": 189, "chunk": [189, 196, 198, 214], "cewithchunkedoutputloss": [189, 199], "upcast": [189, 196, 198], "set_num_output_chunk": [189, 199], "num_chunk": [189, 196, 198], "s_e": [189, 199], "d_e": [189, 199], "arang": [189, 199], "prompt_length": [189, 199], "correspondingli": 189, "padded_prompt_length": 189, "m_": [189, 199], "realloc": [189, 199], "runtimeerror": [189, 216, 232, 236, 238, 243], "num_output_chunk": [189, 196, 198, 199], "transformercrossattentionlay": [189, 199, 201], "fusionlay": [189, 199], "sa_norm": 190, "sa_scal": 190, "token_pos_embed": 191, "pre_tile_pos_emb": 191, "post_tile_pos_emb": 191, "cls_project": 191, "vit": 191, "11929": 191, "convolut": 191, "flatten": 191, "downscal": 191, "800x400": 191, "400x400": 191, "_transform": 191, "whole": [191, 277], "n_token": 191, "101": 191, "pool": 191, "tiledtokenpositionalembed": 191, "tilepositionalembed": 191, "tile_pos_emb": 191, "8x8": 191, "21": 191, "22": 191, "23": [191, 235], "24": [191, 275, 276], "25": [191, 274, 277], "26": 191, "27": [191, 274], "28": [191, 274], "29": [191, 281], "30": [191, 225, 280], "31": [191, 276], "33": 191, "34": 191, "35": [191, 281], "36": 191, "37": 191, "38": [191, 274], "39": 191, "41": 191, "43": 191, "44": 191, "45": 191, "46": 191, "47": 191, "48": [191, 274, 281], "49": 191, "50": [191, 225, 247, 274], "51": 191, "52": [191, 275], "53": 191, "54": 191, "55": [191, 275], "56": 191, "57": [191, 278, 281], "58": 191, "59": [191, 281], "60": 191, "61": [191, 274], "62": 191, "63": 191, "64": [191, 268, 278], "num_patches_per_til": 191, "emb_dim": 191, "constain": 191, "anim": 191, "max_n_img": 191, "n_channel": 191, "vision_util": 191, "tile_crop": 191, "800": 191, "patch_grid_s": 191, "rand": 191, "nch": 191, "tile_cropped_imag": 191, "batch_imag": 191, "unsqueez": 191, "batch_aspect_ratio": 191, "clip_vision_encod": 191, "cache_en": 192, "float32": [192, 193, 194, 236, 277], "1024": [192, 193, 194, 280], "temporarili": [193, 194, 206, 279], "enter": [193, 194], "overhead": [193, 221, 269, 279, 280], "untouch": [193, 273], "yield": [193, 194, 206], "caller": [193, 194, 206], "delete_kv_cach": 194, "offload_to_cpu": 195, "hook": [195, 252, 279, 281], "nf4": [195, 279, 281], "restor": 195, "higher": [195, 276, 277, 279, 280, 281], "offload": [195, 281], "increas": [195, 221, 235, 276, 277, 278, 279, 280], "peak": [195, 239, 245, 274, 276, 278, 281], "gpu": [195, 269, 272, 274, 275, 276, 277, 278, 279, 280, 281], "_register_state_dict_hook": 195, "mymodul": 195, "_after_": 195, "nf4tensor": [195, 281], "unquant": [195, 280, 281], "unus": 195, "ignore_index": [196, 197, 198, 277], "entropi": [196, 198, 277], "bf16": [196, 198, 236, 279, 281], "ce": [196, 277], "better": [196, 198, 224, 267, 273, 274, 277, 279, 280], "accuraci": [196, 198, 269, 274, 276, 277, 278, 279, 280, 281], "doubl": [196, 198, 281], "therefor": [196, 198, 281], "num_token": [196, 197, 198], "consider": [196, 198], "compute_cross_entropi": 196, "gain": [196, 269, 276], "won": [196, 273], "realiz": 196, "pull": [196, 268, 269, 272], "1390": 196, "loss_fn": [196, 198], "chunkedcrossentropyloss": 196, "output_chunk": [196, 198], "kullback": [197, 277], "leibler": [197, 277], "diverg": [197, 198, 220, 277], "jongwooko": [197, 277], "distillm": [197, 277], "17c0f98bc263b1861a02d5df578c84aea652ee65": 197, "student_logit": [197, 198, 277], "teacher_logit": [197, 198, 277], "student": [197, 198], "teacher": [197, 198, 274], "kl": [197, 198, 220, 277], "teacher_chunk": 198, "teacher_model": 198, "model_fus": [199, 200, 201, 202, 203], "deepfus": 199, "evolut": 199, "signatur": 199, "interchang": 199, "fusion_param": [199, 200, 201, 202, 203], "fusionembed": 199, "fusion_lay": [199, 201], "clip_vit_224": [199, 203], "feedforward": [199, 203], "register_fusion_modul": 199, "flamingo": [199, 201, 218], "strict": [199, 200, 201, 209, 278], "freez": [199, 274, 278], "fusion_vocab_s": 200, "necessit": 200, "rout": 200, "128": [200, 268, 276, 278, 279], "fusion_first": 201, "shot": [201, 274, 276, 280], "infus": 201, "interpret": 201, "enocd": 201, "isn": [201, 236, 272], "fused_lay": 201, "mark": [203, 273], "earli": 203, "peft": [204, 205, 206, 207, 208, 209, 210, 228, 268, 278, 281], "adapter_param": [204, 205, 206, 207, 208], "proj": 204, "loralinear": [204, 278, 281], "alpha": [205, 278, 279, 281], "use_bia": 205, "perturb": 205, "decomposit": [205, 278, 279], "matric": [205, 278, 281], "mapsto": 205, "w_0x": 205, "r": [205, 278], "bax": 205, "lora_a": [205, 278, 281], "lora_b": [205, 278, 281], "polici": [206, 220, 221, 222, 223, 224, 226, 237, 246, 253, 270], "neural": [206, 278, 281], "get_adapter_param": [208, 278], "base_miss": 209, "base_unexpect": 209, "lora_miss": 209, "lora_unexpect": 209, "validate_state_dict_for_lora": [209, 278], "unlik": 209, "reli": [209, 216, 274, 276], "unexpect": 209, "nonempti": 209, "full_model_state_dict_kei": 210, "lora_state_dict_kei": 210, "base_model_state_dict_kei": 210, "confirm": [210, 266], "lora_modul": 210, "complement": 210, "disjoint": 210, "overlap": [210, 279], "tiktokenbasetoken": 211, "light": 213, "sentencepieceprocessor": 213, "trim": 213, "whitespac": 213, "spm_model": [213, 273], "tokenized_text": [213, 214], "add_bo": [213, 214, 273], "trim_leading_whitespac": 213, "prefix": [213, 279], "unbatch": 213, "bos_id": [214, 216], "lightweight": [214, 273], "substr": 214, "repetit": 214, "speed": [214, 256, 276, 279, 280, 281], "identif": 214, "regex": 214, "absent": 214, "tt_model": 214, "tokenizer_json_path": 215, "heavili": 216, "concat": 216, "1788": 216, "2643": 216, "465": 216, "22137": 216, "join": 216, "satisfi": [216, 274], "loos": 217, "image_token_id": 218, "particip": [218, 219], "laid": 218, "fig": 218, "2204": 218, "14198": 218, "immedi": [218, 279], "until": [218, 279], "img3": 218, "equal": [218, 262], "gamma": [219, 223, 224], "lmbda": 219, "estim": [219, 220], "1506": 219, "02438": 219, "response_len": [219, 220], "receiv": 219, "discount": 219, "gae": 219, "logprob": [220, 224], "ref_logprob": 220, "kl_coeff": 220, "valid_score_idx": 220, "coeffici": [220, 222], "total_reward": 220, "kl_reward": 220, "beta": [221, 224], "label_smooth": [221, 224], "18290": 221, "intuit": [221, 223, 224], "dispref": 221, "dynam": [221, 280], "degener": 221, "occur": [221, 269], "naiv": 221, "trl": [221, 223, 224], "5d1deb1445828cfd0e947cb3a7925b1c03a283fc": 221, "dpo_train": [221, 223], "l844": 221, "2009": 221, "01325": 221, "regular": [221, 224, 279, 280, 281], "baselin": [221, 222, 274, 277, 278], "uncertainti": [221, 224], "policy_chosen_logp": [221, 223, 224], "policy_rejected_logp": [221, 223, 224], "reference_chosen_logp": [221, 223], "reference_rejected_logp": [221, 223], "chosen_reward": [221, 223, 224], "rejected_reward": [221, 223, 224], "value_clip_rang": 222, "value_coeff": 222, "proxim": [222, 270], "1707": 222, "06347": 222, "eqn": 222, "vwxyzjn": 222, "ccc19538e817e98a60d3253242ac15e2a562cb49": 222, "lm_human_preference_detail": 222, "train_policy_acceler": 222, "l719": 222, "ea25b9e8b234e6ee1bca43083f8f3cf974143998": 222, "ppo2": 222, "l68": 222, "l75": 222, "pi_old_logprob": 222, "pi_logprob": 222, "phi_old_valu": 222, "phi_valu": 222, "value_padding_mask": 222, "old": 222, "participag": 222, "five": 222, "policy_loss": 222, "value_loss": 222, "clipfrac": 222, "fraction": 222, "statist": [223, 279], "rso": 223, "hing": 223, "2309": 223, "06657": 223, "logist": 223, "regress": 223, "slic": 223, "10425": 223, "almost": [223, 278], "svm": 223, "counter": 223, "4dce042a3863db1d375358e8c8092b874b02934b": 223, "l1141": 223, "simpo": 224, "2405": 224, "14734": 224, "averag": [224, 277], "implicit": 224, "margin": 224, "bradlei": 224, "terri": 224, "larger": [224, 230, 274, 276, 277, 279], "win": 224, "lose": 224, "98ad01ddfd1e1b67ec018014b83cba40e0caea66": 224, "cpo_train": 224, "l603": 224, "pretti": [224, 274], "identitc": 224, "elimin": 224, "kind": 224, "ipoloss": 224, "fill_valu": 225, "sequence_length": 225, "stop_token_id": 225, "869": 225, "eos_mask": 225, "truncated_sequ": 225, "datatyp": [226, 279, 281], "denot": 226, "auto_wrap_polici": [226, 237, 253], "submodul": [226, 246], "obei": 226, "contract": 226, "get_fsdp_polici": 226, "modules_to_wrap": [226, 237, 246], "min_num_param": 226, "my_fsdp_polici": 226, "recurs": [226, 246, 250], "sum": [226, 277, 278], "p": [226, 232, 278, 280, 281], "numel": [226, 278], "1000": [226, 280], "stabl": [226, 244, 250, 255, 266, 279], "html": [226, 244, 250, 253, 255, 261, 265], "filename_format": 227, "max_filenam": 227, "concis": 227, "filenam": [227, 248], "file_": 227, "_of_": 227, "n_file": 227, "build_checkpoint_filenam": 227, "file_00001_of_00003": 227, "file_00002_of_00003": 227, "file_00003_of_00003": 227, "safe_seri": 228, "from_pretrain": 228, "0001_of_0003": 228, "0002_of_0003": 228, "todo": 228, "preserv": [228, 281], "weight_map": [228, 274], "convert_weight": 228, "_model_typ": [228, 231], "intermediate_checkpoint": [228, 229, 230], "adapter_onli": [228, 229, 230], "_weight_map": 228, "shard": [229, 276], "wip": 229, "qualnam": 231, "boundari": 231, "distinguish": 231, "llama3_vis": 231, "llama3_2_vision_decod": 231, "mistral_reward_7b": 231, "my_new_model": 231, "my_custom_state_dict_map": 231, "optim_map": 232, "bare": 232, "bone": 232, "optim_dict": [232, 234, 252], "cfg_optim": 232, "ckpt": 232, "optim_ckpt": 232, "placeholder_optim_dict": 232, "optiminbackwardwrapp": 232, "get_last_lr": 232, "rate": [232, 235, 238, 267, 275, 279], "schedul": [232, 235, 256, 275, 279], "get_optim_kei": 232, "arbitrari": [232, 278, 279], "optim_ckpt_map": 232, "set_lr_schedul": 232, "lr_schedul": [232, 235], "lrschedul": 232, "loadabl": 232, "step_lr_schedul": 232, "ac_mod": 233, "ac_opt": 233, "op": [233, 280], "ac": [233, 237], "optimizerinbackwardwrapp": [234, 238], "named_paramet": [234, 257], "num_warmup_step": 235, "num_training_step": 235, "num_cycl": [235, 256], "last_epoch": 235, "lambdalr": 235, "linearli": 235, "decreas": [235, 278, 279, 280, 281], "cosin": 235, "v4": 235, "src": 235, "l104": 235, "warmup": [235, 256], "phase": 235, "wave": 235, "half": [235, 279], "kernel": 236, "memory_efficient_fsdp_wrap": [237, 280], "maxim": [237, 246, 265, 267], "workload": [237, 269, 279, 280], "fullyshardeddataparallel": [237, 246], "fsdppolicytyp": [237, 246], "warpper": 238, "optimizer_in_backward": 238, "reset_stat": 239, "track": [239, 247], "alloc": [239, 245, 246, 276, 281], "reserv": [239, 245, 273, 281], "stat": [239, 245, 281], "int4": [240, 280], "4w": 240, "recogn": 240, "int8dynactint4weightquant": [240, 269, 280], "8da4w": [240, 280], "int4weightonlyquant": [240, 280], "int8dynactint4weightqatquant": [240, 269, 280], "qat": [240, 265, 270], "int4weightonlyqatquant": 240, "exclud": 241, "aka": 242, "master": 244, "port": [244, 272], "address": [244, 277, 279], "hold": [244, 275], "peak_memory_act": 245, "peak_memory_alloc": 245, "peak_memory_reserv": 245, "get_memory_stat": 245, "hierarch": 246, "api_kei": 247, "experiment_kei": 247, "onlin": 247, "log_cod": 247, "comet": 247, "site": [247, 274], "ml": 247, "team": 247, "compar": [247, 250, 262, 274, 276, 277, 278, 280, 281], "sdk": 247, "uncategor": 247, "alphanumer": 247, "charact": 247, "get_or_cr": 247, "fresh": 247, "persist": 247, "hpo": 247, "sweep": 247, "server": 247, "offlin": 247, "auto": [247, 272], "creation": 247, "experimentconfig": 247, "project_nam": 247, "my_workspac": 247, "my_metr": [247, 250, 251], "importerror": [247, 251], "termin": [247, 250, 251], "comet_api_kei": 247, "flush": [247, 248, 249, 250, 251], "ndarrai": [247, 248, 249, 250, 251], "scalar": [247, 248, 249, 250, 251], "record": [247, 248, 249, 250, 251, 256], "log_config": [247, 251], "payload": [247, 248, 249, 250, 251], "log_": 248, "unixtimestamp": 248, "thread": 248, "safe": 248, "organize_log": 250, "tensorboard": 250, "subdirectori": 250, "logdir": 250, "startup": 250, "tree": [250, 274, 276], "tfevent": 250, "encount": 250, "frontend": 250, "organ": [250, 272], "accordingli": [250, 280], "my_log_dir": 250, "view": [250, 277], "entiti": 251, "bias": [251, 278, 281], "usernam": 251, "my_ent": 251, "my_group": 251, "account": [251, 278, 281], "link": [251, 274, 276], "capecap": 251, "6053ofw0": 251, "torchtune_config_j67sb73v": 251, "soon": [252, 279], "readi": [252, 265, 273, 280], "grad": 252, "acwrappolicytyp": 253, "author": [253, 267, 275, 279, 281], "fsdp_adavnced_tutori": 253, "insid": 254, "contextmanag": 254, "debug_mod": 255, "pseudo": 255, "commonli": [255, 278, 279, 281], "numpi": 255, "determinist": 255, "global": [255, 279], "warn": 255, "nondeterminist": 255, "cudnn": 255, "set_deterministic_debug_mod": 255, "profile_memori": 256, "with_stack": 256, "record_shap": 256, "with_flop": 256, "wait_step": 256, "warmup_step": 256, "active_step": 256, "profil": 256, "layout": 256, "trace": 256, "profileract": 256, "gradient_accumul": 256, "sensibl": 256, "default_schedul": 256, "reduct": [256, 269, 278], "scope": 256, "flop": 256, "cycl": 256, "repeat": [256, 279], "model_named_paramet": 257, "force_overrid": 257, "concret": [257, 279], "vocab_dim": 257, "named_param": 258, "inplac": [259, 278], "too": [259, 269, 276], "handler": 261, "_log": 261, "__version__": 262, "generated_examples_python": 263, "galleri": [263, 271], "sphinx": 263, "000": [264, 271, 276], "execut": [264, 271], "generated_exampl": 264, "mem": [264, 271], "mb": [264, 271], "gentl": 265, "introduct": 265, "first_finetune_tutori": 265, "kd": 265, "torchvis": 266, "torchao": [266, 269, 274, 276, 280, 281], "latest": [266, 269, 275, 279, 281], "whl": 266, "cu121": 266, "cu118": 266, "cu124": 266, "And": [266, 274], "welcom": [266, 272], "greatest": [266, 275], "contributor": 266, "dev": 266, "commit": 266, "branch": 266, "therebi": [266, 279, 280, 281], "forc": [266, 277], "reinstal": 266, "opt": [266, 275], "suffix": 266, "On": [267, 278], "pointer": 267, "emphas": 267, "simplic": 267, "component": 267, "prove": 267, "democrat": 267, "zoo": 267, "varieti": [267, 278], "integr": [267, 274, 275, 276, 278, 280, 281], "fsdp2": 267, "excit": 267, "checkout": 267, "quickstart": 267, "attain": 267, "embodi": 267, "philosophi": 267, "usabl": 267, "composit": 267, "hard": 267, "outlin": 267, "unecessari": 267, "never": 267, "thoroughli": 267, "competit": 268, "grant": [268, 269, 275], "interest": [268, 269, 274, 277], "8b_lora_single_devic": [268, 272, 273, 276, 277, 279], "lever": [268, 269], "action": [268, 269], "degrad": [269, 279, 280, 281], "simul": [269, 279, 280], "compromis": 269, "blogpost": [269, 279], "qat_distribut": [269, 280], "8b_qat_ful": [269, 280], "least": [269, 276, 278, 280], "vram": [269, 276, 278, 279, 280], "80gb": [269, 280], "a100": 269, "h100": 269, "delai": 269, "fake": [269, 280], "empir": [269, 280], "potenti": [269, 278, 279], "fake_quant_after_n_step": [269, 280], "idea": [269, 277, 281], "roughli": 269, "total_step": 269, "futur": [269, 280], "plan": [269, 274], "un": 269, "groupsiz": [269, 280], "256": [269, 276, 280], "hackabl": [270, 275], "singularli": [270, 275], "technic": [270, 275], "awar": [270, 279, 280], "tracker": 270, "short": 272, "subcommand": 272, "anytim": 272, "symlink": 272, "wrote": 272, "readm": [272, 274, 276], "md": 272, "lot": [272, 274, 279], "recent": 272, "releas": [272, 276], "agre": 272, "term": [272, 279], "perman": 272, "eat": 272, "bandwith": 272, "storag": [272, 281], "00030": 272, "ootb": 272, "7b_full_low_memori": [272, 274, 275], "8b_full_single_devic": 272, "mini_full_low_memori": 272, "7b_full": [272, 274, 275], "13b_full": [272, 274, 275], "70b_full": 272, "edit": 272, "clobber": 272, "destin": 272, "lora_finetune_distribut": [272, 276, 278], "torchrun": 272, "launch": [272, 273, 275], "nproc": 272, "node": 272, "worker": 272, "nnode": [272, 278, 280], "minimum_nod": 272, "maximum_nod": 272, "fail": 272, "rdzv": 272, "rendezv": 272, "endpoint": 272, "8b_lora": [272, 276], "bypass": 272, "fancy_lora": 272, "8b_fancy_lora": 272, "nice": 273, "meet": 273, "overhaul": 273, "multiturn": 273, "untrain": 273, "accompani": 273, "who": 273, "influenti": 273, "hip": 273, "hop": 273, "artist": 273, "2pac": 273, "rakim": 273, "flavor": 273, "formatted_messag": 273, "nyou": 273, "nwho": 273, "why": [273, 275, 278], "518": 273, "25580": 273, "29962": 273, "3532": 273, "14816": 273, "29903": 273, "6778": 273, "_spm_model": 273, "piece_to_id": 273, "manual": [273, 281], "529": 273, "29879": 273, "29958": 273, "nhere": 273, "pure": 273, "mess": 273, "prime": 273, "strictli": 273, "ask": [273, 279], "though": 273, "robust": 273, "pretend": 273, "zuckerberg": 273, "seem": [273, 274], "good": [273, 278, 279], "altogeth": 273, "honor": 273, "custom_8b_lora_single_devic": 273, "favorit": [274, 278], "seemlessli": 274, "connect": [274, 280], "amount": 274, "natur": 274, "export": 274, "leverag": [274, 276, 281], "percentag": 274, "16gb": [274, 278], "rtx": 274, "3090": 274, "4090": 274, "hour": 274, "7b_qlora_single_devic": [274, 275, 281], "473": 274, "98": [274, 281], "gb": [274, 276, 278, 280, 281], "484": 274, "01": [274, 275], "fact": [274, 276, 278, 279], "third": 274, "But": [274, 278], "realli": 274, "eleuther_ev": [274, 276, 280], "eleuther_evalu": [274, 276, 280], "lm_eval": [274, 276], "custom_eval_config": [274, 276], "truthfulqa_mc2": [274, 276, 277, 278], "measur": [274, 276], "propens": [274, 276], "324": 274, "loglikelihood": 274, "195": 274, "121": 274, "197": 274, "acc": [274, 280], "388": 274, "489": 274, "great": [274, 279], "custom_generation_config": [274, 276], "kick": 274, "300": 274, "bai": 274, "area": 274, "92": 274, "exploratorium": 274, "san": 274, "francisco": 274, "magazin": 274, "awesom": 274, "bridg": 274, "cool": 274, "96": [274, 281], "sec": [274, 276], "83": 274, "99": [274, 278], "72": 274, "littl": 274, "int8_weight_onli": [274, 276], "int8_dynamic_activation_int8_weight": [274, 276], "ao": [274, 276], "quant_api": [274, 276], "quantize_": [274, 276], "int4_weight_onli": [274, 276], "previous": [274, 276, 278], "benefit": 274, "clone": [274, 278, 280, 281], "assumpt": 274, "new_dir": 274, "output_dict": 274, "sd_1": 274, "sd_2": 274, "dump": 274, "convert_hf_checkpoint": 274, "checkpoint_path": 274, "justin": 274, "school": 274, "math": 274, "ws": 274, "94": [274, 276], "bandwidth": [274, 276], "1391": 274, "84": 274, "thats": 274, "seamlessli": 274, "authent": [274, 275], "hopefulli": 274, "gave": 274, "minut": 275, "agreement": 275, "depth": 275, "principl": 275, "boilerpl": 275, "substanti": [275, 278], "custom_config": 275, "replic": 275, "lorafinetunerecipesingledevic": 275, "lora_finetune_output": 275, "log_1713194212": 275, "3697006702423096": 275, "25880": [275, 281], "83it": 275, "monitor": 275, "tqdm": 275, "e2": 275, "focu": 276, "theta": 276, "observ": [276, 280], "consum": [276, 281], "overal": [276, 277], "8b_qlora_single_devic": [276, 279], "coupl": [276, 278, 281], "meta_model_0": [276, 280], "122": 276, "sarah": 276, "busi": 276, "mum": 276, "young": 276, "children": 276, "live": 276, "north": 276, "east": 276, "england": 276, "135": 276, "88": 276, "138": 276, "346": 276, "09": 276, "139": 276, "broader": 276, "teach": [277, 278], "straight": [277, 278], "jump": [277, 278], "compress": 277, "transfer": 277, "capac": 277, "computation": 277, "expens": 277, "deploi": 277, "imit": 277, "diagram": 277, "aim": [277, 279], "minillm": 277, "forwardklloss": 277, "super": 277, "teacher_prob": 277, "student_logprob": 277, "log_softmax": 277, "prod_prob": 277, "forwardklwithchunkedoutputloss": 277, "knowledge_distillation_single_devic": 277, "bit": [277, 278, 279, 280, 281], "alpaca_cleaned_dataset": 277, "hellaswag": [277, 280], "commonsense_qa": 277, "kd_ratio": 277, "teacher_checkpoint": 277, "00004": 277, "truthfulqa": [277, 278], "commonsens": 277, "constant": 277, "boost": 277, "graph": 277, "irrespect": 277, "3e": 277, "slightli": 277, "truthful_qa": 277, "wherea": 277, "unfamiliar": 278, "oppos": [278, 281], "momentum": [278, 279], "aghajanyan": 278, "et": 278, "al": 278, "hypothes": 278, "intrins": 278, "eight": 278, "practic": 278, "blue": 278, "although": [278, 280], "rememb": 278, "approx": 278, "15m": 278, "65k": 278, "requires_grad": [278, 281], "frozen_out": [278, 281], "lora_out": [278, 281], "base_model": 278, "lora_model": 278, "lora_llama_2_7b": [278, 281], "alon": 278, "in_featur": [278, 280], "out_featur": [278, 280], "validate_missing_and_unexpected_for_lora": 278, "peft_util": 278, "set_trainable_param": 278, "lora_param": 278, "total_param": 278, "trainable_param": 278, "2f": 278, "6742609920": 278, "4194304": 278, "7b_lora": 278, "my_model_checkpoint_path": [278, 280, 281], "tokenizer_checkpoint": [278, 280, 281], "my_tokenizer_checkpoint_path": [278, 280, 281], "constraint": 278, "factori": 278, "benefici": 278, "impact": [278, 279], "minor": 278, "lora_experiment_1": 278, "smooth": [278, 281], "curv": [278, 281], "500": 278, "ran": 278, "footprint": [278, 280], "commod": 278, "cogniz": 278, "ax": 278, "parallel": 278, "475": 278, "87": 278, "508": 278, "86": 278, "504": 278, "04": 278, "514": 278, "lowest": 278, "absolut": 278, "4gb": 278, "tradeoff": 278, "salman": 279, "mohammadi": 279, "brief": 279, "glossari": 279, "struggl": 279, "constrain": [279, 280], "particularli": 279, "gradient_accumulation_step": 279, "throughput": 279, "cost": 279, "sebastian": 279, "raschka": 279, "fp16": 279, "sound": 279, "quot": 279, "aliv": 279, "region": 279, "enable_activation_checkpoint": 279, "bring": 279, "autograd": [279, 281], "saved_tensors_hook": 279, "cours": 279, "runtim": 279, "hide": 279, "later": 279, "brought": 279, "enable_activation_offload": 279, "dev20240907": 279, "total_batch_s": 279, "count": 279, "suppos": 279, "log_every_n_step": 279, "translat": 279, "frequent": 279, "slowli": 279, "num_devic": 279, "adamw8bit": 279, "pagedadamw": 279, "modern": 279, "converg": 279, "stateless": 279, "stochast": 279, "descent": 279, "sacrif": 279, "remov": 279, "optimizer_in_bwd": 279, "greatli": 279, "lora_": 279, "lora_llama3": 279, "_lora": 279, "firstli": 279, "secondli": 279, "affect": 279, "fashion": 279, "slower": [279, 281], "jointli": 279, "sens": 279, "novel": 279, "normalfloat": [279, 281], "8x": [279, 281], "worth": 279, "cast": [279, 280], "incur": [279, 280, 281], "penalti": 279, "qlora_": 279, "qlora_llama3_8b": 279, "_qlora": 279, "perplex": 280, "goal": 280, "ptq": 280, "kept": 280, "nois": 280, "henc": 280, "x_q": 280, "int8": 280, "zp": 280, "x_float": 280, "qmin": 280, "qmax": 280, "clamp": 280, "x_fq": 280, "dequant": 280, "proce": 280, "prepared_model": 280, "swap": 280, "int8dynactint4weightqatlinear": 280, "int8dynactint4weightlinear": 280, "train_loop": 280, "converted_model": 280, "qat_distributed_recipe_label": 280, "recov": 280, "custom_8b_qat_ful": 280, "2000": 280, "led": 280, "presum": 280, "mutat": 280, "5gb": 280, "custom_quant": 280, "poorli": 280, "custom_eleuther_evalu": 280, "fullmodeltorchtunecheckpoint": 280, "max_seq_length": 280, "my_eleuther_evalu": 280, "stderr": 280, "word_perplex": 280, "9148": 280, "byte_perplex": 280, "5357": 280, "bits_per_byt": 280, "6189": 280, "5687": 280, "0049": 280, "acc_norm": 280, "7536": 280, "0043": 280, "portion": [280, 281], "74": 280, "048": 280, "190": 280, "7735": 280, "5598": 280, "6413": 280, "5481": 280, "0050": 280, "7390": 280, "0044": 280, "7251": 280, "4994": 280, "5844": 280, "5740": 280, "7610": 280, "outperform": 280, "importantli": 280, "characterist": 280, "187": 280, "958": 280, "halv": 280, "motiv": 280, "edg": 280, "smartphon": 280, "executorch": 280, "xnnpack": 280, "export_llama": 280, "use_sdpa_with_kv_cach": 280, "qmode": 280, "group_siz": 280, "get_bos_id": 280, "get_eos_id": 280, "output_nam": 280, "llama3_8da4w": 280, "pte": 280, "881": 280, "oneplu": 280, "709": 280, "tok": 280, "815": 280, "316": 280, "364": 280, "highli": 281, "vanilla": 281, "held": 281, "bespok": 281, "vast": 281, "major": 281, "normatfloat": 281, "deepdiv": 281, "distinct": 281, "de": 281, "counterpart": 281, "set_default_devic": 281, "qlora_linear": 281, "memory_alloc": 281, "177": 281, "152": 281, "del": 281, "empty_cach": 281, "lora_linear": 281, "081": 281, "344": 281, "qlora_llama2_7b": 281, "qlora_model": 281, "hood": 281, "essenti": 281, "reparametrize_as_dtype_state_dict_post_hook": 281, "149": 281, "9157477021217346": 281, "02": 281, "08": 281, "15it": 281, "nightli": 281, "200": 281, "hundr": 281, "228": 281, "8158286809921265": 281, "95it": 281, "exercis": 281, "linear_nf4": 281, "to_nf4": 281, "linear_weight": 281, "incom": 281}, "objects": {"torchtune.config": [[27, 0, 1, "", "instantiate"], [28, 0, 1, "", "log_config"], [29, 0, 1, "", "parse"], [30, 0, 1, "", "validate"]], "torchtune.data": [[31, 1, 1, "", "AlpacaToMessages"], [32, 1, 1, "", "ChatMLTemplate"], [33, 1, 1, "", "ChosenRejectedToMessages"], [34, 2, 1, "", "GrammarErrorCorrectionTemplate"], [35, 1, 1, "", "InputOutputToMessages"], [36, 1, 1, "", "Message"], [37, 1, 1, "", "OpenAIToMessages"], [38, 1, 1, "", "PromptTemplate"], [39, 1, 1, "", "PromptTemplateInterface"], [40, 2, 1, "", "QuestionAnswerTemplate"], [41, 2, 1, "", "Role"], [42, 1, 1, "", "ShareGPTToMessages"], [43, 2, 1, "", "SummarizeTemplate"], [44, 0, 1, "", "format_content_with_images"], [45, 0, 1, "", "left_pad_sequence"], [46, 0, 1, "", "load_image"], [47, 0, 1, "", "padded_collate"], [48, 0, 1, "", "padded_collate_dpo"], [49, 0, 1, "", "padded_collate_sft"], [50, 0, 1, "", "padded_collate_tiled_images_and_mask"], [51, 0, 1, "", "truncate"], [52, 0, 1, "", "validate_messages"]], "torchtune.data.Message": [[36, 3, 1, "", "contains_media"], [36, 4, 1, "", "from_dict"], [36, 4, 1, "", "get_media"], [36, 3, 1, "", "text_content"]], "torchtune.datasets": [[53, 1, 1, "", "ConcatDataset"], [54, 1, 1, "", "PackedDataset"], [55, 1, 1, "", "PreferenceDataset"], [56, 1, 1, "", "SFTDataset"], [57, 1, 1, "", "TextCompletionDataset"], [58, 0, 1, "", "alpaca_cleaned_dataset"], [59, 0, 1, "", "alpaca_dataset"], [60, 0, 1, "", "chat_dataset"], [61, 0, 1, "", "cnn_dailymail_articles_dataset"], [62, 0, 1, "", "grammar_dataset"], [63, 0, 1, "", "hh_rlhf_helpful_dataset"], [64, 0, 1, "", "instruct_dataset"], [67, 0, 1, "", "preference_dataset"], [68, 0, 1, "", "samsum_dataset"], [69, 0, 1, "", "slimorca_dataset"], [70, 0, 1, "", "stack_exchange_paired_dataset"], [71, 0, 1, "", "text_completion_dataset"], [72, 0, 1, "", "wikitext_dataset"]], "torchtune.datasets.multimodal": [[65, 0, 1, "", "llava_instruct_dataset"], [66, 0, 1, "", "the_cauldron_dataset"]], "torchtune.generation": [[73, 0, 1, "", "generate"], [74, 0, 1, "", "generate_next_token"], [75, 0, 1, "", "get_causal_mask_from_padding_mask"], [76, 0, 1, "", "get_position_ids_from_padding_mask"], [77, 0, 1, "", "sample"]], "torchtune.models.clip": [[78, 1, 1, "", "TilePositionalEmbedding"], [79, 1, 1, "", "TiledTokenPositionalEmbedding"], [80, 1, 1, "", "TokenPositionalEmbedding"], [81, 0, 1, "", "clip_vision_encoder"]], "torchtune.models.clip.TilePositionalEmbedding": [[78, 4, 1, "", "forward"]], "torchtune.models.clip.TiledTokenPositionalEmbedding": [[79, 4, 1, "", "forward"]], "torchtune.models.clip.TokenPositionalEmbedding": [[80, 4, 1, "", "forward"]], "torchtune.models.code_llama2": [[82, 0, 1, "", "code_llama2_13b"], [83, 0, 1, "", "code_llama2_70b"], [84, 0, 1, "", "code_llama2_7b"], [85, 0, 1, "", "lora_code_llama2_13b"], [86, 0, 1, "", "lora_code_llama2_70b"], [87, 0, 1, "", "lora_code_llama2_7b"], [88, 0, 1, "", "qlora_code_llama2_13b"], [89, 0, 1, "", "qlora_code_llama2_70b"], [90, 0, 1, "", "qlora_code_llama2_7b"]], "torchtune.models.gemma": [[91, 0, 1, "", "gemma"], [92, 0, 1, "", "gemma_2b"], [93, 0, 1, "", "gemma_7b"], [94, 0, 1, "", "gemma_tokenizer"], [95, 0, 1, "", "lora_gemma"], [96, 0, 1, "", "lora_gemma_2b"], [97, 0, 1, "", "lora_gemma_7b"], [98, 0, 1, "", "qlora_gemma_2b"], [99, 0, 1, "", "qlora_gemma_7b"]], "torchtune.models.llama2": [[100, 1, 1, "", "Llama2ChatTemplate"], [101, 0, 1, "", "llama2"], [102, 0, 1, "", "llama2_13b"], [103, 0, 1, "", "llama2_70b"], [104, 0, 1, "", "llama2_7b"], [105, 0, 1, "", "llama2_reward_7b"], [106, 0, 1, "", "llama2_tokenizer"], [107, 0, 1, "", "lora_llama2"], [108, 0, 1, "", "lora_llama2_13b"], [109, 0, 1, "", "lora_llama2_70b"], [110, 0, 1, "", "lora_llama2_7b"], [111, 0, 1, "", "lora_llama2_reward_7b"], [112, 0, 1, "", "qlora_llama2_13b"], [113, 0, 1, "", "qlora_llama2_70b"], [114, 0, 1, "", "qlora_llama2_7b"], [115, 0, 1, "", "qlora_llama2_reward_7b"]], "torchtune.models.llama3": [[116, 0, 1, "", "llama3"], [117, 0, 1, "", "llama3_70b"], [118, 0, 1, "", "llama3_8b"], [119, 0, 1, "", "llama3_tokenizer"], [120, 0, 1, "", "lora_llama3"], [121, 0, 1, "", "lora_llama3_70b"], [122, 0, 1, "", "lora_llama3_8b"], [123, 0, 1, "", "qlora_llama3_70b"], [124, 0, 1, "", "qlora_llama3_8b"]], "torchtune.models.llama3_1": [[125, 0, 1, "", "llama3_1"], [126, 0, 1, "", "llama3_1_405b"], [127, 0, 1, "", "llama3_1_70b"], [128, 0, 1, "", "llama3_1_8b"], [129, 0, 1, "", "lora_llama3_1"], [130, 0, 1, "", "lora_llama3_1_405b"], [131, 0, 1, "", "lora_llama3_1_70b"], [132, 0, 1, "", "lora_llama3_1_8b"], [133, 0, 1, "", "qlora_llama3_1_405b"], [134, 0, 1, "", "qlora_llama3_1_70b"], [135, 0, 1, "", "qlora_llama3_1_8b"]], "torchtune.models.llama3_2": [[136, 0, 1, "", "llama3_2_1b"], [137, 0, 1, "", "llama3_2_3b"], [138, 0, 1, "", "lora_llama3_2_1b"], [139, 0, 1, "", "lora_llama3_2_3b"], [140, 0, 1, "", "qlora_llama3_2_1b"], [141, 0, 1, "", "qlora_llama3_2_3b"]], "torchtune.models.llama3_2_vision": [[142, 1, 1, "", "Llama3VisionEncoder"], [143, 1, 1, "", "Llama3VisionProjectionHead"], [144, 1, 1, "", "Llama3VisionTransform"], [145, 0, 1, "", "llama3_2_vision_11b"], [146, 0, 1, "", "llama3_2_vision_decoder"], [147, 0, 1, "", "llama3_2_vision_encoder"], [148, 0, 1, "", "llama3_2_vision_transform"], [149, 0, 1, "", "lora_llama3_2_vision_11b"], [150, 0, 1, "", "lora_llama3_2_vision_decoder"], [151, 0, 1, "", "lora_llama3_2_vision_encoder"], [152, 0, 1, "", "qlora_llama3_2_vision_11b"]], "torchtune.models.llama3_2_vision.Llama3VisionEncoder": [[142, 4, 1, "", "forward"]], "torchtune.models.llama3_2_vision.Llama3VisionProjectionHead": [[143, 4, 1, "", "forward"]], "torchtune.models.llama3_2_vision.Llama3VisionTransform": [[144, 4, 1, "", "decode"], [144, 4, 1, "", "tokenize_message"], [144, 4, 1, "", "tokenize_messages"]], "torchtune.models.mistral": [[153, 1, 1, "", "MistralChatTemplate"], [154, 0, 1, "", "lora_mistral"], [155, 0, 1, "", "lora_mistral_7b"], [156, 0, 1, "", "lora_mistral_classifier"], [157, 0, 1, "", "lora_mistral_reward_7b"], [158, 0, 1, "", "mistral"], [159, 0, 1, "", "mistral_7b"], [160, 0, 1, "", "mistral_classifier"], [161, 0, 1, "", "mistral_reward_7b"], [162, 0, 1, "", "mistral_tokenizer"], [163, 0, 1, "", "qlora_mistral_7b"], [164, 0, 1, "", "qlora_mistral_reward_7b"]], "torchtune.models.phi3": [[165, 0, 1, "", "lora_phi3"], [166, 0, 1, "", "lora_phi3_mini"], [167, 0, 1, "", "phi3"], [168, 0, 1, "", "phi3_mini"], [169, 0, 1, "", "phi3_mini_tokenizer"], [170, 0, 1, "", "qlora_phi3_mini"]], "torchtune.models.qwen2": [[171, 0, 1, "", "lora_qwen2"], [172, 0, 1, "", "lora_qwen2_0_5b"], [173, 0, 1, "", "lora_qwen2_1_5b"], [174, 0, 1, "", "lora_qwen2_7b"], [175, 0, 1, "", "qwen2"], [176, 0, 1, "", "qwen2_0_5b"], [177, 0, 1, "", "qwen2_1_5b"], [178, 0, 1, "", "qwen2_7b"], [179, 0, 1, "", "qwen2_tokenizer"]], "torchtune.modules": [[180, 1, 1, "", "FeedForward"], [181, 1, 1, "", "Fp32LayerNorm"], [182, 1, 1, "", "KVCache"], [183, 1, 1, "", "MultiHeadAttention"], [184, 1, 1, "", "RMSNorm"], [185, 1, 1, "", "RotaryPositionalEmbeddings"], [186, 1, 1, "", "TanhGate"], [187, 1, 1, "", "TiedLinear"], [188, 1, 1, "", "TransformerCrossAttentionLayer"], [189, 1, 1, "", "TransformerDecoder"], [190, 1, 1, "", "TransformerSelfAttentionLayer"], [191, 1, 1, "", "VisionTransformer"]], "torchtune.modules.FeedForward": [[180, 4, 1, "", "forward"]], "torchtune.modules.Fp32LayerNorm": [[181, 4, 1, "", "forward"]], "torchtune.modules.KVCache": [[182, 4, 1, "", "reset"], [182, 4, 1, "", "update"]], "torchtune.modules.MultiHeadAttention": [[183, 4, 1, "", "forward"], [183, 4, 1, "", "reset_cache"], [183, 4, 1, "", "setup_cache"]], "torchtune.modules.RMSNorm": [[184, 4, 1, "", "forward"]], "torchtune.modules.RotaryPositionalEmbeddings": [[185, 4, 1, "", "forward"]], "torchtune.modules.TanhGate": [[186, 4, 1, "", "forward"]], "torchtune.modules.TransformerCrossAttentionLayer": [[188, 4, 1, "", "caches_are_enabled"], [188, 4, 1, "", "caches_are_setup"], [188, 4, 1, "", "forward"], [188, 4, 1, "", "reset_cache"], [188, 4, 1, "", "setup_caches"]], "torchtune.modules.TransformerDecoder": [[189, 4, 1, "", "caches_are_enabled"], [189, 4, 1, "", "caches_are_setup"], [189, 4, 1, "", "chunked_output"], [189, 4, 1, "", "forward"], [189, 4, 1, "", "reset_caches"], [189, 4, 1, "", "set_num_output_chunks"], [189, 4, 1, "", "setup_caches"]], "torchtune.modules.TransformerSelfAttentionLayer": [[190, 4, 1, "", "caches_are_enabled"], [190, 4, 1, "", "caches_are_setup"], [190, 4, 1, "", "forward"], [190, 4, 1, "", "reset_cache"], [190, 4, 1, "", "setup_caches"]], "torchtune.modules.VisionTransformer": [[191, 4, 1, "", "forward"]], "torchtune.modules.common_utils": [[192, 0, 1, "", "delete_kv_caches"], [193, 0, 1, "", "disable_kv_cache"], [194, 0, 1, "", "local_kv_cache"], [195, 0, 1, "", "reparametrize_as_dtype_state_dict_post_hook"]], "torchtune.modules.loss": [[196, 1, 1, "", "CEWithChunkedOutputLoss"], [197, 1, 1, "", "ForwardKLLoss"], [198, 1, 1, "", "ForwardKLWithChunkedOutputLoss"]], "torchtune.modules.loss.CEWithChunkedOutputLoss": [[196, 4, 1, "", "compute_cross_entropy"], [196, 4, 1, "", "forward"]], "torchtune.modules.loss.ForwardKLLoss": [[197, 4, 1, "", "forward"]], "torchtune.modules.loss.ForwardKLWithChunkedOutputLoss": [[198, 4, 1, "", "forward"]], "torchtune.modules.model_fusion": [[199, 1, 1, "", "DeepFusionModel"], [200, 1, 1, "", "FusionEmbedding"], [201, 1, 1, "", "FusionLayer"], [202, 0, 1, "", "get_fusion_params"], [203, 0, 1, "", "register_fusion_module"]], "torchtune.modules.model_fusion.DeepFusionModel": [[199, 4, 1, "", "caches_are_enabled"], [199, 4, 1, "", "caches_are_setup"], [199, 4, 1, "", "forward"], [199, 4, 1, "", "reset_caches"], [199, 4, 1, "", "set_num_output_chunks"], [199, 4, 1, "", "setup_caches"]], "torchtune.modules.model_fusion.FusionEmbedding": [[200, 4, 1, "", "forward"], [200, 4, 1, "", "fusion_params"]], "torchtune.modules.model_fusion.FusionLayer": [[201, 4, 1, "", "caches_are_enabled"], [201, 4, 1, "", "caches_are_setup"], [201, 4, 1, "", "forward"], [201, 4, 1, "", "fusion_params"], [201, 4, 1, "", "reset_cache"], [201, 4, 1, "", "setup_caches"]], "torchtune.modules.peft": [[204, 1, 1, "", "AdapterModule"], [205, 1, 1, "", "LoRALinear"], [206, 0, 1, "", "disable_adapter"], [207, 0, 1, "", "get_adapter_params"], [208, 0, 1, "", "set_trainable_params"], [209, 0, 1, "", "validate_missing_and_unexpected_for_lora"], [210, 0, 1, "", "validate_state_dict_for_lora"]], "torchtune.modules.peft.AdapterModule": [[204, 4, 1, "", "adapter_params"]], "torchtune.modules.peft.LoRALinear": [[205, 4, 1, "", "adapter_params"], [205, 4, 1, "", "forward"]], "torchtune.modules.tokenizers": [[211, 1, 1, "", "BaseTokenizer"], [212, 1, 1, "", "ModelTokenizer"], [213, 1, 1, "", "SentencePieceBaseTokenizer"], [214, 1, 1, "", "TikTokenBaseTokenizer"], [215, 0, 1, "", "parse_hf_tokenizer_json"], [216, 0, 1, "", "tokenize_messages_no_special_tokens"]], "torchtune.modules.tokenizers.BaseTokenizer": [[211, 4, 1, "", "decode"], [211, 4, 1, "", "encode"]], "torchtune.modules.tokenizers.ModelTokenizer": [[212, 4, 1, "", "tokenize_messages"]], "torchtune.modules.tokenizers.SentencePieceBaseTokenizer": [[213, 4, 1, "", "decode"], [213, 4, 1, "", "encode"]], "torchtune.modules.tokenizers.TikTokenBaseTokenizer": [[214, 4, 1, "", "decode"], [214, 4, 1, "", "encode"]], "torchtune.modules.transforms": [[217, 1, 1, "", "Transform"], [218, 1, 1, "", "VisionCrossAttentionMask"]], "torchtune.rlhf": [[219, 0, 1, "", "estimate_advantages"], [220, 0, 1, "", "get_rewards_ppo"], [225, 0, 1, "", "truncate_sequence_at_first_stop_token"]], "torchtune.rlhf.loss": [[221, 1, 1, "", "DPOLoss"], [222, 1, 1, "", "PPOLoss"], [223, 1, 1, "", "RSOLoss"], [224, 1, 1, "", "SimPOLoss"]], "torchtune.rlhf.loss.DPOLoss": [[221, 4, 1, "", "forward"]], "torchtune.rlhf.loss.PPOLoss": [[222, 4, 1, "", "forward"]], "torchtune.rlhf.loss.RSOLoss": [[223, 4, 1, "", "forward"]], "torchtune.rlhf.loss.SimPOLoss": [[224, 4, 1, "", "forward"]], "torchtune.training": [[226, 2, 1, "", "FSDPPolicyType"], [227, 1, 1, "", "FormattedCheckpointFiles"], [228, 1, 1, "", "FullModelHFCheckpointer"], [229, 1, 1, "", "FullModelMetaCheckpointer"], [230, 1, 1, "", "FullModelTorchTuneCheckpointer"], [231, 1, 1, "", "ModelType"], [232, 1, 1, "", "OptimizerInBackwardWrapper"], [233, 0, 1, "", "apply_selective_activation_checkpointing"], [234, 0, 1, "", "create_optim_in_bwd_wrapper"], [235, 0, 1, "", "get_cosine_schedule_with_warmup"], [236, 0, 1, "", "get_dtype"], [237, 0, 1, "", "get_full_finetune_fsdp_wrap_policy"], [238, 0, 1, "", "get_lr"], [239, 0, 1, "", "get_memory_stats"], [240, 0, 1, "", "get_quantizer_mode"], [241, 0, 1, "", "get_unmasked_sequence_lengths"], [242, 0, 1, "", "get_world_size_and_rank"], [243, 0, 1, "", "init_distributed"], [244, 0, 1, "", "is_distributed"], [245, 0, 1, "", "log_memory_stats"], [246, 0, 1, "", "lora_fsdp_wrap_policy"], [252, 0, 1, "", "register_optim_in_bwd_hooks"], [253, 0, 1, "", "set_activation_checkpointing"], [254, 0, 1, "", "set_default_dtype"], [255, 0, 1, "", "set_seed"], [256, 0, 1, "", "setup_torch_profiler"], [257, 0, 1, "", "update_state_dict_for_classifier"], [258, 0, 1, "", "validate_expected_param_dtype"]], "torchtune.training.FormattedCheckpointFiles": [[227, 4, 1, "", "build_checkpoint_filenames"]], "torchtune.training.FullModelHFCheckpointer": [[228, 4, 1, "", "load_checkpoint"], [228, 4, 1, "", "save_checkpoint"]], "torchtune.training.FullModelMetaCheckpointer": [[229, 4, 1, "", "load_checkpoint"], [229, 4, 1, "", "save_checkpoint"]], "torchtune.training.FullModelTorchTuneCheckpointer": [[230, 4, 1, "", "load_checkpoint"], [230, 4, 1, "", "save_checkpoint"]], "torchtune.training.OptimizerInBackwardWrapper": [[232, 4, 1, "", "get_last_lr"], [232, 4, 1, "", "get_optim_key"], [232, 4, 1, "", "load_state_dict"], [232, 4, 1, "", "set_lr_scheduler"], [232, 4, 1, "", "state_dict"], [232, 4, 1, "", "step_lr_scheduler"]], "torchtune.training.metric_logging": [[247, 1, 1, "", "CometLogger"], [248, 1, 1, "", "DiskLogger"], [249, 1, 1, "", "StdoutLogger"], [250, 1, 1, "", "TensorBoardLogger"], [251, 1, 1, "", "WandBLogger"]], "torchtune.training.metric_logging.CometLogger": [[247, 4, 1, "", "close"], [247, 4, 1, "", "log"], [247, 4, 1, "", "log_config"], [247, 4, 1, "", "log_dict"]], "torchtune.training.metric_logging.DiskLogger": [[248, 4, 1, "", "close"], [248, 4, 1, "", "log"], [248, 4, 1, "", "log_dict"]], "torchtune.training.metric_logging.StdoutLogger": [[249, 4, 1, "", "close"], [249, 4, 1, "", "log"], [249, 4, 1, "", "log_dict"]], "torchtune.training.metric_logging.TensorBoardLogger": [[250, 4, 1, "", "close"], [250, 4, 1, "", "log"], [250, 4, 1, "", "log_dict"]], "torchtune.training.metric_logging.WandBLogger": [[251, 4, 1, "", "close"], [251, 4, 1, "", "log"], [251, 4, 1, "", "log_config"], [251, 4, 1, "", "log_dict"]], "torchtune.utils": [[259, 0, 1, "", "batch_to_device"], [260, 0, 1, "", "get_device"], [261, 0, 1, "", "get_logger"], [262, 0, 1, "", "torch_version_ge"]]}, "objtypes": {"0": "py:function", "1": "py:class", "2": "py:data", "3": "py:property", "4": "py:method"}, "objnames": {"0": ["py", "function", "Python function"], "1": ["py", "class", "Python class"], "2": ["py", "data", "Python data"], "3": ["py", "property", "Python property"], "4": ["py", "method", "Python method"]}, "titleterms": {"torchtun": [0, 1, 2, 3, 4, 5, 6, 7, 8, 10, 22, 34, 40, 41, 43, 226, 265, 267, 272, 274, 276, 277, 278, 280, 281], "config": [0, 10, 24, 25, 272, 275], "data": [1, 11, 34, 40, 41, 43, 273], "text": [1, 2, 14, 16, 20, 276], "templat": [1, 9, 12, 14, 19, 21, 273], "type": 1, "messag": [1, 13, 14, 36], "transform": [1, 5, 13, 14, 15, 217], "collat": 1, "helper": 1, "function": 1, "dataset": [2, 9, 11, 12, 16, 18, 20, 273], "imag": [2, 14, 16], "gener": [2, 3, 73, 274, 276], "builder": 2, "class": [2, 19, 25], "model": [4, 5, 15, 21, 26, 272, 274, 275, 276, 277, 278, 279, 280], "llama3": [4, 116, 273, 276, 277, 280], "2": [4, 277], "vision": [4, 5], "1": [4, 277], "llama2": [4, 101, 273, 274, 278, 281], "code": 4, "llama": 4, "qwen": 4, "phi": 4, "3": 4, "mistral": [4, 158], "gemma": [4, 91], "clip": 4, "modul": 5, "compon": [5, 10, 24, 279], "build": [5, 266, 281], "block": 5, "loss": 5, "base": [5, 21], "token": [5, 14, 21, 273], "util": [5, 8], "peft": [5, 279], "fusion": 5, "rlhf": 6, "train": [7, 226, 269, 275], "checkpoint": [7, 22, 26, 274, 279], "reduc": 7, "precis": [7, 279], "distribut": [7, 269], "memori": [7, 278, 279, 281], "manag": 7, "schedul": 7, "metric": [7, 23, 26], "log": [7, 23, 26], "perform": [7, 278], "profil": 7, "miscellan": [7, 8], "chat": [9, 273], "exampl": [9, 12, 13, 15, 16, 18, 20], "format": [9, 12, 14, 16, 18, 20, 22], "load": [9, 12, 16, 18, 20, 21], "from": [9, 12, 16, 18, 20, 21, 273, 281], "hug": [9, 12, 16, 18, 20, 21, 274], "face": [9, 12, 16, 18, 20, 21, 274], "local": [9, 12, 16, 18, 20], "remot": [9, 12, 16], "specifi": 9, "convers": 9, "style": 9, "sharegpt": 9, "openai": 9, "renam": [9, 12], "column": [9, 12], "built": [9, 12, 16, 18, 19, 20, 272], "custom": [10, 13, 19, 273], "recip": [10, 25, 270, 272, 275, 277, 278, 280], "set": [10, 21], "up": [10, 274], "your": [10, 24, 25, 274, 275], "project": 10, "launch": 10, "overview": [11, 22, 267, 270, 274, 279], "pipelin": 11, "instruct": [12, 266, 276], "configur": [13, 24], "creat": [14, 15], "prompt": [14, 19, 21, 273], "access": [14, 276], "content": 14, "multimod": [15, 16], "us": [15, 19, 24, 25, 273, 274, 277, 281], "interleav": 16, "sampl": [17, 77], "pack": 17, "prefer": 18, "defin": 19, "via": [19, 266, 276], "dotpath": 19, "string": 19, "dictionari": 19, "prompttempl": [19, 38], "complet": 20, "json": 20, "txt": 20, "download": [21, 272, 274, 275], "file": 21, "max": 21, "sequenc": 21, "length": 21, "special": [21, 273], "handl": 22, "differ": 22, "hfcheckpoint": 22, "metacheckpoint": 22, "torchtunecheckpoint": 22, "intermedi": 22, "vs": 22, "final": 22, "lora": [22, 268, 274, 278, 279, 281], "put": [22, 281], "thi": 22, "all": [22, 24, 281], "togeth": [22, 281], "comet": 23, "logger": [23, 26], "about": 24, "where": 24, "do": 24, "paramet": [24, 279], "live": 24, "write": 24, "instanti": [24, 27], "referenc": 24, "other": [24, 274], "field": 24, "interpol": 24, "valid": [24, 30, 272], "best": 24, "practic": 24, "airtight": 24, "public": 24, "api": 24, "onli": 24, "command": 24, "line": 24, "overrid": 24, "remov": 24, "what": [25, 267, 277, 278, 280, 281], "ar": 25, "script": 25, "run": [25, 272, 274], "cli": [25, 272], "pars": [25, 29], "weight": 26, "bias": 26, "w": 26, "b": 26, "log_config": 28, "alpacatomessag": 31, "chatmltempl": 32, "chosenrejectedtomessag": 33, "grammarerrorcorrectiontempl": 34, "inputoutputtomessag": 35, "openaitomessag": 37, "prompttemplateinterfac": 39, "questionanswertempl": 40, "role": 41, "sharegpttomessag": 42, "summarizetempl": 43, "format_content_with_imag": 44, "left_pad_sequ": 45, "load_imag": 46, "padded_col": 47, "padded_collate_dpo": 48, "padded_collate_sft": 49, "padded_collate_tiled_images_and_mask": 50, "truncat": 51, "validate_messag": 52, "concatdataset": 53, "packeddataset": 54, "preferencedataset": 55, "sftdataset": 56, "textcompletiondataset": 57, "alpaca_cleaned_dataset": 58, "alpaca_dataset": 59, "chat_dataset": 60, "cnn_dailymail_articles_dataset": 61, "grammar_dataset": 62, "hh_rlhf_helpful_dataset": 63, "instruct_dataset": 64, "llava_instruct_dataset": 65, "the_cauldron_dataset": 66, "preference_dataset": 67, "samsum_dataset": 68, "slimorca_dataset": 69, "stack_exchange_paired_dataset": 70, "text_completion_dataset": 71, "wikitext_dataset": 72, "generate_next_token": 74, "get_causal_mask_from_padding_mask": 75, "get_position_ids_from_padding_mask": 76, "tilepositionalembed": 78, "tiledtokenpositionalembed": 79, "tokenpositionalembed": 80, "clip_vision_encod": 81, "code_llama2_13b": 82, "code_llama2_70b": 83, "code_llama2_7b": 84, "lora_code_llama2_13b": 85, "lora_code_llama2_70b": 86, "lora_code_llama2_7b": 87, "qlora_code_llama2_13b": 88, "qlora_code_llama2_70b": 89, "qlora_code_llama2_7b": 90, "gemma_2b": 92, "gemma_7b": 93, "gemma_token": 94, "lora_gemma": 95, "lora_gemma_2b": 96, "lora_gemma_7b": 97, "qlora_gemma_2b": 98, "qlora_gemma_7b": 99, "llama2chattempl": 100, "llama2_13b": 102, "llama2_70b": 103, "llama2_7b": 104, "llama2_reward_7b": 105, "llama2_token": 106, "lora_llama2": 107, "lora_llama2_13b": 108, "lora_llama2_70b": 109, "lora_llama2_7b": 110, "lora_llama2_reward_7b": 111, "qlora_llama2_13b": 112, "qlora_llama2_70b": 113, "qlora_llama2_7b": 114, "qlora_llama2_reward_7b": 115, "llama3_70b": 117, "llama3_8b": 118, "llama3_token": 119, "lora_llama3": 120, "lora_llama3_70b": 121, "lora_llama3_8b": 122, "qlora_llama3_70b": 123, "qlora_llama3_8b": 124, "llama3_1": 125, "llama3_1_405b": 126, "llama3_1_70b": 127, "llama3_1_8b": 128, "lora_llama3_1": 129, "lora_llama3_1_405b": 130, "lora_llama3_1_70b": 131, "lora_llama3_1_8b": 132, "qlora_llama3_1_405b": 133, "qlora_llama3_1_70b": 134, "qlora_llama3_1_8b": 135, "llama3_2_1b": 136, "llama3_2_3b": 137, "lora_llama3_2_1b": 138, "lora_llama3_2_3b": 139, "qlora_llama3_2_1b": 140, "qlora_llama3_2_3b": 141, "llama3visionencod": 142, "llama3visionprojectionhead": 143, "llama3visiontransform": 144, "llama3_2_vision_11b": 145, "llama3_2_vision_decod": 146, "llama3_2_vision_encod": 147, "llama3_2_vision_transform": 148, "lora_llama3_2_vision_11b": 149, "lora_llama3_2_vision_decod": 150, "lora_llama3_2_vision_encod": 151, "qlora_llama3_2_vision_11b": 152, "mistralchattempl": 153, "lora_mistr": 154, "lora_mistral_7b": 155, "lora_mistral_classifi": 156, "lora_mistral_reward_7b": 157, "mistral_7b": 159, "mistral_classifi": 160, "mistral_reward_7b": 161, "mistral_token": 162, "qlora_mistral_7b": 163, "qlora_mistral_reward_7b": 164, "lora_phi3": 165, "lora_phi3_mini": 166, "phi3": 167, "phi3_mini": 168, "phi3_mini_token": 169, "qlora_phi3_mini": 170, "lora_qwen2": 171, "lora_qwen2_0_5b": 172, "lora_qwen2_1_5b": 173, "lora_qwen2_7b": 174, "qwen2": [175, 277], "qwen2_0_5b": 176, "qwen2_1_5b": 177, "qwen2_7b": 178, "qwen2_token": 179, "feedforward": 180, "fp32layernorm": 181, "kvcach": 182, "multiheadattent": 183, "rmsnorm": 184, "rotarypositionalembed": 185, "tanhgat": 186, "tiedlinear": 187, "transformercrossattentionlay": 188, "transformerdecod": 189, "transformerselfattentionlay": 190, "visiontransform": 191, "delete_kv_cach": 192, "disable_kv_cach": 193, "local_kv_cach": 194, "reparametrize_as_dtype_state_dict_post_hook": 195, "cewithchunkedoutputloss": 196, "forwardklloss": 197, "forwardklwithchunkedoutputloss": 198, "deepfusionmodel": 199, "fusionembed": 200, "fusionlay": 201, "get_fusion_param": 202, "register_fusion_modul": 203, "adaptermodul": 204, "loralinear": 205, "disable_adapt": 206, "get_adapter_param": 207, "set_trainable_param": 208, "validate_missing_and_unexpected_for_lora": 209, "validate_state_dict_for_lora": 210, "basetoken": 211, "modeltoken": 212, "sentencepiecebasetoken": 213, "tiktokenbasetoken": 214, "parse_hf_tokenizer_json": 215, "tokenize_messages_no_special_token": 216, "visioncrossattentionmask": 218, "estimate_advantag": 219, "get_rewards_ppo": 220, "dpoloss": 221, "ppoloss": 222, "rsoloss": 223, "simpoloss": 224, "truncate_sequence_at_first_stop_token": 225, "fsdppolicytyp": 226, "formattedcheckpointfil": 227, "fullmodelhfcheckpoint": 228, "fullmodelmetacheckpoint": 229, "fullmodeltorchtunecheckpoint": 230, "modeltyp": 231, "optimizerinbackwardwrapp": 232, "apply_selective_activation_checkpoint": 233, "create_optim_in_bwd_wrapp": 234, "get_cosine_schedule_with_warmup": 235, "get_dtyp": 236, "get_full_finetune_fsdp_wrap_polici": 237, "get_lr": 238, "get_memory_stat": 239, "get_quantizer_mod": 240, "get_unmasked_sequence_length": 241, "get_world_size_and_rank": 242, "init_distribut": 243, "is_distribut": 244, "log_memory_stat": 245, "lora_fsdp_wrap_polici": 246, "cometlogg": 247, "disklogg": 248, "stdoutlogg": 249, "tensorboardlogg": 250, "wandblogg": 251, "register_optim_in_bwd_hook": 252, "set_activation_checkpoint": 253, "set_default_dtyp": 254, "set_se": 255, "setup_torch_profil": 256, "update_state_dict_for_classifi": 257, "validate_expected_param_dtyp": 258, "batch_to_devic": 259, "get_devic": 260, "get_logg": 261, "torch_version_g": 262, "comput": [264, 271], "time": [264, 271], "welcom": 265, "document": 265, "get": [265, 272, 276], "start": [265, 272], "tutori": 265, "instal": 266, "pre": 266, "requisit": 266, "pypi": 266, "git": 266, "clone": 266, "nightli": 266, "kei": 267, "concept": 267, "design": 267, "principl": 267, "singl": 268, "devic": [268, 280], "finetun": [268, 270, 274, 278, 280, 281], "quantiz": [269, 274, 276, 279, 280], "awar": 269, "qat": [269, 280], "list": 272, "copi": 272, "fine": [273, 275, 276, 277, 278, 279, 280, 281], "tune": [273, 275, 276, 277, 278, 279, 280, 281], "chang": 273, "when": 273, "should": 273, "i": 273, "end": 274, "workflow": 274, "7b": 274, "evalu": [274, 276, 280], "eleutherai": [274, 276], "s": [274, 276], "eval": [274, 276], "har": [274, 276], "speed": 274, "librari": 274, "upload": 274, "hub": 274, "first": 275, "llm": 275, "select": 275, "modifi": 275, "next": 275, "step": [275, 279], "meta": 276, "8b": [276, 277], "our": 276, "faster": 276, "distil": 277, "1b": 277, "knowledg": 277, "how": [277, 278], "doe": [277, 278], "work": [277, 278], "kd": 277, "ablat": 277, "studi": 277, "teacher": 277, "student": 277, "hyperparamet": 277, "learn": 277, "rate": 277, "ratio": 277, "5b": 277, "0": 277, "appli": [278, 280], "trade": 278, "off": 278, "optim": 279, "activ": 279, "offload": 279, "gradient": 279, "accumul": 279, "lower": [279, 280], "fuse": 279, "backward": 279, "pass": 279, "effici": 279, "low": 279, "rank": 279, "adapt": 279, "qlora": [279, 281], "option": 280, "save": 281, "deep": 281, "dive": 281}, "envversion": {"sphinx.domains.c": 2, "sphinx.domains.changeset": 1, "sphinx.domains.citation": 1, "sphinx.domains.cpp": 6, "sphinx.domains.index": 1, "sphinx.domains.javascript": 2, "sphinx.domains.math": 2, "sphinx.domains.python": 3, "sphinx.domains.rst": 2, "sphinx.domains.std": 2, "sphinx.ext.intersphinx": 1, "sphinx.ext.todo": 2, "sphinx.ext.viewcode": 1, "sphinx": 56}})
\ No newline at end of file
+Search.setIndex({"docnames": ["api_ref_config", "api_ref_data", "api_ref_datasets", "api_ref_generation", "api_ref_models", "api_ref_modules", "api_ref_rlhf", "api_ref_training", "api_ref_utilities", "basics/chat_datasets", "basics/custom_components", "basics/datasets_overview", "basics/instruct_datasets", "basics/message_transforms", "basics/messages", "basics/model_transforms", "basics/multimodal_datasets", "basics/packing", "basics/preference_datasets", "basics/prompt_templates", "basics/text_completion_datasets", "basics/tokenizers", "deep_dives/checkpointer", "deep_dives/comet_logging", "deep_dives/configs", "deep_dives/recipe_deepdive", "deep_dives/wandb_logging", "generated/torchtune.config.instantiate", "generated/torchtune.config.log_config", "generated/torchtune.config.parse", "generated/torchtune.config.validate", "generated/torchtune.data.AlpacaToMessages", "generated/torchtune.data.ChatMLTemplate", "generated/torchtune.data.ChosenRejectedToMessages", "generated/torchtune.data.GrammarErrorCorrectionTemplate", "generated/torchtune.data.InputOutputToMessages", "generated/torchtune.data.Message", "generated/torchtune.data.OpenAIToMessages", "generated/torchtune.data.PromptTemplate", "generated/torchtune.data.PromptTemplateInterface", "generated/torchtune.data.QuestionAnswerTemplate", "generated/torchtune.data.Role", "generated/torchtune.data.ShareGPTToMessages", "generated/torchtune.data.SummarizeTemplate", "generated/torchtune.data.format_content_with_images", "generated/torchtune.data.left_pad_sequence", "generated/torchtune.data.load_image", "generated/torchtune.data.padded_collate", "generated/torchtune.data.padded_collate_dpo", "generated/torchtune.data.padded_collate_sft", "generated/torchtune.data.padded_collate_tiled_images_and_mask", "generated/torchtune.data.truncate", "generated/torchtune.data.validate_messages", "generated/torchtune.datasets.ConcatDataset", "generated/torchtune.datasets.PackedDataset", "generated/torchtune.datasets.PreferenceDataset", "generated/torchtune.datasets.SFTDataset", "generated/torchtune.datasets.TextCompletionDataset", "generated/torchtune.datasets.alpaca_cleaned_dataset", "generated/torchtune.datasets.alpaca_dataset", "generated/torchtune.datasets.chat_dataset", "generated/torchtune.datasets.cnn_dailymail_articles_dataset", "generated/torchtune.datasets.grammar_dataset", "generated/torchtune.datasets.hh_rlhf_helpful_dataset", "generated/torchtune.datasets.instruct_dataset", "generated/torchtune.datasets.multimodal.llava_instruct_dataset", "generated/torchtune.datasets.multimodal.the_cauldron_dataset", "generated/torchtune.datasets.preference_dataset", "generated/torchtune.datasets.samsum_dataset", "generated/torchtune.datasets.slimorca_dataset", "generated/torchtune.datasets.stack_exchange_paired_dataset", "generated/torchtune.datasets.text_completion_dataset", "generated/torchtune.datasets.wikitext_dataset", "generated/torchtune.generation.generate", "generated/torchtune.generation.generate_next_token", "generated/torchtune.generation.get_causal_mask_from_padding_mask", "generated/torchtune.generation.get_position_ids_from_padding_mask", "generated/torchtune.generation.sample", "generated/torchtune.models.clip.TilePositionalEmbedding", "generated/torchtune.models.clip.TiledTokenPositionalEmbedding", "generated/torchtune.models.clip.TokenPositionalEmbedding", "generated/torchtune.models.clip.clip_vision_encoder", "generated/torchtune.models.code_llama2.code_llama2_13b", "generated/torchtune.models.code_llama2.code_llama2_70b", "generated/torchtune.models.code_llama2.code_llama2_7b", "generated/torchtune.models.code_llama2.lora_code_llama2_13b", "generated/torchtune.models.code_llama2.lora_code_llama2_70b", "generated/torchtune.models.code_llama2.lora_code_llama2_7b", "generated/torchtune.models.code_llama2.qlora_code_llama2_13b", "generated/torchtune.models.code_llama2.qlora_code_llama2_70b", "generated/torchtune.models.code_llama2.qlora_code_llama2_7b", "generated/torchtune.models.gemma.gemma", "generated/torchtune.models.gemma.gemma_2b", "generated/torchtune.models.gemma.gemma_7b", "generated/torchtune.models.gemma.gemma_tokenizer", "generated/torchtune.models.gemma.lora_gemma", "generated/torchtune.models.gemma.lora_gemma_2b", "generated/torchtune.models.gemma.lora_gemma_7b", "generated/torchtune.models.gemma.qlora_gemma_2b", "generated/torchtune.models.gemma.qlora_gemma_7b", "generated/torchtune.models.llama2.Llama2ChatTemplate", "generated/torchtune.models.llama2.llama2", "generated/torchtune.models.llama2.llama2_13b", "generated/torchtune.models.llama2.llama2_70b", "generated/torchtune.models.llama2.llama2_7b", "generated/torchtune.models.llama2.llama2_reward_7b", "generated/torchtune.models.llama2.llama2_tokenizer", "generated/torchtune.models.llama2.lora_llama2", "generated/torchtune.models.llama2.lora_llama2_13b", "generated/torchtune.models.llama2.lora_llama2_70b", "generated/torchtune.models.llama2.lora_llama2_7b", "generated/torchtune.models.llama2.lora_llama2_reward_7b", "generated/torchtune.models.llama2.qlora_llama2_13b", "generated/torchtune.models.llama2.qlora_llama2_70b", "generated/torchtune.models.llama2.qlora_llama2_7b", "generated/torchtune.models.llama2.qlora_llama2_reward_7b", "generated/torchtune.models.llama3.llama3", "generated/torchtune.models.llama3.llama3_70b", "generated/torchtune.models.llama3.llama3_8b", "generated/torchtune.models.llama3.llama3_tokenizer", "generated/torchtune.models.llama3.lora_llama3", "generated/torchtune.models.llama3.lora_llama3_70b", "generated/torchtune.models.llama3.lora_llama3_8b", "generated/torchtune.models.llama3.qlora_llama3_70b", "generated/torchtune.models.llama3.qlora_llama3_8b", "generated/torchtune.models.llama3_1.llama3_1", "generated/torchtune.models.llama3_1.llama3_1_405b", "generated/torchtune.models.llama3_1.llama3_1_70b", "generated/torchtune.models.llama3_1.llama3_1_8b", "generated/torchtune.models.llama3_1.lora_llama3_1", "generated/torchtune.models.llama3_1.lora_llama3_1_405b", "generated/torchtune.models.llama3_1.lora_llama3_1_70b", "generated/torchtune.models.llama3_1.lora_llama3_1_8b", "generated/torchtune.models.llama3_1.qlora_llama3_1_405b", "generated/torchtune.models.llama3_1.qlora_llama3_1_70b", "generated/torchtune.models.llama3_1.qlora_llama3_1_8b", "generated/torchtune.models.llama3_2.llama3_2_1b", "generated/torchtune.models.llama3_2.llama3_2_3b", "generated/torchtune.models.llama3_2.lora_llama3_2_1b", "generated/torchtune.models.llama3_2.lora_llama3_2_3b", "generated/torchtune.models.llama3_2.qlora_llama3_2_1b", "generated/torchtune.models.llama3_2.qlora_llama3_2_3b", "generated/torchtune.models.llama3_2_vision.Llama3VisionEncoder", "generated/torchtune.models.llama3_2_vision.Llama3VisionProjectionHead", "generated/torchtune.models.llama3_2_vision.Llama3VisionTransform", "generated/torchtune.models.llama3_2_vision.llama3_2_vision_11b", "generated/torchtune.models.llama3_2_vision.llama3_2_vision_decoder", "generated/torchtune.models.llama3_2_vision.llama3_2_vision_encoder", "generated/torchtune.models.llama3_2_vision.llama3_2_vision_transform", "generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_11b", "generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_decoder", "generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_encoder", "generated/torchtune.models.llama3_2_vision.qlora_llama3_2_vision_11b", "generated/torchtune.models.mistral.MistralChatTemplate", "generated/torchtune.models.mistral.lora_mistral", "generated/torchtune.models.mistral.lora_mistral_7b", "generated/torchtune.models.mistral.lora_mistral_classifier", "generated/torchtune.models.mistral.lora_mistral_reward_7b", "generated/torchtune.models.mistral.mistral", "generated/torchtune.models.mistral.mistral_7b", "generated/torchtune.models.mistral.mistral_classifier", "generated/torchtune.models.mistral.mistral_reward_7b", "generated/torchtune.models.mistral.mistral_tokenizer", "generated/torchtune.models.mistral.qlora_mistral_7b", "generated/torchtune.models.mistral.qlora_mistral_reward_7b", "generated/torchtune.models.phi3.lora_phi3", "generated/torchtune.models.phi3.lora_phi3_mini", "generated/torchtune.models.phi3.phi3", "generated/torchtune.models.phi3.phi3_mini", "generated/torchtune.models.phi3.phi3_mini_tokenizer", "generated/torchtune.models.phi3.qlora_phi3_mini", "generated/torchtune.models.qwen2.lora_qwen2", "generated/torchtune.models.qwen2.lora_qwen2_0_5b", "generated/torchtune.models.qwen2.lora_qwen2_1_5b", "generated/torchtune.models.qwen2.lora_qwen2_7b", "generated/torchtune.models.qwen2.qwen2", "generated/torchtune.models.qwen2.qwen2_0_5b", "generated/torchtune.models.qwen2.qwen2_1_5b", "generated/torchtune.models.qwen2.qwen2_7b", "generated/torchtune.models.qwen2.qwen2_tokenizer", "generated/torchtune.modules.FeedForward", "generated/torchtune.modules.Fp32LayerNorm", "generated/torchtune.modules.KVCache", "generated/torchtune.modules.MultiHeadAttention", "generated/torchtune.modules.RMSNorm", "generated/torchtune.modules.RotaryPositionalEmbeddings", "generated/torchtune.modules.TanhGate", "generated/torchtune.modules.TiedLinear", "generated/torchtune.modules.TransformerCrossAttentionLayer", "generated/torchtune.modules.TransformerDecoder", "generated/torchtune.modules.TransformerSelfAttentionLayer", "generated/torchtune.modules.VisionTransformer", "generated/torchtune.modules.common_utils.delete_kv_caches", "generated/torchtune.modules.common_utils.disable_kv_cache", "generated/torchtune.modules.common_utils.local_kv_cache", "generated/torchtune.modules.common_utils.reparametrize_as_dtype_state_dict_post_hook", "generated/torchtune.modules.loss.CEWithChunkedOutputLoss", "generated/torchtune.modules.loss.ForwardKLLoss", "generated/torchtune.modules.loss.ForwardKLWithChunkedOutputLoss", "generated/torchtune.modules.model_fusion.DeepFusionModel", "generated/torchtune.modules.model_fusion.FusionEmbedding", "generated/torchtune.modules.model_fusion.FusionLayer", "generated/torchtune.modules.model_fusion.get_fusion_params", "generated/torchtune.modules.model_fusion.register_fusion_module", "generated/torchtune.modules.peft.AdapterModule", "generated/torchtune.modules.peft.LoRALinear", "generated/torchtune.modules.peft.disable_adapter", "generated/torchtune.modules.peft.get_adapter_params", "generated/torchtune.modules.peft.set_trainable_params", "generated/torchtune.modules.peft.validate_missing_and_unexpected_for_lora", "generated/torchtune.modules.peft.validate_state_dict_for_lora", "generated/torchtune.modules.tokenizers.BaseTokenizer", "generated/torchtune.modules.tokenizers.ModelTokenizer", "generated/torchtune.modules.tokenizers.SentencePieceBaseTokenizer", "generated/torchtune.modules.tokenizers.TikTokenBaseTokenizer", "generated/torchtune.modules.tokenizers.parse_hf_tokenizer_json", "generated/torchtune.modules.tokenizers.tokenize_messages_no_special_tokens", "generated/torchtune.modules.transforms.Transform", "generated/torchtune.modules.transforms.VisionCrossAttentionMask", "generated/torchtune.rlhf.estimate_advantages", "generated/torchtune.rlhf.get_rewards_ppo", "generated/torchtune.rlhf.loss.DPOLoss", "generated/torchtune.rlhf.loss.PPOLoss", "generated/torchtune.rlhf.loss.RSOLoss", "generated/torchtune.rlhf.loss.SimPOLoss", "generated/torchtune.rlhf.truncate_sequence_at_first_stop_token", "generated/torchtune.training.FSDPPolicyType", "generated/torchtune.training.FormattedCheckpointFiles", "generated/torchtune.training.FullModelHFCheckpointer", "generated/torchtune.training.FullModelMetaCheckpointer", "generated/torchtune.training.FullModelTorchTuneCheckpointer", "generated/torchtune.training.ModelType", "generated/torchtune.training.OptimizerInBackwardWrapper", "generated/torchtune.training.apply_selective_activation_checkpointing", "generated/torchtune.training.create_optim_in_bwd_wrapper", "generated/torchtune.training.get_cosine_schedule_with_warmup", "generated/torchtune.training.get_dtype", "generated/torchtune.training.get_full_finetune_fsdp_wrap_policy", "generated/torchtune.training.get_lr", "generated/torchtune.training.get_memory_stats", "generated/torchtune.training.get_quantizer_mode", "generated/torchtune.training.get_unmasked_sequence_lengths", "generated/torchtune.training.get_world_size_and_rank", "generated/torchtune.training.init_distributed", "generated/torchtune.training.is_distributed", "generated/torchtune.training.log_memory_stats", "generated/torchtune.training.lora_fsdp_wrap_policy", "generated/torchtune.training.metric_logging.CometLogger", "generated/torchtune.training.metric_logging.DiskLogger", "generated/torchtune.training.metric_logging.StdoutLogger", "generated/torchtune.training.metric_logging.TensorBoardLogger", "generated/torchtune.training.metric_logging.WandBLogger", "generated/torchtune.training.register_optim_in_bwd_hooks", "generated/torchtune.training.set_activation_checkpointing", "generated/torchtune.training.set_default_dtype", "generated/torchtune.training.set_seed", "generated/torchtune.training.setup_torch_profiler", "generated/torchtune.training.update_state_dict_for_classifier", "generated/torchtune.training.validate_expected_param_dtype", "generated/torchtune.utils.batch_to_device", "generated/torchtune.utils.get_device", "generated/torchtune.utils.get_logger", "generated/torchtune.utils.torch_version_ge", "generated_examples/index", "generated_examples/sg_execution_times", "index", "install", "overview", "recipes/lora_finetune_single_device", "recipes/qat_distributed", "recipes/recipes_overview", "sg_execution_times", "tune_cli", "tutorials/chat", "tutorials/e2e_flow", "tutorials/first_finetune_tutorial", "tutorials/llama3", "tutorials/llama_kd_tutorial", "tutorials/lora_finetune", "tutorials/memory_optimizations", "tutorials/qat_finetune", "tutorials/qlora_finetune"], "filenames": ["api_ref_config.rst", "api_ref_data.rst", "api_ref_datasets.rst", "api_ref_generation.rst", "api_ref_models.rst", "api_ref_modules.rst", "api_ref_rlhf.rst", "api_ref_training.rst", "api_ref_utilities.rst", "basics/chat_datasets.rst", "basics/custom_components.rst", "basics/datasets_overview.rst", "basics/instruct_datasets.rst", "basics/message_transforms.rst", "basics/messages.rst", "basics/model_transforms.rst", "basics/multimodal_datasets.rst", "basics/packing.rst", "basics/preference_datasets.rst", "basics/prompt_templates.rst", "basics/text_completion_datasets.rst", "basics/tokenizers.rst", "deep_dives/checkpointer.rst", "deep_dives/comet_logging.rst", "deep_dives/configs.rst", "deep_dives/recipe_deepdive.rst", "deep_dives/wandb_logging.rst", "generated/torchtune.config.instantiate.rst", "generated/torchtune.config.log_config.rst", "generated/torchtune.config.parse.rst", "generated/torchtune.config.validate.rst", "generated/torchtune.data.AlpacaToMessages.rst", "generated/torchtune.data.ChatMLTemplate.rst", "generated/torchtune.data.ChosenRejectedToMessages.rst", "generated/torchtune.data.GrammarErrorCorrectionTemplate.rst", "generated/torchtune.data.InputOutputToMessages.rst", "generated/torchtune.data.Message.rst", "generated/torchtune.data.OpenAIToMessages.rst", "generated/torchtune.data.PromptTemplate.rst", "generated/torchtune.data.PromptTemplateInterface.rst", "generated/torchtune.data.QuestionAnswerTemplate.rst", "generated/torchtune.data.Role.rst", "generated/torchtune.data.ShareGPTToMessages.rst", "generated/torchtune.data.SummarizeTemplate.rst", "generated/torchtune.data.format_content_with_images.rst", "generated/torchtune.data.left_pad_sequence.rst", "generated/torchtune.data.load_image.rst", "generated/torchtune.data.padded_collate.rst", "generated/torchtune.data.padded_collate_dpo.rst", "generated/torchtune.data.padded_collate_sft.rst", "generated/torchtune.data.padded_collate_tiled_images_and_mask.rst", "generated/torchtune.data.truncate.rst", "generated/torchtune.data.validate_messages.rst", "generated/torchtune.datasets.ConcatDataset.rst", "generated/torchtune.datasets.PackedDataset.rst", "generated/torchtune.datasets.PreferenceDataset.rst", "generated/torchtune.datasets.SFTDataset.rst", "generated/torchtune.datasets.TextCompletionDataset.rst", "generated/torchtune.datasets.alpaca_cleaned_dataset.rst", "generated/torchtune.datasets.alpaca_dataset.rst", "generated/torchtune.datasets.chat_dataset.rst", "generated/torchtune.datasets.cnn_dailymail_articles_dataset.rst", "generated/torchtune.datasets.grammar_dataset.rst", "generated/torchtune.datasets.hh_rlhf_helpful_dataset.rst", "generated/torchtune.datasets.instruct_dataset.rst", "generated/torchtune.datasets.multimodal.llava_instruct_dataset.rst", "generated/torchtune.datasets.multimodal.the_cauldron_dataset.rst", "generated/torchtune.datasets.preference_dataset.rst", "generated/torchtune.datasets.samsum_dataset.rst", "generated/torchtune.datasets.slimorca_dataset.rst", "generated/torchtune.datasets.stack_exchange_paired_dataset.rst", "generated/torchtune.datasets.text_completion_dataset.rst", "generated/torchtune.datasets.wikitext_dataset.rst", "generated/torchtune.generation.generate.rst", "generated/torchtune.generation.generate_next_token.rst", "generated/torchtune.generation.get_causal_mask_from_padding_mask.rst", "generated/torchtune.generation.get_position_ids_from_padding_mask.rst", "generated/torchtune.generation.sample.rst", "generated/torchtune.models.clip.TilePositionalEmbedding.rst", "generated/torchtune.models.clip.TiledTokenPositionalEmbedding.rst", "generated/torchtune.models.clip.TokenPositionalEmbedding.rst", "generated/torchtune.models.clip.clip_vision_encoder.rst", "generated/torchtune.models.code_llama2.code_llama2_13b.rst", "generated/torchtune.models.code_llama2.code_llama2_70b.rst", "generated/torchtune.models.code_llama2.code_llama2_7b.rst", "generated/torchtune.models.code_llama2.lora_code_llama2_13b.rst", "generated/torchtune.models.code_llama2.lora_code_llama2_70b.rst", "generated/torchtune.models.code_llama2.lora_code_llama2_7b.rst", "generated/torchtune.models.code_llama2.qlora_code_llama2_13b.rst", "generated/torchtune.models.code_llama2.qlora_code_llama2_70b.rst", "generated/torchtune.models.code_llama2.qlora_code_llama2_7b.rst", "generated/torchtune.models.gemma.gemma.rst", "generated/torchtune.models.gemma.gemma_2b.rst", "generated/torchtune.models.gemma.gemma_7b.rst", "generated/torchtune.models.gemma.gemma_tokenizer.rst", "generated/torchtune.models.gemma.lora_gemma.rst", "generated/torchtune.models.gemma.lora_gemma_2b.rst", "generated/torchtune.models.gemma.lora_gemma_7b.rst", "generated/torchtune.models.gemma.qlora_gemma_2b.rst", "generated/torchtune.models.gemma.qlora_gemma_7b.rst", "generated/torchtune.models.llama2.Llama2ChatTemplate.rst", "generated/torchtune.models.llama2.llama2.rst", "generated/torchtune.models.llama2.llama2_13b.rst", "generated/torchtune.models.llama2.llama2_70b.rst", "generated/torchtune.models.llama2.llama2_7b.rst", "generated/torchtune.models.llama2.llama2_reward_7b.rst", "generated/torchtune.models.llama2.llama2_tokenizer.rst", "generated/torchtune.models.llama2.lora_llama2.rst", "generated/torchtune.models.llama2.lora_llama2_13b.rst", "generated/torchtune.models.llama2.lora_llama2_70b.rst", "generated/torchtune.models.llama2.lora_llama2_7b.rst", "generated/torchtune.models.llama2.lora_llama2_reward_7b.rst", "generated/torchtune.models.llama2.qlora_llama2_13b.rst", "generated/torchtune.models.llama2.qlora_llama2_70b.rst", "generated/torchtune.models.llama2.qlora_llama2_7b.rst", "generated/torchtune.models.llama2.qlora_llama2_reward_7b.rst", "generated/torchtune.models.llama3.llama3.rst", "generated/torchtune.models.llama3.llama3_70b.rst", "generated/torchtune.models.llama3.llama3_8b.rst", "generated/torchtune.models.llama3.llama3_tokenizer.rst", "generated/torchtune.models.llama3.lora_llama3.rst", "generated/torchtune.models.llama3.lora_llama3_70b.rst", "generated/torchtune.models.llama3.lora_llama3_8b.rst", "generated/torchtune.models.llama3.qlora_llama3_70b.rst", "generated/torchtune.models.llama3.qlora_llama3_8b.rst", "generated/torchtune.models.llama3_1.llama3_1.rst", "generated/torchtune.models.llama3_1.llama3_1_405b.rst", "generated/torchtune.models.llama3_1.llama3_1_70b.rst", "generated/torchtune.models.llama3_1.llama3_1_8b.rst", "generated/torchtune.models.llama3_1.lora_llama3_1.rst", "generated/torchtune.models.llama3_1.lora_llama3_1_405b.rst", "generated/torchtune.models.llama3_1.lora_llama3_1_70b.rst", "generated/torchtune.models.llama3_1.lora_llama3_1_8b.rst", "generated/torchtune.models.llama3_1.qlora_llama3_1_405b.rst", "generated/torchtune.models.llama3_1.qlora_llama3_1_70b.rst", "generated/torchtune.models.llama3_1.qlora_llama3_1_8b.rst", "generated/torchtune.models.llama3_2.llama3_2_1b.rst", "generated/torchtune.models.llama3_2.llama3_2_3b.rst", "generated/torchtune.models.llama3_2.lora_llama3_2_1b.rst", "generated/torchtune.models.llama3_2.lora_llama3_2_3b.rst", "generated/torchtune.models.llama3_2.qlora_llama3_2_1b.rst", "generated/torchtune.models.llama3_2.qlora_llama3_2_3b.rst", "generated/torchtune.models.llama3_2_vision.Llama3VisionEncoder.rst", "generated/torchtune.models.llama3_2_vision.Llama3VisionProjectionHead.rst", "generated/torchtune.models.llama3_2_vision.Llama3VisionTransform.rst", "generated/torchtune.models.llama3_2_vision.llama3_2_vision_11b.rst", "generated/torchtune.models.llama3_2_vision.llama3_2_vision_decoder.rst", "generated/torchtune.models.llama3_2_vision.llama3_2_vision_encoder.rst", "generated/torchtune.models.llama3_2_vision.llama3_2_vision_transform.rst", "generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_11b.rst", "generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_decoder.rst", "generated/torchtune.models.llama3_2_vision.lora_llama3_2_vision_encoder.rst", "generated/torchtune.models.llama3_2_vision.qlora_llama3_2_vision_11b.rst", "generated/torchtune.models.mistral.MistralChatTemplate.rst", "generated/torchtune.models.mistral.lora_mistral.rst", "generated/torchtune.models.mistral.lora_mistral_7b.rst", "generated/torchtune.models.mistral.lora_mistral_classifier.rst", "generated/torchtune.models.mistral.lora_mistral_reward_7b.rst", "generated/torchtune.models.mistral.mistral.rst", "generated/torchtune.models.mistral.mistral_7b.rst", "generated/torchtune.models.mistral.mistral_classifier.rst", "generated/torchtune.models.mistral.mistral_reward_7b.rst", "generated/torchtune.models.mistral.mistral_tokenizer.rst", "generated/torchtune.models.mistral.qlora_mistral_7b.rst", "generated/torchtune.models.mistral.qlora_mistral_reward_7b.rst", "generated/torchtune.models.phi3.lora_phi3.rst", "generated/torchtune.models.phi3.lora_phi3_mini.rst", "generated/torchtune.models.phi3.phi3.rst", "generated/torchtune.models.phi3.phi3_mini.rst", "generated/torchtune.models.phi3.phi3_mini_tokenizer.rst", "generated/torchtune.models.phi3.qlora_phi3_mini.rst", "generated/torchtune.models.qwen2.lora_qwen2.rst", "generated/torchtune.models.qwen2.lora_qwen2_0_5b.rst", "generated/torchtune.models.qwen2.lora_qwen2_1_5b.rst", "generated/torchtune.models.qwen2.lora_qwen2_7b.rst", "generated/torchtune.models.qwen2.qwen2.rst", "generated/torchtune.models.qwen2.qwen2_0_5b.rst", "generated/torchtune.models.qwen2.qwen2_1_5b.rst", "generated/torchtune.models.qwen2.qwen2_7b.rst", "generated/torchtune.models.qwen2.qwen2_tokenizer.rst", "generated/torchtune.modules.FeedForward.rst", "generated/torchtune.modules.Fp32LayerNorm.rst", "generated/torchtune.modules.KVCache.rst", "generated/torchtune.modules.MultiHeadAttention.rst", "generated/torchtune.modules.RMSNorm.rst", "generated/torchtune.modules.RotaryPositionalEmbeddings.rst", "generated/torchtune.modules.TanhGate.rst", "generated/torchtune.modules.TiedLinear.rst", "generated/torchtune.modules.TransformerCrossAttentionLayer.rst", "generated/torchtune.modules.TransformerDecoder.rst", "generated/torchtune.modules.TransformerSelfAttentionLayer.rst", "generated/torchtune.modules.VisionTransformer.rst", "generated/torchtune.modules.common_utils.delete_kv_caches.rst", "generated/torchtune.modules.common_utils.disable_kv_cache.rst", "generated/torchtune.modules.common_utils.local_kv_cache.rst", "generated/torchtune.modules.common_utils.reparametrize_as_dtype_state_dict_post_hook.rst", "generated/torchtune.modules.loss.CEWithChunkedOutputLoss.rst", "generated/torchtune.modules.loss.ForwardKLLoss.rst", "generated/torchtune.modules.loss.ForwardKLWithChunkedOutputLoss.rst", "generated/torchtune.modules.model_fusion.DeepFusionModel.rst", "generated/torchtune.modules.model_fusion.FusionEmbedding.rst", "generated/torchtune.modules.model_fusion.FusionLayer.rst", "generated/torchtune.modules.model_fusion.get_fusion_params.rst", "generated/torchtune.modules.model_fusion.register_fusion_module.rst", "generated/torchtune.modules.peft.AdapterModule.rst", "generated/torchtune.modules.peft.LoRALinear.rst", "generated/torchtune.modules.peft.disable_adapter.rst", "generated/torchtune.modules.peft.get_adapter_params.rst", "generated/torchtune.modules.peft.set_trainable_params.rst", "generated/torchtune.modules.peft.validate_missing_and_unexpected_for_lora.rst", "generated/torchtune.modules.peft.validate_state_dict_for_lora.rst", "generated/torchtune.modules.tokenizers.BaseTokenizer.rst", "generated/torchtune.modules.tokenizers.ModelTokenizer.rst", "generated/torchtune.modules.tokenizers.SentencePieceBaseTokenizer.rst", "generated/torchtune.modules.tokenizers.TikTokenBaseTokenizer.rst", "generated/torchtune.modules.tokenizers.parse_hf_tokenizer_json.rst", "generated/torchtune.modules.tokenizers.tokenize_messages_no_special_tokens.rst", "generated/torchtune.modules.transforms.Transform.rst", "generated/torchtune.modules.transforms.VisionCrossAttentionMask.rst", "generated/torchtune.rlhf.estimate_advantages.rst", "generated/torchtune.rlhf.get_rewards_ppo.rst", "generated/torchtune.rlhf.loss.DPOLoss.rst", "generated/torchtune.rlhf.loss.PPOLoss.rst", "generated/torchtune.rlhf.loss.RSOLoss.rst", "generated/torchtune.rlhf.loss.SimPOLoss.rst", "generated/torchtune.rlhf.truncate_sequence_at_first_stop_token.rst", "generated/torchtune.training.FSDPPolicyType.rst", "generated/torchtune.training.FormattedCheckpointFiles.rst", "generated/torchtune.training.FullModelHFCheckpointer.rst", "generated/torchtune.training.FullModelMetaCheckpointer.rst", "generated/torchtune.training.FullModelTorchTuneCheckpointer.rst", "generated/torchtune.training.ModelType.rst", "generated/torchtune.training.OptimizerInBackwardWrapper.rst", "generated/torchtune.training.apply_selective_activation_checkpointing.rst", "generated/torchtune.training.create_optim_in_bwd_wrapper.rst", "generated/torchtune.training.get_cosine_schedule_with_warmup.rst", "generated/torchtune.training.get_dtype.rst", "generated/torchtune.training.get_full_finetune_fsdp_wrap_policy.rst", "generated/torchtune.training.get_lr.rst", "generated/torchtune.training.get_memory_stats.rst", "generated/torchtune.training.get_quantizer_mode.rst", "generated/torchtune.training.get_unmasked_sequence_lengths.rst", "generated/torchtune.training.get_world_size_and_rank.rst", "generated/torchtune.training.init_distributed.rst", "generated/torchtune.training.is_distributed.rst", "generated/torchtune.training.log_memory_stats.rst", "generated/torchtune.training.lora_fsdp_wrap_policy.rst", "generated/torchtune.training.metric_logging.CometLogger.rst", "generated/torchtune.training.metric_logging.DiskLogger.rst", "generated/torchtune.training.metric_logging.StdoutLogger.rst", "generated/torchtune.training.metric_logging.TensorBoardLogger.rst", "generated/torchtune.training.metric_logging.WandBLogger.rst", "generated/torchtune.training.register_optim_in_bwd_hooks.rst", "generated/torchtune.training.set_activation_checkpointing.rst", "generated/torchtune.training.set_default_dtype.rst", "generated/torchtune.training.set_seed.rst", "generated/torchtune.training.setup_torch_profiler.rst", "generated/torchtune.training.update_state_dict_for_classifier.rst", "generated/torchtune.training.validate_expected_param_dtype.rst", "generated/torchtune.utils.batch_to_device.rst", "generated/torchtune.utils.get_device.rst", "generated/torchtune.utils.get_logger.rst", "generated/torchtune.utils.torch_version_ge.rst", "generated_examples/index.rst", "generated_examples/sg_execution_times.rst", "index.rst", "install.rst", "overview.rst", "recipes/lora_finetune_single_device.rst", "recipes/qat_distributed.rst", "recipes/recipes_overview.rst", "sg_execution_times.rst", "tune_cli.rst", "tutorials/chat.rst", "tutorials/e2e_flow.rst", "tutorials/first_finetune_tutorial.rst", "tutorials/llama3.rst", "tutorials/llama_kd_tutorial.rst", "tutorials/lora_finetune.rst", "tutorials/memory_optimizations.rst", "tutorials/qat_finetune.rst", "tutorials/qlora_finetune.rst"], "titles": ["torchtune.config", "torchtune.data", "torchtune.datasets", "torchtune.generation", "torchtune.models", "torchtune.modules", "torchtune.rlhf", "torchtune.training", "torchtune.utils", "Chat Datasets", "Custom Components and Recipes", "Datasets Overview", "Instruct Datasets", "Message Transforms", "Messages", "Multimodal Transforms", "Multimodal Datasets", "Sample packing", "Preference Datasets", "Prompt Templates", "Text-completion Datasets", "Tokenizers", "Checkpointing in torchtune", "Logging to Comet", "All About Configs", "What Are Recipes?", "Logging to Weights &amp; Biases", "instantiate", "log_config", "parse", "validate", "AlpacaToMessages", "ChatMLTemplate", "ChosenRejectedToMessages", "torchtune.data.GrammarErrorCorrectionTemplate", "InputOutputToMessages", "Message", "OpenAIToMessages", "PromptTemplate", "PromptTemplateInterface", "torchtune.data.QuestionAnswerTemplate", "torchtune.data.Role", "ShareGPTToMessages", "torchtune.data.SummarizeTemplate", "format_content_with_images", "left_pad_sequence", "load_image", "padded_collate", "padded_collate_dpo", "padded_collate_sft", "padded_collate_tiled_images_and_mask", "truncate", "validate_messages", "ConcatDataset", "PackedDataset", "PreferenceDataset", "SFTDataset", "TextCompletionDataset", "alpaca_cleaned_dataset", "alpaca_dataset", "chat_dataset", "cnn_dailymail_articles_dataset", "grammar_dataset", "hh_rlhf_helpful_dataset", "instruct_dataset", "llava_instruct_dataset", "the_cauldron_dataset", "preference_dataset", "samsum_dataset", "slimorca_dataset", "stack_exchange_paired_dataset", "text_completion_dataset", "wikitext_dataset", "generate", "generate_next_token", "get_causal_mask_from_padding_mask", "get_position_ids_from_padding_mask", "sample", "TilePositionalEmbedding", "TiledTokenPositionalEmbedding", "TokenPositionalEmbedding", "clip_vision_encoder", "code_llama2_13b", "code_llama2_70b", "code_llama2_7b", "lora_code_llama2_13b", "lora_code_llama2_70b", "lora_code_llama2_7b", "qlora_code_llama2_13b", "qlora_code_llama2_70b", "qlora_code_llama2_7b", "gemma", "gemma_2b", "gemma_7b", "gemma_tokenizer", "lora_gemma", "lora_gemma_2b", "lora_gemma_7b", "qlora_gemma_2b", "qlora_gemma_7b", "Llama2ChatTemplate", "llama2", "llama2_13b", "llama2_70b", "llama2_7b", "llama2_reward_7b", "llama2_tokenizer", "lora_llama2", "lora_llama2_13b", "lora_llama2_70b", "lora_llama2_7b", "lora_llama2_reward_7b", "qlora_llama2_13b", "qlora_llama2_70b", "qlora_llama2_7b", "qlora_llama2_reward_7b", "llama3", "llama3_70b", "llama3_8b", "llama3_tokenizer", "lora_llama3", "lora_llama3_70b", "lora_llama3_8b", "qlora_llama3_70b", "qlora_llama3_8b", "llama3_1", "llama3_1_405b", "llama3_1_70b", "llama3_1_8b", "lora_llama3_1", "lora_llama3_1_405b", "lora_llama3_1_70b", "lora_llama3_1_8b", "qlora_llama3_1_405b", "qlora_llama3_1_70b", "qlora_llama3_1_8b", "llama3_2_1b", "llama3_2_3b", "lora_llama3_2_1b", "lora_llama3_2_3b", "qlora_llama3_2_1b", "qlora_llama3_2_3b", "Llama3VisionEncoder", "Llama3VisionProjectionHead", "Llama3VisionTransform", "llama3_2_vision_11b", "llama3_2_vision_decoder", "llama3_2_vision_encoder", "llama3_2_vision_transform", "lora_llama3_2_vision_11b", "lora_llama3_2_vision_decoder", "lora_llama3_2_vision_encoder", "qlora_llama3_2_vision_11b", "MistralChatTemplate", "lora_mistral", "lora_mistral_7b", "lora_mistral_classifier", "lora_mistral_reward_7b", "mistral", "mistral_7b", "mistral_classifier", "mistral_reward_7b", "mistral_tokenizer", "qlora_mistral_7b", "qlora_mistral_reward_7b", "lora_phi3", "lora_phi3_mini", "phi3", "phi3_mini", "phi3_mini_tokenizer", "qlora_phi3_mini", "lora_qwen2", "lora_qwen2_0_5b", "lora_qwen2_1_5b", "lora_qwen2_7b", "qwen2", "qwen2_0_5b", "qwen2_1_5b", "qwen2_7b", "qwen2_tokenizer", "FeedForward", "Fp32LayerNorm", "KVCache", "MultiHeadAttention", "RMSNorm", "RotaryPositionalEmbeddings", "TanhGate", "TiedLinear", "TransformerCrossAttentionLayer", "TransformerDecoder", "TransformerSelfAttentionLayer", "VisionTransformer", "delete_kv_caches", "disable_kv_cache", "local_kv_cache", "reparametrize_as_dtype_state_dict_post_hook", "CEWithChunkedOutputLoss", "ForwardKLLoss", "ForwardKLWithChunkedOutputLoss", "DeepFusionModel", "FusionEmbedding", "FusionLayer", "get_fusion_params", "register_fusion_module", "AdapterModule", "LoRALinear", "disable_adapter", "get_adapter_params", "set_trainable_params", "validate_missing_and_unexpected_for_lora", "validate_state_dict_for_lora", "BaseTokenizer", "ModelTokenizer", "SentencePieceBaseTokenizer", "TikTokenBaseTokenizer", "parse_hf_tokenizer_json", "tokenize_messages_no_special_tokens", "Transform", "VisionCrossAttentionMask", "estimate_advantages", "get_rewards_ppo", "DPOLoss", "PPOLoss", "RSOLoss", "SimPOLoss", "truncate_sequence_at_first_stop_token", "torchtune.training.FSDPPolicyType", "FormattedCheckpointFiles", "FullModelHFCheckpointer", "FullModelMetaCheckpointer", "FullModelTorchTuneCheckpointer", "ModelType", "OptimizerInBackwardWrapper", "apply_selective_activation_checkpointing", "create_optim_in_bwd_wrapper", "get_cosine_schedule_with_warmup", "get_dtype", "get_full_finetune_fsdp_wrap_policy", "get_lr", "get_memory_stats", "get_quantizer_mode", "get_unmasked_sequence_lengths", "get_world_size_and_rank", "init_distributed", "is_distributed", "log_memory_stats", "lora_fsdp_wrap_policy", "CometLogger", "DiskLogger", "StdoutLogger", "TensorBoardLogger", "WandBLogger", "register_optim_in_bwd_hooks", "set_activation_checkpointing", "set_default_dtype", "set_seed", "setup_torch_profiler", "update_state_dict_for_classifier", "validate_expected_param_dtype", "batch_to_device", "get_device", "get_logger", "torch_version_ge", "&lt;no title&gt;", "Computation times", "Welcome to the torchtune Documentation", "Install Instructions", "torchtune Overview", "LoRA Single Device Finetuning", "Distributed Quantization-Aware Training (QAT)", "Recipes Overview", "Computation times", "torchtune CLI", "Fine-Tuning Llama3 with Chat Data", "End-to-End Workflow with torchtune", "Fine-Tune Your First LLM", "Meta Llama3 in torchtune", "Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation", "Fine-Tuning Llama2 with LoRA", "Memory Optimization Overview", "Fine-Tuning Llama3 with QAT", "Fine-Tuning Llama2 with QLoRA"], "terms": {"instruct": [1, 2, 4, 9, 10, 11, 13, 15, 16, 17, 18, 19, 21, 31, 32, 33, 35, 37, 42, 54, 56, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 71, 145, 148, 149, 153, 161, 167, 168, 169, 176, 177, 178, 265, 268, 269, 272, 273, 275, 277, 278, 280, 281], "prompt": [1, 9, 10, 11, 12, 13, 18, 31, 33, 34, 35, 36, 37, 38, 39, 40, 42, 43, 55, 56, 59, 60, 62, 63, 64, 67, 68, 69, 70, 73, 74, 94, 100, 106, 119, 144, 148, 153, 162, 169, 179, 189, 199, 216, 274, 276], "chat": [1, 2, 11, 13, 16, 18, 32, 37, 42, 56, 60, 100, 169, 268], "includ": [1, 9, 11, 12, 16, 18, 19, 21, 22, 24, 25, 38, 39, 56, 77, 81, 91, 101, 116, 125, 146, 147, 148, 150, 151, 158, 169, 175, 189, 205, 211, 228, 229, 267, 270, 272, 273, 274, 275, 276, 277, 278, 281], "some": [1, 17, 18, 20, 21, 22, 24, 32, 156, 200, 202, 207, 208, 265, 267, 268, 269, 272, 273, 274, 275, 277, 278, 279, 280, 281], "specif": [1, 5, 11, 12, 15, 19, 21, 24, 25, 27, 55, 56, 65, 66, 144, 212, 237, 269, 273, 274, 279, 280, 281], "format": [1, 2, 7, 11, 19, 21, 36, 45, 46, 55, 56, 59, 60, 63, 64, 67, 100, 144, 153, 212, 227, 228, 229, 230, 231, 272, 273, 274, 275, 276, 278, 279], "differ": [1, 9, 10, 17, 18, 19, 21, 24, 26, 48, 53, 60, 64, 78, 79, 80, 144, 191, 213, 221, 231, 258, 267, 268, 269, 272, 273, 274, 276, 277, 278, 279, 280, 281], "dataset": [1, 10, 13, 14, 15, 17, 19, 24, 31, 33, 35, 36, 37, 42, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 221, 267, 275, 276, 277, 280], "model": [1, 2, 9, 10, 11, 12, 13, 14, 16, 17, 18, 19, 20, 22, 23, 24, 25, 27, 31, 32, 33, 35, 36, 37, 42, 53, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 182, 183, 184, 185, 187, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 215, 216, 217, 219, 220, 221, 222, 223, 224, 228, 229, 230, 231, 233, 234, 237, 239, 246, 247, 252, 253, 257, 265, 267, 268, 269, 273, 281], "convert": [1, 9, 11, 14, 21, 22, 33, 35, 37, 42, 49, 55, 56, 60, 65, 66, 67, 75, 142, 228, 274, 280, 281], "from": [1, 2, 4, 10, 11, 13, 14, 15, 17, 19, 22, 23, 24, 25, 26, 27, 31, 33, 36, 37, 42, 45, 46, 47, 50, 53, 54, 55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 76, 77, 78, 79, 80, 81, 82, 83, 84, 92, 93, 100, 102, 103, 104, 105, 119, 143, 144, 148, 159, 161, 169, 176, 177, 178, 179, 180, 183, 188, 189, 190, 191, 192, 193, 194, 196, 197, 198, 201, 202, 203, 204, 207, 210, 213, 215, 218, 221, 223, 224, 227, 228, 229, 230, 232, 234, 235, 247, 250, 251, 252, 257, 264, 266, 269, 271, 272, 274, 275, 276, 277, 278, 279, 280], "common": [1, 2, 5, 9, 14, 15, 24, 216, 272, 273, 276, 278, 279, 280], "schema": [1, 9, 11, 12, 16], "convers": [1, 13, 16, 18, 19, 21, 22, 33, 42, 52, 55, 56, 60, 65, 67, 69, 228, 230, 231, 267, 273, 274, 278, 279, 281], "json": [1, 9, 12, 13, 16, 18, 21, 22, 37, 42, 55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 119, 148, 169, 179, 215, 228, 272, 273, 274, 280], "list": [1, 9, 11, 14, 15, 18, 19, 21, 22, 24, 33, 36, 38, 44, 45, 47, 48, 49, 50, 51, 52, 53, 55, 56, 60, 61, 65, 66, 67, 72, 73, 81, 85, 86, 87, 88, 89, 90, 94, 95, 96, 97, 98, 99, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 119, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 143, 144, 147, 148, 149, 150, 151, 152, 154, 155, 156, 157, 162, 163, 164, 165, 166, 169, 170, 171, 172, 173, 174, 189, 191, 196, 198, 199, 200, 201, 204, 205, 209, 210, 211, 212, 213, 214, 216, 218, 227, 228, 229, 230, 247, 261, 270, 273, 274, 275, 276, 279, 280], "us": [1, 2, 4, 5, 9, 10, 11, 12, 13, 14, 16, 17, 18, 20, 21, 22, 23, 26, 27, 29, 32, 35, 36, 38, 44, 47, 50, 53, 54, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 77, 79, 80, 81, 100, 101, 107, 116, 119, 120, 125, 129, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 165, 169, 171, 175, 179, 180, 182, 183, 184, 185, 187, 189, 190, 191, 192, 193, 194, 195, 196, 199, 200, 203, 206, 209, 213, 214, 218, 219, 220, 221, 222, 224, 226, 228, 229, 231, 232, 236, 237, 239, 246, 247, 248, 249, 250, 251, 255, 257, 259, 260, 265, 266, 267, 268, 269, 270, 272, 275, 276, 278, 279, 280], "collect": [1, 24, 275], "sampl": [1, 9, 11, 12, 13, 14, 15, 16, 19, 20, 21, 23, 26, 33, 35, 36, 37, 42, 44, 50, 54, 55, 56, 57, 62, 63, 65, 66, 67, 68, 69, 71, 73, 74, 183, 185, 189, 190, 191, 199, 217, 218, 223, 273, 274, 279], "batch": [1, 11, 17, 25, 47, 48, 49, 50, 54, 59, 62, 65, 66, 68, 79, 142, 143, 182, 183, 185, 188, 189, 190, 191, 194, 199, 201, 219, 220, 221, 223, 224, 241, 256, 259, 267, 275, 276, 278, 279], "handl": [1, 13, 16, 17, 24, 29, 31, 53, 56, 144, 213, 214, 273, 274, 278, 279, 281], "ani": [1, 5, 10, 11, 13, 14, 15, 16, 17, 21, 22, 24, 25, 27, 29, 30, 33, 36, 37, 38, 42, 44, 47, 50, 51, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 76, 80, 181, 189, 195, 199, 201, 202, 207, 208, 209, 210, 211, 212, 213, 216, 228, 229, 230, 232, 243, 246, 247, 255, 258, 272, 273, 275, 278, 279, 280], "pad": [1, 45, 47, 48, 49, 50, 54, 73, 75, 76, 189, 191, 218, 220, 222, 225, 241], "miscellan": 1, "modifi": [1, 10, 21, 24, 25, 26, 193, 195, 232, 267, 274, 276, 277, 278, 279, 280, 281], "For": [2, 7, 9, 10, 12, 14, 16, 18, 19, 20, 21, 22, 24, 25, 33, 35, 36, 37, 38, 42, 50, 53, 54, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 78, 79, 80, 81, 91, 95, 101, 107, 116, 120, 125, 129, 143, 146, 147, 150, 151, 154, 156, 158, 160, 165, 167, 171, 175, 183, 189, 191, 196, 199, 200, 203, 206, 217, 228, 234, 240, 247, 251, 253, 255, 266, 268, 269, 270, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "detail": [2, 9, 10, 12, 13, 16, 21, 22, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 78, 79, 80, 81, 160, 182, 191, 196, 222, 226, 237, 246, 255, 268, 269, 272, 274, 275, 276, 277, 278, 279, 280, 281], "usag": [2, 21, 195, 196, 198, 227, 231, 232, 256, 266, 272, 274, 275, 276, 279, 280, 281], "guid": [2, 23, 24, 26, 33, 35, 37, 42, 60, 62, 63, 64, 65, 66, 67, 68, 69, 224, 247, 267, 273, 275, 277, 278], "pleas": [2, 7, 34, 40, 43, 78, 79, 80, 81, 88, 89, 90, 98, 99, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 152, 163, 164, 170, 191, 196, 226, 237, 246, 253, 266, 269, 270, 274, 276, 281], "see": [2, 7, 9, 10, 11, 12, 13, 14, 16, 18, 19, 21, 22, 23, 26, 34, 40, 43, 47, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 74, 88, 89, 90, 98, 99, 100, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 152, 153, 160, 163, 164, 170, 182, 188, 190, 191, 201, 204, 211, 212, 217, 226, 231, 237, 246, 247, 251, 253, 255, 261, 266, 267, 268, 269, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "overview": [2, 7, 24, 26, 199, 265, 268, 269, 275, 277, 278, 281], "support": [2, 4, 10, 11, 15, 16, 17, 18, 21, 22, 23, 25, 26, 27, 36, 37, 54, 55, 56, 59, 60, 61, 62, 65, 66, 67, 68, 69, 72, 77, 95, 107, 120, 129, 142, 149, 150, 151, 153, 154, 156, 165, 168, 169, 171, 181, 183, 191, 200, 201, 205, 223, 229, 230, 232, 236, 239, 240, 267, 268, 269, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "sever": [2, 279], "wide": [2, 9, 183, 277], "onli": [2, 4, 10, 16, 18, 22, 23, 26, 35, 36, 42, 54, 55, 56, 61, 67, 73, 77, 81, 95, 107, 120, 129, 144, 149, 150, 151, 153, 154, 156, 165, 171, 183, 187, 189, 191, 196, 198, 202, 205, 207, 209, 213, 228, 229, 230, 232, 236, 237, 239, 240, 246, 272, 274, 275, 277, 278, 279, 280, 281], "help": [2, 11, 18, 19, 22, 63, 100, 189, 191, 199, 228, 247, 265, 266, 267, 272, 273, 274, 275, 277, 279, 280, 281], "quickli": [2, 11, 24, 38, 57, 268, 273, 279], "bootstrap": [2, 11], "your": [2, 7, 9, 11, 12, 13, 14, 16, 17, 18, 21, 23, 26, 27, 38, 57, 60, 64, 67, 79, 80, 81, 147, 151, 191, 200, 247, 250, 251, 257, 265, 266, 267, 268, 269, 272, 273, 276, 277, 278, 279, 280, 281], "fine": [2, 9, 10, 11, 12, 16, 18, 19, 20, 22, 23, 25, 26, 36, 54, 55, 56, 71, 257, 265, 267, 268, 269, 270, 274], "tune": [2, 4, 9, 10, 11, 12, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 29, 36, 54, 55, 56, 71, 257, 265, 266, 267, 268, 269, 270, 272, 274], "also": [2, 9, 10, 12, 14, 16, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 53, 60, 64, 67, 71, 73, 74, 91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 169, 171, 175, 183, 189, 192, 205, 224, 237, 239, 246, 247, 251, 257, 260, 266, 269, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "like": [2, 6, 12, 22, 23, 24, 25, 26, 169, 191, 196, 198, 200, 230, 266, 272, 273, 274, 275, 277, 278, 279, 280], "These": [2, 5, 10, 13, 15, 18, 19, 21, 22, 24, 25, 27, 54, 55, 67, 191, 218, 268, 270, 273, 274, 275, 276, 278, 279, 280, 281], "ar": [2, 5, 9, 10, 11, 12, 13, 14, 15, 16, 18, 19, 20, 21, 22, 23, 24, 26, 27, 31, 35, 38, 39, 42, 45, 47, 48, 52, 54, 55, 56, 59, 60, 64, 65, 66, 67, 73, 75, 76, 79, 85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 100, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 144, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 182, 188, 189, 190, 191, 193, 199, 200, 201, 205, 206, 209, 210, 218, 220, 226, 228, 229, 231, 232, 234, 236, 238, 239, 244, 246, 256, 257, 266, 267, 268, 270, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "especi": [2, 267, 272, 274, 279], "specifi": [2, 10, 12, 16, 18, 20, 22, 24, 25, 27, 31, 33, 35, 37, 42, 44, 60, 62, 63, 64, 65, 66, 67, 68, 69, 73, 75, 77, 94, 101, 106, 107, 116, 119, 120, 125, 129, 146, 148, 150, 162, 169, 171, 175, 179, 183, 189, 190, 197, 198, 199, 226, 237, 240, 246, 251, 253, 256, 269, 270, 272, 273, 274, 275, 276, 279, 280, 281], "yaml": [2, 10, 17, 18, 20, 24, 25, 27, 28, 29, 53, 60, 64, 67, 71, 251, 267, 270, 272, 273, 274, 275, 276, 278, 280, 281], "config": [2, 9, 12, 13, 16, 17, 18, 19, 20, 21, 22, 23, 26, 27, 28, 29, 30, 53, 60, 64, 67, 71, 183, 209, 228, 232, 247, 251, 256, 267, 268, 269, 270, 273, 274, 276, 277, 278, 279, 280, 281], "represent": [2, 227, 277, 278, 280, 281], "abov": [2, 4, 9, 16, 17, 18, 20, 22, 55, 195, 244, 266, 269, 274, 276, 278, 279, 280, 281], "text": [4, 5, 9, 11, 12, 15, 18, 19, 21, 35, 36, 37, 38, 39, 42, 44, 50, 54, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 144, 200, 201, 211, 213, 214, 216, 218, 273, 274, 280], "version": [4, 58, 73, 95, 107, 120, 129, 149, 154, 156, 165, 171, 183, 262, 266, 276, 279, 280, 281], "famili": [4, 22, 25, 59, 61, 65, 66, 69, 70, 72, 231, 267, 272, 276, 277], "import": [4, 9, 10, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 24, 27, 60, 64, 65, 66, 67, 71, 77, 191, 192, 193, 194, 221, 247, 250, 251, 273, 274, 275, 276, 277, 278, 279, 280, 281], "you": [4, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 36, 38, 55, 56, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 100, 182, 189, 191, 194, 196, 198, 201, 203, 231, 247, 250, 251, 257, 265, 266, 267, 268, 269, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "need": [4, 9, 10, 12, 14, 16, 18, 19, 20, 22, 23, 24, 25, 26, 38, 54, 56, 183, 189, 191, 199, 200, 224, 246, 247, 250, 251, 252, 266, 268, 269, 270, 272, 273, 274, 275, 276, 278, 279, 281], "request": [4, 236, 274], "access": [4, 10, 22, 24, 25, 53, 228, 234, 268, 269, 272, 274, 275], "hug": [4, 11, 22, 32, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 119, 148, 169, 179, 215, 235, 267, 272, 275, 276], "face": [4, 11, 22, 32, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 119, 148, 169, 179, 215, 235, 267, 272, 275, 276], "befor": [4, 19, 22, 38, 52, 54, 65, 78, 79, 81, 91, 95, 147, 151, 183, 188, 189, 190, 191, 196, 198, 199, 201, 205, 214, 228, 247, 269, 272, 274, 279, 280], "download": [4, 10, 11, 16, 22, 65, 263, 266, 268, 269, 273, 276, 277, 278, 280, 281], "To": [4, 9, 12, 13, 14, 16, 17, 18, 19, 20, 22, 24, 25, 26, 54, 65, 189, 191, 201, 228, 257, 266, 267, 269, 270, 272, 274, 275, 276, 277, 278, 279, 280, 281], "1b": [4, 10, 17, 136, 138, 140, 265], "meta": [4, 10, 15, 16, 20, 21, 22, 100, 185, 228, 229, 268, 269, 272, 273, 274, 275, 277], "output": [4, 10, 12, 13, 14, 20, 21, 22, 31, 35, 45, 53, 55, 56, 59, 62, 64, 68, 69, 73, 81, 85, 86, 87, 91, 95, 101, 105, 107, 108, 109, 110, 111, 116, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 142, 143, 146, 147, 149, 150, 151, 154, 155, 156, 157, 158, 161, 165, 166, 171, 174, 175, 180, 181, 183, 185, 186, 188, 189, 190, 191, 196, 198, 199, 200, 201, 205, 208, 209, 210, 218, 230, 237, 249, 256, 257, 266, 268, 269, 272, 274, 275, 276, 277, 278, 279, 281], "dir": [4, 10, 21, 22, 251, 266, 268, 269, 272, 274, 275, 276, 277, 280], "tmp": [4, 9, 10, 12, 14, 15, 16, 17, 18, 19, 20, 21, 24, 232, 268, 269, 273, 275, 277], "ignor": [4, 9, 10, 12, 22, 42, 71, 187, 188, 190, 197, 198, 233, 257, 268, 269, 272, 277], "pattern": [4, 10, 19, 214, 268, 269, 272, 277], "origin": [4, 10, 15, 16, 17, 20, 21, 22, 58, 59, 63, 195, 200, 201, 205, 268, 269, 273, 274, 276, 277, 278, 279, 280, 281], "consolid": [4, 10, 22, 268, 269, 277], "00": [4, 10, 16, 22, 60, 64, 264, 268, 269, 271, 275, 277], "pth": [4, 10, 22, 227, 268, 269, 274, 277], "hf": [4, 9, 18, 20, 21, 22, 221, 223, 228, 272, 273, 274, 275, 276], "token": [4, 9, 10, 11, 12, 13, 15, 16, 17, 18, 19, 20, 22, 24, 25, 36, 42, 47, 49, 50, 51, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 91, 94, 95, 101, 106, 107, 116, 119, 120, 125, 129, 144, 146, 147, 148, 150, 151, 154, 156, 158, 160, 162, 165, 167, 169, 171, 175, 179, 183, 185, 188, 189, 190, 191, 196, 198, 199, 200, 201, 211, 212, 213, 214, 215, 216, 218, 220, 222, 225, 237, 241, 268, 272, 274, 275, 276, 277, 278, 279, 280, 281], "hf_token": [4, 21, 269, 277], "3b": [4, 137, 139, 141], "The": [4, 9, 11, 12, 15, 16, 18, 19, 20, 21, 22, 23, 24, 25, 26, 29, 30, 32, 36, 46, 47, 52, 53, 54, 55, 56, 60, 63, 64, 65, 66, 67, 70, 78, 79, 80, 81, 85, 86, 87, 95, 96, 97, 107, 108, 109, 110, 111, 120, 121, 122, 129, 130, 131, 132, 138, 139, 142, 144, 147, 149, 150, 151, 154, 156, 165, 166, 171, 172, 173, 174, 181, 184, 185, 186, 187, 191, 195, 196, 197, 198, 199, 200, 201, 206, 211, 212, 213, 214, 215, 216, 218, 219, 221, 222, 223, 224, 226, 228, 230, 232, 235, 236, 238, 240, 247, 251, 254, 256, 260, 261, 262, 266, 267, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "reus": [4, 267], "llama3_token": [4, 15, 17, 20, 21, 65, 66, 73, 273, 276], "class": [4, 10, 13, 14, 15, 21, 24, 26, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 42, 43, 53, 54, 55, 56, 57, 65, 66, 78, 79, 80, 81, 94, 100, 105, 106, 119, 142, 143, 144, 148, 153, 156, 160, 161, 162, 169, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 196, 197, 198, 199, 200, 201, 202, 204, 205, 207, 208, 211, 212, 213, 214, 217, 218, 221, 222, 223, 224, 227, 228, 229, 230, 231, 232, 247, 248, 249, 250, 251, 270, 273, 275, 277, 278, 279, 281], "languag": [4, 10, 16, 32, 73, 150, 200, 201, 205, 221, 257, 278, 279], "11b": [4, 145, 152], "8b": [4, 15, 16, 20, 21, 118, 122, 124, 128, 130, 132, 135, 166, 265, 268, 269, 272, 273, 280], "70b": [4, 83, 86, 89, 103, 109, 113, 117, 121, 123, 127, 131, 134, 276], "405b": [4, 126, 130, 133], "weight": [4, 21, 22, 25, 50, 85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 145, 148, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 187, 195, 204, 205, 209, 213, 221, 228, 229, 230, 231, 240, 251, 257, 265, 269, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "can": [4, 5, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 30, 33, 35, 36, 37, 38, 39, 42, 50, 53, 55, 56, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 78, 79, 81, 144, 147, 151, 184, 185, 187, 188, 189, 191, 196, 198, 199, 201, 203, 206, 213, 214, 226, 228, 231, 233, 237, 246, 247, 250, 251, 253, 256, 265, 266, 267, 268, 269, 270, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "instead": [4, 9, 12, 14, 16, 22, 25, 31, 45, 54, 55, 71, 81, 129, 150, 151, 182, 187, 191, 205, 224, 272, 276, 278, 279, 280], "builder": [4, 9, 10, 11, 12, 13, 15, 16, 17, 22, 58, 60, 61, 64, 67, 82, 83, 84, 85, 86, 87, 88, 89, 90, 92, 93, 96, 97, 98, 99, 102, 103, 104, 105, 108, 109, 110, 111, 112, 113, 114, 115, 117, 118, 121, 122, 123, 124, 126, 127, 128, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 152, 155, 157, 159, 161, 163, 164, 166, 168, 170, 172, 173, 174, 176, 177, 178, 273, 279, 281], "all": [4, 5, 10, 11, 14, 15, 19, 21, 25, 30, 35, 36, 38, 42, 45, 47, 50, 53, 54, 55, 56, 81, 119, 142, 148, 169, 179, 183, 187, 189, 191, 192, 193, 194, 195, 199, 200, 201, 203, 206, 217, 228, 232, 234, 238, 244, 252, 258, 259, 263, 265, 267, 268, 269, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280], "7b": [4, 9, 12, 14, 18, 19, 20, 21, 22, 61, 72, 84, 87, 90, 93, 97, 104, 105, 110, 111, 114, 115, 155, 157, 159, 161, 164, 174, 178, 228, 229, 273, 275, 276, 278, 281], "13b": [4, 22, 82, 85, 88, 102, 108, 112], "codellama": 4, "size": [4, 14, 15, 16, 22, 25, 27, 45, 50, 59, 62, 65, 66, 68, 79, 80, 81, 142, 143, 144, 145, 147, 148, 149, 151, 182, 183, 184, 185, 188, 189, 190, 191, 194, 196, 198, 199, 200, 201, 218, 219, 220, 241, 242, 244, 267, 269, 272, 274, 275, 276, 278, 279, 280], "0": [4, 9, 10, 12, 14, 15, 16, 18, 20, 22, 25, 45, 47, 48, 49, 50, 54, 60, 64, 67, 73, 74, 76, 77, 81, 85, 86, 87, 88, 89, 90, 91, 95, 96, 97, 98, 99, 101, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 120, 121, 122, 123, 124, 125, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 143, 144, 146, 147, 149, 150, 151, 152, 154, 155, 156, 157, 158, 160, 163, 164, 165, 166, 167, 170, 171, 172, 173, 174, 175, 176, 177, 182, 183, 189, 191, 192, 193, 194, 200, 205, 216, 221, 222, 223, 224, 225, 235, 241, 247, 250, 251, 255, 260, 262, 264, 269, 271, 273, 274, 275, 276, 278, 279, 280, 281], "5b": [4, 172, 173, 176, 177, 279], "qwen2": [4, 10, 171, 172, 173, 174, 176, 177, 178, 179, 231, 279], "exampl": [4, 10, 19, 21, 22, 23, 24, 25, 26, 27, 29, 33, 35, 37, 38, 42, 44, 45, 46, 47, 48, 49, 50, 53, 54, 56, 59, 60, 61, 62, 64, 65, 66, 67, 68, 69, 71, 72, 73, 75, 76, 77, 81, 143, 144, 147, 151, 182, 183, 191, 192, 193, 194, 196, 198, 199, 200, 201, 203, 204, 206, 211, 212, 213, 214, 216, 217, 221, 223, 224, 225, 226, 227, 228, 229, 231, 232, 240, 241, 247, 250, 251, 254, 257, 260, 261, 262, 263, 264, 266, 268, 269, 271, 272, 273, 274, 276, 277, 278, 279, 280, 281], "none": [4, 9, 16, 25, 26, 28, 30, 31, 33, 35, 37, 42, 50, 51, 52, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 77, 81, 94, 101, 106, 107, 116, 119, 120, 125, 129, 142, 143, 144, 146, 148, 150, 162, 169, 179, 180, 182, 183, 185, 188, 189, 190, 191, 192, 193, 194, 199, 201, 206, 208, 209, 210, 213, 216, 218, 219, 220, 222, 228, 229, 230, 231, 232, 233, 236, 240, 245, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 258, 259, 260, 261, 272, 274, 280], "mini": [4, 21, 166, 167, 168, 169, 170], "4k": [4, 21, 167, 168, 169], "microsoft": [4, 168, 169], "ai": [4, 10, 12, 14, 19, 55, 56, 159, 251, 273, 276], "thi": [4, 9, 10, 12, 13, 14, 16, 17, 18, 19, 20, 21, 23, 24, 25, 26, 27, 31, 33, 34, 35, 36, 37, 42, 43, 44, 45, 47, 48, 50, 53, 54, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 75, 79, 80, 81, 91, 95, 101, 107, 116, 120, 125, 129, 142, 144, 146, 147, 150, 151, 153, 154, 156, 158, 160, 165, 167, 168, 169, 171, 175, 180, 182, 183, 185, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 199, 200, 201, 203, 204, 206, 209, 210, 211, 212, 213, 214, 216, 217, 218, 220, 221, 222, 224, 226, 227, 228, 229, 230, 232, 235, 236, 239, 241, 244, 246, 247, 248, 250, 251, 252, 253, 255, 257, 259, 260, 265, 266, 267, 268, 269, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "v0": [4, 9, 14, 18, 19, 21, 153], "mistralai": [4, 21, 272], "2b": [4, 92, 96], "gemma2": 4, "googl": [4, 92, 93], "gguf": 4, "compon": [4, 6, 14, 21, 22, 25, 30, 48, 55, 56, 65, 66, 267, 270, 275, 277, 278, 281], "multimod": [4, 11, 14, 36, 42, 56, 65, 66, 199, 266], "encod": [4, 5, 15, 21, 50, 56, 73, 74, 81, 142, 143, 145, 146, 147, 149, 150, 151, 183, 188, 189, 190, 194, 199, 200, 201, 203, 211, 213, 214, 216, 218, 221, 224, 273], "perform": [5, 12, 13, 17, 19, 20, 21, 22, 54, 73, 191, 196, 206, 217, 224, 267, 268, 269, 273, 274, 276, 277, 279, 280, 281], "direct": [5, 18, 25, 48, 85, 86, 95, 96, 97, 107, 108, 109, 110, 120, 121, 122, 131, 132, 138, 139, 154, 155, 156, 157, 165, 166, 221, 266, 270], "id": [5, 14, 17, 21, 22, 47, 48, 49, 50, 54, 61, 65, 66, 72, 73, 74, 76, 77, 144, 183, 185, 189, 190, 199, 211, 212, 213, 214, 215, 216, 218, 228, 230, 247, 273, 274], "decod": [5, 9, 12, 14, 15, 16, 18, 20, 21, 60, 64, 67, 73, 91, 95, 101, 107, 116, 120, 125, 129, 143, 144, 145, 146, 147, 149, 150, 151, 154, 156, 158, 160, 165, 167, 171, 175, 183, 188, 189, 190, 194, 199, 201, 203, 211, 213, 214, 273], "typic": [5, 9, 12, 20, 24, 33, 37, 42, 50, 54, 55, 56, 57, 71, 169, 203, 221, 224, 279, 280, 281], "byte": [5, 21, 214, 279, 281], "pair": [5, 10, 18, 21, 24, 48, 49, 63, 67, 70, 214], "underli": [5, 13, 18, 21, 213, 279, 281], "helper": 5, "method": [5, 13, 14, 15, 19, 21, 22, 24, 25, 26, 29, 46, 55, 57, 59, 60, 61, 62, 63, 64, 67, 68, 69, 70, 71, 72, 144, 189, 195, 196, 199, 202, 203, 204, 207, 209, 211, 212, 232, 240, 266, 267, 278, 281], "two": [5, 15, 18, 19, 22, 24, 35, 50, 52, 65, 66, 73, 74, 79, 191, 200, 203, 218, 225, 227, 267, 269, 274, 275, 276, 278, 279, 280, 281], "pre": [5, 9, 11, 12, 17, 18, 19, 20, 54, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 100, 145, 148, 149, 191, 199, 201, 203, 269, 273, 279], "train": [5, 9, 10, 11, 12, 13, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 31, 33, 35, 50, 53, 54, 55, 56, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 100, 144, 145, 148, 149, 181, 183, 185, 189, 190, 195, 196, 198, 199, 200, 201, 203, 221, 224, 227, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 265, 267, 268, 270, 272, 273, 274, 276, 277, 278, 279, 280, 281], "function": [5, 10, 22, 24, 25, 27, 29, 45, 46, 47, 48, 60, 64, 67, 73, 79, 80, 81, 147, 151, 180, 183, 191, 192, 195, 206, 209, 210, 221, 222, 226, 228, 242, 255, 257, 259, 260, 267, 277, 281], "preprocess": [5, 54, 191], "imag": [5, 11, 15, 35, 36, 37, 42, 44, 46, 50, 56, 65, 66, 78, 79, 80, 81, 142, 143, 144, 145, 147, 148, 149, 151, 191, 200, 218, 278], "loss": [6, 9, 12, 14, 24, 25, 36, 38, 55, 56, 59, 60, 62, 64, 67, 68, 69, 196, 197, 198, 221, 222, 223, 224, 275, 277, 278, 281], "algorithm": [6, 21, 219, 224, 255], "ppo": [6, 219, 220, 221, 222, 270], "dpo": [6, 18, 48, 55, 206, 221, 223, 224, 270], "offer": 7, "allow": [7, 10, 53, 201, 209, 250, 269, 272, 279, 280, 281], "seamless": 7, "transit": 7, "between": [7, 9, 18, 19, 21, 22, 55, 60, 67, 146, 150, 188, 189, 193, 199, 220, 222, 224, 228, 231, 247, 274, 276, 277, 278, 279, 280, 281], "interoper": [7, 22, 25, 267, 274, 281], "rest": [7, 273, 279, 281], "ecosystem": [7, 22, 25, 267, 274, 276, 281], "comprehens": [7, 279], "deep": [7, 22, 23, 24, 25, 26, 201, 203, 267, 270, 275, 276, 279], "dive": [7, 22, 23, 24, 25, 26, 267, 269, 270, 275, 276, 279], "util": [7, 14, 16, 22, 24, 25, 27, 45, 47, 50, 142, 233, 250, 252, 253, 259, 260, 261, 262, 267, 274, 275, 279, 281], "work": [7, 22, 25, 35, 42, 187, 200, 201, 267, 269, 272, 274, 276, 279, 281], "set": [7, 9, 12, 17, 18, 20, 22, 23, 24, 25, 26, 33, 36, 37, 42, 50, 54, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 71, 72, 101, 107, 116, 120, 125, 129, 144, 146, 149, 150, 154, 156, 158, 160, 165, 167, 171, 175, 183, 185, 188, 189, 192, 193, 194, 199, 206, 208, 226, 232, 237, 244, 246, 247, 253, 254, 255, 256, 259, 260, 267, 270, 272, 273, 274, 275, 276, 277, 278, 279, 280], "enabl": [7, 10, 11, 17, 21, 23, 24, 25, 26, 53, 85, 86, 87, 88, 89, 90, 96, 97, 98, 99, 108, 109, 110, 111, 112, 113, 114, 115, 121, 122, 123, 124, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 152, 155, 157, 163, 164, 166, 170, 172, 173, 174, 176, 177, 183, 188, 189, 190, 192, 193, 194, 199, 201, 205, 255, 256, 269, 276, 278, 279, 281], "consumpt": [7, 53, 75, 268, 279], "dure": [7, 10, 11, 22, 54, 59, 60, 62, 64, 67, 68, 69, 182, 183, 185, 189, 190, 191, 195, 199, 200, 218, 224, 239, 268, 269, 273, 274, 276, 278, 279, 280, 281], "control": [7, 13, 18, 21, 25, 36, 59, 60, 62, 64, 67, 68, 69, 193, 194, 201, 206, 247, 255, 269, 274, 279], "lr": [7, 24, 232, 235, 238, 277, 279], "process": [7, 11, 14, 15, 17, 25, 26, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 81, 147, 151, 191, 195, 242, 243, 255, 275, 280, 281], "variou": 7, "provid": [7, 10, 11, 12, 14, 22, 24, 25, 27, 32, 33, 35, 37, 42, 46, 47, 51, 53, 54, 73, 75, 81, 183, 187, 189, 191, 199, 206, 216, 221, 230, 237, 247, 251, 256, 260, 267, 268, 269, 272, 273, 274, 275, 276, 279], "debug": [7, 22, 24, 25, 247, 272], "finetun": [7, 10, 22, 24, 25, 85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 172, 173, 174, 199, 265, 267, 269, 275, 276, 279], "job": [7, 10, 26, 255, 275], "involv": [9, 12, 17, 20, 56, 280], "multi": [9, 18, 25, 183, 276], "turn": [9, 18, 25, 33, 36, 37, 42, 52, 55, 67, 273, 279], "multipl": [9, 16, 17, 18, 22, 24, 25, 33, 36, 37, 42, 48, 53, 56, 67, 142, 143, 183, 189, 190, 191, 199, 205, 247, 248, 249, 250, 251, 256, 275, 276, 277, 279], "back": [9, 21, 22, 52, 206, 228, 278, 279, 281], "forth": [9, 52], "user": [9, 12, 13, 14, 15, 16, 18, 19, 21, 25, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 46, 52, 55, 56, 60, 64, 67, 94, 101, 106, 107, 116, 119, 120, 125, 129, 146, 148, 150, 154, 156, 158, 160, 162, 165, 167, 169, 171, 175, 179, 183, 216, 270, 273, 275, 280], "assist": [9, 12, 13, 14, 15, 16, 18, 19, 21, 31, 32, 33, 35, 36, 37, 38, 39, 41, 42, 44, 52, 55, 56, 60, 67, 73, 94, 100, 106, 119, 148, 162, 169, 179, 216, 273], "role": [9, 13, 14, 15, 16, 18, 19, 21, 33, 36, 37, 38, 39, 42, 44, 55, 56, 60, 67, 94, 106, 119, 144, 148, 162, 169, 179, 216, 273], "content": [9, 13, 15, 16, 18, 19, 21, 22, 33, 36, 37, 38, 39, 42, 44, 55, 56, 60, 67, 216, 273], "what": [9, 14, 15, 16, 18, 22, 23, 24, 26, 36, 37, 55, 56, 60, 64, 67, 100, 153, 191, 265, 270, 273, 274, 275, 276, 279], "answer": [9, 15, 16, 19, 40, 64, 274, 276], "ultim": [9, 280], "question": [9, 15, 16, 19, 40, 64, 274, 276], "life": 9, "42": [9, 73, 191], "That": [9, 273], "s": [9, 10, 12, 13, 14, 16, 17, 18, 19, 20, 22, 24, 25, 26, 27, 29, 32, 37, 42, 52, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 85, 86, 87, 100, 107, 108, 109, 110, 111, 120, 121, 122, 129, 130, 131, 132, 138, 139, 142, 143, 144, 149, 150, 151, 153, 154, 155, 156, 157, 165, 166, 169, 171, 174, 175, 182, 183, 185, 189, 190, 191, 195, 199, 202, 203, 204, 207, 209, 210, 214, 221, 223, 224, 225, 226, 228, 229, 232, 237, 239, 241, 246, 247, 250, 253, 254, 257, 259, 260, 267, 272, 273, 275, 277, 278, 279, 280, 281], "ridicul": 9, "oh": 9, "i": [9, 12, 14, 18, 19, 20, 25, 36, 67, 73, 100, 142, 143, 153, 183, 188, 189, 190, 191, 195, 199, 208, 227, 232, 274, 276, 279, 280, 281], "know": [9, 273, 274, 277, 278], "more": [9, 10, 11, 12, 13, 14, 16, 18, 19, 21, 22, 24, 25, 38, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 182, 191, 196, 203, 209, 226, 227, 230, 247, 251, 253, 255, 259, 267, 268, 269, 270, 272, 274, 275, 276, 277, 278, 279, 280, 281], "structur": [9, 12, 13, 14, 19, 25, 37, 39, 42, 60, 119, 144, 148, 169, 179, 218, 273, 274, 280], "than": [9, 10, 12, 16, 18, 24, 50, 52, 73, 75, 182, 183, 191, 221, 226, 230, 231, 258, 259, 262, 273, 274, 275, 276, 277, 278, 279, 281], "freeform": [9, 12, 57, 71], "associ": [9, 10, 11, 12, 22, 24, 25, 73, 74, 81, 91, 101, 116, 125, 146, 150, 158, 175, 247, 274, 278], "where": [9, 10, 12, 14, 16, 18, 19, 20, 36, 38, 45, 48, 59, 73, 75, 76, 79, 105, 142, 143, 161, 180, 183, 189, 191, 193, 196, 198, 199, 205, 213, 218, 219, 221, 222, 225, 237, 241, 246, 277, 279], "thei": [9, 11, 12, 19, 21, 24, 25, 53, 65, 66, 81, 142, 147, 151, 189, 191, 201, 210, 237, 272, 273, 278, 279, 280], "learn": [9, 12, 25, 53, 200, 201, 203, 232, 235, 238, 267, 268, 269, 270, 273, 275, 276, 278, 279, 280, 281], "simpli": [9, 12, 13, 14, 16, 20, 22, 24, 54, 56, 221, 272, 273, 274, 276, 277, 279, 281], "predict": [9, 12, 73, 74, 77, 219, 220, 222, 268], "next": [9, 12, 22, 54, 71, 73, 74, 81, 191, 218, 268, 276, 281], "respond": 9, "accur": 9, "primari": [9, 12, 16, 18, 20, 22, 24, 25, 55, 56, 270, 275], "entri": [9, 12, 16, 18, 20, 24, 25, 47, 50, 270, 275, 279], "point": [9, 10, 12, 16, 18, 20, 21, 24, 25, 46, 60, 216, 270, 274, 275, 276, 278, 280, 281], "torchtun": [9, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 35, 36, 37, 38, 39, 42, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 227, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 262, 266, 268, 269, 270, 273, 275, 279], "chat_dataset": [9, 12, 13, 18, 273], "let": [9, 10, 11, 12, 16, 18, 22, 24, 26, 272, 273, 274, 275, 276, 277, 278, 279, 281], "follow": [9, 10, 11, 12, 15, 16, 19, 22, 25, 36, 37, 38, 42, 50, 54, 55, 56, 64, 67, 144, 183, 188, 218, 222, 230, 231, 232, 235, 244, 251, 256, 265, 266, 269, 270, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "data": [9, 10, 12, 13, 14, 15, 16, 19, 21, 23, 31, 32, 33, 35, 36, 37, 38, 39, 42, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 100, 148, 179, 191, 217, 221, 223, 239, 247, 248, 249, 250, 251, 259, 268, 269, 274, 279, 280, 281], "directli": [9, 10, 12, 13, 14, 16, 22, 24, 25, 27, 31, 55, 56, 60, 64, 65, 67, 71, 221, 226, 228, 272, 274, 275, 276, 278, 279, 280, 281], "llm": [9, 10, 11, 12, 21, 25, 199, 201, 265, 266, 267, 268, 270, 274, 276, 277, 278], "my_data": [9, 12, 13, 16, 273], "human": [9, 16, 18, 36, 42, 60, 100, 221, 222, 223, 273], "valu": [9, 16, 22, 24, 33, 35, 37, 42, 45, 47, 48, 50, 59, 60, 62, 63, 64, 67, 68, 69, 70, 73, 74, 76, 77, 82, 83, 84, 91, 92, 93, 95, 101, 102, 103, 104, 105, 107, 116, 117, 118, 120, 125, 126, 127, 128, 129, 136, 137, 144, 146, 150, 154, 156, 158, 159, 160, 161, 165, 167, 171, 175, 176, 177, 178, 182, 183, 184, 188, 189, 190, 197, 198, 199, 201, 209, 219, 220, 222, 225, 228, 231, 232, 235, 241, 247, 248, 249, 250, 251, 255, 269, 272, 273, 275, 276, 278, 279, 280], "gpt": [9, 16, 42, 60, 74, 273, 274], "mistral": [9, 14, 18, 19, 21, 144, 153, 154, 155, 156, 157, 159, 160, 161, 162, 163, 164, 231, 272, 273, 274, 275], "mistral_token": [9, 14, 18, 19, 21], "m_token": [9, 14, 18, 19, 20, 21], "path": [9, 10, 12, 14, 15, 16, 17, 18, 19, 20, 21, 24, 25, 26, 27, 35, 42, 46, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 94, 106, 119, 144, 148, 162, 169, 179, 213, 214, 215, 228, 229, 230, 256, 272, 273, 274, 276, 278], "1": [9, 14, 16, 18, 19, 20, 21, 22, 25, 35, 42, 45, 47, 48, 49, 50, 54, 69, 73, 74, 76, 77, 78, 79, 101, 107, 116, 120, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 143, 144, 146, 150, 153, 154, 156, 158, 160, 165, 167, 171, 172, 173, 175, 176, 177, 182, 183, 189, 191, 192, 193, 194, 196, 197, 198, 213, 214, 216, 221, 222, 223, 224, 229, 231, 235, 241, 244, 247, 250, 251, 254, 255, 267, 268, 272, 273, 274, 275, 278, 279, 280, 281], "prompt_templ": [9, 12, 14, 16, 18, 19, 94, 106, 119, 144, 148, 162, 169, 179], "mistralchattempl": [9, 14, 18, 19, 162, 273], "max_seq_len": [9, 10, 12, 14, 16, 17, 18, 20, 21, 24, 27, 47, 50, 51, 54, 59, 60, 61, 62, 64, 65, 66, 68, 69, 71, 72, 91, 94, 95, 101, 106, 107, 116, 119, 120, 125, 129, 144, 146, 148, 150, 154, 156, 158, 160, 162, 165, 167, 169, 171, 175, 179, 182, 183, 185, 189, 194, 280], "8192": [9, 12, 14, 16, 17, 18, 20, 21, 148, 278, 280], "ds": [9, 10, 12, 15, 16, 18, 20, 54, 69, 273], "sourc": [9, 10, 12, 13, 16, 18, 20, 22, 24, 27, 28, 29, 30, 31, 32, 33, 35, 36, 37, 38, 39, 42, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 91, 92, 93, 94, 95, 96, 97, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 116, 117, 118, 119, 120, 121, 122, 125, 126, 127, 128, 129, 130, 131, 132, 136, 137, 138, 139, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 165, 166, 167, 168, 169, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 227, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 262, 263, 273, 274, 280], "data_fil": [9, 12, 13, 16, 18, 20, 55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 273], "split": [9, 10, 12, 13, 14, 16, 18, 20, 22, 44, 53, 54, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 214, 273, 274, 280], "conversation_column": [9, 60, 273], "conversation_styl": [9, 60, 273], "By": [9, 12, 22, 269, 272, 277, 278, 279, 280, 281], "default": [9, 10, 12, 16, 22, 24, 31, 32, 33, 35, 36, 37, 42, 45, 48, 49, 50, 51, 54, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 77, 81, 82, 83, 84, 85, 86, 87, 91, 92, 93, 94, 95, 96, 97, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 116, 117, 118, 119, 120, 121, 122, 125, 126, 127, 128, 129, 130, 131, 132, 136, 137, 138, 139, 144, 145, 148, 149, 150, 151, 154, 155, 156, 157, 158, 159, 160, 161, 162, 165, 166, 167, 169, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 183, 184, 185, 188, 189, 190, 195, 197, 198, 199, 205, 209, 213, 214, 216, 218, 219, 220, 221, 224, 228, 229, 230, 232, 235, 236, 242, 246, 247, 248, 251, 254, 255, 256, 266, 269, 272, 273, 274, 276, 277, 278, 279, 280, 281], "true": [9, 10, 12, 13, 14, 15, 16, 17, 22, 24, 31, 36, 45, 53, 54, 57, 58, 59, 60, 62, 64, 65, 66, 67, 68, 69, 71, 72, 75, 76, 81, 88, 89, 90, 91, 95, 98, 99, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 144, 145, 152, 163, 164, 170, 183, 188, 189, 190, 192, 193, 194, 195, 196, 197, 199, 201, 206, 213, 214, 216, 218, 219, 222, 225, 226, 228, 229, 230, 237, 238, 239, 241, 243, 244, 247, 250, 256, 262, 268, 272, 273, 274, 276, 278, 279, 280, 281], "train_on_input": [9, 12, 13, 18, 24, 31, 33, 35, 37, 42, 53, 58, 59, 60, 62, 63, 64, 67, 68, 69, 70], "new_system_prompt": [9, 12, 13, 33, 35, 37, 42, 60, 62, 63, 64, 65, 66, 67, 68, 69], "tokenized_dict": [9, 12, 15, 16, 18, 20], "label": [9, 12, 20, 25, 47, 48, 49, 50, 54, 61, 69, 72, 196, 197, 198, 221, 224, 277], "print": [9, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 26, 44, 50, 53, 59, 62, 65, 66, 68, 69, 73, 144, 191, 192, 193, 194, 213, 214, 216, 262, 273, 275, 278, 280, 281], "inst": [9, 14, 19, 21, 100, 144, 153, 273], "733": [9, 14, 21], "16289": [9, 14, 21], "28793": [9, 14, 21], "1824": 9, "349": 9, "272": 9, "4372": 9, "In": [9, 10, 12, 13, 14, 16, 17, 18, 19, 20, 21, 22, 24, 25, 55, 79, 80, 81, 147, 151, 185, 189, 191, 205, 226, 246, 250, 251, 269, 273, 274, 276, 277, 278, 279, 280, 281], "_component_": [9, 10, 12, 13, 16, 17, 18, 19, 20, 21, 22, 23, 24, 26, 27, 53, 60, 64, 67, 71, 256, 269, 273, 274, 276, 277, 278, 279, 280], "null": [9, 22, 24, 280], "have": [9, 10, 13, 14, 18, 21, 22, 24, 27, 35, 36, 55, 60, 67, 75, 79, 80, 81, 142, 147, 151, 181, 182, 183, 184, 187, 189, 191, 192, 193, 194, 196, 198, 199, 204, 210, 218, 224, 227, 230, 232, 237, 238, 250, 258, 266, 273, 274, 275, 276, 277, 278, 279, 280, 281], "singl": [9, 10, 16, 17, 18, 19, 22, 24, 27, 33, 35, 37, 42, 47, 53, 54, 55, 56, 57, 60, 67, 71, 79, 80, 81, 94, 105, 106, 119, 142, 143, 144, 147, 148, 151, 161, 162, 169, 183, 189, 191, 199, 228, 229, 230, 231, 232, 234, 270, 272, 273, 274, 275, 276, 277, 278, 279, 281], "name": [9, 12, 13, 14, 16, 18, 20, 22, 23, 24, 26, 28, 31, 33, 35, 37, 42, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 204, 208, 210, 214, 228, 229, 230, 231, 232, 234, 247, 248, 249, 250, 251, 257, 258, 260, 272, 273, 274, 276, 279, 280], "messag": [9, 11, 12, 15, 16, 18, 19, 21, 31, 32, 33, 35, 37, 38, 39, 42, 44, 52, 55, 56, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 94, 106, 119, 144, 148, 162, 169, 212, 216, 266, 272, 273], "contain": [9, 11, 13, 14, 15, 16, 18, 20, 22, 33, 35, 36, 42, 47, 48, 49, 50, 54, 55, 56, 57, 60, 65, 71, 119, 144, 148, 169, 179, 182, 183, 185, 189, 190, 199, 202, 204, 207, 208, 209, 214, 216, 219, 225, 228, 229, 230, 232, 234, 239, 245, 250, 256, 257, 259, 273, 274, 276, 278], "topic": [9, 265], "per": [9, 16, 47, 88, 89, 90, 98, 99, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 143, 144, 152, 163, 164, 170, 182, 191, 195, 218, 220, 221, 272, 279, 280, 281], "could": [9, 18, 19, 238, 277, 278], "system": [9, 12, 13, 18, 19, 32, 33, 35, 36, 37, 38, 39, 41, 42, 44, 52, 55, 56, 60, 62, 63, 64, 65, 66, 67, 68, 69, 94, 100, 106, 119, 148, 153, 162, 169, 179, 216, 273], "tool": [9, 18, 19, 22, 36, 38, 56, 153, 247, 274, 275], "call": [9, 14, 18, 21, 22, 27, 36, 38, 56, 65, 66, 153, 183, 189, 191, 192, 193, 195, 199, 209, 247, 248, 249, 250, 251, 252, 256, 257, 273, 278, 281], "return": [9, 10, 13, 15, 18, 19, 21, 27, 29, 36, 38, 44, 45, 46, 47, 48, 49, 50, 51, 54, 55, 56, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 91, 92, 93, 94, 95, 96, 97, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 116, 117, 118, 119, 120, 121, 122, 125, 126, 127, 128, 129, 130, 131, 132, 136, 137, 138, 139, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 154, 155, 156, 157, 158, 159, 160, 161, 162, 165, 166, 167, 168, 169, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 188, 189, 190, 191, 193, 194, 196, 197, 198, 199, 200, 201, 202, 204, 205, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, 230, 232, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 254, 255, 256, 260, 261, 262, 277, 278, 281], "dai": [9, 20], "todai": 9, "It": [9, 10, 14, 16, 32, 36, 38, 55, 56, 60, 62, 64, 65, 66, 68, 70, 144, 147, 151, 153, 187, 189, 191, 199, 221, 224, 247, 272, 273, 277, 281], "tuesdai": 9, "about": [9, 10, 13, 14, 18, 22, 25, 65, 66, 191, 221, 224, 247, 251, 267, 268, 269, 270, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "tomorrow": 9, "wednesdai": 9, "As": [9, 12, 16, 22, 24, 25, 26, 205, 267, 274, 279, 281], "an": [9, 10, 12, 14, 16, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 46, 50, 52, 53, 57, 60, 62, 64, 65, 66, 67, 68, 71, 72, 78, 79, 80, 107, 120, 129, 144, 147, 149, 151, 154, 156, 160, 165, 171, 172, 173, 176, 177, 183, 187, 189, 191, 199, 200, 201, 203, 204, 206, 207, 208, 212, 217, 218, 221, 226, 227, 228, 229, 230, 232, 233, 237, 238, 247, 251, 256, 260, 267, 268, 269, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "slimorca": [9, 69], "pass": [9, 10, 11, 12, 14, 15, 16, 17, 18, 19, 20, 21, 24, 27, 36, 38, 53, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 91, 95, 101, 107, 116, 120, 125, 129, 149, 154, 156, 158, 160, 165, 167, 171, 175, 183, 189, 193, 194, 195, 199, 206, 210, 214, 222, 226, 230, 236, 237, 239, 243, 246, 247, 250, 251, 253, 256, 272, 273, 278, 280, 281], "repo": [9, 10, 12, 16, 18, 20, 22, 65, 228, 229, 231, 272, 274], "select": [9, 233], "one": [9, 10, 11, 12, 13, 16, 18, 22, 25, 33, 35, 37, 42, 47, 50, 52, 60, 66, 67, 191, 196, 198, 216, 230, 247, 274, 275, 276, 279, 281], "most": [9, 12, 13, 16, 18, 20, 22, 24, 36, 38, 273, 275, 278, 279, 281], "gemma": [9, 12, 18, 20, 92, 93, 94, 95, 96, 97, 98, 99, 187, 231, 279], "gemma_token": [9, 12, 18, 20], "g_token": [9, 12, 18, 20], "open": [9, 20, 46, 69, 92, 93, 274], "orca": [9, 69], "dedup": [9, 69], "recip": [9, 11, 12, 16, 18, 20, 22, 23, 24, 26, 27, 28, 29, 144, 189, 199, 228, 229, 230, 267, 268, 269, 273, 274, 276, 279, 281], "via": [9, 12, 14, 16, 17, 18, 20, 23, 24, 26, 55, 60, 64, 67, 71, 183, 189, 190, 205, 228, 278, 281], "http": [9, 12, 16, 27, 46, 57, 61, 63, 65, 71, 72, 74, 82, 83, 84, 85, 86, 87, 88, 89, 90, 92, 93, 95, 96, 97, 98, 99, 102, 103, 104, 105, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 152, 154, 155, 156, 157, 159, 161, 163, 164, 165, 166, 168, 169, 170, 172, 173, 174, 176, 177, 178, 183, 184, 185, 191, 196, 197, 218, 219, 221, 222, 223, 224, 226, 228, 229, 235, 244, 247, 250, 251, 253, 255, 261, 266, 274, 276, 277], "ha": [9, 18, 22, 64, 73, 143, 186, 188, 189, 191, 194, 196, 198, 199, 202, 204, 206, 207, 210, 225, 230, 232, 257, 258, 273, 274, 275, 276, 277, 278, 279, 281], "addition": [9, 22, 213, 214, 224, 255, 273, 278, 279], "argument": [9, 10, 12, 16, 22, 24, 27, 34, 40, 43, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 88, 89, 90, 98, 99, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 152, 163, 164, 170, 226, 237, 243, 247, 248, 250, 251, 253, 272, 273, 278, 279, 280], "load_dataset": [9, 12, 16, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 273], "document": [9, 12, 16, 17, 78, 79, 80, 81, 183, 189, 190, 226, 237, 246, 268, 270, 272, 279], "file": [9, 10, 11, 12, 16, 22, 23, 24, 25, 26, 27, 28, 29, 46, 55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 119, 144, 148, 169, 179, 213, 214, 215, 228, 229, 230, 248, 251, 256, 264, 267, 269, 271, 272, 273, 274, 275, 276, 278, 279, 280, 281], "raw": [9, 11, 13, 14, 16, 21, 44], "vari": [9, 50, 54, 189], "field": [9, 10, 14, 15, 27, 31, 35, 36, 42, 44, 54, 55, 56, 59, 65, 66, 245], "indic": [9, 14, 16, 18, 19, 50, 53, 54, 75, 76, 81, 147, 151, 183, 185, 189, 190, 191, 199, 200, 218, 219, 222, 225, 226, 241, 244, 273], "There": [9, 24, 52, 79, 273, 275, 276, 277, 278, 279], "few": [9, 10, 201, 276, 278, 281], "standard": [9, 12, 14, 15, 17, 19, 22, 34, 55, 56, 60, 63, 101, 107, 116, 120, 125, 129, 144, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 183, 249, 267, 273, 274, 276, 277], "across": [9, 22, 25, 50, 53, 228, 250, 255, 274, 276, 277, 280], "mani": [9, 14, 16, 19, 24, 54, 268, 269, 274, 277], "we": [9, 10, 11, 12, 18, 19, 20, 21, 22, 23, 24, 25, 26, 47, 50, 54, 55, 56, 60, 61, 67, 72, 73, 77, 182, 183, 185, 189, 190, 191, 193, 196, 198, 199, 205, 221, 224, 228, 229, 230, 236, 240, 246, 252, 257, 267, 268, 269, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "ipython": [9, 14, 19, 36, 38, 41, 55, 56, 94, 106, 119, 148, 162, 169, 179], "transform": [9, 10, 11, 16, 22, 25, 31, 33, 35, 55, 56, 59, 60, 62, 63, 65, 66, 67, 68, 69, 81, 85, 86, 87, 91, 95, 96, 97, 101, 107, 108, 109, 110, 111, 116, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 143, 144, 146, 147, 148, 149, 150, 151, 154, 155, 156, 157, 158, 160, 165, 166, 167, 171, 172, 173, 174, 175, 188, 189, 190, 191, 201, 218, 235, 253, 278, 279, 280], "sharegpttomessag": [9, 13, 60, 69], "expect": [9, 12, 13, 15, 16, 18, 19, 20, 22, 24, 27, 31, 33, 35, 36, 37, 42, 46, 50, 55, 56, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 143, 144, 185, 199, 210, 232, 247, 251, 258, 273, 278, 279, 280], "code": [9, 10, 12, 13, 16, 19, 21, 22, 25, 82, 83, 84, 85, 86, 87, 88, 89, 90, 189, 247, 263, 267, 275, 279], "openaitomessag": [9, 13, 60, 67], "If": [9, 10, 13, 14, 16, 17, 19, 21, 22, 24, 30, 33, 35, 36, 37, 42, 44, 46, 47, 50, 51, 52, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 71, 73, 75, 77, 81, 94, 101, 106, 107, 116, 119, 120, 125, 129, 142, 144, 146, 148, 150, 162, 169, 171, 175, 179, 182, 183, 185, 187, 189, 190, 191, 193, 194, 195, 196, 198, 199, 205, 210, 216, 228, 229, 230, 231, 232, 233, 236, 237, 238, 239, 240, 243, 247, 250, 251, 255, 256, 258, 260, 266, 272, 273, 274, 275, 276, 277, 278, 279, 280], "doe": [9, 17, 22, 44, 50, 54, 67, 71, 91, 153, 158, 168, 183, 187, 189, 190, 192, 193, 194, 197, 198, 199, 204, 216, 228, 230, 232, 257, 272, 274, 280], "fit": [9, 25, 54, 61, 71, 72, 191, 221, 273], "creat": [9, 10, 13, 16, 19, 22, 24, 27, 38, 54, 56, 60, 67, 75, 82, 83, 84, 85, 86, 87, 88, 89, 90, 92, 93, 96, 97, 98, 99, 102, 103, 104, 105, 108, 109, 110, 111, 112, 113, 114, 115, 117, 118, 121, 122, 123, 124, 126, 127, 128, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 152, 155, 157, 159, 161, 163, 164, 166, 168, 170, 172, 173, 174, 176, 177, 178, 182, 183, 189, 190, 191, 199, 226, 228, 229, 230, 234, 235, 247, 248, 250, 272, 274, 281], "custom": [9, 15, 16, 21, 24, 25, 31, 38, 55, 56, 60, 64, 65, 66, 67, 71, 94, 106, 119, 148, 162, 169, 179, 253, 267, 268, 269, 272, 275, 276, 278, 279], "dialogu": [9, 16, 43, 68, 273], "defin": [9, 10, 17, 22, 24, 25, 38, 55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 183, 188, 189, 199, 202, 204, 205, 207, 220, 275, 278], "same": [9, 10, 11, 15, 18, 22, 24, 38, 45, 78, 79, 85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 143, 166, 172, 173, 174, 181, 182, 184, 186, 187, 188, 190, 191, 194, 199, 201, 216, 222, 224, 225, 232, 237, 238, 251, 257, 259, 269, 272, 273, 274, 276, 277, 278, 279, 280, 281], "wai": [9, 14, 19, 22, 24, 55, 56, 209, 227, 272, 274, 275, 276, 277], "instruct_dataset": [9, 12, 13, 53], "info": [9, 261, 275], "slimorca_dataset": [9, 24], "command": [10, 12, 17, 21, 23, 25, 26, 266, 269, 270, 272, 273, 274, 275, 276, 277, 278, 280, 281], "line": [10, 17, 22, 23, 25, 270, 272, 275, 276], "both": [10, 14, 15, 21, 22, 37, 50, 53, 63, 67, 180, 199, 201, 203, 210, 272, 274, 277, 278, 279, 280, 281], "built": [10, 11, 13, 23, 24, 26, 63, 67, 70, 266, 273, 275, 281], "done": [10, 17, 54, 189, 209, 236, 246, 257, 278, 280, 281], "run": [10, 17, 22, 23, 24, 26, 29, 91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 182, 183, 189, 195, 196, 228, 229, 230, 232, 233, 234, 244, 247, 250, 251, 252, 266, 267, 268, 269, 270, 273, 275, 276, 277, 278, 279, 280, 281], "cli": [10, 24, 26, 28, 29, 266, 268, 274, 275, 279], "which": [10, 11, 12, 14, 16, 17, 18, 19, 20, 21, 22, 24, 25, 46, 47, 53, 54, 57, 59, 60, 62, 64, 67, 68, 69, 71, 76, 77, 85, 86, 87, 94, 95, 96, 97, 106, 107, 108, 109, 110, 111, 119, 120, 121, 122, 129, 130, 131, 132, 138, 139, 144, 147, 148, 149, 150, 151, 153, 154, 155, 156, 157, 162, 165, 166, 169, 171, 172, 173, 174, 182, 183, 185, 189, 190, 191, 192, 193, 194, 199, 201, 209, 210, 213, 228, 229, 230, 232, 235, 236, 248, 251, 253, 257, 267, 268, 269, 270, 272, 273, 274, 275, 277, 278, 279, 280, 281], "folder": [10, 22], "first": [10, 17, 22, 24, 27, 42, 52, 54, 65, 76, 81, 147, 151, 189, 191, 192, 193, 199, 225, 228, 265, 267, 268, 273, 274, 276, 277, 278, 280, 281], "ensur": [10, 19, 21, 22, 24, 30, 52, 55, 56, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 183, 192, 228, 230, 236, 267, 275], "instal": [10, 23, 24, 26, 244, 247, 250, 251, 265, 272, 274, 275, 276, 277, 278, 279, 280, 281], "environ": [10, 25, 244, 247, 260, 266, 270, 272, 274, 275, 280], "so": [10, 13, 18, 19, 20, 22, 24, 54, 65, 187, 191, 228, 266, 267, 273, 274, 275, 276, 278, 279, 280, 281], "directori": [10, 22, 24, 35, 42, 65, 228, 229, 230, 248, 250, 251, 256, 272, 274, 275, 276], "new": [10, 14, 15, 16, 19, 21, 25, 37, 42, 59, 61, 62, 63, 65, 67, 68, 69, 159, 182, 200, 201, 231, 247, 248, 250, 273, 274, 275, 276, 277, 278, 281], "librari": [10, 221, 223, 236, 255, 261, 265, 266, 267, 272, 279, 281], "mkdir": 10, "my_project": [10, 247, 251], "cd": [10, 21, 266, 274], "llama": [10, 15, 16, 17, 20, 21, 22, 100, 142, 144, 145, 146, 147, 148, 150, 151, 184, 185, 228, 229, 268, 269, 272, 273, 274, 275, 276, 277, 278], "3": [10, 15, 16, 17, 20, 21, 22, 45, 47, 48, 49, 50, 54, 76, 77, 81, 142, 144, 145, 146, 147, 148, 150, 151, 153, 166, 168, 169, 191, 231, 241, 254, 261, 268, 269, 272, 273, 274, 275, 276, 277, 280, 281], "2": [10, 14, 15, 17, 21, 22, 26, 45, 47, 48, 49, 50, 52, 54, 69, 76, 77, 78, 79, 136, 137, 138, 139, 140, 141, 142, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 182, 183, 191, 213, 214, 216, 222, 224, 225, 228, 229, 231, 241, 254, 255, 256, 262, 269, 273, 274, 275, 276, 278, 279, 280], "lora": [10, 24, 85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 205, 206, 209, 210, 228, 246, 265, 267, 270, 273, 275, 276, 277], "devic": [10, 17, 24, 25, 194, 232, 236, 239, 259, 260, 270, 272, 273, 274, 275, 276, 278, 279, 281], "lora_finetune_single_devic": [10, 24, 268, 272, 273, 274, 275, 276, 277, 278, 279, 281], "llama3_2": [10, 17, 136, 137, 138, 139, 140, 141, 192, 193, 194, 231, 277], "1b_lora_single_devic": 10, "often": [10, 278, 279], "ll": [10, 18, 20, 22, 24, 25, 73, 240, 267, 269, 273, 274, 275, 276, 277, 279, 280, 281], "want": [10, 12, 19, 22, 24, 25, 26, 27, 50, 55, 56, 73, 203, 266, 272, 273, 274, 275, 276, 277, 278, 279], "start": [10, 23, 25, 26, 46, 76, 216, 231, 247, 266, 267, 273, 274, 275, 277, 279, 280], "our": [10, 12, 13, 20, 22, 25, 267, 268, 269, 270, 273, 274, 275, 277, 278, 279, 280, 281], "particular": [10, 11, 13, 19, 21, 24, 53, 144, 226, 278, 281], "adjust": [10, 268, 269, 277, 279, 280], "hyperparamet": [10, 23, 224, 232, 267, 275, 278, 281], "cp": [10, 24, 266, 272, 273, 274, 275, 276, 280], "copi": [10, 273, 274, 275, 276, 279, 280, 281], "make": [10, 17, 19, 22, 23, 24, 25, 26, 145, 191, 267, 272, 274, 275, 276, 277, 278, 279, 280, 281], "modif": [10, 280], "show": [10, 144, 218, 266, 268, 269, 272, 273, 277, 278], "each": [10, 12, 15, 18, 19, 20, 22, 25, 38, 39, 42, 47, 48, 50, 53, 54, 55, 56, 78, 79, 80, 81, 85, 86, 87, 95, 96, 97, 107, 108, 109, 110, 111, 120, 121, 122, 129, 130, 131, 132, 138, 139, 143, 144, 147, 149, 150, 151, 154, 155, 156, 157, 165, 166, 171, 172, 173, 174, 183, 185, 189, 190, 191, 196, 198, 199, 201, 209, 210, 216, 218, 219, 220, 221, 223, 224, 241, 255, 256, 267, 269, 270, 272, 274, 275, 278, 279, 280], "ls": [10, 21, 266, 270, 272, 274, 275, 276], "full": [10, 11, 13, 16, 22, 24, 25, 34, 40, 43, 55, 72, 88, 89, 90, 98, 99, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 149, 152, 163, 164, 170, 199, 209, 210, 216, 233, 266, 267, 270, 272, 274, 276, 278, 279, 280], "5b_full_single_devic": 10, "qwen_config": 10, "now": [10, 19, 22, 182, 193, 232, 234, 269, 273, 274, 275, 276, 277, 278, 280, 281], "sure": [10, 17, 22, 24, 274, 275, 276, 277, 278, 279, 280, 281], "correct": [10, 12, 14, 19, 25, 34, 62, 184, 185, 189, 260, 267, 273], "ve": [10, 18, 21, 24, 182, 269, 272, 273, 274, 276, 277, 278, 279], "even": [10, 191, 257, 266, 272, 273, 276, 277, 278, 279, 281], "didn": 10, "t": [10, 13, 14, 18, 19, 20, 22, 24, 25, 45, 142, 143, 196, 201, 236, 251, 255, 272, 273, 274, 275, 277, 279, 281], "complet": [10, 11, 12, 18, 22, 25, 37, 54, 61, 71, 169, 273, 274, 275, 276, 279], "note": [10, 16, 21, 22, 24, 95, 199, 204, 232, 252, 255, 257, 269, 273, 274, 277, 278, 279, 280, 281], "must": [10, 13, 17, 27, 38, 53, 65, 66, 183, 193, 204, 227, 247, 281], "extens": [10, 25, 230, 267], "full_finetune_single_devic": [10, 17, 238, 272, 274, 275], "Or": [10, 199, 266], "rel": [10, 16, 17, 54, 183, 185, 189, 190, 199, 221, 239, 277, 278], "discuss": [10, 14, 19, 21, 24, 274, 275, 276, 278], "workflow": [10, 11, 265, 275, 278], "write": [10, 16, 22, 25, 228, 229, 230, 248, 275], "own": [10, 13, 18, 21, 22, 38, 246, 255, 272, 273, 274, 276, 277, 278], "loop": 10, "logic": [10, 15, 25, 31, 56, 212, 231, 267, 270, 275, 278], "case": [10, 14, 16, 22, 25, 26, 36, 38, 55, 79, 80, 81, 147, 151, 191, 193, 228, 232, 236, 240, 246, 248, 253, 267, 272, 273, 274, 276, 278, 279, 281], "similar": [10, 13, 16, 60, 61, 63, 65, 66, 67, 70, 71, 72, 209, 221, 274, 276, 277, 278, 279, 281], "scratch": 10, "local": [10, 11, 14, 46, 55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 119, 148, 169, 179, 247, 251, 255, 266, 272, 273, 274, 275], "single_devic": 10, "py": [10, 13, 24, 27, 74, 85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 172, 173, 174, 182, 184, 185, 197, 221, 222, 223, 224, 235, 272, 274, 276], "recommend": [10, 60, 61, 62, 67, 68, 70, 72, 153, 189, 196, 247, 250, 273, 274, 279, 281], "python": [10, 24, 247, 251, 255, 261, 263, 272, 274, 280], "convent": [10, 188], "main": [10, 27, 29, 169, 184, 185, 266, 269, 274, 276], "script": [10, 22, 26, 270, 272, 274, 275, 276], "decor": [10, 25, 29], "pars": [10, 24, 27, 28, 215, 270, 275], "omegaconf": [10, 27], "dictconfig": [10, 24, 25, 27, 28, 29, 30, 247, 251, 256], "def": [10, 13, 15, 19, 21, 24, 25, 26, 29, 65, 66, 226, 231, 277, 278, 281], "cfg": [10, 24, 25, 28, 29, 30], "add": [10, 12, 13, 14, 16, 19, 21, 23, 24, 26, 50, 54, 57, 71, 144, 153, 191, 203, 214, 216, 230, 231, 273, 274, 276, 278, 281], "here": [10, 12, 14, 15, 16, 18, 20, 21, 22, 23, 24, 26, 32, 62, 65, 66, 184, 185, 238, 268, 269, 270, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281], "attribut": [10, 187, 206, 216, 224, 234], "__name__": 10, "__main__": 10, "don": [10, 13, 14, 18, 19, 20, 22, 24, 25, 251, 255, 272, 273, 274, 275, 277, 279, 281], "experiment": [10, 21, 24], "optim": [10, 18, 19, 22, 24, 25, 48, 53, 55, 91, 158, 168, 221, 222, 223, 224, 230, 232, 234, 235, 238, 239, 252, 256, 268, 269, 270, 273, 274, 275, 276, 277, 278, 281], "them": [10, 12, 15, 18, 19, 22, 24, 53, 67, 191, 195, 201, 216, 259, 269, 272, 273, 274, 278, 279, 280, 281], "when": [10, 16, 17, 18, 20, 21, 22, 24, 25, 29, 53, 54, 55, 56, 57, 67, 71, 73, 75, 182, 183, 185, 189, 190, 191, 193, 194, 195, 196, 198, 199, 200, 206, 209, 220, 235, 237, 250, 252, 257, 268, 272, 274, 276, 277, 278, 279, 280, 281], "mean": [10, 24, 144, 183, 184, 188, 189, 190, 199, 219, 246, 272, 273, 275, 278, 280], "high": [10, 53, 55, 56, 267, 277, 278, 279], "level": [10, 25, 55, 56, 196, 198, 217, 234, 246, 261, 267, 277, 281], "paramet": [10, 13, 14, 15, 16, 25, 27, 28, 29, 30, 31, 33, 35, 36, 37, 38, 42, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 91, 92, 93, 94, 95, 96, 97, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 116, 117, 118, 119, 120, 121, 122, 125, 126, 127, 128, 129, 130, 131, 132, 136, 137, 138, 139, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 154, 155, 156, 157, 158, 159, 160, 161, 162, 165, 166, 167, 169, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, 229, 230, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 243, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 262, 265, 267, 268, 269, 270, 272, 273, 274, 275, 276, 277, 278, 280, 281], "easili": [10, 16, 22, 24, 267, 277, 278, 280, 281], "custom_decod": 10, "customtransformerdecod": 10, "nn": [10, 27, 45, 47, 50, 81, 142, 143, 180, 182, 183, 187, 188, 189, 190, 191, 192, 193, 194, 195, 199, 200, 201, 202, 203, 204, 206, 207, 208, 226, 233, 234, 246, 252, 253, 257, 258, 277, 278, 281], "modul": [10, 13, 15, 21, 24, 27, 65, 66, 78, 79, 80, 81, 142, 143, 144, 147, 151, 156, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 217, 218, 221, 222, 223, 226, 231, 233, 234, 237, 246, 252, 253, 255, 275, 277, 278, 281], "A": [10, 13, 15, 19, 25, 26, 33, 34, 37, 40, 42, 43, 47, 48, 49, 50, 53, 54, 67, 81, 179, 183, 187, 188, 189, 190, 191, 195, 199, 205, 209, 213, 214, 216, 218, 219, 220, 221, 222, 223, 224, 225, 226, 231, 232, 238, 239, 240, 245, 246, 264, 265, 271, 272, 273, 278, 279, 280, 281], "architectur": [10, 25, 100, 153, 189, 191, 199, 201, 231, 272], "present": [10, 33, 37, 42, 62, 63, 65, 66, 67, 68, 69, 214, 230, 257], "custom_model": 10, "num_lay": [10, 27, 81, 91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 189, 191, 199, 201], "int": [10, 15, 21, 24, 26, 47, 48, 49, 50, 51, 54, 61, 65, 66, 72, 73, 74, 75, 77, 78, 79, 80, 81, 85, 86, 87, 88, 89, 90, 91, 94, 95, 96, 97, 98, 99, 101, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 119, 120, 121, 122, 123, 124, 125, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 154, 155, 156, 157, 158, 160, 162, 163, 164, 165, 166, 167, 169, 170, 171, 172, 173, 174, 175, 179, 182, 183, 184, 185, 188, 189, 190, 191, 194, 196, 197, 198, 199, 200, 201, 205, 211, 212, 213, 214, 215, 216, 218, 225, 226, 228, 229, 230, 232, 233, 235, 237, 242, 246, 247, 248, 249, 250, 251, 253, 255, 256, 272, 277, 278, 279, 281], "classification_head": 10, "bool": [10, 15, 19, 21, 24, 31, 33, 35, 36, 37, 42, 45, 54, 57, 58, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 81, 85, 86, 87, 88, 89, 90, 91, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 144, 145, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 175, 183, 188, 189, 190, 195, 196, 197, 199, 201, 205, 209, 210, 212, 213, 214, 216, 219, 225, 226, 228, 229, 230, 237, 239, 243, 244, 246, 247, 250, 253, 256, 257, 262, 279, 281], "fals": [10, 13, 14, 15, 16, 18, 19, 22, 24, 33, 35, 36, 37, 42, 45, 53, 54, 58, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 75, 76, 81, 85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 145, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 175, 183, 189, 190, 192, 193, 194, 199, 200, 201, 205, 206, 209, 213, 225, 228, 229, 230, 241, 244, 256, 257, 272, 273, 274, 276, 278, 280, 281], "setup": [10, 22, 24, 25, 75, 182, 183, 188, 189, 190, 192, 193, 194, 199, 201, 233, 256, 272, 274, 278, 281], "expos": [10, 13, 24, 25, 230, 270, 275], "friendli": [10, 60, 64, 67, 71, 73, 273], "manner": [10, 20], "rather": [10, 221, 279], "everi": [10, 12, 22, 25, 62, 63, 67, 68, 69, 78, 79, 80, 146, 150, 191, 193, 250, 256, 266, 272, 279, 281], "construct": [10, 36, 63, 218, 270, 278], "care": [10, 22, 228, 230, 274, 276, 278], "how": [10, 13, 14, 18, 22, 23, 24, 25, 26, 191, 226, 247, 253, 265, 268, 269, 272, 273, 274, 275, 276, 279, 280, 281], "implement": [10, 19, 21, 22, 25, 55, 57, 59, 60, 61, 62, 63, 64, 67, 68, 69, 70, 71, 72, 180, 184, 185, 186, 191, 197, 204, 205, 211, 212, 217, 221, 222, 223, 224, 228, 235, 240, 250, 267, 269, 277, 278, 279, 280, 281], "llama3_2_vision_11b": 10, "custom_dataset": [10, 13], "sftdataset": [10, 13, 24, 55, 58, 59, 60, 62, 64, 65, 66, 68, 69], "packeddataset": [10, 17, 53, 58, 59, 60, 62, 64, 68, 69, 71, 72], "inputoutputtomessag": [10, 13, 14, 62, 68], "modeltoken": [10, 15, 21, 24, 36, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 67, 68, 69, 70, 71, 72, 216], "build": [10, 25, 71, 81, 91, 101, 116, 125, 146, 147, 150, 151, 158, 160, 175, 227, 267, 276, 278], "block": [10, 25, 54, 85, 86, 87, 91, 95, 96, 97, 101, 107, 108, 109, 110, 111, 116, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 146, 149, 150, 151, 154, 155, 156, 157, 158, 165, 166, 171, 172, 173, 174, 175, 183, 189, 190, 209, 210, 267], "tiny_cod": 10, "pack": [10, 54, 58, 59, 60, 62, 64, 65, 66, 68, 69, 71, 72, 183, 185, 189, 190, 199, 280], "subset": [10, 15, 16, 47, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 95, 107, 120, 129, 149, 154, 156, 165, 171, 202, 207], "nampdn": 10, "tini": 10, "respons": [10, 12, 13, 18, 19, 21, 32, 33, 35, 36, 37, 42, 55, 56, 60, 62, 63, 64, 65, 66, 67, 68, 69, 216, 219, 220, 221, 223, 224, 274, 275, 276], "model_transform": [10, 13, 15, 16, 55, 56, 62, 65, 66, 68, 69, 144], "message_transform": [10, 13, 55, 56], "column_map": [10, 12, 13, 16, 18, 31, 33, 35, 37, 42, 53, 58, 59, 62, 63, 64, 65, 66, 67, 68, 69, 70], "input": [10, 11, 12, 13, 14, 15, 20, 21, 22, 31, 35, 47, 48, 49, 50, 54, 55, 56, 59, 61, 62, 64, 65, 66, 68, 69, 72, 78, 79, 80, 81, 94, 106, 119, 142, 143, 144, 147, 148, 151, 162, 169, 171, 175, 180, 181, 183, 184, 185, 186, 187, 188, 189, 190, 191, 197, 198, 199, 200, 201, 205, 213, 214, 218, 228, 230, 238, 255, 258, 273, 278, 281], "filter_fn": [10, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72], "lambda": [10, 219], "x": [10, 22, 45, 73, 74, 75, 78, 79, 80, 142, 143, 180, 181, 183, 184, 185, 186, 188, 189, 190, 191, 199, 200, 201, 205, 241, 254, 277, 278, 280, 281], "split_across_pack": [10, 54, 71], "els": [10, 11, 12, 19, 25, 251, 267, 281], "posit": [10, 17, 24, 27, 54, 74, 76, 78, 79, 80, 81, 91, 95, 125, 129, 142, 147, 151, 154, 156, 158, 160, 165, 167, 182, 183, 185, 188, 189, 190, 191, 199, 200, 276], "automat": [10, 12, 16, 17, 19, 21, 23, 24, 26, 27, 59, 60, 272, 274, 281], "instanti": [10, 30, 38, 82, 83, 84, 85, 86, 87, 91, 92, 93, 94, 95, 96, 97, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 116, 117, 118, 119, 120, 121, 122, 125, 126, 127, 128, 129, 130, 131, 132, 136, 137, 138, 139, 145, 146, 147, 148, 149, 150, 151, 154, 155, 156, 157, 158, 159, 160, 161, 162, 165, 166, 167, 168, 169, 171, 172, 173, 174, 175, 176, 177, 178, 179, 232], "separ": [10, 55, 201, 216, 228, 273, 275, 276, 278, 281], "under": [10, 24, 256, 279, 281], "best": [10, 16, 18, 25, 269, 273, 277, 279], "root": [10, 184, 250, 251], "custom_finetun": 10, "32": [10, 27, 182, 191, 199, 201, 247, 276, 278, 279, 280, 281], "option": [10, 12, 18, 21, 22, 24, 25, 31, 33, 35, 37, 42, 50, 51, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 77, 80, 81, 85, 86, 87, 94, 95, 96, 97, 101, 106, 107, 108, 109, 110, 111, 116, 119, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 142, 143, 144, 146, 147, 148, 149, 150, 151, 154, 155, 156, 157, 162, 165, 166, 169, 171, 172, 173, 174, 175, 179, 180, 183, 185, 188, 189, 190, 191, 194, 195, 199, 209, 210, 211, 213, 216, 218, 219, 220, 222, 228, 229, 230, 232, 233, 236, 240, 247, 248, 251, 255, 256, 260, 261, 266, 267, 272, 273, 274, 279], "param": [10, 22, 25, 85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 145, 149, 166, 172, 173, 174, 202, 203, 205, 207, 208, 210, 228, 278, 280, 281], "omit": [10, 277, 278, 279], "being": [10, 19, 22, 56, 193, 228, 229, 230, 234, 260, 279, 280, 281], "found": [10, 11, 22, 23, 24, 26, 184, 185, 228, 229, 230, 269, 272, 277, 278, 281], "correctli": [10, 21, 22, 25, 30, 209, 228, 266, 270, 273, 275, 281], "try": [10, 22, 24, 273, 274, 275, 276, 281], "after": [10, 19, 20, 23, 25, 38, 56, 65, 66, 94, 106, 119, 144, 148, 162, 169, 182, 183, 186, 187, 189, 190, 199, 201, 225, 246, 247, 248, 249, 250, 251, 269, 273, 274, 276, 280, 281], "pythonpath": 10, "pwd": 10, "vlm": [11, 16], "hub": [11, 22, 55, 56, 272, 275], "remot": [11, 14, 35, 42, 46, 55, 56], "url": [11, 16, 35, 37, 42, 46, 266], "project": [11, 23, 26, 81, 85, 86, 87, 91, 95, 101, 105, 107, 108, 109, 110, 111, 116, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 142, 143, 146, 147, 149, 150, 151, 154, 155, 156, 157, 158, 161, 165, 166, 171, 174, 175, 180, 183, 189, 191, 199, 203, 209, 210, 231, 237, 247, 251, 265, 278, 279, 281], "prefer": [11, 13, 25, 48, 55, 63, 67, 70, 221, 222, 223, 224, 267, 270, 272, 279], "align": [11, 65, 66, 221, 273, 277], "continu": [11, 20, 54, 191, 247], "pretrain": [11, 142, 143, 144, 199, 201, 203, 213, 214, 272, 273, 275, 278, 281], "beyond": [11, 274, 279, 281], "those": [11, 22, 231, 274, 276, 278], "customiz": 11, "task": [11, 12, 16, 18, 19, 34, 40, 43, 53, 61, 144, 268, 273, 274, 276, 277, 278, 279, 280, 281], "supervis": [11, 20, 56], "rlhf": [11, 55, 63, 219, 220, 221, 222, 223, 224, 225], "queri": [11, 91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 182, 183, 189, 190, 199, 276, 279], "time": [11, 16, 17, 22, 60, 64, 91, 158, 193, 196, 198, 216, 219, 248, 250, 256, 269, 272, 273, 274, 276, 281], "take": [11, 12, 13, 16, 18, 22, 24, 25, 27, 48, 55, 56, 65, 66, 67, 142, 182, 191, 195, 201, 228, 230, 259, 260, 269, 273, 274, 275, 276, 277, 278, 279, 281], "object": [11, 13, 14, 15, 19, 21, 24, 27, 28, 81, 183, 221, 224, 226, 240], "appli": [11, 12, 15, 19, 22, 25, 47, 55, 56, 59, 65, 66, 85, 86, 87, 88, 89, 90, 91, 95, 96, 97, 98, 99, 101, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 120, 121, 122, 123, 124, 125, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 146, 149, 150, 151, 152, 154, 155, 156, 157, 158, 163, 164, 165, 166, 170, 171, 172, 173, 174, 175, 183, 187, 188, 189, 190, 199, 209, 210, 253, 267, 268, 277, 279, 281], "templat": [11, 31, 32, 34, 38, 39, 40, 43, 55, 56, 59, 62, 68, 94, 100, 106, 119, 144, 148, 153, 162, 169, 179], "anyth": [11, 61, 259], "requir": [11, 15, 17, 19, 21, 22, 24, 47, 48, 53, 55, 56, 57, 65, 66, 67, 71, 144, 187, 189, 200, 228, 230, 232, 243, 244, 246, 247, 250, 251, 255, 256, 266, 269, 272, 273, 275, 279, 280, 281], "collat": [11, 47, 49, 50, 54], "packag": [11, 23, 26, 247, 250, 251, 266], "togeth": [11, 25, 54, 196, 251, 270, 275, 278, 279, 280], "form": [12, 18, 22, 24, 25, 31, 44, 52, 55, 56, 272], "along": [12, 22, 278], "describ": [12, 253], "hand": [12, 36], "grammar": [12, 19, 34, 62], "head": [12, 81, 91, 95, 101, 107, 116, 120, 125, 129, 142, 146, 147, 150, 151, 154, 156, 158, 160, 165, 167, 171, 175, 182, 183, 185, 189, 199, 203, 231, 276], "csv": [12, 55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72], "incorrect": [12, 19], "cat": [12, 16, 19, 218], "grammarerrorcorrectiontempl": [12, 19, 62], "prepend": [12, 14, 16, 19, 33, 35, 37, 38, 39, 42, 60, 62, 63, 64, 65, 66, 67, 68, 69, 94, 106, 119, 144, 148, 162, 169, 179, 213], "english": [12, 19, 34], "ncorrect": [12, 34], "mask": [12, 13, 14, 15, 17, 19, 21, 36, 38, 50, 54, 56, 59, 60, 62, 64, 65, 66, 67, 68, 69, 74, 75, 76, 144, 183, 188, 189, 190, 199, 212, 216, 218, 219, 222, 241, 273, 277], "out": [12, 15, 18, 20, 22, 24, 25, 59, 60, 62, 64, 67, 68, 69, 75, 76, 218, 228, 229, 241, 265, 267, 268, 269, 270, 272, 273, 274, 275, 276, 278, 279, 281], "100": [12, 18, 25, 48, 49, 50, 59, 60, 62, 64, 67, 68, 69, 73, 196, 197, 198, 200, 277, 278, 281], "27957": 12, "736": 12, "577": 12, "anoth": [12, 13, 16, 24, 56, 187, 247, 274, 279], "c4": [12, 71, 280], "200m": 12, "liweili": [12, 62], "c4_200m": [12, 62], "chang": [12, 13, 16, 21, 22, 23, 24, 26, 31, 33, 35, 64, 66, 70, 230, 266, 272, 274, 275, 276, 277, 278, 279, 280, 281], "remap": 12, "someth": [12, 22, 25, 26, 273, 274, 280], "hello": [12, 13, 14, 19, 21, 44, 213, 214, 261, 273, 274, 276], "world": [12, 13, 14, 19, 21, 44, 213, 214, 242, 244, 261, 274], "bye": [12, 13], "robot": [12, 15], "am": [12, 14, 16, 60, 64, 100, 153, 273, 274, 276], "prompttempl": [12, 31, 34, 40, 43, 144], "relev": [12, 14, 25, 188, 189, 190, 199, 272, 274, 278, 279], "inform": [12, 14, 22, 247, 251, 253, 267, 272, 274, 275], "mai": [12, 16, 17, 24, 26, 60, 73, 191, 194, 200, 237, 257, 268, 269, 273, 275, 277, 278, 279], "alpaca_dataset": [12, 17, 24, 58], "grammar_dataset": 12, "samsum_dataset": 12, "dictionari": [13, 14, 15, 36, 38, 44, 47, 48, 49, 54, 55, 56, 94, 106, 119, 148, 162, 169, 179, 239, 245, 247, 248, 249, 250, 251, 259, 274], "onc": [13, 21, 24, 38, 189, 199, 274, 275, 276, 278, 281], "repres": [13, 36, 48, 78, 79, 191, 227, 233, 273, 279, 280], "prepar": [13, 15, 273, 280], "ad": [13, 16, 19, 21, 25, 38, 50, 78, 79, 80, 146, 150, 160, 191, 199, 200, 203, 213, 216, 230, 231, 273, 278, 279, 280, 281], "column": [13, 16, 18, 20, 31, 33, 35, 37, 42, 55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 183, 189, 190, 199, 280], "worri": [13, 22, 273, 275], "itself": [13, 24], "do": [13, 15, 18, 21, 22, 23, 25, 36, 47, 65, 67, 193, 209, 216, 247, 251, 257, 272, 274, 275, 276, 278, 279, 280], "well": [13, 18, 22, 24, 25, 267, 272, 274, 276, 277, 279, 281], "flexibl": [13, 24, 53, 279], "inherit": [13, 14, 19, 25, 267], "__call__": [13, 15, 19, 65, 66, 144], "simpl": [13, 22, 25, 191, 224, 265, 275, 278, 280, 281], "contriv": [13, 19], "would": [13, 15, 19, 22, 24, 26, 38, 54, 189, 191, 199, 266, 273, 274, 278, 279, 281], "inde": [13, 236, 274], "quit": [13, 279, 281], "type": [13, 14, 15, 16, 21, 26, 27, 29, 36, 37, 44, 45, 46, 47, 48, 49, 50, 51, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 91, 92, 93, 94, 95, 96, 97, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 116, 117, 118, 119, 120, 121, 122, 125, 126, 127, 128, 129, 130, 131, 132, 136, 137, 138, 139, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 154, 155, 156, 157, 158, 159, 160, 161, 162, 165, 166, 167, 168, 169, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 188, 189, 190, 191, 194, 195, 196, 197, 198, 199, 200, 201, 202, 205, 207, 211, 212, 213, 214, 215, 216, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, 229, 230, 231, 232, 234, 236, 237, 238, 239, 240, 241, 242, 243, 244, 246, 253, 254, 255, 256, 258, 260, 261, 262, 269, 274, 278, 279, 280, 281], "map": [13, 15, 19, 21, 22, 31, 33, 35, 37, 38, 42, 47, 53, 54, 59, 62, 63, 64, 65, 66, 67, 68, 69, 70, 94, 106, 119, 144, 148, 162, 169, 179, 208, 214, 215, 228, 232, 234, 247, 248, 249, 250, 251, 252, 256, 274, 278], "messagetransform": 13, "self": [13, 15, 18, 19, 20, 21, 25, 26, 54, 65, 66, 85, 86, 87, 91, 95, 96, 97, 101, 107, 108, 109, 110, 111, 116, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 146, 149, 150, 151, 154, 155, 156, 157, 158, 160, 165, 166, 167, 171, 172, 173, 174, 175, 183, 188, 189, 190, 196, 198, 199, 201, 204, 209, 210, 228, 231, 232, 277, 278, 281], "str": [13, 15, 21, 24, 27, 28, 31, 33, 35, 36, 37, 38, 42, 44, 46, 47, 48, 49, 50, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 94, 106, 119, 144, 148, 149, 152, 162, 169, 179, 195, 200, 201, 202, 204, 205, 207, 208, 209, 210, 211, 212, 213, 214, 215, 227, 228, 229, 230, 231, 232, 233, 236, 239, 240, 243, 245, 247, 248, 249, 250, 251, 255, 256, 257, 258, 260, 261, 262, 279], "eot": [13, 14, 19, 36, 144], "_messag": 13, "0x7fb0a10094e0": 13, "0x7fb0a100a290": 13, "msg": [13, 14, 16, 19, 21, 273], "text_cont": [13, 14, 16, 19, 36, 273], "manipul": 13, "load_dataset_kwarg": [13, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72], "mymessagetransform": 13, "chosenrejectedtomessag": [13, 63, 67], "core": [14, 25, 55, 56, 267, 270, 275, 281], "govern": [14, 273], "serv": [14, 19, 24, 33, 35, 37, 42, 60, 62, 63, 64, 65, 66, 67, 68, 69, 216, 226, 278], "interfac": [14, 25, 38, 39, 53, 204, 217], "api": [14, 25, 26, 34, 40, 43, 55, 56, 59, 65, 66, 88, 89, 90, 98, 99, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 152, 163, 164, 170, 209, 247, 266, 270, 272, 273, 275, 276, 281], "oper": [14, 25, 191, 206, 217, 255, 280], "send": 14, "other": [14, 15, 18, 20, 22, 25, 27, 35, 38, 53, 230, 237, 256, 259, 268, 269, 273, 275, 276, 277, 278, 279, 280], "special": [14, 16, 19, 36, 42, 119, 144, 146, 148, 150, 169, 179, 191, 200, 211, 212, 214, 215, 216, 218, 232], "individu": [14, 36, 54, 199, 239, 251, 253, 273], "ref": [14, 55, 56, 59, 65, 66, 168, 169, 251], "constructor": [14, 21], "ident": [14, 18, 20, 45, 47, 54, 65, 67, 153, 189, 274, 280], "from_dict": [14, 36, 273], "becaus": [14, 21, 55, 56, 95, 182, 189, 191, 199, 230, 272, 273, 280], "correspond": [14, 18, 21, 36, 48, 74, 75, 76, 202, 204, 207, 219, 222, 236, 269, 275, 276, 279, 280], "begin": [14, 22, 54, 71, 191, 214, 216, 273, 276, 281], "pil": [14, 15, 16, 36, 37, 44, 46], "img_msg": 14, "place": [14, 16, 20, 257, 273, 279], "mode": [14, 15, 16, 194, 233, 240, 247, 274], "rgb": [14, 15, 16, 142], "4": [14, 15, 16, 22, 24, 45, 47, 48, 49, 50, 76, 81, 144, 147, 151, 182, 183, 191, 241, 262, 267, 269, 272, 274, 276, 277, 278, 279, 280, 281], "appropri": [14, 36, 53, 76, 100, 200, 228, 235, 281], "load_imag": [14, 16], "image_path": [14, 16], "jpg": [14, 16, 35, 42, 46], "tag": [14, 16, 19, 21, 38, 42, 44, 94, 100, 106, 119, 144, 148, 153, 162, 169, 179, 247, 248, 249, 250, 251, 273], "placehold": [14, 16, 42, 227], "should": [14, 15, 16, 18, 20, 22, 24, 25, 33, 35, 36, 37, 38, 42, 47, 54, 59, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 75, 76, 85, 86, 87, 95, 96, 97, 100, 101, 107, 108, 109, 110, 111, 116, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 142, 146, 149, 150, 151, 153, 154, 155, 156, 157, 158, 160, 165, 166, 167, 171, 172, 173, 174, 175, 182, 183, 189, 191, 194, 199, 204, 209, 210, 219, 222, 226, 227, 245, 247, 248, 249, 250, 251, 266, 267, 274, 275, 276, 277, 278, 279, 280, 281], "insert": [14, 201, 280], "format_content_with_imag": [14, 16], "image_tag": [14, 16, 42, 44], "conveni": [14, 24, 25, 46, 272], "prompttemplateinterfac": [14, 19, 94, 106, 119, 148, 162, 169, 179], "templated_msg": [14, 19], "contains_media": [14, 16, 36], "get_media": [14, 15, 16, 36], "4x4": 14, "0x7f8d27e72740": 14, "tokenize_messsag": 14, "hi": [14, 20, 73, 273], "tokenize_messag": [14, 15, 21, 36, 55, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 144, 212, 216, 273], "22557": 14, "1526": [14, 21], "28808": 14, "28705": [14, 21], "28748": [14, 21], "15359": 14, "28725": 14, "315": [14, 20], "837": 14, "396": 14, "16107": 14, "13892": 14, "28723": 14, "modal": [15, 16, 56, 144, 201], "current": [15, 16, 18, 22, 35, 42, 54, 67, 75, 91, 95, 107, 120, 129, 149, 150, 151, 154, 156, 158, 165, 168, 171, 182, 183, 185, 189, 190, 199, 222, 229, 230, 232, 237, 240, 242, 248, 250, 252, 255, 269, 270, 275, 276, 277, 279, 280], "intend": [15, 259, 273], "drop": [15, 144, 200, 277, 280], "replac": [15, 16, 42, 51, 59, 60, 62, 64, 67, 68, 69, 144, 195, 200, 257, 278], "llama3_2_vis": [15, 16, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152], "llama3visiontransform": [15, 16, 148], "__init__": [15, 24, 25, 65, 66, 277, 278, 281], "transform_imag": 15, "clipimagetransform": [15, 65, 66, 144, 191], "xattn_mask": 15, "visioncrossattentionmask": [15, 144, 217], "224": [15, 16, 144], "tile_s": [15, 79, 80, 81, 144, 147, 151, 191, 218], "patch_siz": [15, 79, 80, 81, 144, 147, 151, 191, 218], "14": [15, 48, 144, 191, 280, 281], "skip_special_token": [15, 16, 67, 144], "begin_of_text": [15, 16, 21, 273], "start_header_id": [15, 16, 273], "end_header_id": [15, 16, 273], "n": [15, 16, 18, 19, 21, 34, 38, 40, 43, 183, 191, 216, 264, 271, 272, 273, 280], "eot_id": [15, 16, 21, 273], "na": [15, 273], "encoder_input": [15, 16, 50, 188, 189, 199], "shape": [15, 16, 22, 47, 50, 73, 74, 75, 76, 78, 79, 80, 81, 142, 143, 144, 147, 151, 180, 181, 182, 183, 184, 185, 186, 188, 189, 190, 191, 196, 197, 198, 199, 200, 201, 205, 218, 219, 220, 221, 222, 223, 224, 225, 241, 256, 257, 277], "num_til": [15, 16, 142, 143, 191], "num_channel": [15, 16, 191], "tile_height": [15, 16], "tile_width": [15, 16], "torch": [15, 16, 22, 24, 45, 47, 48, 49, 50, 73, 74, 75, 76, 77, 78, 79, 80, 81, 142, 143, 144, 180, 181, 182, 183, 184, 185, 186, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 205, 219, 220, 221, 222, 223, 224, 225, 230, 232, 234, 235, 236, 238, 239, 241, 243, 244, 250, 252, 253, 254, 255, 256, 257, 258, 259, 260, 262, 266, 269, 274, 275, 276, 277, 278, 279, 281], "just": [15, 19, 22, 267, 269, 272, 273, 275, 276, 278, 279, 280], "the_cauldron_dataset": [15, 16], "ai2d": [15, 66], "respir": 15, "combust": 15, "give": [15, 21, 24, 227, 277, 278, 279], "choic": [15, 18], "oxygen": 15, "b": [15, 25, 45, 47, 142, 143, 182, 183, 185, 189, 190, 199, 205, 219, 220, 224, 241, 251, 278, 281], "carbon": 15, "dioxid": 15, "c": [15, 45, 47, 50, 65, 142, 273], "nitrogen": 15, "d": [15, 24, 36, 65, 142, 143, 182, 183, 189, 199, 272, 273, 277, 278, 280], "heat": 15, "letter": 15, "mymultimodaltransform": 15, "my_tokenizer_build": 15, "myimagetransform": 15, "add_eo": [15, 57, 71, 144, 213, 214, 273], "tupl": [15, 19, 21, 24, 27, 38, 48, 73, 74, 80, 94, 106, 119, 144, 148, 162, 169, 179, 182, 191, 195, 212, 216, 219, 220, 221, 222, 223, 224, 225, 226, 242, 256, 257, 258], "infer": [15, 19, 22, 50, 56, 91, 100, 158, 182, 183, 185, 189, 190, 199, 218, 260, 265, 269, 270, 273, 274, 275, 276, 280, 281], "vision": [15, 16, 56, 81, 142, 144, 145, 146, 147, 148, 149, 150, 151, 152, 200, 231], "aspect_ratio": [15, 50, 78, 79, 142, 191], "append": [15, 19, 38, 39, 94, 106, 119, 144, 148, 162, 169, 179, 189, 199, 213, 247, 266], "addit": [15, 21, 22, 24, 25, 27, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 144, 146, 147, 150, 151, 153, 193, 200, 201, 209, 221, 226, 228, 229, 230, 236, 237, 243, 246, 247, 248, 250, 251, 253, 267, 273, 275, 278, 279], "kei": [15, 21, 22, 24, 26, 33, 35, 37, 42, 47, 48, 55, 56, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 182, 183, 188, 189, 190, 199, 201, 208, 209, 210, 224, 228, 230, 232, 247, 256, 272, 274, 275, 278, 279, 281], "e": [16, 18, 19, 36, 46, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 78, 79, 80, 81, 142, 143, 144, 147, 151, 183, 191, 195, 199, 204, 208, 218, 227, 228, 232, 239, 256, 260, 266, 269, 274, 276, 278, 279, 280, 281], "g": [16, 18, 46, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 78, 79, 80, 81, 142, 143, 144, 147, 151, 183, 191, 199, 204, 218, 227, 228, 239, 256, 260, 269, 276, 278, 279, 280, 281], "base": [16, 18, 20, 22, 27, 36, 38, 85, 86, 87, 88, 89, 90, 91, 95, 96, 97, 98, 99, 101, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 125, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 144, 145, 148, 149, 150, 151, 152, 154, 155, 156, 157, 158, 160, 163, 164, 165, 166, 167, 170, 171, 172, 173, 174, 175, 185, 205, 206, 208, 209, 210, 220, 221, 223, 224, 228, 235, 237, 238, 246, 248, 257, 260, 265, 273, 274, 275, 276, 277, 278, 279, 281], "multimodal_chat_dataset": 16, "visual": [16, 201], "get": [16, 22, 23, 24, 25, 26, 50, 144, 232, 236, 239, 242, 247, 261, 266, 267, 268, 269, 273, 274, 275, 277, 278, 279, 280], "below": [16, 23, 26, 47, 226, 276, 277, 278, 281], "clock": 16, "10": [16, 45, 47, 48, 49, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 191, 200, 274, 276, 279, 280, 281], "llama3_2_vision_transform": 16, "questionanswertempl": [16, 19, 70], "image_s": [16, 145, 148, 149, 152, 191], "560": [16, 145, 148, 149, 152], "image_dir": [16, 35, 42, 65], "home": [16, 24, 35, 42, 46], "nquestion": 16, "nit": 16, "00am": 16, "sharegpt": [16, 42, 60, 273], "q1": [16, 33, 55, 60, 67], "a1": [16, 33, 55, 60], "sharegpt4v": 16, "lin": 16, "chen": 16, "renam": 16, "themselv": [16, 281], "pathlib": 16, "pil_imag": 16, "Then": [16, 20, 26, 206, 275, 277, 279], "relat": [16, 188, 189, 199, 278], "user_messag": [16, 34, 40, 43, 144, 273], "locat": [16, 21, 24, 35, 42, 272, 276, 278, 280, 281], "long": [16, 54, 214, 273, 278], "image_dog": 16, "image_cat": 16, "image_bird": 16, "dog": [16, 218], "bird": [16, 46], "pet": 16, "three": [16, 22, 25, 50, 144, 221, 223, 224, 270, 275], "referenc": 16, "huggingfac": [16, 57, 61, 63, 71, 72, 161, 168, 169, 176, 177, 178, 221, 223, 224, 228, 229, 235, 272, 274], "co": [16, 57, 61, 63, 71, 72, 161, 168, 169, 176, 177, 178, 228, 229, 274], "img": 16, "llava_instruct_dataset": 16, "concaten": [17, 21, 48, 53, 147, 151, 212, 216], "sequenc": [17, 45, 47, 48, 49, 50, 54, 57, 61, 65, 66, 71, 72, 75, 76, 91, 94, 95, 101, 106, 107, 116, 119, 120, 125, 129, 142, 143, 144, 146, 148, 150, 154, 156, 158, 160, 162, 165, 167, 169, 171, 175, 179, 182, 183, 185, 188, 189, 190, 191, 194, 199, 201, 214, 216, 218, 220, 224, 225, 241, 273], "upto": [17, 185], "maximum": [17, 24, 47, 50, 51, 54, 61, 72, 75, 78, 79, 81, 91, 94, 95, 101, 106, 107, 116, 119, 120, 125, 129, 144, 146, 147, 148, 150, 151, 154, 156, 158, 160, 162, 165, 167, 169, 171, 175, 182, 183, 185, 188, 189, 190, 194, 199, 201, 218, 227, 272], "length": [17, 45, 47, 49, 50, 51, 52, 53, 54, 61, 72, 75, 91, 94, 95, 101, 106, 107, 116, 119, 120, 125, 129, 142, 143, 144, 146, 148, 150, 154, 156, 158, 160, 162, 165, 167, 168, 169, 171, 175, 179, 182, 183, 185, 188, 189, 190, 194, 196, 198, 199, 201, 214, 218, 219, 220, 229, 241, 247, 279], "slow": [17, 279, 281], "down": [17, 191, 230, 278, 279, 281], "introduc": [17, 85, 86, 95, 96, 97, 107, 108, 109, 110, 120, 121, 122, 131, 132, 138, 139, 154, 155, 156, 157, 165, 166, 183, 184, 201, 205, 224, 269, 273, 277, 278, 279, 280, 281], "signific": [17, 279, 280], "speedup": [17, 274, 276], "depend": [17, 25, 26, 228, 256, 272, 274, 277, 278, 279, 281], "iter": [17, 256, 257, 258, 281], "through": [17, 18, 22, 23, 24, 25, 26, 55, 81, 147, 151, 180, 182, 191, 201, 206, 267, 268, 269, 270, 272, 273, 274, 275, 277, 279, 280, 281], "greedi": [17, 54], "upon": [17, 25, 53, 189, 193, 199, 276], "initi": [17, 22, 25, 29, 53, 54, 82, 83, 84, 92, 93, 102, 103, 104, 105, 117, 118, 126, 127, 128, 136, 137, 159, 161, 176, 177, 178, 221, 232, 243, 244, 257, 269, 275, 278, 281], "max": [17, 50, 54, 179, 189, 191, 199, 214, 227, 235, 272, 278], "llama3": [17, 20, 21, 24, 65, 66, 73, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 144, 146, 148, 149, 150, 152, 165, 196, 198, 231, 237, 265, 267, 268, 269, 272, 274, 279], "load": [17, 22, 25, 35, 42, 46, 53, 54, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 199, 209, 228, 229, 230, 232, 250, 257, 274, 276, 278], "isinst": [17, 226], "1b_full_single_devic": 17, "prevent": [17, 22, 54, 221, 272, 279], "irrelev": 17, "cross": [17, 50, 54, 146, 150, 188, 196, 198, 199, 201, 218, 277], "attend": [17, 54, 183, 188, 189, 190, 199, 218], "pytorch": [17, 24, 25, 74, 189, 195, 196, 226, 244, 250, 253, 255, 256, 265, 266, 267, 269, 274, 276, 278, 279, 280, 281], "flex": 17, "attent": [17, 50, 54, 74, 75, 76, 81, 85, 86, 87, 91, 95, 96, 97, 101, 107, 108, 109, 110, 111, 116, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 146, 147, 149, 150, 151, 154, 155, 156, 157, 158, 160, 165, 166, 167, 168, 171, 172, 173, 174, 175, 182, 183, 185, 188, 189, 190, 192, 199, 201, 209, 210, 218, 276, 278, 279, 281], "flash": 17, "non": [17, 197, 198, 210, 220, 277], "causal": [17, 54, 75, 183, 189, 190, 199], "hardwar": [17, 236, 267, 274, 278, 279], "cuda": [17, 24, 236, 239, 256, 260, 266, 274, 279, 281], "ture": 17, "sdpa": 17, "memori": [17, 21, 25, 53, 54, 57, 61, 71, 72, 187, 189, 195, 196, 198, 199, 209, 237, 239, 245, 246, 256, 265, 267, 268, 269, 274, 275, 276, 277, 280], "effici": [17, 209, 237, 265, 267, 268, 274, 275, 278, 280], "fallback": 17, "while": [17, 24, 25, 85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 172, 173, 174, 200, 267, 269, 274, 279, 280, 281], "retain": [17, 221, 279, 281], "reward": [18, 105, 111, 115, 157, 161, 164, 219, 220, 221, 223, 224, 231], "downstream": 18, "captur": 18, "ground": [18, 196, 197, 198, 279], "truth": [18, 24, 196, 197, 198, 274, 276], "usual": [18, 21, 22, 185, 189, 225, 228, 241, 251, 272, 274, 278, 279], "outcom": 18, "binari": 18, "comparison": [18, 25, 278, 281], "annot": 18, "accord": [18, 19, 65, 66, 76, 153, 273], "criterion": 18, "style": [18, 31, 54, 58, 59, 60, 69, 201, 281], "interact": [18, 25, 55, 67, 265, 270, 275], "free": [18, 224, 270, 278], "preference_dataset": 18, "my_preference_dataset": [18, 67], "chosen_convers": [18, 67], "hole": [18, 67], "my": [18, 19, 23, 67, 73, 272, 273, 274, 276], "trouser": [18, 67], "fix": [18, 20, 67, 280], "rejected_convers": [18, 67], "off": [18, 25, 38, 67, 268, 269, 274, 280], "chosen": [18, 33, 55, 63, 67, 70, 221, 223, 224, 256], "reject": [18, 33, 55, 63, 67, 70, 221, 223, 224], "rejected_input_id": [18, 48, 67], "nwhat": 18, "ntake": 18, "rejected_label": [18, 48], "128006": 18, "78191": 18, "128007": 18, "271": 18, "18293": 18, "1124": 18, "1022": 18, "13": [18, 20, 21, 48, 191, 216, 225, 281], "128009": [18, 273], "accomplish": [18, 20, 53, 60, 64, 67, 71], "shown": [18, 274, 280], "di": 18, "look": [18, 19, 22, 24, 25, 234, 250, 266, 273, 274, 275, 276, 277, 278, 280], "anthrop": [18, 63], "harmless": [18, 63], "granni": 18, "her": [18, 20], "mobil": [18, 274], "phone": [18, 274], "issu": [18, 270, 280], "grandmoth": 18, "manag": [18, 22, 53, 193, 194, 206, 247, 254, 273], "behavior": [18, 22, 246, 273], "thing": [18, 279, 281], "grandma": 18, "feel": [18, 270, 278], "box": [18, 267, 269, 281], "hh_rlhf_helpful_dataset": 18, "hendrydong": 18, "preference_700k": 18, "stack_exchange_paired_dataset": 18, "purpos": [19, 65, 66, 275, 276], "whenev": [19, 144, 196, 278], "llama2": [19, 22, 24, 25, 27, 61, 72, 82, 83, 84, 85, 86, 87, 88, 89, 90, 100, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 144, 180, 189, 190, 231, 265, 268, 272, 275, 276, 279, 280], "were": [19, 20, 21, 53, 142, 191, 206, 222, 275, 280], "gear": [19, 144], "summar": [19, 43, 68, 273, 279], "summarizetempl": [19, 68, 273], "commun": [19, 144, 274, 279], "chatmltempl": [19, 144, 179], "gec_templ": 19, "extend": [19, 21, 22, 25, 267], "customprompttempl": 19, "achiev": [19, 38, 252, 269, 274, 276, 277, 278, 280, 281], "prepend_tag": [19, 38], "append_tag": [19, 38], "thu": [19, 31, 38, 55, 56, 189, 279, 280], "empti": [19, 47, 50, 52, 77, 272], "standalon": [19, 182], "my_custom_templ": 19, "Is": 19, "overhyp": 19, "advanc": [19, 79, 80, 81, 147, 151, 191], "configur": [19, 21, 25, 55, 56, 59, 60, 61, 62, 64, 65, 66, 67, 68, 69, 70, 71, 72, 95, 107, 120, 129, 149, 154, 165, 171, 247, 267, 269, 270, 273, 275, 276, 277, 278, 279, 280, 281], "doesn": [19, 274], "neatli": 19, "fall": 19, "protocol": [19, 21, 204, 211, 212, 217], "arg": [19, 21, 24, 27, 39, 80, 181, 189, 195, 201, 204, 211, 212, 217, 249, 256, 269, 280], "whether": [19, 31, 33, 35, 36, 37, 42, 47, 50, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 85, 86, 87, 91, 95, 96, 97, 107, 108, 109, 110, 111, 120, 121, 122, 129, 130, 131, 132, 138, 139, 144, 145, 149, 150, 151, 154, 155, 156, 157, 165, 166, 171, 172, 173, 174, 175, 195, 197, 199, 205, 209, 210, 213, 214, 226, 236, 238, 239, 247, 257, 273, 277], "sai": [19, 272, 273, 275], "eureka": 19, "eurekatempl": 19, "formatted_dialogu": 19, "llama2chattempl": [19, 106, 153, 179, 273], "paradigm": [20, 25, 268, 279], "unstructur": [20, 57, 71, 72], "unlabel": 20, "text_complet": 20, "odyssei": 20, "clear": [20, 279], "river": 20, "oceanu": 20, "had": 20, "got": [20, 50], "sea": 20, "went": 20, "till": 20, "reach": 20, "aeaean": 20, "island": 20, "dawn": 20, "sunris": 20, "drew": 20, "ship": 20, "sand": 20, "shore": 20, "sleep": 20, "wait": [20, 256], "break": [20, 144, 214], "child": 20, "morn": 20, "rosi": 20, "finger": 20, "appear": [20, 279], "sent": [20, 251], "men": 20, "circ": 20, "hous": 20, "fetch": [20, 278], "bodi": 20, "elpenor": 20, "cut": 20, "firewood": 20, "wood": 20, "headland": 20, "jut": 20, "wept": 20, "over": [20, 21, 25, 36, 56, 197, 198, 221, 235, 267, 269, 272, 274, 277, 278, 279, 281], "him": 20, "lament": 20, "funer": 20, "rite": 20, "armour": 20, "been": [20, 73, 75, 182, 189, 199, 225, 232, 237, 273, 280], "burn": 20, "ash": 20, "rais": [20, 22, 27, 30, 33, 35, 37, 42, 44, 46, 47, 50, 52, 53, 59, 60, 62, 64, 65, 66, 68, 69, 71, 75, 81, 171, 182, 183, 187, 188, 189, 191, 192, 193, 194, 209, 210, 216, 228, 229, 230, 232, 236, 238, 239, 243, 247, 251, 255, 257, 258, 259], "cairn": 20, "stone": 20, "top": [20, 74, 77, 147, 151, 234, 279, 281], "oar": 20, "he": 20, "row": [20, 55, 56, 183, 189, 190, 199], "text_completion_dataset": [20, 280], "128000": [20, 273, 280], "6153": 20, "584": 20, "1051": 20, "2867": 20, "279": 20, "15140": 20, "22302": 20, "355": 20, "11": [20, 22, 45, 47, 48, 191, 274, 280, 281], "323": 20, "1047": 20, "2751": 20, "704": 20, "1139": 20, "1825": 20, "9581": 20, "4024": 20, "389": 20, "12222": 20, "8813": 20, "362": 20, "12791": 20, "5420": 20, "13218": 20, "1405": 20, "1070": 20, "374": 20, "39493": 20, "64919": 20, "439": 20, "304": 20, "1023": 20, "7634": 20, "1226": 20, "1243": 20, "24465": 20, "1057": 20, "8448": 20, "311": 20, "70163": 20, "1077": 20, "31284": 20, "6212": 20, "30315": 20, "1938": 20, "1288": 20, "1464": 20, "128001": [20, 280], "similarli": [20, 119, 148, 169, 179, 280], "wikimedia": 20, "wikipedia": [20, 46, 72], "cnn_dailymail_articles_dataset": 20, "index": [21, 48, 49, 50, 53, 54, 183, 185, 190, 197, 199, 220, 235, 260, 266, 273, 274], "embed": [21, 22, 78, 79, 80, 81, 91, 95, 101, 107, 116, 120, 125, 129, 142, 143, 144, 146, 147, 150, 151, 154, 156, 158, 160, 165, 167, 171, 175, 182, 183, 184, 185, 188, 189, 191, 199, 200, 201, 203, 237, 273, 276, 279, 280], "vector": [21, 223, 273], "understood": 21, "plai": [21, 274, 279], "necessari": [21, 22, 55, 56, 247, 248, 249, 250, 251, 273, 278], "phi3": [21, 22, 165, 166, 168, 169, 170, 231, 272], "phi3_mini_token": 21, "p_token": 21, "phi": [21, 168, 169, 231], "32010": 21, "29871": 21, "1792": [21, 216], "9508": [21, 216], "32007": 21, "32001": 21, "4299": 21, "2933": [21, 216], "nuser": 21, "nmodel": 21, "sentencepiec": [21, 213, 276], "tiktoken": [21, 144, 214, 276], "host": [21, 266, 272, 275, 279], "distribut": [21, 77, 232, 243, 244, 253, 255, 260, 267, 270, 272, 275, 276, 277, 279], "alongsid": [21, 237, 279], "alreadi": [21, 24, 33, 37, 42, 62, 63, 65, 66, 67, 68, 69, 182, 183, 193, 194, 199, 231, 243, 266, 272, 274, 277, 278], "_token": [21, 25], "mistraltoken": [21, 162, 273], "adher": [21, 37, 42], "arbitrarili": 21, "small": [21, 184, 274], "seq": [21, 189, 199], "len": [21, 22, 50, 53, 59, 62, 65, 66, 68, 189, 191, 199], "demonstr": [21, 279, 280], "7": [21, 22, 45, 47, 48, 49, 50, 182, 191, 218, 222], "6312": 21, "28709": 21, "assign": [21, 24, 55, 56], "uniqu": [21, 55, 56, 231], "abil": 21, "NOT": [21, 22, 91, 144, 158], "presenc": [21, 31], "certain": [21, 22, 24, 256, 273], "proper": [21, 266, 275], "end_of_text": 21, "special_token": [21, 144, 214, 273], "added_token": 21, "128257": 21, "128258": 21, "remain": [21, 37, 42, 235, 277, 278, 279], "special_tokens_path": [21, 119, 148, 169, 179], "basetoken": 21, "actual": [21, 23, 24, 26, 31, 33, 35, 55, 56, 59, 62, 63, 64, 66, 67, 68, 70, 144, 269, 273, 280], "string": [21, 22, 35, 36, 38, 44, 60, 61, 94, 106, 119, 144, 148, 162, 169, 179, 204, 211, 213, 214, 216, 227, 233, 236, 240, 247, 260, 272, 279], "kwarg": [21, 24, 27, 39, 179, 181, 188, 190, 195, 201, 204, 211, 212, 217, 243, 247, 248, 249, 250, 251, 253, 256], "dict": [21, 22, 24, 25, 26, 27, 31, 33, 35, 36, 37, 38, 42, 44, 47, 48, 49, 50, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 94, 106, 119, 144, 148, 162, 169, 179, 188, 190, 195, 199, 201, 202, 207, 208, 209, 210, 211, 212, 214, 215, 217, 228, 229, 230, 232, 234, 239, 243, 245, 247, 252, 257, 259], "given": [21, 25, 27, 44, 47, 52, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 76, 144, 193, 194, 205, 206, 211, 212, 220, 236, 240, 246, 252, 260, 262, 267, 278], "token_id": [21, 144, 211, 214], "its": [21, 54, 100, 153, 156, 183, 185, 189, 190, 199, 201, 232, 252, 255, 272, 273, 274, 276, 278, 279], "sentencepiecebasetoken": [21, 211], "bpe": 21, "sp_token": 21, "reason": [21, 25, 73, 274, 279, 280], "walk": [22, 25, 250, 267, 273, 274, 275, 280, 281], "design": [22, 25, 224], "cover": [22, 23, 24, 25, 26, 273, 274, 281], "scenario": [22, 53, 144], "compos": [22, 191], "plug": [22, 279], "evalu": [22, 25, 265, 267, 269, 270, 275, 277, 278, 281], "gener": [22, 25, 47, 54, 61, 71, 74, 75, 76, 77, 144, 193, 194, 206, 219, 238, 247, 254, 255, 256, 263, 265, 269, 273, 277, 278, 279, 280, 281], "easi": [22, 25, 267, 278, 279], "understand": [22, 24, 25, 201, 265, 267, 268, 273, 278, 279, 281], "concept": [22, 270, 274, 275, 279], "talk": 22, "close": [22, 25, 247, 248, 249, 250, 251, 278], "veri": [22, 53, 189, 199, 272, 274, 279], "dictat": 22, "state_dict": [22, 195, 200, 201, 209, 228, 229, 230, 231, 232, 257, 278, 281], "store": [22, 55, 56, 247, 248, 251, 278, 279, 281], "disk": [22, 57, 248], "identifi": [22, 247], "state": [22, 25, 143, 189, 191, 193, 195, 199, 202, 207, 208, 209, 210, 219, 221, 228, 229, 230, 232, 234, 257, 274, 276, 278, 281], "match": [22, 44, 210, 247, 257, 266, 272, 274, 276, 278], "up": [22, 23, 25, 26, 50, 54, 61, 72, 144, 189, 193, 194, 199, 214, 218, 234, 247, 256, 268, 269, 270, 272, 273, 275, 276, 278, 279, 281], "exactli": [22, 210, 227, 280], "definit": [22, 278], "either": [22, 47, 55, 56, 73, 183, 189, 190, 210, 228, 247, 253, 266, 272, 278, 279, 280, 281], "explicit": 22, "error": [22, 24, 34, 52, 228, 255, 272], "except": [22, 36, 153, 216, 277], "wors": [22, 279], "silent": 22, "succe": 22, "popular": [22, 199, 267, 274], "offici": [22, 100, 273, 275, 276], "websit": 22, "inspect": [22, 274, 278, 281], "mmap": [22, 274], "weights_onli": [22, 230], "map_loc": [22, 274], "cpu": [22, 25, 194, 195, 236, 256, 260, 266, 272, 274, 281], "tensor": [22, 45, 47, 48, 49, 50, 73, 74, 75, 76, 77, 78, 79, 80, 81, 142, 143, 180, 181, 182, 183, 184, 185, 186, 188, 189, 190, 191, 195, 196, 197, 198, 199, 200, 201, 205, 219, 220, 221, 222, 223, 224, 225, 228, 241, 247, 248, 249, 250, 251, 254, 257, 259, 277, 278, 279, 281], "item": 22, "f": [22, 26, 59, 62, 65, 66, 68, 227, 273, 274, 277, 278, 281], "tok_embed": [22, 189, 199, 200], "32000": [22, 27, 278], "4096": [22, 27, 61, 72, 183, 185, 278, 280], "292": 22, "tabl": [22, 200, 273, 274, 276, 277, 279, 281], "layer": [22, 25, 81, 85, 86, 87, 88, 89, 90, 91, 95, 96, 97, 98, 99, 101, 105, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 120, 121, 122, 123, 124, 125, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 143, 146, 147, 149, 150, 151, 152, 154, 155, 156, 157, 158, 160, 161, 163, 164, 165, 166, 167, 170, 171, 172, 173, 174, 175, 183, 186, 187, 188, 189, 190, 191, 192, 193, 194, 199, 201, 203, 205, 209, 210, 226, 233, 237, 267, 268, 276, 278, 279, 280, 281], "dim": [22, 50, 142, 143, 180, 183, 184, 185, 189, 196, 198, 199, 277], "within": [22, 24, 27, 54, 73, 77, 78, 95, 107, 120, 129, 149, 150, 151, 154, 156, 165, 171, 191, 193, 194, 250, 255, 256, 272, 278, 281], "big": 22, "bin": [22, 272, 274], "piec": 22, "pytorch_model": [22, 274], "00001": [22, 272, 277], "00002": [22, 272, 277], "embed_token": 22, "241": 22, "Not": 22, "fewer": [22, 183], "sinc": [22, 24, 27, 55, 56, 198, 228, 230, 273, 274, 276, 279, 280], "mismatch": 22, "caus": [22, 213], "re": [22, 24, 193, 201, 224, 230, 267, 268, 269, 273, 274, 275, 278, 279], "end": [22, 25, 36, 57, 71, 144, 214, 216, 265, 267, 273, 276, 278, 280], "number": [22, 25, 44, 50, 54, 61, 72, 73, 78, 79, 81, 91, 95, 101, 107, 116, 120, 125, 129, 142, 143, 144, 146, 147, 150, 151, 154, 156, 158, 160, 165, 167, 171, 175, 182, 183, 189, 191, 197, 198, 218, 228, 229, 230, 232, 233, 235, 242, 255, 256, 272, 275, 277, 278, 279], "save": [22, 25, 26, 189, 195, 196, 198, 199, 228, 229, 230, 232, 237, 246, 251, 265, 269, 272, 273, 274, 276, 278, 279, 280], "less": [22, 50, 73, 274, 275, 276, 279, 281], "prone": 22, "invari": 22, "accept": [22, 24, 226, 275, 279, 281], "explicitli": [22, 204, 267, 278], "produc": [22, 232, 269, 280, 281], "One": [22, 50, 280], "advantag": [22, 219, 222, 269, 278], "abl": [22, 25, 274, 275, 280], "post": [22, 191, 252, 256, 269, 274, 276, 280, 281], "quantiz": [22, 85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 205, 230, 240, 265, 266, 268, 270, 275, 281], "eval": [22, 265, 267, 280], "without": [22, 24, 26, 183, 187, 189, 193, 199, 209, 266, 267, 269, 273, 274, 278, 279, 280], "OR": 22, "surround": [22, 25, 267], "load_checkpoint": [22, 25, 228, 229, 230, 231], "save_checkpoint": [22, 25, 26, 228, 229, 230], "permut": 22, "behav": 22, "further": [22, 191, 224, 272, 277, 278, 279, 280, 281], "illustr": [22, 65, 66, 276], "whilst": [22, 268, 279], "read": [22, 228, 229, 230, 267, 279], "compat": [22, 228, 230, 279, 280], "framework": [22, 25, 267], "mention": [22, 274, 279, 281], "assum": [22, 35, 42, 45, 47, 65, 94, 106, 119, 148, 162, 169, 179, 182, 183, 185, 190, 199, 200, 202, 207, 214, 232, 234, 235, 236, 238, 273, 274, 278], "checkpoint_dir": [22, 24, 228, 229, 230, 274, 276, 277, 280], "easiest": [22, 274, 275], "everyth": [22, 25, 267, 270, 275], "flow": [22, 54, 280, 281], "safetensor": [22, 227, 228, 272, 277], "output_dir": [22, 24, 228, 229, 230, 256, 274, 276, 278, 280, 281], "snippet": 22, "explain": [22, 279], "fullmodelhfcheckpoint": [22, 274, 277], "sort": [22, 228, 230], "order": [22, 23, 25, 228, 230, 250, 251, 275, 279], "matter": [22, 228, 230, 272, 278], "checkpoint_fil": [22, 24, 26, 228, 229, 230, 274, 276, 277, 278, 280, 281], "restart": [22, 272], "previou": [22, 54, 228, 229, 230, 277], "section": [22, 25, 239, 265, 274, 276, 279, 281], "recipe_checkpoint": [22, 228, 229, 230, 280], "model_typ": [22, 228, 229, 230, 274, 276, 280], "resume_from_checkpoint": [22, 228, 229, 230], "discrep": [22, 228], "github": [22, 27, 74, 85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 172, 173, 174, 184, 185, 196, 197, 221, 222, 223, 224, 235, 266, 274, 276, 277], "repositori": [22, 55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 100, 268, 269, 274, 275], "fullmodelmetacheckpoint": [22, 276, 280], "test": [22, 24, 25, 73, 267, 269, 273, 279], "written": [22, 24, 25, 228, 229, 247, 248, 249, 250, 251, 267], "partit": [22, 228, 281], "key_1": [22, 230], "weight_1": 22, "key_2": 22, "weight_2": 22, "mid": 22, "chekpoint": 22, "middl": [22, 201, 274, 279], "subsequ": [22, 25, 182, 189, 191, 218], "recipe_st": [22, 228, 229, 230], "pt": [22, 26, 228, 229, 230, 274, 276, 277, 280], "epoch": [22, 25, 26, 228, 229, 230, 232, 235, 272, 273, 274, 275, 276, 280], "etc": [22, 25, 143, 228, 239, 275], "flood": 22, "overwritten": 22, "updat": [22, 24, 25, 38, 182, 183, 189, 199, 217, 221, 222, 228, 232, 256, 259, 266, 273, 274, 275, 276, 278, 279, 280, 281], "hf_model_0001_0": [22, 274, 277], "hf_model_0002_0": [22, 274], "adapt": [22, 85, 86, 95, 96, 97, 107, 108, 109, 110, 120, 121, 122, 131, 132, 138, 139, 143, 154, 155, 156, 157, 165, 166, 199, 201, 202, 204, 205, 206, 207, 208, 228, 229, 230, 246, 268, 273, 274, 278, 281], "merg": [22, 27, 28, 179, 228, 274, 276, 281], "tutori": [22, 253, 267, 268, 269, 273, 274, 275, 276, 277, 278, 279, 280, 281], "save_adapter_weights_onli": 22, "choos": [22, 60, 278], "resum": [22, 25, 228, 229, 230, 235, 281], "frozen": [22, 143, 149, 152, 200, 221, 278, 279, 281], "learnt": [22, 273, 274], "refer": [22, 24, 25, 184, 185, 187, 191, 196, 206, 220, 221, 222, 223, 224, 247, 267, 278, 279, 280], "adapter_checkpoint": [22, 228, 229, 230], "adapter_0": [22, 274], "knowledg": [22, 265], "forward": [22, 25, 78, 79, 80, 142, 143, 180, 181, 183, 184, 185, 186, 188, 189, 190, 191, 193, 194, 196, 197, 198, 199, 200, 201, 205, 221, 222, 223, 224, 239, 256, 276, 277, 278, 279, 281], "modeltyp": [22, 228, 229, 230], "llama2_13b": [22, 108], "right": [22, 47, 50, 76, 189, 228, 274, 276, 278], "pytorch_fil": 22, "00003": [22, 227, 277], "torchtune_sd": 22, "load_state_dict": [22, 199, 200, 201, 209, 232, 257, 278], "successfulli": [22, 272, 275], "vocab": [22, 27, 179, 189, 199, 200, 276], "70": [22, 117], "randint": 22, "no_grad": 22, "6": [22, 45, 47, 48, 49, 50, 54, 91, 95, 184, 191, 241, 269, 280, 281], "3989": 22, "9": [22, 45, 47, 48, 50, 182, 191, 241, 274, 280, 281], "0531": 22, "2375": 22, "5": [22, 24, 45, 47, 48, 49, 50, 75, 191, 221, 224, 225, 235, 274, 275, 276, 277, 279], "2822": 22, "4872": 22, "7469": 22, "8": [22, 45, 47, 48, 50, 59, 62, 65, 66, 68, 85, 86, 87, 88, 89, 90, 96, 97, 98, 99, 108, 109, 110, 111, 112, 113, 114, 115, 121, 122, 123, 124, 125, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 155, 157, 163, 164, 166, 170, 172, 173, 174, 182, 191, 196, 198, 274, 277, 278, 279, 280, 281], "6737": 22, "0023": 22, "8235": 22, "6819": 22, "2424": 22, "0109": 22, "6915": 22, "3618": 22, "1628": 22, "8594": 22, "5857": 22, "1151": 22, "7808": 22, "2322": 22, "8850": 22, "9604": 22, "7624": 22, "6040": 22, "3159": 22, "5849": 22, "8039": 22, "9322": 22, "2010": [22, 191], "6824": 22, "8929": 22, "8465": 22, "3794": 22, "3500": 22, "6145": 22, "5931": 22, "find": [22, 23, 25, 26, 221, 272, 274, 275, 277, 278, 279], "hope": 22, "deeper": [22, 268, 269, 275, 279], "insight": [22, 274], "happi": [22, 274], "cometlogg": 23, "checkpoint": [23, 24, 25, 195, 199, 201, 214, 227, 228, 229, 230, 231, 232, 233, 251, 253, 257, 267, 269, 272, 276, 277, 278, 280, 281], "workspac": [23, 26, 247], "seen": [23, 26, 278, 281], "screenshot": [23, 26], "comet_ml": [23, 247], "featur": [23, 25, 26, 266, 267, 268, 269, 274, 275, 279], "pip": [23, 26, 247, 250, 251, 266, 274, 276, 279], "login": [23, 26, 247, 251, 272, 274], "metric_logg": [23, 24, 25, 26], "metric_log": [23, 24, 26, 247, 248, 249, 250, 251], "experiment_nam": [23, 247], "experi": [23, 24, 247, 251, 265, 267, 276, 277, 278], "grab": [23, 26, 276], "tab": [23, 26], "asset": 23, "artifact": [23, 26, 256], "click": [23, 26], "effect": [24, 224, 277, 279, 280], "prerequisit": [24, 273, 274, 275, 276, 277, 278, 280, 281], "Be": [24, 273, 274, 275, 276, 277, 278, 279, 280, 281], "familiar": [24, 273, 274, 275, 276, 277, 278, 280, 281], "fundament": [24, 280], "reproduc": [24, 247], "overridden": [24, 256], "quick": 24, "seed": [24, 25, 26, 255, 275, 280], "shuffl": [24, 54, 280], "dtype": [24, 25, 77, 182, 183, 188, 189, 190, 192, 193, 194, 195, 199, 201, 236, 254, 258, 274, 277, 279, 280, 281], "fp32": [24, 189, 196, 198, 279, 280, 281], "enable_fsdp": 24, "keyword": [24, 27, 55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 195, 273], "subfield": 24, "dotpath": [24, 94, 106, 119, 148, 162, 169, 179], "wish": [24, 182, 193, 257], "exact": [24, 27, 274], "normal": [24, 54, 144, 181, 183, 184, 188, 189, 190, 196, 197, 198, 213, 273, 278, 280, 281], "instanc": [24, 27, 53, 107, 120, 129, 149, 154, 156, 165, 171, 172, 173, 176, 177, 195, 202, 207, 208, 278], "preced": [24, 27, 272, 276, 278], "throw": 24, "notic": [24, 78, 79, 80, 191, 273, 278], "miss": [24, 209, 210, 256, 278], "llama2_token": [24, 273, 274], "llama2token": [24, 106], "512": [24, 81, 281], "overwrit": [24, 230, 257, 266, 272], "duplic": [24, 25, 267, 272], "sometim": 24, "resolv": [24, 28, 275], "alpaca": [24, 31, 53, 58, 59, 85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 172, 173, 174, 277], "disklogg": 24, "log_dir": [24, 248, 250, 251], "verifi": [24, 236, 237, 260, 273, 275, 278], "properli": [24, 209, 244, 272], "wa": [24, 35, 42, 50, 75, 79, 80, 81, 147, 151, 191, 209, 273, 278, 280, 281], "7b_lora_single_devic": [24, 274, 275, 278, 281], "my_config": [24, 272], "guidelin": 24, "tempt": 24, "put": [24, 25, 270, 275, 278, 280], "much": [24, 200, 224, 274, 276, 278, 280, 281], "switch": 24, "encourag": [24, 224, 278, 279], "clariti": 24, "significantli": [24, 221, 268, 269, 279], "easier": [24, 274, 275], "dont": 24, "privat": 24, "parent": [24, 272], "guarante": 24, "stabil": [24, 196, 198, 267, 269, 279, 280, 281], "underscor": 24, "_alpaca": 24, "k1": [24, 25], "v1": [24, 25, 72], "k2": [24, 25], "v2": [24, 25, 247], "my_model_checkpoint": 24, "file_1": 24, "file_2": 24, "my_tokenizer_path": 24, "nest": [24, 259], "dot": 24, "notat": [24, 50, 142, 143, 183, 185, 189, 199, 219, 220, 241], "flag": [24, 25, 36, 59, 60, 62, 64, 67, 68, 69, 226, 230, 237, 272, 279, 281], "bitsandbyt": [24, 279], "pagedadamw8bit": [24, 279], "delet": [24, 189, 192, 193, 194, 199], "foreach": 24, "8b_full": [24, 272], "adamw": [24, 278, 279], "2e": [24, 279], "fuse": [24, 146, 150, 199, 200, 201, 202, 252, 280], "nproc_per_nod": [24, 269, 276, 278, 280], "full_finetune_distribut": [24, 238, 272, 274, 275], "thought": [25, 267, 270, 275, 281], "target": [25, 75, 197, 198, 224, 267, 277], "pipelin": [25, 267, 269], "eg": [25, 189, 199, 228, 267], "meaning": [25, 267, 274], "fsdp": [25, 187, 226, 232, 237, 246, 275, 276, 279], "activ": [25, 81, 180, 233, 239, 245, 253, 256, 267, 269, 280, 281], "gradient": [25, 197, 198, 246, 252, 256, 267, 269, 274, 276, 278, 281], "accumul": [25, 252, 256, 267, 269], "mix": [25, 181, 272, 274, 279], "precis": [25, 181, 195, 236, 267, 269, 275, 281], "complex": 25, "becom": [25, 191, 266], "harder": 25, "anticip": 25, "methodolog": 25, "possibl": [25, 54, 227, 272, 279], "trade": [25, 279], "vs": [25, 275], "qualiti": [25, 274, 278, 280], "believ": 25, "suit": [25, 275, 279], "solut": 25, "result": [25, 65, 81, 147, 151, 191, 198, 216, 218, 256, 269, 274, 276, 277, 278, 279, 280, 281], "meant": [25, 195, 232], "expertis": 25, "routin": 25, "yourself": [25, 272, 276, 278], "exist": [25, 194, 201, 232, 247, 266, 272, 274, 275, 276, 281], "ones": [25, 50, 182], "modular": [25, 267], "wandb": [25, 26, 251, 275], "log": [25, 28, 221, 222, 223, 224, 239, 245, 247, 248, 249, 250, 251, 261, 274, 275, 276, 277, 278, 279, 281], "fulli": [25, 53, 149], "nativ": [25, 265, 267, 278, 280, 281], "numer": [25, 66, 267, 269, 280], "pariti": [25, 267], "verif": [25, 184], "benchmark": [25, 255, 267, 274, 276, 278, 280], "limit": [25, 232, 277, 279, 280], "hidden": [25, 81, 143, 147, 151, 180, 189, 191], "behind": 25, "unnecessari": 25, "abstract": [25, 211, 212, 267, 275, 281], "No": [25, 230, 267], "go": [25, 81, 100, 147, 151, 153, 191, 216, 267, 274, 275, 277, 279, 281], "figur": [25, 278, 281], "spectrum": 25, "decid": 25, "avail": [25, 35, 42, 72, 199, 201, 236, 244, 260, 267, 272, 274, 276, 278, 279], "consist": [25, 33, 37, 42, 65, 66, 72, 270, 275], "overrid": [25, 28, 29, 33, 37, 42, 62, 63, 65, 66, 67, 68, 69, 257, 270, 272, 274, 275, 276, 277, 281], "valid": [25, 52, 76, 197, 209, 210, 220, 238, 257, 258, 266, 270, 274, 275], "closer": [25, 277, 278], "monolith": [25, 267], "trainer": [25, 221, 223, 224], "wrapper": [25, 181, 213, 214, 232, 234, 272, 278], "around": [25, 144, 181, 213, 214, 239, 272, 273, 274, 278, 279, 280, 281], "extern": 25, "primarili": [25, 53, 278], "eleutherai": [25, 72, 267, 277, 278, 280], "har": [25, 267, 277, 278, 280], "stage": [25, 191], "distil": [25, 265], "dataload": [25, 54, 59, 62, 65, 66, 68], "applic": [25, 228, 229, 251], "clean": [25, 26, 58, 277], "group": [25, 183, 242, 243, 247, 248, 249, 250, 251, 272, 276, 280], "init_process_group": [25, 243], "backend": [25, 272, 280], "gloo": 25, "nccl": 25, "fullfinetunerecipedistribut": 25, "cleanup": 25, "stuff": 25, "carri": [25, 56], "metric": [25, 275, 277, 279, 280], "logger": [25, 245, 247, 248, 249, 250, 251, 261, 275], "_devic": 25, "get_devic": 25, "_dtype": 25, "get_dtyp": 25, "ckpt_dict": 25, "wrap": [25, 201, 226, 233, 237, 246, 253, 273, 279], "_model": [25, 232], "_setup_model": 25, "_setup_token": 25, "_optim": 25, "_setup_optim": 25, "_loss_fn": 25, "_setup_loss": 25, "_sampler": 25, "_dataload": 25, "_setup_data": 25, "backward": [25, 232, 234, 252, 256, 281], "zero_grad": 25, "curr_epoch": 25, "rang": [25, 200, 221, 222, 224, 255, 272, 276, 280], "epochs_run": [25, 26], "total_epoch": [25, 26], "idx": [25, 54], "enumer": 25, "_autocast": 25, "logit": [25, 73, 74, 77, 196, 197, 198, 241, 277], "global_step": 25, "_log_every_n_step": 25, "_metric_logg": 25, "log_dict": [25, 247, 248, 249, 250, 251], "step": [25, 54, 55, 56, 65, 66, 189, 199, 219, 232, 234, 235, 247, 248, 249, 250, 251, 252, 256, 265, 269, 274, 278, 280, 281], "recipe_main": [25, 29], "fullfinetunerecip": 25, "wandblogg": [26, 278, 281], "tip": 26, "straggler": 26, "background": 26, "crash": 26, "otherwis": [26, 45, 47, 50, 79, 80, 81, 147, 151, 187, 189, 191, 244, 247, 273, 280], "exit": [26, 193, 194, 206, 266, 272], "resourc": [26, 247, 248, 249, 250, 251, 279, 280], "kill": 26, "ps": 26, "aux": 26, "grep": 26, "awk": 26, "xarg": 26, "desir": [26, 55, 56, 254, 273, 279], "suggest": [26, 277], "approach": [26, 53, 277], "full_finetun": 26, "joinpath": 26, "_checkpoint": [26, 274], "_output_dir": [26, 228, 229, 230], "torchtune_model_": 26, "with_suffix": 26, "wandb_at": 26, "descript": [26, 272], "whatev": 26, "metadata": [26, 280], "seed_kei": 26, "epochs_kei": 26, "total_epochs_kei": 26, "max_steps_kei": 26, "max_steps_per_epoch": [26, 280], "add_fil": 26, "log_artifact": 26, "hydra": 27, "facebook": 27, "research": 27, "com": [27, 74, 85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 172, 173, 174, 184, 185, 196, 197, 221, 222, 223, 224, 235, 247, 266, 274, 276, 277], "facebookresearch": [27, 184], "blob": [27, 74, 85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 169, 172, 173, 174, 184, 185, 197, 221, 222, 223, 224, 235], "_intern": 27, "_instantiate2": 27, "l148": 27, "num_head": [27, 81, 91, 95, 101, 107, 116, 120, 125, 129, 146, 147, 150, 151, 154, 156, 158, 160, 165, 167, 171, 175, 182, 183, 185, 189], "num_kv_head": [27, 91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 182, 183], "vocab_s": [27, 73, 74, 91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 196, 197, 198, 200], "parsed_yaml": 27, "embed_dim": [27, 78, 79, 80, 81, 91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 183, 185, 188, 189, 190, 191, 200, 201, 257, 278], "valueerror": [27, 33, 35, 37, 42, 44, 46, 47, 50, 52, 53, 59, 60, 62, 64, 65, 66, 68, 69, 71, 171, 182, 183, 191, 192, 193, 194, 228, 229, 230, 236, 239, 255, 258], "recipe_nam": 28, "rank": [28, 85, 86, 87, 95, 96, 97, 107, 108, 109, 110, 111, 120, 121, 122, 129, 130, 131, 132, 138, 139, 149, 150, 151, 154, 155, 156, 157, 165, 166, 171, 172, 173, 174, 205, 242, 244, 255, 268, 275, 278, 281], "zero": [28, 50, 182, 184, 189, 199, 227, 274, 276, 280], "displai": 28, "callabl": [29, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 81, 189, 206, 226, 237, 240, 246, 253], "With": [29, 274, 277, 278, 280, 281], "my_recip": 29, "foo": 29, "bar": [29, 267, 275, 279], "configerror": 30, "cannot": [30, 46, 230, 276], "equival": [31, 35, 79, 223, 224], "condit": [31, 73, 244, 272], "dedic": 31, "due": [31, 213, 278, 279, 281], "keep": [31, 33, 35, 37, 42, 63, 64, 66, 67, 70, 200, 274, 278, 279], "openai": [32, 37, 60, 222], "markup": 32, "im_start": 32, "context": [32, 168, 193, 194, 206, 254, 256, 279], "im_end": 32, "goe": [32, 206], "a2": [33, 55], "functool": [34, 40, 43, 226], "partial": [34, 40, 43, 226], "_prompt_templ": [34, 40, 43], "assistant_messag": [34, 40, 43], "respect": [35, 53, 100, 182, 208, 256, 273], "final": [35, 42, 55, 56, 85, 86, 87, 91, 95, 101, 107, 108, 109, 110, 111, 116, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 146, 147, 149, 150, 151, 154, 155, 156, 157, 158, 165, 166, 171, 174, 175, 180, 189, 199, 209, 210, 274, 276, 277, 278, 279, 281], "leav": [35, 42, 279], "liter": [36, 38, 41, 85, 86, 87, 88, 89, 90, 94, 95, 96, 97, 98, 99, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 119, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 148, 149, 150, 151, 152, 154, 155, 156, 157, 162, 163, 164, 165, 166, 169, 170, 171, 172, 173, 174, 179, 209, 210], "union": [36, 46, 47, 58, 59, 60, 62, 64, 68, 69, 71, 72, 94, 106, 119, 148, 162, 169, 179, 189, 199, 210, 228, 233, 238, 247, 248, 249, 250, 251, 253, 255], "interleav": [36, 218], "attach": 36, "writer": 36, "calcul": [36, 38, 76, 142, 144, 183, 188, 190, 191, 219, 220, 222, 276], "consecut": [36, 52, 182, 218], "last": [36, 51, 54, 71, 189, 220, 232, 235], "properti": [36, 278, 279], "media": [36, 56], "classmethod": 36, "image_url": 37, "unmask": [37, 42, 197], "consid": [38, 53, 55, 56, 79, 80, 81, 147, 151, 191, 279], "come": [38, 52, 204, 278, 279], "nanswer": 40, "alia": [41, 226], "alwai": [42, 247, 257, 273, 279], "nsummari": [43, 273], "summari": [43, 53, 68, 191, 239], "batch_first": 45, "padding_valu": 45, "float": [45, 73, 74, 77, 85, 86, 87, 88, 89, 90, 91, 95, 96, 97, 98, 99, 101, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 120, 121, 122, 123, 124, 125, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 144, 149, 150, 151, 152, 154, 155, 156, 157, 158, 160, 163, 164, 165, 166, 167, 170, 171, 172, 173, 174, 175, 183, 184, 205, 219, 220, 221, 222, 223, 224, 232, 235, 238, 239, 245, 247, 248, 249, 250, 251, 278, 279, 280, 281], "rnn": [45, 47, 50], "pad_sequ": [45, 47, 50], "variabl": [45, 231, 244, 247, 279, 281], "left": [45, 47, 50, 144, 189, 278], "longest": [45, 49, 50], "trail": 45, "dimens": [45, 50, 91, 95, 101, 107, 116, 120, 125, 129, 142, 146, 147, 150, 151, 154, 156, 158, 160, 165, 167, 171, 175, 180, 182, 183, 185, 189, 191, 200, 205, 276, 278, 279, 281], "element": [45, 47, 50, 53, 197, 241, 274], "12": [45, 47, 48, 69, 191, 266, 280], "image_loc": 46, "www": [46, 247], "org": [46, 65, 82, 83, 84, 85, 86, 88, 89, 90, 95, 96, 97, 98, 99, 102, 103, 104, 105, 107, 108, 109, 110, 112, 113, 114, 115, 120, 121, 122, 123, 124, 131, 132, 133, 134, 135, 138, 139, 140, 141, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 183, 184, 185, 191, 218, 219, 221, 222, 223, 224, 226, 244, 250, 253, 255, 261, 266], "en": [46, 57, 61, 63, 71, 72, 280], "pad_direct": [47, 50], "keys_to_pad": 47, "padding_idx": [47, 48, 49, 50, 54], "left_pad_sequ": [47, 50], "integ": [47, 49, 200, 226, 227, 233, 255], "batch_siz": [47, 59, 62, 65, 66, 68, 182, 183, 188, 189, 190, 192, 193, 194, 196, 197, 198, 199, 200, 201, 221, 223, 225, 274, 279, 280], "ignore_idx": [48, 49, 50], "input_id": [48, 241], "chosen_input_id": [48, 67], "chosen_label": 48, "15": [48, 191, 237, 273, 274, 278, 281], "16": [48, 85, 86, 87, 88, 89, 90, 96, 97, 98, 99, 108, 109, 110, 111, 112, 113, 114, 115, 121, 122, 123, 124, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 155, 157, 163, 164, 166, 170, 172, 173, 174, 182, 191, 278, 281], "17": [48, 191, 278], "18": [48, 191, 276], "19": [48, 191, 281], "20": [48, 191, 225, 280], "token_pair": 49, "padded_col": 49, "pad_max_til": 50, "pad_max_imag": 50, "tile": [50, 78, 79, 80, 81, 142, 143, 144, 145, 147, 148, 149, 151, 191, 218], "aspect": [50, 78, 79, 267], "ratio": [50, 78, 79, 221, 222], "text_seq_len": [50, 218], "n_tile": [50, 78, 79, 191], "h": [50, 142, 182, 191, 196, 198, 266, 272], "w": [50, 82, 83, 84, 92, 93, 102, 103, 104, 105, 117, 118, 126, 127, 128, 136, 137, 142, 159, 161, 176, 177, 178, 191, 247, 250, 251, 273, 274, 278, 281], "h_ratio": 50, "w_ratio": 50, "encoder_mask": [50, 188, 189, 199], "image_seq_len": [50, 218], "channel": [50, 81, 142, 144, 147, 151, 191, 280], "height": [50, 142], "largest": 50, "bsz": [50, 73, 74, 75, 76, 78, 79, 191, 196, 198], "max_num_imag": 50, "max_num_til": [50, 78, 79, 81, 144, 147, 151, 191, 218], "tokens_per_til": 50, "image_id": 50, "four": [50, 278], "model_input": 50, "max_text_seq_len": 50, "40": [50, 79, 80, 81, 147, 151, 191, 218, 279, 281], "did": [50, 276, 281], "extra": [50, 144, 199, 266, 273, 278, 279, 280, 281], "second": [50, 183, 200, 274, 278, 279, 281], "eos_id": [51, 144, 214, 216], "shorter": [52, 189], "min": [52, 278], "invalid": 52, "sub": [53, 250], "unifi": [53, 161], "simplifi": [53, 221, 272, 277, 278], "simultan": 53, "intern": 53, "aggreg": 53, "transpar": 53, "howev": [53, 169, 266, 277, 279], "constitu": 53, "might": [53, 193, 200, 203, 272, 274, 279], "larg": [53, 196, 198, 205, 256, 272, 279, 281], "comput": [53, 55, 56, 101, 107, 116, 120, 125, 129, 142, 143, 146, 150, 171, 175, 183, 185, 189, 190, 196, 198, 199, 218, 221, 223, 224, 239, 255, 269, 274, 277, 279, 280, 281], "cumul": 53, "maintain": [53, 201, 268, 279, 281], "deleg": 53, "retriev": [53, 55, 56, 189, 237], "lead": [53, 213, 227, 269], "scale": [53, 73, 74, 77, 85, 86, 87, 95, 96, 97, 107, 108, 109, 110, 111, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 149, 150, 151, 154, 155, 156, 157, 165, 166, 171, 172, 173, 174, 184, 186, 188, 190, 205, 220, 224, 278, 279, 280, 281], "strategi": [53, 269], "stream": [53, 261, 279], "demand": 53, "deriv": [53, 180, 189, 190], "instans": 53, "dataset1": 53, "mycustomdataset": 53, "params1": 53, "dataset2": 53, "params2": 53, "concat_dataset": 53, "total": [53, 220, 222, 235, 242, 264, 271, 274, 276, 277, 278, 279], "data_point": 53, "1500": 53, "vicgal": 53, "gpt4": 53, "samsum": [53, 68], "focus": [53, 270, 275, 279], "enhanc": [53, 191, 224, 279, 281], "divers": 53, "machin": [53, 223, 260, 272, 274], "max_pack": 54, "outsid": [54, 255, 256, 278], "sampler": [54, 275], "part": [54, 200, 223, 273, 281], "buffer": [54, 189, 199, 279], "enough": [54, 273], "lower": [54, 269, 277, 278], "triangular": 54, "wise": 54, "made": [54, 60, 64, 67, 71, 144, 274], "smaller": [54, 200, 274, 276, 277, 278, 279, 280, 281], "jam": 54, "s1": [54, 213], "s2": [54, 213], "s3": 54, "s4": 54, "contamin": 54, "input_po": [54, 74, 183, 185, 189, 190, 199], "matrix": [54, 188, 189, 199], "increment": 54, "move": [54, 71, 189, 259, 279], "entir": [54, 71, 196, 203, 246, 273, 281], "avoid": [54, 71, 184, 191, 195, 255, 272, 280, 281], "truncat": [54, 61, 71, 72, 94, 106, 119, 144, 148, 162, 169, 179, 214, 225], "sentenc": [54, 71], "techniqu": [55, 267, 268, 269, 274, 275, 276, 277, 278, 279, 280], "repons": 55, "At": [55, 56, 189, 199], "extract": [55, 56, 61, 215], "against": [55, 56, 224, 262, 280, 281], "unit": [55, 56, 246, 267], "filepath": [55, 56, 57, 59, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72], "filter": [55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 280], "prior": [55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 257], "doc": [55, 56, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 226, 244, 247, 250, 251, 255, 261, 272, 274], "round": [56, 280], "incorpor": [56, 221], "happen": [56, 196, 198], "ti": [56, 95, 171, 175, 187, 279], "agnost": 56, "treat": [56, 191, 206, 273], "minimum": [56, 65, 66], "corpu": [57, 61, 71, 72], "package_refer": [57, 61, 63, 71, 72], "loading_method": [57, 61, 63, 71, 72], "tabular": [57, 71], "txt": [57, 71, 179, 248, 275], "eo": [57, 71, 169, 213, 216, 273], "yahma": 58, "variant": [58, 62, 68], "page": [58, 72, 266, 267, 272, 275, 276, 279], "tatsu": 59, "lab": [59, 74], "codebas": [59, 274], "independ": 59, "contribut": [59, 60, 62, 64, 67, 68, 69, 197, 198, 220, 222], "alpacatomessag": 59, "alpaca_d": 59, "altern": [60, 64, 67, 193, 275, 279], "toward": [60, 224], "my_dataset": [60, 64], "london": [60, 64], "ccdv": 61, "cnn_dailymail": 61, "textcompletiondataset": [61, 71, 72], "cnn": 61, "dailymail": 61, "articl": [61, 72], "highlight": [61, 281], "disabl": [61, 72, 189, 193, 199, 206, 255, 280], "highest": [61, 72], "conjunct": [62, 68, 70, 189, 279], "grammar_d": 62, "rlhflow": 63, "hh": 63, "preferencedataset": [63, 67, 70], "liuhaotian": 65, "llava": 65, "150k": 65, "coco": 65, "train2017": 65, "llava_instruct_150k": 65, "2017": 65, "visit": [65, 274], "cocodataset": 65, "wget": 65, "zip": [65, 263], "unzip": 65, "minim": [65, 66, 275, 277, 278, 279, 280, 281], "clip": [65, 66, 78, 79, 80, 81, 142, 143, 144, 147, 151, 191, 222], "mymodeltransform": [65, 66], "tokenizer_path": [65, 66], "image_transform": [65, 66], "yet": [65, 66, 153, 273, 274], "llava_instruct_d": 65, "huggingfacem4": 66, "the_cauldron": 66, "cauldron": 66, "card": 66, "cauldron_d": 66, "compris": 67, "share": [67, 183, 187, 274], "c1": 67, "r1": 67, "chosen_messag": 67, "rejected_messag": 67, "samsung": 68, "samsum_d": 68, "351": 69, "82": 69, "391": 69, "221": 69, "220": 69, "193": 69, "471": 69, "lvwerra": 70, "stack": [70, 191, 256], "exchang": 70, "allenai": [71, 280], "data_dir": 71, "realnewslik": 71, "wikitext_document_level": 72, "wikitext": [72, 280], "103": [72, 274], "transformerdecod": [73, 74, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 95, 96, 97, 98, 99, 101, 102, 103, 104, 105, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 146, 150, 154, 155, 156, 157, 158, 159, 160, 161, 163, 164, 165, 166, 167, 168, 170, 171, 172, 173, 174, 175, 176, 177, 178, 188, 190, 199, 200, 201, 278], "max_generated_token": 73, "pad_id": [73, 225], "temperatur": [73, 74, 77, 221, 223, 224, 274], "top_k": [73, 74, 77, 274], "stop_token": [73, 225], "rng": 73, "custom_generate_next_token": 73, "seq_length": [73, 74, 75, 188, 190, 200, 201], "prune": [73, 77, 281], "probabl": [73, 77, 85, 86, 87, 95, 96, 97, 107, 109, 110, 111, 120, 121, 122, 129, 130, 131, 132, 138, 139, 149, 150, 151, 154, 155, 156, 157, 165, 166, 171, 172, 173, 174, 205, 221, 222, 223, 224, 274, 277], "stop": [73, 225], "random": [73, 191, 255, 275], "compil": [73, 196, 274, 276, 279, 281], "generate_next_token": 73, "llama3_8b": [73, 122, 130, 199, 276, 279, 280], "manual_se": 73, "tolist": 73, "jeremi": 73, "m": [73, 195, 273, 280], "seq_len": [73, 75, 76, 189], "num_generated_token": 73, "q": [74, 77, 183, 278], "randomli": [74, 77, 257], "softmax": [74, 77, 183, 189, 190, 199, 277], "trick": [74, 77], "fast": [74, 274], "32971d3129541c5bfb4f715abc33d1c5f408d204": 74, "l40": 74, "k": [74, 77, 79, 183, 278], "padding_mask": [75, 76, 222, 225], "target_seq_len": 75, "suitabl": 75, "scaled_dot_product_attent": [75, 91, 95, 101, 107, 116, 120, 125, 129, 154, 156, 158, 160, 165, 167, 171, 175, 183], "static": 75, "kv": [75, 182, 183, 189, 190, 192, 193, 194, 199, 280], "cach": [75, 182, 183, 185, 188, 189, 190, 192, 193, 194, 199, 201, 266, 272], "longer": [75, 182, 279], "boolean": [75, 76, 81, 183, 188, 189, 190, 199, 201, 226, 241], "assertionerror": [75, 81, 182, 188, 189, 209, 210, 257], "shift": [76, 189], "uniform_": 77, "int32": 77, "patch": [78, 79, 80, 81, 143, 144, 147, 151, 191, 218], "check": [78, 79, 80, 81, 188, 189, 190, 191, 199, 201, 209, 236, 244, 262, 265, 267, 268, 269, 270, 273, 274, 275, 278, 279], "vision_transform": [78, 79, 80, 81], "visiontransform": [78, 79, 80, 81], "divid": [78, 79, 80, 81, 144, 147, 151, 191, 197, 198, 218], "dimension": [78, 79, 80, 81, 147, 151, 191], "n_img": [78, 79, 191], "n_tokens_per_til": [78, 79, 80], "crop": [78, 79, 80, 81, 142, 147, 151, 191], "local_token_positional_embed": 79, "_position_embed": [79, 191], "tokenpositionalembed": [79, 191], "gate": [79, 186, 231, 268, 269, 272, 275], "global_token_positional_embed": 79, "400": [79, 80, 81, 147, 151, 191, 218], "10x10": [79, 80, 81, 147, 151, 191, 218], "grid": [79, 80, 81, 147, 151, 191, 218], "th": [79, 182], "silu": [81, 180], "cls_output_dim": [81, 191], "attn_bia": 81, "out_indic": [81, 191], "output_cls_project": 81, "in_channel": [81, 147, 151, 191], "intermediate_act": 81, "transformerencoderlay": 81, "cl": [81, 143, 191], "mlp": [81, 85, 86, 87, 91, 95, 96, 97, 101, 107, 108, 109, 110, 111, 116, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 146, 149, 150, 151, 154, 155, 156, 157, 158, 160, 165, 166, 167, 171, 172, 173, 174, 175, 188, 189, 190, 209, 210, 276, 278, 279], "bia": [81, 187, 204, 205, 257, 278, 280, 281], "intermedi": [81, 91, 95, 101, 107, 116, 120, 125, 129, 146, 147, 150, 151, 154, 156, 158, 160, 165, 167, 171, 175, 191, 230, 253, 276, 281], "fourth": [81, 147, 151, 191], "determin": [81, 147, 151, 210], "divis": [81, 184], "code_llama2": [82, 83, 84, 85, 86, 87, 88, 89, 90, 272], "arxiv": [82, 83, 84, 85, 86, 88, 89, 90, 95, 96, 97, 98, 99, 102, 103, 104, 105, 107, 108, 109, 110, 112, 113, 114, 115, 120, 121, 122, 123, 124, 131, 132, 133, 134, 135, 138, 139, 140, 141, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 183, 184, 185, 191, 218, 219, 221, 222, 223, 224], "pdf": [82, 83, 84, 218, 219], "2308": [82, 83, 84], "12950": [82, 83, 84], "lora_attn_modul": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 209, 210, 268, 278, 279, 281], "q_proj": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 183, 209, 210, 268, 278, 279, 280, 281], "k_proj": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 183, 209, 210, 268, 278, 279, 280, 281], "v_proj": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 183, 209, 210, 268, 278, 279, 280, 281], "output_proj": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 183, 209, 210, 278, 279, 280, 281], "apply_lora_to_mlp": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 209, 210, 268, 278, 279], "apply_lora_to_output": [85, 86, 87, 88, 89, 90, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 174, 209, 210, 278, 279], "lora_rank": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 268, 278, 279], "lora_alpha": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 268, 278, 279], "lora_dropout": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 279], "use_dora": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 131, 132, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174], "quantize_bas": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 205, 281], "code_llama2_13b": 85, "tloen": [85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 172, 173, 174], "8bb8579e403dc78e37fe81ffbb253c413007323f": [85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 172, 173, 174], "l41": [85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 172, 173, 174], "l43": [85, 86, 87, 96, 97, 108, 109, 110, 111, 121, 122, 130, 131, 132, 138, 139, 166, 172, 173, 174], "linear": [85, 86, 87, 88, 89, 90, 95, 96, 97, 98, 99, 107, 108, 109, 110, 111, 112, 113, 114, 115, 120, 121, 122, 123, 124, 129, 130, 131, 132, 133, 134, 135, 138, 139, 140, 141, 143, 149, 150, 151, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 171, 172, 173, 174, 187, 189, 204, 205, 209, 210, 278, 279, 280, 281], "low": [85, 86, 87, 95, 96, 97, 107, 108, 109, 110, 111, 120, 121, 122, 129, 130, 131, 132, 138, 139, 149, 150, 151, 154, 155, 156, 157, 165, 166, 171, 172, 173, 174, 205, 268, 274, 277, 278, 281], "approxim": [85, 86, 87, 95, 96, 97, 107, 108, 109, 110, 111, 120, 121, 122, 129, 130, 131, 132, 138, 139, 149, 150, 151, 154, 155, 156, 157, 165, 166, 171, 172, 173, 174, 205, 278], "factor": [85, 86, 87, 95, 96, 97, 107, 108, 109, 110, 111, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 149, 150, 151, 154, 155, 156, 157, 165, 166, 171, 172, 173, 174, 205, 219, 274], "dropout": [85, 86, 87, 91, 95, 96, 97, 101, 107, 109, 110, 111, 116, 120, 121, 122, 125, 129, 130, 131, 132, 138, 139, 149, 150, 151, 154, 155, 156, 157, 158, 160, 165, 166, 167, 171, 172, 173, 174, 175, 183, 205, 278, 279, 281], "decompos": [85, 86, 95, 96, 97, 107, 108, 109, 110, 120, 121, 122, 131, 132, 138, 139, 154, 155, 156, 157, 165, 166], "magnitud": [85, 86, 95, 96, 97, 107, 108, 109, 110, 120, 121, 122, 131, 132, 138, 139, 154, 155, 156, 157, 165, 166, 279], "dora": [85, 86, 95, 96, 97, 107, 108, 109, 110, 120, 121, 122, 129, 131, 132, 138, 139, 150, 151, 154, 155, 156, 157, 165, 166], "ab": [85, 86, 88, 89, 90, 95, 96, 97, 98, 99, 102, 103, 104, 105, 107, 108, 109, 110, 112, 113, 114, 115, 120, 121, 122, 123, 124, 131, 132, 133, 134, 135, 138, 139, 140, 141, 152, 154, 155, 156, 157, 163, 164, 165, 166, 170, 183, 184, 185, 191, 221, 222, 223, 224], "2402": [85, 86, 95, 96, 97, 107, 108, 109, 110, 120, 121, 122, 131, 132, 138, 139, 154, 155, 156, 157, 165, 166], "09353": [85, 86, 95, 96, 97, 107, 108, 109, 110, 120, 121, 122, 131, 132, 138, 139, 154, 155, 156, 157, 165, 166], "code_llama2_70b": 86, "code_llama2_7b": 87, "qlora": [88, 89, 90, 98, 99, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 152, 163, 164, 170, 195, 265, 267, 268, 276, 278], "paper": [88, 89, 90, 98, 99, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 152, 163, 164, 170, 218, 221, 223, 224, 277, 278, 281], "2305": [88, 89, 90, 98, 99, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 152, 163, 164, 170, 183, 221, 223], "14314": [88, 89, 90, 98, 99, 112, 113, 114, 115, 123, 124, 133, 134, 135, 140, 141, 152, 163, 164, 170], "lora_code_llama2_13b": 88, "lora_code_llama2_70b": 89, "lora_code_llama2_7b": 90, "head_dim": [91, 95, 182, 183, 189], "intermediate_dim": [91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175], "attn_dropout": [91, 95, 101, 107, 116, 120, 125, 129, 154, 156, 158, 160, 165, 167, 171, 175, 183, 189], "norm_ep": [91, 95, 101, 107, 116, 120, 125, 129, 154, 156, 158, 160, 165, 167, 171, 175], "1e": [91, 95, 101, 107, 116, 120, 125, 129, 154, 156, 158, 160, 165, 167, 171, 175, 184, 277, 279], "06": [91, 95, 184, 278], "rope_bas": [91, 95, 101, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175], "10000": [91, 95, 101, 154, 156, 158, 160, 165, 167, 185], "norm_embed": [91, 95], "transformerselfattentionlay": [91, 101, 116, 125, 158, 175, 188, 189, 199, 201], "rm": [91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175], "norm": [91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 189], "space": [91, 101, 116, 125, 146, 150, 158, 175, 189, 203, 279], "slide": [91, 158, 168], "window": [91, 158, 168], "vocabulari": [91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 196, 198, 278, 279], "mha": [91, 95, 101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 183, 189], "onto": [91, 95, 101, 107, 116, 120, 125, 129, 154, 156, 158, 160, 165, 167, 171, 175, 183, 203], "epsilon": [91, 95, 101, 107, 116, 120, 125, 129, 154, 156, 158, 160, 165, 167, 171, 175, 222], "rotari": [91, 95, 101, 125, 129, 154, 156, 158, 160, 165, 167, 185, 276], "10_000": [91, 95, 154, 156, 158, 160, 167], "blog": [92, 93], "technolog": [92, 93], "develop": [92, 93, 266, 281], "gemmatoken": 94, "_templatetyp": [94, 106, 119, 148, 162, 169, 179], "gemma_2b": 96, "gemma_7b": 97, "lora_gemma_2b": 98, "lora_gemma_7b": 99, "taken": [100, 278, 281], "sy": [100, 273], "honest": [100, 273], "pari": [100, 153], "capit": [100, 153], "franc": [100, 153], "known": [100, 153, 240, 280], "stun": [100, 153], "05": [101, 107, 116, 120, 125, 129, 154, 156, 158, 160, 165, 167, 171, 175], "gqa": [101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 183], "mqa": [101, 107, 116, 120, 125, 129, 146, 150, 154, 156, 158, 160, 165, 167, 171, 175, 183], "kvcach": [101, 107, 116, 120, 125, 129, 146, 150, 165, 171, 175, 183, 189, 192, 193, 194, 199], "scale_hidden_dim_for_mlp": [101, 107, 116, 120, 125, 129, 146, 150, 171, 175], "2307": [102, 103, 104, 105], "09288": [102, 103, 104, 105], "classif": [105, 156, 160, 161, 231], "llama2_70b": 109, "llama2_7b": [110, 278], "classifi": [111, 156, 160, 161, 257, 279], "llama2_reward_7b": [111, 231], "lora_llama2_13b": 112, "lora_llama2_70b": 113, "lora_llama2_7b": [114, 278], "lora_llama2_reward_7b": 115, "500000": [116, 120, 125, 129, 146, 150], "llama3token": [119, 144, 212], "regist": [119, 144, 148, 169, 179, 195, 252, 281], "canon": [119, 144, 148, 169, 179], "llama3_70b": 121, "lora_llama3_70b": 123, "lora_llama3_8b": 124, "scale_factor": [125, 129], "500_000": [125, 129], "rope": [125, 129, 171, 175, 183, 185], "llama3_1": [126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 268, 277], "llama3_1_70b": 131, "llama3_1_8b": 132, "lora_llama3_1_405b": 133, "lora_llama3_1_70b": 134, "lora_llama3_1_8b": 135, "llama3_2_1b": [138, 192, 193, 194], "llama3_2_3b": 139, "lora_llama3_2_1b": 140, "lora_llama3_2_3b": 141, "projection_head": [142, 199, 203], "combin": [142, 144, 147, 151, 189, 199, 201, 203, 220, 277], "learnabl": [142, 186, 199, 201, 274], "fusion": [142, 145, 146, 147, 149, 150, 151, 199, 200, 201, 202, 203], "encoder_dim": [142, 143], "decoder_dim": [142, 143], "num_img": [142, 143], "num_emb": [142, 143], "broken": [142, 143, 191, 201], "width": [142, 280], "clip_embeds_per_til": 142, "emb": [142, 143, 183, 188, 189, 199], "num_hidden_input": 143, "sequenti": [143, 199, 203], "num_hidden": 143, "hidden_st": [143, 191], "image_mean": 144, "image_std": 144, "tranform": 144, "possible_resolut": 144, "448": [144, 145, 148, 149], "deviat": 144, "still": [144, 196, 198, 200, 201, 268, 278, 280, 281], "transformed_data": 144, "img1": [144, 218], "img2": [144, 218], "31587": [144, 213, 214], "29644": [144, 213, 214], "102": [144, 213, 214], "truncate_at_eo": [144, 214], "skip": [144, 183], "tokenize_head": 144, "tokenize_end": 144, "header": 144, "eom": 144, "wether": 144, "decoder_train": [145, 149, 152, 199], "encoder_train": [145, 149, 152, 199], "fusion_train": [145, 149, 152, 199], "deepfusionmodel": [145, 149, 152], "trainabl": [145, 149, 201, 205, 208, 246, 278, 279, 281], "resiz": [145, 148, 149], "fusion_interv": [146, 150], "num_special_token": [146, 150], "encoder_max_seq_len": [146, 150, 188, 189, 190, 194, 199, 201], "causalselfattent": [146, 150], "interv": [146, 150, 275], "clip_embed_dim": [147, 151], "clip_num_lay": [147, 151], "clip_hidden_st": [147, 151], "num_layers_project": [147, 151], "decoder_embed_dim": [147, 151], "llama3visionencod": [147, 151], "spatial": [147, 151], "backbon": [147, 151], "trainbl": 149, "decoder_lora": 150, "fusion_lora": [150, 151], "encoder_lora": 151, "lora_llama3_2_vision_11b": 152, "num_class": [156, 160, 257], "announc": 159, "ray2333": 161, "feedback": [161, 221], "lora_mistral_7b": 163, "lora_mistral_reward_7b": 164, "phi3_mini": [166, 231], "128k": 168, "nor": 168, "phi3minitoken": 169, "tokenizer_config": 169, "spm": 169, "lm": [169, 222, 277], "bo": [169, 213, 216, 273], "unk": 169, "augment": [169, 281], "endoftext": 169, "phi3minisentencepiecebasetoken": 169, "lora_phi3_mini": 170, "1000000": [171, 175], "tie_word_embed": [171, 172, 173, 175, 176, 177], "qwen2transformerdecod": 171, "period": [171, 175], "word": [171, 175, 279, 280], "qwen2_0_5b": [172, 187], "qwen2_1_5b": [173, 187], "qwen2_7b": 174, "qwen": [176, 177, 178], "merges_fil": 179, "qwen2token": 179, "gate_proj": 180, "down_proj": 180, "up_proj": 180, "feed": [180, 188, 190], "network": [180, 206, 278, 281], "fed": [180, 273], "multipli": [180, 279], "in_dim": [180, 204, 205, 278, 279, 281], "out_dim": [180, 189, 204, 205, 278, 279, 281], "layernorm": 181, "past": 182, "expand": 182, "dpython": [182, 183, 188, 189, 190, 194, 195, 199, 201, 254, 258], "reset": [182, 183, 188, 189, 190, 199, 201, 239], "k_val": 182, "v_val": 182, "fill": 182, "bfloat16": [182, 195, 254, 274, 275, 276, 278, 279, 280], "greater": [182, 191, 262], "pos_embed": [183, 188, 278, 280], "q_norm": 183, "k_norm": 183, "kv_cach": [183, 192, 193, 194], "is_caus": 183, "13245v1": 183, "multihead": 183, "extrem": 183, "credit": 183, "litgpt": 183, "v": [183, 189, 199, 278], "n_kv_head": 183, "rotarypositionalembed": [183, 278, 280], "rmsnorm": 183, "vice": [183, 272], "versa": [183, 272], "y": 183, "s_x": 183, "s_y": 183, "_masktyp": [183, 189, 190], "score": [183, 189, 190, 220], "encoder_max_cache_seq_len": [183, 189, 190], "j": [183, 188, 189, 190, 199], "blockmask": [183, 189, 190], "create_block_mask": [183, 189, 190], "flex_attent": [183, 189, 190], "n_h": [183, 185], "num": [183, 185], "n_kv": 183, "h_d": [183, 185], "reset_cach": [183, 188, 189, 190, 199, 201], "setup_cach": [183, 188, 189, 190, 192, 193, 199, 201], "ep": 184, "squar": 184, "1910": 184, "07467": 184, "propos": [185, 279], "2104": 185, "09864": 185, "verfic": 185, "l80": 185, "init": [185, 239, 251, 281], "exceed": 185, "freq": 185, "recomput": [185, 279], "geometr": 185, "progress": [185, 270, 275, 279], "rotat": 185, "angl": 185, "basic": [186, 276], "tied_modul": 187, "lost": 187, "whose": [187, 206, 247, 252], "attributeerror": [187, 259], "attn": [188, 190, 192, 193, 194, 278, 280, 281], "multiheadattent": [188, 190, 278, 280], "ca_norm": 188, "mlp_norm": [188, 190], "ca_scal": 188, "mlp_scale": [188, 190], "ff": [188, 190], "caches_are_en": [188, 189, 190, 192, 193, 194, 199, 201], "func": [188, 190, 201], "caches_are_setup": [188, 189, 190, 192, 193, 194, 199, 201], "token_sequ": 188, "embed_sequ": 188, "decoder_max_seq_len": [188, 189, 190, 192, 193, 194, 199, 201], "modulelist": 189, "output_hidden_st": [189, 199], "belong": [189, 234], "reduc": [189, 221, 267, 268, 269, 277, 278, 279, 280, 281], "statement": 189, "improv": [189, 214, 223, 237, 269, 276, 277, 278, 279], "readabl": [189, 274], "behaviour": [189, 199, 257], "alter": [189, 199], "common_util": [189, 192, 193, 194, 195], "disable_kv_cach": [189, 199], "chunked_output": 189, "last_hidden_st": 189, "chunk": [189, 196, 198, 214], "cewithchunkedoutputloss": [189, 199], "upcast": [189, 196, 198], "set_num_output_chunk": [189, 199], "num_chunk": [189, 196, 198], "s_e": [189, 199], "d_e": [189, 199], "arang": [189, 199], "prompt_length": [189, 199], "correspondingli": 189, "padded_prompt_length": 189, "m_": [189, 199], "realloc": [189, 199], "runtimeerror": [189, 216, 232, 236, 238, 243], "num_output_chunk": [189, 196, 198, 199], "transformercrossattentionlay": [189, 199, 201], "fusionlay": [189, 199], "sa_norm": 190, "sa_scal": 190, "token_pos_embed": 191, "pre_tile_pos_emb": 191, "post_tile_pos_emb": 191, "cls_project": 191, "vit": 191, "11929": 191, "convolut": 191, "flatten": 191, "downscal": 191, "800x400": 191, "400x400": 191, "_transform": 191, "whole": [191, 277], "n_token": 191, "101": 191, "pool": 191, "tiledtokenpositionalembed": 191, "tilepositionalembed": 191, "tile_pos_emb": 191, "8x8": 191, "21": 191, "22": 191, "23": [191, 235], "24": [191, 275, 276], "25": [191, 274, 277], "26": 191, "27": [191, 274], "28": [191, 274], "29": [191, 281], "30": [191, 225, 280], "31": [191, 276], "33": 191, "34": 191, "35": [191, 281], "36": 191, "37": 191, "38": [191, 274], "39": 191, "41": 191, "43": 191, "44": 191, "45": 191, "46": 191, "47": 191, "48": [191, 274, 281], "49": 191, "50": [191, 225, 247, 274], "51": 191, "52": [191, 275], "53": 191, "54": 191, "55": [191, 275], "56": 191, "57": [191, 278, 281], "58": 191, "59": [191, 281], "60": 191, "61": [191, 274], "62": 191, "63": 191, "64": [191, 268, 278], "num_patches_per_til": 191, "emb_dim": 191, "constain": 191, "anim": 191, "max_n_img": 191, "n_channel": 191, "vision_util": 191, "tile_crop": 191, "800": 191, "patch_grid_s": 191, "rand": 191, "nch": 191, "tile_cropped_imag": 191, "batch_imag": 191, "unsqueez": 191, "batch_aspect_ratio": 191, "clip_vision_encod": 191, "cache_en": 192, "float32": [192, 193, 194, 236, 277], "1024": [192, 193, 194, 280], "temporarili": [193, 194, 206, 279], "enter": [193, 194], "overhead": [193, 221, 269, 279, 280], "untouch": [193, 273], "yield": [193, 194, 206], "caller": [193, 194, 206], "delete_kv_cach": 194, "offload_to_cpu": 195, "hook": [195, 252, 279, 281], "nf4": [195, 279, 281], "restor": 195, "higher": [195, 276, 277, 279, 280, 281], "offload": [195, 281], "increas": [195, 221, 235, 276, 277, 278, 279, 280], "peak": [195, 239, 245, 274, 276, 278, 281], "gpu": [195, 269, 272, 274, 275, 276, 277, 278, 279, 280, 281], "_register_state_dict_hook": 195, "mymodul": 195, "_after_": 195, "nf4tensor": [195, 281], "unquant": [195, 280, 281], "unus": 195, "ignore_index": [196, 197, 198, 277], "entropi": [196, 198, 277], "bf16": [196, 198, 236, 279, 281], "ce": [196, 277], "better": [196, 198, 224, 267, 273, 274, 277, 279, 280], "accuraci": [196, 198, 269, 274, 276, 277, 278, 279, 280, 281], "doubl": [196, 198, 281], "therefor": [196, 198, 279, 281], "num_token": [196, 197, 198], "consider": [196, 198], "compute_cross_entropi": 196, "gain": [196, 269, 276], "won": [196, 273], "realiz": 196, "pull": [196, 268, 269, 272], "1390": 196, "loss_fn": [196, 198], "chunkedcrossentropyloss": 196, "output_chunk": [196, 198], "kullback": [197, 277], "leibler": [197, 277], "diverg": [197, 198, 220, 277], "jongwooko": [197, 277], "distillm": [197, 277], "17c0f98bc263b1861a02d5df578c84aea652ee65": 197, "student_logit": [197, 198, 277], "teacher_logit": [197, 198, 277], "student": [197, 198], "teacher": [197, 198, 274], "kl": [197, 198, 220, 277], "teacher_chunk": 198, "teacher_model": 198, "model_fus": [199, 200, 201, 202, 203], "deepfus": 199, "evolut": 199, "signatur": 199, "interchang": 199, "fusion_param": [199, 200, 201, 202, 203], "fusionembed": 199, "fusion_lay": [199, 201], "clip_vit_224": [199, 203], "feedforward": [199, 203], "register_fusion_modul": 199, "flamingo": [199, 201, 218], "strict": [199, 200, 201, 209, 278], "freez": [199, 274, 278], "fusion_vocab_s": 200, "necessit": 200, "rout": 200, "128": [200, 268, 276, 278, 279], "fusion_first": 201, "shot": [201, 274, 276, 280], "infus": 201, "interpret": 201, "enocd": 201, "isn": [201, 236, 272], "fused_lay": 201, "mark": [203, 273], "earli": 203, "peft": [204, 205, 206, 207, 208, 209, 210, 228, 268, 278, 281], "adapter_param": [204, 205, 206, 207, 208], "proj": 204, "loralinear": [204, 278, 281], "alpha": [205, 278, 279, 281], "use_bia": 205, "perturb": 205, "decomposit": [205, 278, 279], "matric": [205, 278, 281], "mapsto": 205, "w_0x": 205, "r": [205, 278], "bax": 205, "lora_a": [205, 278, 281], "lora_b": [205, 278, 281], "polici": [206, 220, 221, 222, 223, 224, 226, 237, 246, 253, 270], "neural": [206, 278, 281], "get_adapter_param": [208, 278], "base_miss": 209, "base_unexpect": 209, "lora_miss": 209, "lora_unexpect": 209, "validate_state_dict_for_lora": [209, 278], "unlik": 209, "reli": [209, 216, 274, 276], "unexpect": 209, "nonempti": 209, "full_model_state_dict_kei": 210, "lora_state_dict_kei": 210, "base_model_state_dict_kei": 210, "confirm": [210, 266], "lora_modul": 210, "complement": 210, "disjoint": 210, "overlap": [210, 279], "tiktokenbasetoken": 211, "light": 213, "sentencepieceprocessor": 213, "trim": 213, "whitespac": 213, "spm_model": [213, 273], "tokenized_text": [213, 214], "add_bo": [213, 214, 273], "trim_leading_whitespac": 213, "prefix": [213, 279], "unbatch": 213, "bos_id": [214, 216], "lightweight": [214, 273], "substr": 214, "repetit": 214, "speed": [214, 256, 276, 279, 280, 281], "identif": 214, "regex": 214, "absent": 214, "tt_model": 214, "tokenizer_json_path": 215, "heavili": 216, "concat": 216, "1788": 216, "2643": 216, "465": 216, "22137": 216, "join": 216, "satisfi": [216, 274], "loos": 217, "image_token_id": 218, "particip": [218, 219], "laid": 218, "fig": 218, "2204": 218, "14198": 218, "immedi": [218, 279], "until": [218, 279], "img3": 218, "equal": [218, 262], "gamma": [219, 223, 224], "lmbda": 219, "estim": [219, 220], "1506": 219, "02438": 219, "response_len": [219, 220], "receiv": 219, "discount": 219, "gae": 219, "logprob": [220, 224], "ref_logprob": 220, "kl_coeff": 220, "valid_score_idx": 220, "coeffici": [220, 222], "total_reward": 220, "kl_reward": 220, "beta": [221, 224], "label_smooth": [221, 224], "18290": 221, "intuit": [221, 223, 224], "dispref": 221, "dynam": [221, 280], "degener": 221, "occur": [221, 269], "naiv": 221, "trl": [221, 223, 224], "5d1deb1445828cfd0e947cb3a7925b1c03a283fc": 221, "dpo_train": [221, 223], "l844": 221, "2009": 221, "01325": 221, "regular": [221, 224, 279, 280, 281], "baselin": [221, 222, 274, 277, 278], "uncertainti": [221, 224], "policy_chosen_logp": [221, 223, 224], "policy_rejected_logp": [221, 223, 224], "reference_chosen_logp": [221, 223], "reference_rejected_logp": [221, 223], "chosen_reward": [221, 223, 224], "rejected_reward": [221, 223, 224], "value_clip_rang": 222, "value_coeff": 222, "proxim": [222, 270], "1707": 222, "06347": 222, "eqn": 222, "vwxyzjn": 222, "ccc19538e817e98a60d3253242ac15e2a562cb49": 222, "lm_human_preference_detail": 222, "train_policy_acceler": 222, "l719": 222, "ea25b9e8b234e6ee1bca43083f8f3cf974143998": 222, "ppo2": 222, "l68": 222, "l75": 222, "pi_old_logprob": 222, "pi_logprob": 222, "phi_old_valu": 222, "phi_valu": 222, "value_padding_mask": 222, "old": 222, "participag": 222, "five": 222, "policy_loss": 222, "value_loss": 222, "clipfrac": 222, "fraction": 222, "statist": [223, 279], "rso": 223, "hing": 223, "2309": 223, "06657": 223, "logist": 223, "regress": 223, "slic": 223, "10425": 223, "almost": [223, 278], "svm": 223, "counter": 223, "4dce042a3863db1d375358e8c8092b874b02934b": 223, "l1141": 223, "simpo": 224, "2405": 224, "14734": 224, "averag": [224, 277], "implicit": 224, "margin": 224, "bradlei": 224, "terri": 224, "larger": [224, 230, 274, 276, 277, 279], "win": 224, "lose": 224, "98ad01ddfd1e1b67ec018014b83cba40e0caea66": 224, "cpo_train": 224, "l603": 224, "pretti": [224, 274], "identitc": 224, "elimin": 224, "kind": 224, "ipoloss": 224, "fill_valu": 225, "sequence_length": 225, "stop_token_id": 225, "869": 225, "eos_mask": 225, "truncated_sequ": 225, "datatyp": [226, 279, 281], "denot": 226, "auto_wrap_polici": [226, 237, 253], "submodul": [226, 246], "obei": 226, "contract": 226, "get_fsdp_polici": 226, "modules_to_wrap": [226, 237, 246], "min_num_param": 226, "my_fsdp_polici": 226, "recurs": [226, 246, 250], "sum": [226, 277, 278], "p": [226, 232, 278, 280, 281], "numel": [226, 278], "1000": [226, 280], "stabl": [226, 244, 250, 255, 266, 279], "html": [226, 244, 250, 253, 255, 261, 265], "filename_format": 227, "max_filenam": 227, "concis": 227, "filenam": [227, 248], "file_": 227, "_of_": 227, "n_file": 227, "build_checkpoint_filenam": 227, "file_00001_of_00003": 227, "file_00002_of_00003": 227, "file_00003_of_00003": 227, "safe_seri": 228, "from_pretrain": 228, "0001_of_0003": 228, "0002_of_0003": 228, "todo": 228, "preserv": [228, 281], "weight_map": [228, 274], "convert_weight": 228, "_model_typ": [228, 231], "intermediate_checkpoint": [228, 229, 230], "adapter_onli": [228, 229, 230], "_weight_map": 228, "shard": [229, 276], "wip": 229, "qualnam": 231, "boundari": 231, "distinguish": 231, "llama3_vis": 231, "llama3_2_vision_decod": 231, "mistral_reward_7b": 231, "my_new_model": 231, "my_custom_state_dict_map": 231, "optim_map": 232, "bare": 232, "bone": 232, "optim_dict": [232, 234, 252], "cfg_optim": 232, "ckpt": 232, "optim_ckpt": 232, "placeholder_optim_dict": 232, "optiminbackwardwrapp": 232, "get_last_lr": 232, "rate": [232, 235, 238, 267, 275, 279], "schedul": [232, 235, 256, 275, 279], "get_optim_kei": 232, "arbitrari": [232, 278, 279], "optim_ckpt_map": 232, "set_lr_schedul": 232, "lr_schedul": [232, 235], "lrschedul": 232, "loadabl": 232, "step_lr_schedul": 232, "ac_mod": 233, "ac_opt": 233, "op": [233, 280], "ac": [233, 237], "optimizerinbackwardwrapp": [234, 238], "named_paramet": [234, 257], "num_warmup_step": 235, "num_training_step": 235, "num_cycl": [235, 256], "last_epoch": 235, "lambdalr": 235, "linearli": 235, "decreas": [235, 278, 279, 280, 281], "cosin": 235, "v4": 235, "src": 235, "l104": 235, "warmup": [235, 256], "phase": 235, "wave": 235, "half": [235, 279], "kernel": 236, "memory_efficient_fsdp_wrap": [237, 280], "maxim": [237, 246, 265, 267], "workload": [237, 269, 279, 280], "fullyshardeddataparallel": [237, 246, 279], "fsdppolicytyp": [237, 246], "warpper": 238, "optimizer_in_backward": 238, "reset_stat": 239, "track": [239, 247], "alloc": [239, 245, 246, 276, 279, 281], "reserv": [239, 245, 273, 281], "stat": [239, 245, 281], "int4": [240, 280], "4w": 240, "recogn": 240, "int8dynactint4weightquant": [240, 269, 280], "8da4w": [240, 280], "int4weightonlyquant": [240, 280], "int8dynactint4weightqatquant": [240, 269, 280], "qat": [240, 265, 270], "int4weightonlyqatquant": 240, "exclud": 241, "aka": 242, "master": 244, "port": [244, 272], "address": [244, 277, 279], "hold": [244, 275], "peak_memory_act": 245, "peak_memory_alloc": 245, "peak_memory_reserv": 245, "get_memory_stat": 245, "hierarch": 246, "api_kei": 247, "experiment_kei": 247, "onlin": 247, "log_cod": 247, "comet": 247, "site": [247, 274], "ml": 247, "team": 247, "compar": [247, 250, 262, 274, 276, 277, 278, 280, 281], "sdk": 247, "uncategor": 247, "alphanumer": 247, "charact": 247, "get_or_cr": 247, "fresh": 247, "persist": 247, "hpo": 247, "sweep": 247, "server": 247, "offlin": 247, "auto": [247, 272], "creation": 247, "experimentconfig": 247, "project_nam": 247, "my_workspac": 247, "my_metr": [247, 250, 251], "importerror": [247, 251], "termin": [247, 250, 251], "comet_api_kei": 247, "flush": [247, 248, 249, 250, 251], "ndarrai": [247, 248, 249, 250, 251], "scalar": [247, 248, 249, 250, 251], "record": [247, 248, 249, 250, 251, 256], "log_config": [247, 251], "payload": [247, 248, 249, 250, 251], "log_": 248, "unixtimestamp": 248, "thread": 248, "safe": 248, "organize_log": 250, "tensorboard": 250, "subdirectori": 250, "logdir": 250, "startup": 250, "tree": [250, 274, 276], "tfevent": 250, "encount": 250, "frontend": 250, "organ": [250, 272], "accordingli": [250, 280], "my_log_dir": 250, "view": [250, 277], "entiti": 251, "bias": [251, 278, 281], "usernam": 251, "my_ent": 251, "my_group": 251, "account": [251, 278, 281], "link": [251, 274, 276], "capecap": 251, "6053ofw0": 251, "torchtune_config_j67sb73v": 251, "soon": [252, 279], "readi": [252, 265, 273, 280], "grad": 252, "acwrappolicytyp": 253, "author": [253, 267, 275, 279, 281], "fsdp_adavnced_tutori": 253, "insid": 254, "contextmanag": 254, "debug_mod": 255, "pseudo": 255, "commonli": [255, 278, 279, 281], "numpi": 255, "determinist": 255, "global": [255, 279], "warn": 255, "nondeterminist": 255, "cudnn": 255, "set_deterministic_debug_mod": 255, "profile_memori": 256, "with_stack": 256, "record_shap": 256, "with_flop": 256, "wait_step": 256, "warmup_step": 256, "active_step": 256, "profil": 256, "layout": 256, "trace": 256, "profileract": 256, "gradient_accumul": 256, "sensibl": 256, "default_schedul": 256, "reduct": [256, 269, 278], "scope": 256, "flop": 256, "cycl": 256, "repeat": [256, 279], "model_named_paramet": 257, "force_overrid": 257, "concret": [257, 279], "vocab_dim": 257, "named_param": 258, "inplac": [259, 278], "too": [259, 269, 276], "handler": 261, "_log": 261, "__version__": 262, "generated_examples_python": 263, "galleri": [263, 271], "sphinx": 263, "000": [264, 271, 276], "execut": [264, 271], "generated_exampl": 264, "mem": [264, 271], "mb": [264, 271], "gentl": 265, "introduct": 265, "first_finetune_tutori": 265, "kd": 265, "torchvis": 266, "torchao": [266, 269, 274, 276, 279, 280, 281], "latest": [266, 269, 275, 279, 281], "whl": 266, "cu121": 266, "cu118": 266, "cu124": 266, "And": [266, 274], "welcom": [266, 272], "greatest": [266, 275], "contributor": 266, "dev": 266, "commit": 266, "branch": 266, "therebi": [266, 279, 280, 281], "forc": [266, 277], "reinstal": 266, "opt": [266, 275], "suffix": 266, "On": [267, 278], "pointer": 267, "emphas": 267, "simplic": 267, "component": 267, "prove": 267, "democrat": 267, "zoo": 267, "varieti": [267, 278], "integr": [267, 274, 275, 276, 278, 280, 281], "fsdp2": 267, "excit": 267, "checkout": 267, "quickstart": 267, "attain": 267, "embodi": 267, "philosophi": 267, "usabl": 267, "composit": 267, "hard": 267, "outlin": 267, "unecessari": 267, "never": 267, "thoroughli": 267, "competit": 268, "grant": [268, 269, 275], "interest": [268, 269, 274, 277], "8b_lora_single_devic": [268, 272, 273, 276, 277, 279], "lever": [268, 269], "action": [268, 269], "degrad": [269, 279, 280, 281], "simul": [269, 279, 280], "compromis": 269, "blogpost": [269, 279], "qat_distribut": [269, 280], "8b_qat_ful": [269, 280], "least": [269, 276, 278, 280], "vram": [269, 276, 278, 279, 280], "80gb": [269, 280], "a100": 269, "h100": 269, "delai": 269, "fake": [269, 280], "empir": [269, 280], "potenti": [269, 278, 279], "fake_quant_after_n_step": [269, 280], "idea": [269, 277, 281], "roughli": 269, "total_step": 269, "futur": [269, 280], "plan": [269, 274], "un": 269, "groupsiz": [269, 280], "256": [269, 276, 280], "hackabl": [270, 275], "singularli": [270, 275], "technic": [270, 275], "awar": [270, 279, 280], "tracker": 270, "short": 272, "subcommand": 272, "anytim": 272, "symlink": 272, "wrote": 272, "readm": [272, 274, 276], "md": 272, "lot": [272, 274, 279], "recent": 272, "releas": [272, 276], "agre": 272, "term": [272, 279], "perman": 272, "eat": 272, "bandwith": 272, "storag": [272, 281], "00030": 272, "ootb": 272, "7b_full_low_memori": [272, 274, 275], "8b_full_single_devic": 272, "mini_full_low_memori": 272, "7b_full": [272, 274, 275], "13b_full": [272, 274, 275], "70b_full": 272, "edit": 272, "clobber": 272, "destin": 272, "lora_finetune_distribut": [272, 276, 278], "torchrun": 272, "launch": [272, 273, 275], "nproc": 272, "node": 272, "worker": 272, "nnode": [272, 278, 280], "minimum_nod": 272, "maximum_nod": 272, "fail": 272, "rdzv": 272, "rendezv": 272, "endpoint": 272, "8b_lora": [272, 276], "bypass": 272, "fancy_lora": 272, "8b_fancy_lora": 272, "nice": 273, "meet": 273, "overhaul": 273, "multiturn": 273, "untrain": 273, "accompani": 273, "who": 273, "influenti": 273, "hip": 273, "hop": 273, "artist": 273, "2pac": 273, "rakim": 273, "flavor": 273, "formatted_messag": 273, "nyou": 273, "nwho": 273, "why": [273, 275, 278], "518": 273, "25580": 273, "29962": 273, "3532": 273, "14816": 273, "29903": 273, "6778": 273, "_spm_model": 273, "piece_to_id": 273, "manual": [273, 281], "529": 273, "29879": 273, "29958": 273, "nhere": 273, "pure": 273, "mess": 273, "prime": 273, "strictli": 273, "ask": [273, 279], "though": 273, "robust": 273, "pretend": 273, "zuckerberg": 273, "seem": [273, 274], "good": [273, 278, 279], "altogeth": 273, "honor": 273, "custom_8b_lora_single_devic": 273, "favorit": [274, 278], "seemlessli": 274, "connect": [274, 280], "amount": 274, "natur": 274, "export": 274, "leverag": [274, 276, 281], "percentag": 274, "16gb": [274, 278], "rtx": 274, "3090": 274, "4090": 274, "hour": 274, "7b_qlora_single_devic": [274, 275, 281], "473": 274, "98": [274, 281], "gb": [274, 276, 278, 280, 281], "484": 274, "01": [274, 275], "fact": [274, 276, 278, 279], "third": 274, "But": [274, 278], "realli": 274, "eleuther_ev": [274, 276, 280], "eleuther_evalu": [274, 276, 280], "lm_eval": [274, 276], "custom_eval_config": [274, 276], "truthfulqa_mc2": [274, 276, 277, 278], "measur": [274, 276], "propens": [274, 276], "324": 274, "loglikelihood": 274, "195": 274, "121": 274, "197": 274, "acc": [274, 280], "388": 274, "489": 274, "great": [274, 279], "custom_generation_config": [274, 276], "kick": 274, "300": 274, "bai": 274, "area": 274, "92": 274, "exploratorium": 274, "san": 274, "francisco": 274, "magazin": 274, "awesom": 274, "bridg": 274, "cool": 274, "96": [274, 281], "sec": [274, 276], "83": 274, "99": [274, 278], "72": 274, "littl": 274, "int8_weight_onli": [274, 276], "int8_dynamic_activation_int8_weight": [274, 276], "ao": [274, 276], "quant_api": [274, 276], "quantize_": [274, 276], "int4_weight_onli": [274, 276], "previous": [274, 276, 278], "benefit": 274, "clone": [274, 278, 280, 281], "assumpt": 274, "new_dir": 274, "output_dict": 274, "sd_1": 274, "sd_2": 274, "dump": 274, "convert_hf_checkpoint": 274, "checkpoint_path": 274, "justin": 274, "school": 274, "math": 274, "ws": 274, "94": [274, 276], "bandwidth": [274, 276], "1391": 274, "84": 274, "thats": 274, "seamlessli": 274, "authent": [274, 275], "hopefulli": 274, "gave": 274, "minut": 275, "agreement": 275, "depth": 275, "principl": 275, "boilerpl": 275, "substanti": [275, 278], "custom_config": 275, "replic": 275, "lorafinetunerecipesingledevic": 275, "lora_finetune_output": 275, "log_1713194212": 275, "3697006702423096": 275, "25880": [275, 281], "83it": 275, "monitor": 275, "tqdm": 275, "e2": 275, "focu": 276, "theta": 276, "observ": [276, 280], "consum": [276, 281], "overal": [276, 277], "8b_qlora_single_devic": [276, 279], "coupl": [276, 278, 281], "meta_model_0": [276, 280], "122": 276, "sarah": 276, "busi": 276, "mum": 276, "young": 276, "children": 276, "live": 276, "north": 276, "east": 276, "england": 276, "135": 276, "88": 276, "138": 276, "346": 276, "09": 276, "139": 276, "broader": 276, "teach": [277, 278], "straight": [277, 278], "jump": [277, 278], "compress": 277, "transfer": 277, "capac": 277, "computation": 277, "expens": 277, "deploi": 277, "imit": 277, "diagram": 277, "aim": [277, 279], "minillm": 277, "forwardklloss": 277, "super": 277, "teacher_prob": 277, "student_logprob": 277, "log_softmax": 277, "prod_prob": 277, "forwardklwithchunkedoutputloss": 277, "knowledge_distillation_single_devic": 277, "bit": [277, 278, 279, 280, 281], "alpaca_cleaned_dataset": 277, "hellaswag": [277, 280], "commonsense_qa": 277, "kd_ratio": 277, "teacher_checkpoint": 277, "00004": 277, "truthfulqa": [277, 278], "commonsens": 277, "constant": 277, "boost": 277, "graph": [277, 279], "irrespect": 277, "3e": 277, "slightli": 277, "truthful_qa": 277, "wherea": 277, "unfamiliar": 278, "oppos": [278, 281], "momentum": [278, 279], "aghajanyan": 278, "et": 278, "al": 278, "hypothes": 278, "intrins": 278, "eight": 278, "practic": 278, "blue": 278, "although": [278, 280], "rememb": 278, "approx": 278, "15m": 278, "65k": 278, "requires_grad": [278, 281], "frozen_out": [278, 281], "lora_out": [278, 281], "base_model": 278, "lora_model": 278, "lora_llama_2_7b": [278, 281], "alon": 278, "in_featur": [278, 280], "out_featur": [278, 280], "validate_missing_and_unexpected_for_lora": 278, "peft_util": 278, "set_trainable_param": 278, "lora_param": 278, "total_param": 278, "trainable_param": 278, "2f": 278, "6742609920": 278, "4194304": 278, "7b_lora": 278, "my_model_checkpoint_path": [278, 280, 281], "tokenizer_checkpoint": [278, 280, 281], "my_tokenizer_checkpoint_path": [278, 280, 281], "constraint": 278, "factori": 278, "benefici": 278, "impact": [278, 279], "minor": 278, "lora_experiment_1": 278, "smooth": [278, 281], "curv": [278, 281], "500": 278, "ran": 278, "footprint": [278, 280], "commod": 278, "cogniz": 278, "ax": 278, "parallel": 278, "475": 278, "87": 278, "508": 278, "86": 278, "504": 278, "04": 278, "514": 278, "lowest": 278, "absolut": 278, "4gb": 278, "tradeoff": 278, "salman": 279, "mohammadi": 279, "brief": 279, "glossari": 279, "struggl": 279, "constrain": [279, 280], "cost": 279, "particularli": 279, "gradient_accumulation_step": 279, "throughput": 279, "ram": 279, "bottleneck": 279, "sebastian": 279, "raschka": 279, "fp16": 279, "sound": 279, "quot": 279, "aliv": 279, "region": 279, "enable_activation_checkpoint": 279, "bring": 279, "autograd": [279, 281], "saved_tensors_hook": 279, "cours": 279, "runtim": 279, "hide": 279, "later": 279, "brought": 279, "enable_activation_offload": 279, "dev20240907": 279, "total_batch_s": 279, "count": 279, "suppos": 279, "log_every_n_step": 279, "translat": 279, "frequent": 279, "slowli": 279, "num_devic": 279, "adamw8bit": 279, "pagedadamw": 279, "modern": 279, "converg": 279, "stateless": 279, "stochast": 279, "descent": 279, "sacrif": 279, "remov": 279, "optimizer_in_bwd": 279, "cpuoffloadoptim": 279, "offload_gradi": 279, "prototyp": 279, "low_bit_optim": 279, "4e": 279, "adam": 279, "hint": 279, "slowdown": 279, "4x": 279, "fsdp_cpu_offload": 279, "greatli": 279, "lora_": 279, "lora_llama3": 279, "_lora": 279, "firstli": 279, "secondli": 279, "affect": 279, "fashion": 279, "slower": [279, 281], "jointli": 279, "sens": 279, "novel": 279, "normalfloat": [279, 281], "8x": [279, 281], "worth": 279, "cast": [279, 280], "incur": [279, 280, 281], "penalti": 279, "qlora_": 279, "qlora_llama3_8b": 279, "_qlora": 279, "perplex": 280, "goal": 280, "ptq": 280, "kept": 280, "nois": 280, "henc": 280, "x_q": 280, "int8": 280, "zp": 280, "x_float": 280, "qmin": 280, "qmax": 280, "clamp": 280, "x_fq": 280, "dequant": 280, "proce": 280, "prepared_model": 280, "swap": 280, "int8dynactint4weightqatlinear": 280, "int8dynactint4weightlinear": 280, "train_loop": 280, "converted_model": 280, "qat_distributed_recipe_label": 280, "recov": 280, "custom_8b_qat_ful": 280, "2000": 280, "led": 280, "presum": 280, "mutat": 280, "5gb": 280, "custom_quant": 280, "poorli": 280, "custom_eleuther_evalu": 280, "fullmodeltorchtunecheckpoint": 280, "max_seq_length": 280, "my_eleuther_evalu": 280, "stderr": 280, "word_perplex": 280, "9148": 280, "byte_perplex": 280, "5357": 280, "bits_per_byt": 280, "6189": 280, "5687": 280, "0049": 280, "acc_norm": 280, "7536": 280, "0043": 280, "portion": [280, 281], "74": 280, "048": 280, "190": 280, "7735": 280, "5598": 280, "6413": 280, "5481": 280, "0050": 280, "7390": 280, "0044": 280, "7251": 280, "4994": 280, "5844": 280, "5740": 280, "7610": 280, "outperform": 280, "importantli": 280, "characterist": 280, "187": 280, "958": 280, "halv": 280, "motiv": 280, "edg": 280, "smartphon": 280, "executorch": 280, "xnnpack": 280, "export_llama": 280, "use_sdpa_with_kv_cach": 280, "qmode": 280, "group_siz": 280, "get_bos_id": 280, "get_eos_id": 280, "output_nam": 280, "llama3_8da4w": 280, "pte": 280, "881": 280, "oneplu": 280, "709": 280, "tok": 280, "815": 280, "316": 280, "364": 280, "highli": 281, "vanilla": 281, "held": 281, "bespok": 281, "vast": 281, "major": 281, "normatfloat": 281, "deepdiv": 281, "distinct": 281, "de": 281, "counterpart": 281, "set_default_devic": 281, "qlora_linear": 281, "memory_alloc": 281, "177": 281, "152": 281, "del": 281, "empty_cach": 281, "lora_linear": 281, "081": 281, "344": 281, "qlora_llama2_7b": 281, "qlora_model": 281, "hood": 281, "essenti": 281, "reparametrize_as_dtype_state_dict_post_hook": 281, "149": 281, "9157477021217346": 281, "02": 281, "08": 281, "15it": 281, "nightli": 281, "200": 281, "hundr": 281, "228": 281, "8158286809921265": 281, "95it": 281, "exercis": 281, "linear_nf4": 281, "to_nf4": 281, "linear_weight": 281, "incom": 281}, "objects": {"torchtune.config": [[27, 0, 1, "", "instantiate"], [28, 0, 1, "", "log_config"], [29, 0, 1, "", "parse"], [30, 0, 1, "", "validate"]], "torchtune.data": [[31, 1, 1, "", "AlpacaToMessages"], [32, 1, 1, "", "ChatMLTemplate"], [33, 1, 1, "", "ChosenRejectedToMessages"], [34, 2, 1, "", "GrammarErrorCorrectionTemplate"], [35, 1, 1, "", "InputOutputToMessages"], [36, 1, 1, "", "Message"], [37, 1, 1, "", "OpenAIToMessages"], [38, 1, 1, "", "PromptTemplate"], [39, 1, 1, "", "PromptTemplateInterface"], [40, 2, 1, "", "QuestionAnswerTemplate"], [41, 2, 1, "", "Role"], [42, 1, 1, "", "ShareGPTToMessages"], [43, 2, 1, "", "SummarizeTemplate"], [44, 0, 1, "", "format_content_with_images"], [45, 0, 1, "", "left_pad_sequence"], [46, 0, 1, "", "load_image"], [47, 0, 1, "", "padded_collate"], [48, 0, 1, "", "padded_collate_dpo"], [49, 0, 1, "", "padded_collate_sft"], [50, 0, 1, "", "padded_collate_tiled_images_and_mask"], [51, 0, 1, "", "truncate"], [52, 0, 1, "", "validate_messages"]], "torchtune.data.Message": [[36, 3, 1, "", "contains_media"], [36, 4, 1, "", "from_dict"], [36, 4, 1, "", "get_media"], [36, 3, 1, "", "text_content"]], "torchtune.datasets": [[53, 1, 1, "", "ConcatDataset"], [54, 1, 1, "", "PackedDataset"], [55, 1, 1, "", "PreferenceDataset"], [56, 1, 1, "", "SFTDataset"], [57, 1, 1, "", "TextCompletionDataset"], [58, 0, 1, "", "alpaca_cleaned_dataset"], [59, 0, 1, "", "alpaca_dataset"], [60, 0, 1, "", "chat_dataset"], [61, 0, 1, "", "cnn_dailymail_articles_dataset"], [62, 0, 1, "", "grammar_dataset"], [63, 0, 1, "", "hh_rlhf_helpful_dataset"], [64, 0, 1, "", "instruct_dataset"], [67, 0, 1, "", "preference_dataset"], [68, 0, 1, "", "samsum_dataset"], [69, 0, 1, "", "slimorca_dataset"], [70, 0, 1, "", "stack_exchange_paired_dataset"], [71, 0, 1, "", "text_completion_dataset"], [72, 0, 1, "", "wikitext_dataset"]], "torchtune.datasets.multimodal": [[65, 0, 1, "", "llava_instruct_dataset"], [66, 0, 1, "", "the_cauldron_dataset"]], "torchtune.generation": [[73, 0, 1, "", "generate"], [74, 0, 1, "", "generate_next_token"], [75, 0, 1, "", "get_causal_mask_from_padding_mask"], [76, 0, 1, "", "get_position_ids_from_padding_mask"], [77, 0, 1, "", "sample"]], "torchtune.models.clip": [[78, 1, 1, "", "TilePositionalEmbedding"], [79, 1, 1, "", "TiledTokenPositionalEmbedding"], [80, 1, 1, "", "TokenPositionalEmbedding"], [81, 0, 1, "", "clip_vision_encoder"]], "torchtune.models.clip.TilePositionalEmbedding": [[78, 4, 1, "", "forward"]], "torchtune.models.clip.TiledTokenPositionalEmbedding": [[79, 4, 1, "", "forward"]], "torchtune.models.clip.TokenPositionalEmbedding": [[80, 4, 1, "", "forward"]], "torchtune.models.code_llama2": [[82, 0, 1, "", "code_llama2_13b"], [83, 0, 1, "", "code_llama2_70b"], [84, 0, 1, "", "code_llama2_7b"], [85, 0, 1, "", "lora_code_llama2_13b"], [86, 0, 1, "", "lora_code_llama2_70b"], [87, 0, 1, "", "lora_code_llama2_7b"], [88, 0, 1, "", "qlora_code_llama2_13b"], [89, 0, 1, "", "qlora_code_llama2_70b"], [90, 0, 1, "", "qlora_code_llama2_7b"]], "torchtune.models.gemma": [[91, 0, 1, "", "gemma"], [92, 0, 1, "", "gemma_2b"], [93, 0, 1, "", "gemma_7b"], [94, 0, 1, "", "gemma_tokenizer"], [95, 0, 1, "", "lora_gemma"], [96, 0, 1, "", "lora_gemma_2b"], [97, 0, 1, "", "lora_gemma_7b"], [98, 0, 1, "", "qlora_gemma_2b"], [99, 0, 1, "", "qlora_gemma_7b"]], "torchtune.models.llama2": [[100, 1, 1, "", "Llama2ChatTemplate"], [101, 0, 1, "", "llama2"], [102, 0, 1, "", "llama2_13b"], [103, 0, 1, "", "llama2_70b"], [104, 0, 1, "", "llama2_7b"], [105, 0, 1, "", "llama2_reward_7b"], [106, 0, 1, "", "llama2_tokenizer"], [107, 0, 1, "", "lora_llama2"], [108, 0, 1, "", "lora_llama2_13b"], [109, 0, 1, "", "lora_llama2_70b"], [110, 0, 1, "", "lora_llama2_7b"], [111, 0, 1, "", "lora_llama2_reward_7b"], [112, 0, 1, "", "qlora_llama2_13b"], [113, 0, 1, "", "qlora_llama2_70b"], [114, 0, 1, "", "qlora_llama2_7b"], [115, 0, 1, "", "qlora_llama2_reward_7b"]], "torchtune.models.llama3": [[116, 0, 1, "", "llama3"], [117, 0, 1, "", "llama3_70b"], [118, 0, 1, "", "llama3_8b"], [119, 0, 1, "", "llama3_tokenizer"], [120, 0, 1, "", "lora_llama3"], [121, 0, 1, "", "lora_llama3_70b"], [122, 0, 1, "", "lora_llama3_8b"], [123, 0, 1, "", "qlora_llama3_70b"], [124, 0, 1, "", "qlora_llama3_8b"]], "torchtune.models.llama3_1": [[125, 0, 1, "", "llama3_1"], [126, 0, 1, "", "llama3_1_405b"], [127, 0, 1, "", "llama3_1_70b"], [128, 0, 1, "", "llama3_1_8b"], [129, 0, 1, "", "lora_llama3_1"], [130, 0, 1, "", "lora_llama3_1_405b"], [131, 0, 1, "", "lora_llama3_1_70b"], [132, 0, 1, "", "lora_llama3_1_8b"], [133, 0, 1, "", "qlora_llama3_1_405b"], [134, 0, 1, "", "qlora_llama3_1_70b"], [135, 0, 1, "", "qlora_llama3_1_8b"]], "torchtune.models.llama3_2": [[136, 0, 1, "", "llama3_2_1b"], [137, 0, 1, "", "llama3_2_3b"], [138, 0, 1, "", "lora_llama3_2_1b"], [139, 0, 1, "", "lora_llama3_2_3b"], [140, 0, 1, "", "qlora_llama3_2_1b"], [141, 0, 1, "", "qlora_llama3_2_3b"]], "torchtune.models.llama3_2_vision": [[142, 1, 1, "", "Llama3VisionEncoder"], [143, 1, 1, "", "Llama3VisionProjectionHead"], [144, 1, 1, "", "Llama3VisionTransform"], [145, 0, 1, "", "llama3_2_vision_11b"], [146, 0, 1, "", "llama3_2_vision_decoder"], [147, 0, 1, "", "llama3_2_vision_encoder"], [148, 0, 1, "", "llama3_2_vision_transform"], [149, 0, 1, "", "lora_llama3_2_vision_11b"], [150, 0, 1, "", "lora_llama3_2_vision_decoder"], [151, 0, 1, "", "lora_llama3_2_vision_encoder"], [152, 0, 1, "", "qlora_llama3_2_vision_11b"]], "torchtune.models.llama3_2_vision.Llama3VisionEncoder": [[142, 4, 1, "", "forward"]], "torchtune.models.llama3_2_vision.Llama3VisionProjectionHead": [[143, 4, 1, "", "forward"]], "torchtune.models.llama3_2_vision.Llama3VisionTransform": [[144, 4, 1, "", "decode"], [144, 4, 1, "", "tokenize_message"], [144, 4, 1, "", "tokenize_messages"]], "torchtune.models.mistral": [[153, 1, 1, "", "MistralChatTemplate"], [154, 0, 1, "", "lora_mistral"], [155, 0, 1, "", "lora_mistral_7b"], [156, 0, 1, "", "lora_mistral_classifier"], [157, 0, 1, "", "lora_mistral_reward_7b"], [158, 0, 1, "", "mistral"], [159, 0, 1, "", "mistral_7b"], [160, 0, 1, "", "mistral_classifier"], [161, 0, 1, "", "mistral_reward_7b"], [162, 0, 1, "", "mistral_tokenizer"], [163, 0, 1, "", "qlora_mistral_7b"], [164, 0, 1, "", "qlora_mistral_reward_7b"]], "torchtune.models.phi3": [[165, 0, 1, "", "lora_phi3"], [166, 0, 1, "", "lora_phi3_mini"], [167, 0, 1, "", "phi3"], [168, 0, 1, "", "phi3_mini"], [169, 0, 1, "", "phi3_mini_tokenizer"], [170, 0, 1, "", "qlora_phi3_mini"]], "torchtune.models.qwen2": [[171, 0, 1, "", "lora_qwen2"], [172, 0, 1, "", "lora_qwen2_0_5b"], [173, 0, 1, "", "lora_qwen2_1_5b"], [174, 0, 1, "", "lora_qwen2_7b"], [175, 0, 1, "", "qwen2"], [176, 0, 1, "", "qwen2_0_5b"], [177, 0, 1, "", "qwen2_1_5b"], [178, 0, 1, "", "qwen2_7b"], [179, 0, 1, "", "qwen2_tokenizer"]], "torchtune.modules": [[180, 1, 1, "", "FeedForward"], [181, 1, 1, "", "Fp32LayerNorm"], [182, 1, 1, "", "KVCache"], [183, 1, 1, "", "MultiHeadAttention"], [184, 1, 1, "", "RMSNorm"], [185, 1, 1, "", "RotaryPositionalEmbeddings"], [186, 1, 1, "", "TanhGate"], [187, 1, 1, "", "TiedLinear"], [188, 1, 1, "", "TransformerCrossAttentionLayer"], [189, 1, 1, "", "TransformerDecoder"], [190, 1, 1, "", "TransformerSelfAttentionLayer"], [191, 1, 1, "", "VisionTransformer"]], "torchtune.modules.FeedForward": [[180, 4, 1, "", "forward"]], "torchtune.modules.Fp32LayerNorm": [[181, 4, 1, "", "forward"]], "torchtune.modules.KVCache": [[182, 4, 1, "", "reset"], [182, 4, 1, "", "update"]], "torchtune.modules.MultiHeadAttention": [[183, 4, 1, "", "forward"], [183, 4, 1, "", "reset_cache"], [183, 4, 1, "", "setup_cache"]], "torchtune.modules.RMSNorm": [[184, 4, 1, "", "forward"]], "torchtune.modules.RotaryPositionalEmbeddings": [[185, 4, 1, "", "forward"]], "torchtune.modules.TanhGate": [[186, 4, 1, "", "forward"]], "torchtune.modules.TransformerCrossAttentionLayer": [[188, 4, 1, "", "caches_are_enabled"], [188, 4, 1, "", "caches_are_setup"], [188, 4, 1, "", "forward"], [188, 4, 1, "", "reset_cache"], [188, 4, 1, "", "setup_caches"]], "torchtune.modules.TransformerDecoder": [[189, 4, 1, "", "caches_are_enabled"], [189, 4, 1, "", "caches_are_setup"], [189, 4, 1, "", "chunked_output"], [189, 4, 1, "", "forward"], [189, 4, 1, "", "reset_caches"], [189, 4, 1, "", "set_num_output_chunks"], [189, 4, 1, "", "setup_caches"]], "torchtune.modules.TransformerSelfAttentionLayer": [[190, 4, 1, "", "caches_are_enabled"], [190, 4, 1, "", "caches_are_setup"], [190, 4, 1, "", "forward"], [190, 4, 1, "", "reset_cache"], [190, 4, 1, "", "setup_caches"]], "torchtune.modules.VisionTransformer": [[191, 4, 1, "", "forward"]], "torchtune.modules.common_utils": [[192, 0, 1, "", "delete_kv_caches"], [193, 0, 1, "", "disable_kv_cache"], [194, 0, 1, "", "local_kv_cache"], [195, 0, 1, "", "reparametrize_as_dtype_state_dict_post_hook"]], "torchtune.modules.loss": [[196, 1, 1, "", "CEWithChunkedOutputLoss"], [197, 1, 1, "", "ForwardKLLoss"], [198, 1, 1, "", "ForwardKLWithChunkedOutputLoss"]], "torchtune.modules.loss.CEWithChunkedOutputLoss": [[196, 4, 1, "", "compute_cross_entropy"], [196, 4, 1, "", "forward"]], "torchtune.modules.loss.ForwardKLLoss": [[197, 4, 1, "", "forward"]], "torchtune.modules.loss.ForwardKLWithChunkedOutputLoss": [[198, 4, 1, "", "forward"]], "torchtune.modules.model_fusion": [[199, 1, 1, "", "DeepFusionModel"], [200, 1, 1, "", "FusionEmbedding"], [201, 1, 1, "", "FusionLayer"], [202, 0, 1, "", "get_fusion_params"], [203, 0, 1, "", "register_fusion_module"]], "torchtune.modules.model_fusion.DeepFusionModel": [[199, 4, 1, "", "caches_are_enabled"], [199, 4, 1, "", "caches_are_setup"], [199, 4, 1, "", "forward"], [199, 4, 1, "", "reset_caches"], [199, 4, 1, "", "set_num_output_chunks"], [199, 4, 1, "", "setup_caches"]], "torchtune.modules.model_fusion.FusionEmbedding": [[200, 4, 1, "", "forward"], [200, 4, 1, "", "fusion_params"]], "torchtune.modules.model_fusion.FusionLayer": [[201, 4, 1, "", "caches_are_enabled"], [201, 4, 1, "", "caches_are_setup"], [201, 4, 1, "", "forward"], [201, 4, 1, "", "fusion_params"], [201, 4, 1, "", "reset_cache"], [201, 4, 1, "", "setup_caches"]], "torchtune.modules.peft": [[204, 1, 1, "", "AdapterModule"], [205, 1, 1, "", "LoRALinear"], [206, 0, 1, "", "disable_adapter"], [207, 0, 1, "", "get_adapter_params"], [208, 0, 1, "", "set_trainable_params"], [209, 0, 1, "", "validate_missing_and_unexpected_for_lora"], [210, 0, 1, "", "validate_state_dict_for_lora"]], "torchtune.modules.peft.AdapterModule": [[204, 4, 1, "", "adapter_params"]], "torchtune.modules.peft.LoRALinear": [[205, 4, 1, "", "adapter_params"], [205, 4, 1, "", "forward"]], "torchtune.modules.tokenizers": [[211, 1, 1, "", "BaseTokenizer"], [212, 1, 1, "", "ModelTokenizer"], [213, 1, 1, "", "SentencePieceBaseTokenizer"], [214, 1, 1, "", "TikTokenBaseTokenizer"], [215, 0, 1, "", "parse_hf_tokenizer_json"], [216, 0, 1, "", "tokenize_messages_no_special_tokens"]], "torchtune.modules.tokenizers.BaseTokenizer": [[211, 4, 1, "", "decode"], [211, 4, 1, "", "encode"]], "torchtune.modules.tokenizers.ModelTokenizer": [[212, 4, 1, "", "tokenize_messages"]], "torchtune.modules.tokenizers.SentencePieceBaseTokenizer": [[213, 4, 1, "", "decode"], [213, 4, 1, "", "encode"]], "torchtune.modules.tokenizers.TikTokenBaseTokenizer": [[214, 4, 1, "", "decode"], [214, 4, 1, "", "encode"]], "torchtune.modules.transforms": [[217, 1, 1, "", "Transform"], [218, 1, 1, "", "VisionCrossAttentionMask"]], "torchtune.rlhf": [[219, 0, 1, "", "estimate_advantages"], [220, 0, 1, "", "get_rewards_ppo"], [225, 0, 1, "", "truncate_sequence_at_first_stop_token"]], "torchtune.rlhf.loss": [[221, 1, 1, "", "DPOLoss"], [222, 1, 1, "", "PPOLoss"], [223, 1, 1, "", "RSOLoss"], [224, 1, 1, "", "SimPOLoss"]], "torchtune.rlhf.loss.DPOLoss": [[221, 4, 1, "", "forward"]], "torchtune.rlhf.loss.PPOLoss": [[222, 4, 1, "", "forward"]], "torchtune.rlhf.loss.RSOLoss": [[223, 4, 1, "", "forward"]], "torchtune.rlhf.loss.SimPOLoss": [[224, 4, 1, "", "forward"]], "torchtune.training": [[226, 2, 1, "", "FSDPPolicyType"], [227, 1, 1, "", "FormattedCheckpointFiles"], [228, 1, 1, "", "FullModelHFCheckpointer"], [229, 1, 1, "", "FullModelMetaCheckpointer"], [230, 1, 1, "", "FullModelTorchTuneCheckpointer"], [231, 1, 1, "", "ModelType"], [232, 1, 1, "", "OptimizerInBackwardWrapper"], [233, 0, 1, "", "apply_selective_activation_checkpointing"], [234, 0, 1, "", "create_optim_in_bwd_wrapper"], [235, 0, 1, "", "get_cosine_schedule_with_warmup"], [236, 0, 1, "", "get_dtype"], [237, 0, 1, "", "get_full_finetune_fsdp_wrap_policy"], [238, 0, 1, "", "get_lr"], [239, 0, 1, "", "get_memory_stats"], [240, 0, 1, "", "get_quantizer_mode"], [241, 0, 1, "", "get_unmasked_sequence_lengths"], [242, 0, 1, "", "get_world_size_and_rank"], [243, 0, 1, "", "init_distributed"], [244, 0, 1, "", "is_distributed"], [245, 0, 1, "", "log_memory_stats"], [246, 0, 1, "", "lora_fsdp_wrap_policy"], [252, 0, 1, "", "register_optim_in_bwd_hooks"], [253, 0, 1, "", "set_activation_checkpointing"], [254, 0, 1, "", "set_default_dtype"], [255, 0, 1, "", "set_seed"], [256, 0, 1, "", "setup_torch_profiler"], [257, 0, 1, "", "update_state_dict_for_classifier"], [258, 0, 1, "", "validate_expected_param_dtype"]], "torchtune.training.FormattedCheckpointFiles": [[227, 4, 1, "", "build_checkpoint_filenames"]], "torchtune.training.FullModelHFCheckpointer": [[228, 4, 1, "", "load_checkpoint"], [228, 4, 1, "", "save_checkpoint"]], "torchtune.training.FullModelMetaCheckpointer": [[229, 4, 1, "", "load_checkpoint"], [229, 4, 1, "", "save_checkpoint"]], "torchtune.training.FullModelTorchTuneCheckpointer": [[230, 4, 1, "", "load_checkpoint"], [230, 4, 1, "", "save_checkpoint"]], "torchtune.training.OptimizerInBackwardWrapper": [[232, 4, 1, "", "get_last_lr"], [232, 4, 1, "", "get_optim_key"], [232, 4, 1, "", "load_state_dict"], [232, 4, 1, "", "set_lr_scheduler"], [232, 4, 1, "", "state_dict"], [232, 4, 1, "", "step_lr_scheduler"]], "torchtune.training.metric_logging": [[247, 1, 1, "", "CometLogger"], [248, 1, 1, "", "DiskLogger"], [249, 1, 1, "", "StdoutLogger"], [250, 1, 1, "", "TensorBoardLogger"], [251, 1, 1, "", "WandBLogger"]], "torchtune.training.metric_logging.CometLogger": [[247, 4, 1, "", "close"], [247, 4, 1, "", "log"], [247, 4, 1, "", "log_config"], [247, 4, 1, "", "log_dict"]], "torchtune.training.metric_logging.DiskLogger": [[248, 4, 1, "", "close"], [248, 4, 1, "", "log"], [248, 4, 1, "", "log_dict"]], "torchtune.training.metric_logging.StdoutLogger": [[249, 4, 1, "", "close"], [249, 4, 1, "", "log"], [249, 4, 1, "", "log_dict"]], "torchtune.training.metric_logging.TensorBoardLogger": [[250, 4, 1, "", "close"], [250, 4, 1, "", "log"], [250, 4, 1, "", "log_dict"]], "torchtune.training.metric_logging.WandBLogger": [[251, 4, 1, "", "close"], [251, 4, 1, "", "log"], [251, 4, 1, "", "log_config"], [251, 4, 1, "", "log_dict"]], "torchtune.utils": [[259, 0, 1, "", "batch_to_device"], [260, 0, 1, "", "get_device"], [261, 0, 1, "", "get_logger"], [262, 0, 1, "", "torch_version_ge"]]}, "objtypes": {"0": "py:function", "1": "py:class", "2": "py:data", "3": "py:property", "4": "py:method"}, "objnames": {"0": ["py", "function", "Python function"], "1": ["py", "class", "Python class"], "2": ["py", "data", "Python data"], "3": ["py", "property", "Python property"], "4": ["py", "method", "Python method"]}, "titleterms": {"torchtun": [0, 1, 2, 3, 4, 5, 6, 7, 8, 10, 22, 34, 40, 41, 43, 226, 265, 267, 272, 274, 276, 277, 278, 280, 281], "config": [0, 10, 24, 25, 272, 275], "data": [1, 11, 34, 40, 41, 43, 273], "text": [1, 2, 14, 16, 20, 276], "templat": [1, 9, 12, 14, 19, 21, 273], "type": 1, "messag": [1, 13, 14, 36], "transform": [1, 5, 13, 14, 15, 217], "collat": 1, "helper": 1, "function": 1, "dataset": [2, 9, 11, 12, 16, 18, 20, 273], "imag": [2, 14, 16], "gener": [2, 3, 73, 274, 276], "builder": 2, "class": [2, 19, 25], "model": [4, 5, 15, 21, 26, 272, 274, 275, 276, 277, 278, 279, 280], "llama3": [4, 116, 273, 276, 277, 280], "2": [4, 277], "vision": [4, 5], "1": [4, 277], "llama2": [4, 101, 273, 274, 278, 281], "code": 4, "llama": 4, "qwen": 4, "phi": 4, "3": 4, "mistral": [4, 158], "gemma": [4, 91], "clip": 4, "modul": 5, "compon": [5, 10, 24, 279], "build": [5, 266, 281], "block": 5, "loss": 5, "base": [5, 21], "token": [5, 14, 21, 273], "util": [5, 8], "peft": [5, 279], "fusion": 5, "rlhf": 6, "train": [7, 226, 269, 275], "checkpoint": [7, 22, 26, 274, 279], "reduc": 7, "precis": [7, 279], "distribut": [7, 269], "memori": [7, 278, 279, 281], "manag": 7, "schedul": 7, "metric": [7, 23, 26], "log": [7, 23, 26], "perform": [7, 278], "profil": 7, "miscellan": [7, 8], "chat": [9, 273], "exampl": [9, 12, 13, 15, 16, 18, 20], "format": [9, 12, 14, 16, 18, 20, 22], "load": [9, 12, 16, 18, 20, 21], "from": [9, 12, 16, 18, 20, 21, 273, 281], "hug": [9, 12, 16, 18, 20, 21, 274], "face": [9, 12, 16, 18, 20, 21, 274], "local": [9, 12, 16, 18, 20], "remot": [9, 12, 16], "specifi": 9, "convers": 9, "style": 9, "sharegpt": 9, "openai": 9, "renam": [9, 12], "column": [9, 12], "built": [9, 12, 16, 18, 19, 20, 272], "custom": [10, 13, 19, 273], "recip": [10, 25, 270, 272, 275, 277, 278, 280], "set": [10, 21], "up": [10, 274], "your": [10, 24, 25, 274, 275], "project": 10, "launch": 10, "overview": [11, 22, 267, 270, 274, 279], "pipelin": 11, "instruct": [12, 266, 276], "configur": [13, 24], "creat": [14, 15], "prompt": [14, 19, 21, 273], "access": [14, 276], "content": 14, "multimod": [15, 16], "us": [15, 19, 24, 25, 273, 274, 277, 281], "interleav": 16, "sampl": [17, 77], "pack": 17, "prefer": 18, "defin": 19, "via": [19, 266, 276], "dotpath": 19, "string": 19, "dictionari": 19, "prompttempl": [19, 38], "complet": 20, "json": 20, "txt": 20, "download": [21, 272, 274, 275], "file": 21, "max": 21, "sequenc": 21, "length": 21, "special": [21, 273], "handl": 22, "differ": 22, "hfcheckpoint": 22, "metacheckpoint": 22, "torchtunecheckpoint": 22, "intermedi": 22, "vs": 22, "final": 22, "lora": [22, 268, 274, 278, 279, 281], "put": [22, 281], "thi": 22, "all": [22, 24, 281], "togeth": [22, 281], "comet": 23, "logger": [23, 26], "about": 24, "where": 24, "do": 24, "paramet": [24, 279], "live": 24, "write": 24, "instanti": [24, 27], "referenc": 24, "other": [24, 274], "field": 24, "interpol": 24, "valid": [24, 30, 272], "best": 24, "practic": 24, "airtight": 24, "public": 24, "api": 24, "onli": 24, "command": 24, "line": 24, "overrid": 24, "remov": 24, "what": [25, 267, 277, 278, 280, 281], "ar": 25, "script": 25, "run": [25, 272, 274], "cli": [25, 272], "pars": [25, 29], "weight": 26, "bias": 26, "w": 26, "b": 26, "log_config": 28, "alpacatomessag": 31, "chatmltempl": 32, "chosenrejectedtomessag": 33, "grammarerrorcorrectiontempl": 34, "inputoutputtomessag": 35, "openaitomessag": 37, "prompttemplateinterfac": 39, "questionanswertempl": 40, "role": 41, "sharegpttomessag": 42, "summarizetempl": 43, "format_content_with_imag": 44, "left_pad_sequ": 45, "load_imag": 46, "padded_col": 47, "padded_collate_dpo": 48, "padded_collate_sft": 49, "padded_collate_tiled_images_and_mask": 50, "truncat": 51, "validate_messag": 52, "concatdataset": 53, "packeddataset": 54, "preferencedataset": 55, "sftdataset": 56, "textcompletiondataset": 57, "alpaca_cleaned_dataset": 58, "alpaca_dataset": 59, "chat_dataset": 60, "cnn_dailymail_articles_dataset": 61, "grammar_dataset": 62, "hh_rlhf_helpful_dataset": 63, "instruct_dataset": 64, "llava_instruct_dataset": 65, "the_cauldron_dataset": 66, "preference_dataset": 67, "samsum_dataset": 68, "slimorca_dataset": 69, "stack_exchange_paired_dataset": 70, "text_completion_dataset": 71, "wikitext_dataset": 72, "generate_next_token": 74, "get_causal_mask_from_padding_mask": 75, "get_position_ids_from_padding_mask": 76, "tilepositionalembed": 78, "tiledtokenpositionalembed": 79, "tokenpositionalembed": 80, "clip_vision_encod": 81, "code_llama2_13b": 82, "code_llama2_70b": 83, "code_llama2_7b": 84, "lora_code_llama2_13b": 85, "lora_code_llama2_70b": 86, "lora_code_llama2_7b": 87, "qlora_code_llama2_13b": 88, "qlora_code_llama2_70b": 89, "qlora_code_llama2_7b": 90, "gemma_2b": 92, "gemma_7b": 93, "gemma_token": 94, "lora_gemma": 95, "lora_gemma_2b": 96, "lora_gemma_7b": 97, "qlora_gemma_2b": 98, "qlora_gemma_7b": 99, "llama2chattempl": 100, "llama2_13b": 102, "llama2_70b": 103, "llama2_7b": 104, "llama2_reward_7b": 105, "llama2_token": 106, "lora_llama2": 107, "lora_llama2_13b": 108, "lora_llama2_70b": 109, "lora_llama2_7b": 110, "lora_llama2_reward_7b": 111, "qlora_llama2_13b": 112, "qlora_llama2_70b": 113, "qlora_llama2_7b": 114, "qlora_llama2_reward_7b": 115, "llama3_70b": 117, "llama3_8b": 118, "llama3_token": 119, "lora_llama3": 120, "lora_llama3_70b": 121, "lora_llama3_8b": 122, "qlora_llama3_70b": 123, "qlora_llama3_8b": 124, "llama3_1": 125, "llama3_1_405b": 126, "llama3_1_70b": 127, "llama3_1_8b": 128, "lora_llama3_1": 129, "lora_llama3_1_405b": 130, "lora_llama3_1_70b": 131, "lora_llama3_1_8b": 132, "qlora_llama3_1_405b": 133, "qlora_llama3_1_70b": 134, "qlora_llama3_1_8b": 135, "llama3_2_1b": 136, "llama3_2_3b": 137, "lora_llama3_2_1b": 138, "lora_llama3_2_3b": 139, "qlora_llama3_2_1b": 140, "qlora_llama3_2_3b": 141, "llama3visionencod": 142, "llama3visionprojectionhead": 143, "llama3visiontransform": 144, "llama3_2_vision_11b": 145, "llama3_2_vision_decod": 146, "llama3_2_vision_encod": 147, "llama3_2_vision_transform": 148, "lora_llama3_2_vision_11b": 149, "lora_llama3_2_vision_decod": 150, "lora_llama3_2_vision_encod": 151, "qlora_llama3_2_vision_11b": 152, "mistralchattempl": 153, "lora_mistr": 154, "lora_mistral_7b": 155, "lora_mistral_classifi": 156, "lora_mistral_reward_7b": 157, "mistral_7b": 159, "mistral_classifi": 160, "mistral_reward_7b": 161, "mistral_token": 162, "qlora_mistral_7b": 163, "qlora_mistral_reward_7b": 164, "lora_phi3": 165, "lora_phi3_mini": 166, "phi3": 167, "phi3_mini": 168, "phi3_mini_token": 169, "qlora_phi3_mini": 170, "lora_qwen2": 171, "lora_qwen2_0_5b": 172, "lora_qwen2_1_5b": 173, "lora_qwen2_7b": 174, "qwen2": [175, 277], "qwen2_0_5b": 176, "qwen2_1_5b": 177, "qwen2_7b": 178, "qwen2_token": 179, "feedforward": 180, "fp32layernorm": 181, "kvcach": 182, "multiheadattent": 183, "rmsnorm": 184, "rotarypositionalembed": 185, "tanhgat": 186, "tiedlinear": 187, "transformercrossattentionlay": 188, "transformerdecod": 189, "transformerselfattentionlay": 190, "visiontransform": 191, "delete_kv_cach": 192, "disable_kv_cach": 193, "local_kv_cach": 194, "reparametrize_as_dtype_state_dict_post_hook": 195, "cewithchunkedoutputloss": 196, "forwardklloss": 197, "forwardklwithchunkedoutputloss": 198, "deepfusionmodel": 199, "fusionembed": 200, "fusionlay": 201, "get_fusion_param": 202, "register_fusion_modul": 203, "adaptermodul": 204, "loralinear": 205, "disable_adapt": 206, "get_adapter_param": 207, "set_trainable_param": 208, "validate_missing_and_unexpected_for_lora": 209, "validate_state_dict_for_lora": 210, "basetoken": 211, "modeltoken": 212, "sentencepiecebasetoken": 213, "tiktokenbasetoken": 214, "parse_hf_tokenizer_json": 215, "tokenize_messages_no_special_token": 216, "visioncrossattentionmask": 218, "estimate_advantag": 219, "get_rewards_ppo": 220, "dpoloss": 221, "ppoloss": 222, "rsoloss": 223, "simpoloss": 224, "truncate_sequence_at_first_stop_token": 225, "fsdppolicytyp": 226, "formattedcheckpointfil": 227, "fullmodelhfcheckpoint": 228, "fullmodelmetacheckpoint": 229, "fullmodeltorchtunecheckpoint": 230, "modeltyp": 231, "optimizerinbackwardwrapp": 232, "apply_selective_activation_checkpoint": 233, "create_optim_in_bwd_wrapp": 234, "get_cosine_schedule_with_warmup": 235, "get_dtyp": 236, "get_full_finetune_fsdp_wrap_polici": 237, "get_lr": 238, "get_memory_stat": 239, "get_quantizer_mod": 240, "get_unmasked_sequence_length": 241, "get_world_size_and_rank": 242, "init_distribut": 243, "is_distribut": 244, "log_memory_stat": 245, "lora_fsdp_wrap_polici": 246, "cometlogg": 247, "disklogg": 248, "stdoutlogg": 249, "tensorboardlogg": 250, "wandblogg": 251, "register_optim_in_bwd_hook": 252, "set_activation_checkpoint": 253, "set_default_dtyp": 254, "set_se": 255, "setup_torch_profil": 256, "update_state_dict_for_classifi": 257, "validate_expected_param_dtyp": 258, "batch_to_devic": 259, "get_devic": 260, "get_logg": 261, "torch_version_g": 262, "comput": [264, 271], "time": [264, 271], "welcom": 265, "document": 265, "get": [265, 272, 276], "start": [265, 272], "tutori": 265, "instal": 266, "pre": 266, "requisit": 266, "pypi": 266, "git": 266, "clone": 266, "nightli": 266, "kei": 267, "concept": 267, "design": 267, "principl": 267, "singl": 268, "devic": [268, 280], "finetun": [268, 270, 274, 278, 280, 281], "quantiz": [269, 274, 276, 279, 280], "awar": 269, "qat": [269, 280], "list": 272, "copi": 272, "fine": [273, 275, 276, 277, 278, 279, 280, 281], "tune": [273, 275, 276, 277, 278, 279, 280, 281], "chang": 273, "when": 273, "should": 273, "i": 273, "end": 274, "workflow": 274, "7b": 274, "evalu": [274, 276, 280], "eleutherai": [274, 276], "s": [274, 276], "eval": [274, 276], "har": [274, 276], "speed": 274, "librari": 274, "upload": 274, "hub": 274, "first": 275, "llm": 275, "select": 275, "modifi": 275, "next": 275, "step": [275, 279], "meta": 276, "8b": [276, 277], "our": 276, "faster": 276, "distil": 277, "1b": 277, "knowledg": 277, "how": [277, 278], "doe": [277, 278], "work": [277, 278], "kd": 277, "ablat": 277, "studi": 277, "teacher": 277, "student": 277, "hyperparamet": 277, "learn": 277, "rate": 277, "ratio": 277, "5b": 277, "0": 277, "appli": [278, 280], "trade": 278, "off": 278, "optim": 279, "activ": 279, "offload": 279, "gradient": 279, "accumul": 279, "lower": [279, 280], "fuse": 279, "backward": 279, "pass": 279, "state": 279, "cpu": 279, "effici": 279, "low": 279, "rank": 279, "adapt": 279, "qlora": [279, 281], "option": 280, "save": 281, "deep": 281, "dive": 281}, "envversion": {"sphinx.domains.c": 2, "sphinx.domains.changeset": 1, "sphinx.domains.citation": 1, "sphinx.domains.cpp": 6, "sphinx.domains.index": 1, "sphinx.domains.javascript": 2, "sphinx.domains.math": 2, "sphinx.domains.python": 3, "sphinx.domains.rst": 2, "sphinx.domains.std": 2, "sphinx.ext.intersphinx": 1, "sphinx.ext.todo": 2, "sphinx.ext.viewcode": 1, "sphinx": 56}})
\ No newline at end of file
diff --git a/main/sg_execution_times.html b/main/sg_execution_times.html
index fa1f8ca49e..e8cdb2153e 100644
--- a/main/sg_execution_times.html
+++ b/main/sg_execution_times.html
@@ -331,8 +331,8 @@
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/tune_cli.html b/main/tune_cli.html
index 4b860c7c2e..83ff2b136d 100644
--- a/main/tune_cli.html
+++ b/main/tune_cli.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="tutorials/llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="tutorials/memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/tutorials/chat.html b/main/tutorials/chat.html
index 0ed25b93a9..580f0fdf38 100644
--- a/main/tutorials/chat.html
+++ b/main/tutorials/chat.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/tutorials/e2e_flow.html b/main/tutorials/e2e_flow.html
index b8def79339..6768ecbe66 100644
--- a/main/tutorials/e2e_flow.html
+++ b/main/tutorials/e2e_flow.html
@@ -40,7 +40,7 @@
   <link rel="stylesheet" href="../_static/css/custom_torchtune.css" type="text/css" />
     <link rel="index" title="Index" href="../genindex.html" />
     <link rel="search" title="Search" href="../search.html" />
-    <link rel="next" title="Memory Optimization Overview" href="memory_optimizations.html" />
+    <link rel="next" title="Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation" href="llama_kd_tutorial.html" />
     <link rel="prev" title="Fine-Tuning Llama3 with QAT" href="qat_finetune.html" />
   <!-- Google Tag Manager -->
     <script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1 current"><a class="current reference internal" href="#">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
@@ -828,7 +828,7 @@ <h2>Uploading your model to the Hugging Face Hub<a class="headerlink" href="#upl
   
     <div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
       
-        <a href="memory_optimizations.html" class="btn btn-neutral float-right" title="Memory Optimization Overview" accesskey="n" rel="next">Next <img src="../_static/images/chevron-right-orange.svg" class="next-page"></a>
+        <a href="llama_kd_tutorial.html" class="btn btn-neutral float-right" title="Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation" accesskey="n" rel="next">Next <img src="../_static/images/chevron-right-orange.svg" class="next-page"></a>
       
       
         <a href="qat_finetune.html" class="btn btn-neutral" title="Fine-Tuning Llama3 with QAT" accesskey="p" rel="prev"><img src="../_static/images/chevron-right-orange.svg" class="previous-page"> Previous</a>
diff --git a/main/tutorials/first_finetune_tutorial.html b/main/tutorials/first_finetune_tutorial.html
index e671c1dba3..60c6a30ac9 100644
--- a/main/tutorials/first_finetune_tutorial.html
+++ b/main/tutorials/first_finetune_tutorial.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/tutorials/llama3.html b/main/tutorials/llama3.html
index 40a523b2e2..f9218d1850 100644
--- a/main/tutorials/llama3.html
+++ b/main/tutorials/llama3.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/tutorials/llama_kd_tutorial.html b/main/tutorials/llama_kd_tutorial.html
index 52a19dd326..6fe325611a 100644
--- a/main/tutorials/llama_kd_tutorial.html
+++ b/main/tutorials/llama_kd_tutorial.html
@@ -40,8 +40,8 @@
   <link rel="stylesheet" href="../_static/css/custom_torchtune.css" type="text/css" />
     <link rel="index" title="Index" href="../genindex.html" />
     <link rel="search" title="Search" href="../search.html" />
-    <link rel="next" title="Checkpointing in torchtune" href="../deep_dives/checkpointer.html" />
-    <link rel="prev" title="Memory Optimization Overview" href="memory_optimizations.html" />
+    <link rel="next" title="Memory Optimization Overview" href="memory_optimizations.html" />
+    <link rel="prev" title="End-to-End Workflow with torchtune" href="e2e_flow.html" />
   <!-- Google Tag Manager -->
     <script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':
     new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1 current"><a class="current reference internal" href="#">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
@@ -648,10 +648,10 @@ <h3>Qwen2 1.5B to 0.5B<a class="headerlink" href="#qwen2-1-5b-to-0-5b" title="Pe
   
     <div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
       
-        <a href="../deep_dives/checkpointer.html" class="btn btn-neutral float-right" title="Checkpointing in torchtune" accesskey="n" rel="next">Next <img src="../_static/images/chevron-right-orange.svg" class="next-page"></a>
+        <a href="memory_optimizations.html" class="btn btn-neutral float-right" title="Memory Optimization Overview" accesskey="n" rel="next">Next <img src="../_static/images/chevron-right-orange.svg" class="next-page"></a>
       
       
-        <a href="memory_optimizations.html" class="btn btn-neutral" title="Memory Optimization Overview" accesskey="p" rel="prev"><img src="../_static/images/chevron-right-orange.svg" class="previous-page"> Previous</a>
+        <a href="e2e_flow.html" class="btn btn-neutral" title="End-to-End Workflow with torchtune" accesskey="p" rel="prev"><img src="../_static/images/chevron-right-orange.svg" class="previous-page"> Previous</a>
       
     </div>
   
diff --git a/main/tutorials/lora_finetune.html b/main/tutorials/lora_finetune.html
index 71428d326e..2dbd75b69a 100644
--- a/main/tutorials/lora_finetune.html
+++ b/main/tutorials/lora_finetune.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/tutorials/memory_optimizations.html b/main/tutorials/memory_optimizations.html
index 0a383de6b0..783feba307 100644
--- a/main/tutorials/memory_optimizations.html
+++ b/main/tutorials/memory_optimizations.html
@@ -40,8 +40,8 @@
   <link rel="stylesheet" href="../_static/css/custom_torchtune.css" type="text/css" />
     <link rel="index" title="Index" href="../genindex.html" />
     <link rel="search" title="Search" href="../search.html" />
-    <link rel="next" title="Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation" href="llama_kd_tutorial.html" />
-    <link rel="prev" title="End-to-End Workflow with torchtune" href="e2e_flow.html" />
+    <link rel="next" title="Checkpointing in torchtune" href="../deep_dives/checkpointer.html" />
+    <link rel="prev" title="Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation" href="llama_kd_tutorial.html" />
   <!-- Google Tag Manager -->
     <script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':
     new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1 current"><a class="current reference internal" href="#">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1 current"><a class="current reference internal" href="#">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
@@ -451,15 +451,21 @@
 <tr class="row-odd"><td><p><a class="reference internal" href="#glossary-act-ckpt"><span class="std std-ref">Activation Checkpointing</span></a></p></td>
 <td><p>Use when you’re memory constrained and need to handle larger batch sizes or longer context lengths. Be aware that it may slow down training speed.</p></td>
 </tr>
-<tr class="row-even"><td><p><a class="reference internal" href="#glossary-grad-accm"><span class="std std-ref">Gradient Accumulation</span></a></p></td>
+<tr class="row-even"><td><p><a class="reference internal" href="#glossary-act-off"><span class="std std-ref">Activation Offloading</span></a></p></td>
+<td><p>Similar to activation checkpointing, this can be used when memory constrained, but comes at the cost of training speed due to the overhead of moving tensors between GPU VRAM and CPU. This can also be used alongside activation checkpointing.</p></td>
+</tr>
+<tr class="row-odd"><td><p><a class="reference internal" href="#glossary-grad-accm"><span class="std std-ref">Gradient Accumulation</span></a></p></td>
 <td><p>Helpful when memory-constrained to simulate larger batch sizes. Often preferable to activation checkpointing for better training speed.</p></td>
 </tr>
-<tr class="row-odd"><td><p><a class="reference internal" href="#glossary-low-precision-opt"><span class="std std-ref">Lower Precision Optimizers</span></a></p></td>
+<tr class="row-even"><td><p><a class="reference internal" href="#glossary-low-precision-opt"><span class="std std-ref">Lower Precision Optimizers</span></a></p></td>
 <td><p>When you need to further reduce memory usage beyond using <code class="docutils literal notranslate"><span class="pre">bf16</span></code> by reducing the precision in the optimizer states. Note that lower precision optimizers may reduce training stability/accuracy.</p></td>
 </tr>
-<tr class="row-even"><td><p><a class="reference internal" href="#glossary-opt-in-bwd"><span class="std std-ref">Fusing Optimizer Step into Backward Pass</span></a></p></td>
+<tr class="row-odd"><td><p><a class="reference internal" href="#glossary-opt-in-bwd"><span class="std std-ref">Fusing Optimizer Step into Backward Pass</span></a></p></td>
 <td><p>Helps reduce memory usage when using stateful optimizers, particularly when full-finetuning large models with high gradient memory usage. This is not compatible with <code class="docutils literal notranslate"><span class="pre">gradient_accumulation_steps</span></code>, so training may slow down due to reduced model throughput.</p></td>
 </tr>
+<tr class="row-even"><td><p><a class="reference internal" href="#glossary-cpu-offload"><span class="std std-ref">Offloading Optimizer/Gradient states to CPU</span></a></p></td>
+<td><p>Offloads optimizer states and (optionally) gradients to CPU, and performs optimizer steps on CPU. This can be used to significantly reduce GPU memory usage at the cost of CPU RAM and training speed, as CPU optimizer steps can be slow and bottleneck training performance.</p></td>
+</tr>
 <tr class="row-odd"><td><p><a class="reference internal" href="#glossary-lora"><span class="std std-ref">Low Rank Adaptation (LoRA)</span></a></p></td>
 <td><p>When you want to significantly reduce the number of trainable parameters, saving gradient and optimizer memory during training, and significantly speeding up training.</p></td>
 </tr>
@@ -523,7 +529,7 @@
 efficiency technique that allows saving GPU VRAM by temporarily moving activations to CPU and bringing
 them back when needed in the backward pass.</p>
 <p>See <a class="reference external" href="https://pytorch.org/tutorials/intermediate/autograd_saved_tensors_hooks_tutorial.html#saving-tensors-to-cpu">PyTorch autograd hook tutorial</a>
-for more details about how this is implemented through saved_tensors_hooks.</p>
+for more details about how this is implemented through <code class="xref py py-func docutils literal notranslate"><span class="pre">torch.autograd.graph.saved_tensors_hooks()</span></code>.</p>
 <p>This setting is especially helpful for larger batch sizes, or longer context lengths when you’re memory constrained.
 While of course it takes runtime and resources to move Tensors from GPU to CPU and back, the implementation in
 torchtune uses multiple CUDA streams (when available) in order to overlap the extra communication with the computation
@@ -571,8 +577,10 @@
 <p>Gradient accumulation should always be set to 1 when <a class="reference internal" href="#glossary-opt-in-bwd"><span class="std std-ref">fusing the optimizer step into the backward pass</span></a>.</p>
 </div>
 </section>
+<section id="optimizers">
+<h2>Optimizers<a class="headerlink" href="#optimizers" title="Permalink to this heading">¶</a></h2>
 <section id="lower-precision-optimizers">
-<span id="glossary-low-precision-opt"></span><h2>Lower Precision Optimizers<a class="headerlink" href="#lower-precision-optimizers" title="Permalink to this heading">¶</a></h2>
+<span id="glossary-low-precision-opt"></span><h3>Lower Precision Optimizers<a class="headerlink" href="#lower-precision-optimizers" title="Permalink to this heading">¶</a></h3>
 <p><em>What’s going on here?</em></p>
 <p>In addition to <a class="reference internal" href="#glossary-precision"><span class="std std-ref">reducing model and optimizer precision</span></a> during training, we can further reduce precision in our optimizer states.
 All of our single-device fine-tuning recipes support lower-precision optimizers from the <a class="reference external" href="https://huggingface.co/docs/bitsandbytes/main/en/index">bitsandbytes</a> library -
@@ -592,7 +600,7 @@
 </div>
 </section>
 <section id="fusing-optimizer-step-into-backward-pass">
-<span id="glossary-opt-in-bwd"></span><h2>Fusing Optimizer Step into Backward Pass<a class="headerlink" href="#fusing-optimizer-step-into-backward-pass" title="Permalink to this heading">¶</a></h2>
+<span id="glossary-opt-in-bwd"></span><h3>Fusing Optimizer Step into Backward Pass<a class="headerlink" href="#fusing-optimizer-step-into-backward-pass" title="Permalink to this heading">¶</a></h3>
 <p><em>What’s going on here?</em></p>
 <p>Stateful optimizers (e.g. optimizers which use momentum) are the default in modern deep learning due to their stable convergence properties.
 However, maintaining a state of gradient statistics comes at the cost of additional memory usage. An immediate alternative might be to
@@ -605,10 +613,57 @@
 <a class="reference external" href="https://pytorch.org/tutorials/intermediate/optimizer_step_in_backward_tutorial.html">How to save memory by fusing the optimizer step into the backward pass</a>.</p>
 <p><em>Sounds great! How do I use it?</em></p>
 <p>In torchtune, you can enable this feature using the <code class="docutils literal notranslate"><span class="pre">optimizer_in_bwd</span></code> flag, which is currently only supported in our
-single-device full finetune recipe. This feature works best when gradient memory is particularly large;
+single-device full finetune recipe. This feature works best when optimizer memory is particularly large;
 e.g. when using a stateful optimizer with a model with a lot of parameters, and when you don’t need to use
 <a class="reference internal" href="#glossary-grad-accm"><span class="std std-ref">gradient accumulation</span></a>.</p>
 </section>
+<section id="offloading-optimizer-gradient-states-to-cpu">
+<span id="glossary-cpu-offload"></span><h3>Offloading Optimizer/Gradient states to CPU<a class="headerlink" href="#offloading-optimizer-gradient-states-to-cpu" title="Permalink to this heading">¶</a></h3>
+<p><em>What’s going on here?</em></p>
+<p>We’ve mentioned above the concept of optimizer states - memory used by the stateful optimizers to maintain a state of gradient statistics, and
+model gradients - tensors used to store gradients when we perform model backwards passes. We support using CPU offloading in our single-device recipes
+through the <a class="reference external" href="https://github.com/pytorch/ao/tree/main/torchao/prototype/low_bit_optim#optimizer-cpu-offload">CPUOffloadOptimizer</a> from <code class="docutils literal notranslate"><span class="pre">torchao</span></code>.</p>
+<p>This optimizer can wrap any base optimizer and works by keeping the optimizer states and performing the optimizer step on CPU, thus reducing
+GPU memory usage by the size of the optimizer states. Additionally, we can also offload gradients to the CPU by using <cite>offload_gradients=True</cite>.</p>
+<p><em>Sounds great! How do I use it?</em></p>
+<p>To use this optimizer in your recipes, set the <code class="docutils literal notranslate"><span class="pre">optimizer</span></code> key in your config to <code class="xref py py-class docutils literal notranslate"><span class="pre">torchao.prototype.low_bit_optim.CPUOffloadOptimizer</span></code>, which
+will use the <a class="reference external" href="https://pytorch.org/docs/stable/generated/torch.optim.AdamW.html#torch.optim.AdamW" title="(in PyTorch v2.5)"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.optim.AdamW</span></code></a> optimizer with <code class="docutils literal notranslate"><span class="pre">fused=True</span></code> as the base optimizer. For example, to use this optimizer to offload
+both optimizer states and gradients to CPU:</p>
+<div class="highlight-bash notranslate"><div class="highlight"><pre><span></span>tune<span class="w"> </span>run<span class="w"> </span>&lt;RECIPE&gt;<span class="w"> </span>--config<span class="w"> </span>&lt;CONFIG&gt;<span class="w"> </span><span class="se">\</span>
+<span class="nv">optimizer</span><span class="o">=</span><span class="nv">optimizer</span><span class="o">=</span>torchao.prototype.low_bit_optim.CPUOffloadOptimizer<span class="w"> </span><span class="se">\</span>
+optimizer.offload_gradients<span class="o">=</span>True<span class="w"> </span><span class="se">\</span>
+<span class="nv">lr</span><span class="o">=</span>4e-5
+</pre></div>
+</div>
+<p>or by directly <a class="reference internal" href="../deep_dives/configs.html#config-tutorial-label"><span class="std std-ref">modifying a config file</span></a>:</p>
+<div class="highlight-yaml notranslate"><div class="highlight"><pre><span></span><span class="nt">optimizer</span><span class="p">:</span>
+<span class="w">  </span><span class="nt">_component_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">torchao.prototype.low_bit_optim.CPUOffloadOptimizer</span>
+<span class="w">  </span><span class="nt">offload_gradients</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">True</span>
+<span class="w">  </span><span class="c1"># additional key-word arguments can be passed to torch.optim.AdamW</span>
+<span class="w">  </span><span class="nt">lr</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">4e-5</span>
+</pre></div>
+</div>
+<p>or using it directly in your code, which allows you to change the base optimizer:</p>
+<div class="highlight-python notranslate"><div class="highlight"><pre><span></span><span class="kn">from</span> <span class="nn">torchao.prototype.low_bit_optim</span> <span class="kn">import</span> <span class="n">CPUOffloadOptimizer</span>
+<span class="kn">from</span> <span class="nn">torch.optim</span> <span class="kn">import</span> <span class="n">Adam</span>
+
+<span class="n">optimizer</span> <span class="o">=</span> <span class="n">CPUOffloadOptimizer</span><span class="p">(</span>
+    <span class="n">model</span><span class="o">.</span><span class="n">parameters</span><span class="p">(),</span> <span class="c1"># your model here</span>
+    <span class="n">Adam</span><span class="p">,</span>
+    <span class="n">lr</span><span class="o">=</span><span class="mf">1e-5</span><span class="p">,</span>
+    <span class="n">fused</span><span class="o">=</span><span class="kc">True</span>
+<span class="p">)</span>
+</pre></div>
+</div>
+<p>Some helpful hints from the <code class="docutils literal notranslate"><span class="pre">torchao</span></code> <a class="reference external" href="https://github.com/pytorch/ao/tree/main/torchao/prototype/low_bit_optim#optimizer-cpu-offload">CPUOffloadOptimizer page</a>:</p>
+<ul class="simple">
+<li><p>The CPU optimizer step is often the bottleneck when optimizer CPU offload is used. To minimize the slowdown, it is recommended to (1) use full <code class="docutils literal notranslate"><span class="pre">bf16</span></code> training so that parameters, gradients, and optimizer states are in <code class="docutils literal notranslate"><span class="pre">bf16</span></code>; and (2) give GPU more work per optimizer step (e.g. larger batch size with activation checkpointing, gradient accumulation).</p></li>
+<li><p>Gradient accumulation should always be set to 1 when <code class="docutils literal notranslate"><span class="pre">offload_gradients=True</span></code>, as gradients are cleared on GPU every backward pass.</p></li>
+<li><p>This optimizer works by keeping a copy of parameters and pre-allocating gradient memory on CPU. Therefore, expect your RAM usage to increase by 4x model size.</p></li>
+<li><p>This optimizer is only supported for single-device recipes. To use CPU-offloading in distributed recipes, use <code class="docutils literal notranslate"><span class="pre">fsdp_cpu_offload=True</span></code> in any distributed recipe. See <a class="reference external" href="https://pytorch.org/docs/stable/fsdp.html#torch.distributed.fsdp.FullyShardedDataParallel" title="(in PyTorch v2.5)"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.distributed.fsdp.FullyShardedDataParallel</span></code></a> for more details</p></li>
+</ul>
+</section>
+</section>
 <section id="parameter-efficient-fine-tuning-peft">
 <span id="glossary-peft"></span><h2>Parameter Efficient Fine-Tuning (PEFT)<a class="headerlink" href="#parameter-efficient-fine-tuning-peft" title="Permalink to this heading">¶</a></h2>
 <section id="low-rank-adaptation-lora">
@@ -715,10 +770,10 @@
   
     <div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
       
-        <a href="llama_kd_tutorial.html" class="btn btn-neutral float-right" title="Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation" accesskey="n" rel="next">Next <img src="../_static/images/chevron-right-orange.svg" class="next-page"></a>
+        <a href="../deep_dives/checkpointer.html" class="btn btn-neutral float-right" title="Checkpointing in torchtune" accesskey="n" rel="next">Next <img src="../_static/images/chevron-right-orange.svg" class="next-page"></a>
       
       
-        <a href="e2e_flow.html" class="btn btn-neutral" title="End-to-End Workflow with torchtune" accesskey="p" rel="prev"><img src="../_static/images/chevron-right-orange.svg" class="previous-page"> Previous</a>
+        <a href="llama_kd_tutorial.html" class="btn btn-neutral" title="Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation" accesskey="p" rel="prev"><img src="../_static/images/chevron-right-orange.svg" class="previous-page"> Previous</a>
       
     </div>
   
@@ -755,8 +810,12 @@
 <li><a class="reference internal" href="#activation-checkpointing">Activation Checkpointing</a></li>
 <li><a class="reference internal" href="#activation-offloading">Activation Offloading</a></li>
 <li><a class="reference internal" href="#gradient-accumulation">Gradient Accumulation</a></li>
+<li><a class="reference internal" href="#optimizers">Optimizers</a><ul>
 <li><a class="reference internal" href="#lower-precision-optimizers">Lower Precision Optimizers</a></li>
 <li><a class="reference internal" href="#fusing-optimizer-step-into-backward-pass">Fusing Optimizer Step into Backward Pass</a></li>
+<li><a class="reference internal" href="#offloading-optimizer-gradient-states-to-cpu">Offloading Optimizer/Gradient states to CPU</a></li>
+</ul>
+</li>
 <li><a class="reference internal" href="#parameter-efficient-fine-tuning-peft">Parameter Efficient Fine-Tuning (PEFT)</a><ul>
 <li><a class="reference internal" href="#low-rank-adaptation-lora">Low Rank Adaptation (LoRA)</a></li>
 <li><a class="reference internal" href="#quantized-low-rank-adaptation-qlora">Quantized Low Rank Adaptation (QLoRA)</a></li>
diff --git a/main/tutorials/qat_finetune.html b/main/tutorials/qat_finetune.html
index e9dabb864f..5ade0cd36b 100644
--- a/main/tutorials/qat_finetune.html
+++ b/main/tutorials/qat_finetune.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1"><a class="reference internal" href="qlora_finetune.html">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1 current"><a class="current reference internal" href="#">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>
diff --git a/main/tutorials/qlora_finetune.html b/main/tutorials/qlora_finetune.html
index 6157a15d93..1861acef12 100644
--- a/main/tutorials/qlora_finetune.html
+++ b/main/tutorials/qlora_finetune.html
@@ -333,8 +333,8 @@
 <li class="toctree-l1 current"><a class="current reference internal" href="#">Fine-Tuning Llama2 with QLoRA</a></li>
 <li class="toctree-l1"><a class="reference internal" href="qat_finetune.html">Fine-Tuning Llama3 with QAT</a></li>
 <li class="toctree-l1"><a class="reference internal" href="e2e_flow.html">End-to-End Workflow with torchtune</a></li>
-<li class="toctree-l1"><a class="reference internal" href="memory_optimizations.html">Memory Optimization Overview</a></li>
 <li class="toctree-l1"><a class="reference internal" href="llama_kd_tutorial.html">Distilling Llama3.1 8B into Llama3.2 1B using Knowledge Distillation</a></li>
+<li class="toctree-l1"><a class="reference internal" href="memory_optimizations.html">Memory Optimization Overview</a></li>
 </ul>
 <p class="caption" role="heading"><span class="caption-text">Deep-Dives</span></p>
 <ul>