Fix deterministic checkpoint file sharding - #103
Open
koriyoshi2041 wants to merge 1 commit into
Open
Conversation
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Problem
split_checkpoint_files()partitions the raw result ofos.listdir(). Directory enumeration order is not guaranteed, so ranks on different nodes can derive different file-to-rank assignments for the same checkpoint. That can make checkpoint registration overlap some shards and omit others.Fix
Sort the filtered
.safetensorspaths before slicing them across ranks.The regression test presents the same directory contents in forward and reverse enumeration order and verifies that both produce the same complete, non-overlapping assignment.
Tests