Error with checking args.eval_accumulation_steps to gather tensors #25819
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
…n_steps to gather tensors) While the deprecated code has the correct check (line 3772): "if args.eval_accumulation_steps is not None and (step + 1) % args.eval_accumulation_steps == 0:" The current code does not (line 3196): "if args.eval_accumulation_steps is not None and self.accelerator.sync_gradients:" We need to check "(step + 1) % args.eval_accumulation_steps == 0". Hence, the line 3196 should be modified to: "if args.eval_accumulation_steps is not None and (step + 1) % args.eval_accumulation_steps == 0 and self.accelerator.sync_gradients:"
|
cc @muellerzr |
muellerzr
approved these changes
Aug 29, 2023
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. |
|
@muellerzr Thank you. I have run the make commands. All tests passed. |
amyeroberts
approved these changes
Aug 29, 2023
4 tasks
parambharat
pushed a commit
to parambharat/transformers
that referenced
this pull request
Sep 26, 2023
…uggingface#25819) * Update trainer.py (error with checking steps in args.eval_accumulation_steps to gather tensors) While the deprecated code has the correct check (line 3772): "if args.eval_accumulation_steps is not None and (step + 1) % args.eval_accumulation_steps == 0:" The current code does not (line 3196): "if args.eval_accumulation_steps is not None and self.accelerator.sync_gradients:" We need to check "(step + 1) % args.eval_accumulation_steps == 0". Hence, the line 3196 should be modified to: "if args.eval_accumulation_steps is not None and (step + 1) % args.eval_accumulation_steps == 0 and self.accelerator.sync_gradients:" * Fix error with checking args.eval_accumulation_steps to gather tensors
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.
The error is in
trainer.py.While the deprecated (legacy) code has the correct check (line 3772):
The current code does not (line 3196):
We need to check
(step + 1) % args.eval_accumulation_steps == 0, because otherwise, if we set args.eval_accumulation_steps = 10, the code may still gather tensors for each step.Hence, the line 3196 should be modified to: