Speed up model load by skipping init on strict load - #12
Merged
Conversation
Model load spends significant time on initializing layers with a slow truncnorm plus an implicit kaiminig_uniform_ of torch.nn.Linear. But this work is discarded at construction when the parameters are overwritten by the actual model weights with load_state_dict. This commit monkey patches trunc_normal_init_ and nn.Linear's reset_parameters to noops during inference via a skip_random_init context manager.
Collaborator
|
Hi rizsp, thank you for the contribution! We validated the change with a real checkpoint. It significantly reduces the unnecessary parameter-initialization overhead before loading the checkpoint. The change fits our inference use case well. This should be ready to merge. Thanks again! |
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Model load spends significant time on initializing layers with a slow truncnorm plus an implicit kaiminig_uniform_ of torch.nn.Linear. But this work is discarded at construction when the parameters are overwritten by the actual model weights with load_state_dict.
Changes
This commit monkey patches trunc_normal_init_ and nn.Linear's reset_parameters to noops during inference via a skip_random_init context manager.
On my machine, this leads to a reduction of ~1 minute. A similar patch has already been merged into openfold-3 (PR 220).