Added attention layers to wrap for fsdp #30735
Open
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Outdated suggestions cannot be applied.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.
What does this PR do?
This PR defines the
fsdp_transformer_layer_cls_to_wrap
value in the Mistral config. This way user can easily load the config to figure out what values to use for FSDP, e.g.[out]:
Context: Users has been asking when and which layer to wrap, there shouldn't be a need to load the model to figure it out by going through the state_dict of model summary,
Fixes: https://discuss.huggingface.co/t/accelerate-fsdp-config-prompts/21262/3
Currently, this information is also available through https://github.com/huggingface/transformers/blob/main/src/transformers/models/mistral/modeling_mistral.py#L809
Who can review?
Models:
Integrations:
Documentation: @stevhliu and @MKhalusova