fsdp for bert #140
Draft
fsdp for bert #140
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
abhi-mosaic
reviewed
Feb 9, 2023
examples/bert/src/bert_layers.py
Outdated
| @@ -545,6 +546,70 @@ def forward( | |||
| all_encoder_layers.append(hidden_states) | |||
| return all_encoder_layers | |||
|
|
|||
| # Param Initialization, needed for device='meta' fast initialization | |||
| def param_init_fn(self, module): | |||
There was a problem hiding this comment.
@jacobfulano a few comments:
param_init_fnis not needed unless you plan to make your modelsmetainitializable. For a first draft, I would remove entirelyfsdp_wrap_fnandactivation_checkpointing_fnneed to be defined on a root module underneath theComposerModel. So they have to be defined for say,BertForMaskedLM. I would recommend moving the definition there, and I think it will work!
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This is a draft PR for FSDP implementation in BERT