Compiling MLX Examples using Reactant

[avik-pal]

Original Repository: https://github.com/ml-explore/mlx-examples/

Listing out examples from there which would be nice to have. We don't expect the models to work out the moment they are translated to julia but these also provide us an understanding of what currently works and what doesn't in Reactant

Text Models

• Transformer Language Model Training.
• Large Scale Text Generation
  • LLaMA
  • Mistral
  • Other LLMs in https://github.com/ml-explore/mlx-examples/tree/main/llms
• Mixture of Experts language model with Mixtral 8x7B.
• Parameter efficient fine-tuning with LoRA or QLoRA.
• Text-to-text multi-task Transformers with T5.
• Bidirectional language understanding with BERT.

Image Models

• Image Generation
  • FLUX (not the julia package Flux.jl)
  • Stable Diffusion or SDXL
• Convolutional variational autoencoder (CVAE) on MNIST.

Audio Models

• Speech recognition with OpenAI's Whisper.
• Audio compression and generation with Meta's EnCodec.

Multimodal models

• Joint text and image embeddings with CLIP.
• Text generation from image and text inputs with LLaVA.
• Image segmentation with Segment Anything (SAM).

Other Models

• Semi-supervised learning on graph-structured data with GCN.
• Real NVP normalizing flow for density estimation and sampling.

[avik-pal]

Relevant Materials to get starting:

1. https://lux.csail.mit.edu/stable/manual/compiling_lux_models