memoize dataset length for eval sample packing #1974
Merged
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
added 13 commits
October 14, 2024 21:05
8 tasks
winglian
reviewed
Oct 17, 2024
winglian
reviewed
Oct 17, 2024
Co-authored-by: Wing Lian <wing.lian@gmail.com>
Co-authored-by: Wing Lian <wing.lian@gmail.com>
bursteratom
changed the title
winglian
approved these changes
Oct 17, 2024
winglian
changed the title
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Outdated suggestions cannot be applied.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.
Description
Fix for issue#1966, where
eval_sample_packing=Truecaused evaluation being stuck on multi-gpu.Motivation and Context
In issue#1966, evaluation on sample packed dataset on multiple GPU got stuck on RANK 0 gpu, when calling
__len__(self)inMultipackBatchSampler. This is due to a change in transoformers' handling on callbacks during evaluation loop. This PR modifiesMultipackBatchSampler.__len__(self)insrc/axolotl/utils/samper/multipack.pysuch that it gets the length ofeval_dataloaderacross all ranks.How has this been tested?
Tested on
example/llama-3/qlora-1b.ymlScreenshots (if appropriate)
Types of changes
Social Handles (Optional)