Fix streaming validation infinite loop (#42)#45
Open
vominh1919 wants to merge 1 commit intoPrimeIntellect-ai:mainfrom
Open
Fix streaming validation infinite loop (#42)#45vominh1919 wants to merge 1 commit intoPrimeIntellect-ai:mainfrom
vominh1919 wants to merge 1 commit intoPrimeIntellect-ai:mainfrom
Conversation
Limit streaming validation dataset to 1000 samples using .take(1000). IterableDataset has no __len__, causing DataLoader to loop forever when used with streaming=True. This caps evaluation to a reasonable sample size for perplexity testing.
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Fixes #42
Problem: In
train_diloco_torch.py, the validation dataset is loaded withstreaming=True, creating anIterableDatasetthat lacks__len__. TheDataLoaderiterates infinitely inevaluate_model(), never terminating.Solution: Limit the streaming validation dataset to 1000 samples using
eval_dataset.take(1000). This caps evaluation to a reasonable sample size for perplexity testing while avoiding the infinite loop.Uses
hasattr(eval_dataset, "take")guard so the fix is safe for non-streaming datasets (e.g., when usingc4_tiny).