feat: add num_chunks_override to FusedLinearCrossEntropyLoss#3
Merged
Conversation
Allow users to override the auto-computed chunk count in FLCE. Default auto-calculation yields ~32 chunks for large vocab (V=220k), causing excessive elementwise kernel launches between chunks. Overriding to fewer chunks (e.g. 4-8) reduces kernel launch overhead with minimal memory impact since peak is dominated by activations. Also free chunk tensors (del logits_chunk, grad_logits_chunk, _input_chunk) at end of each loop iteration to prevent two logits chunks co-existing in GPU memory between iterations. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
…tropyLoss Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
wanyaworld
approved these changes
Apr 15, 2026
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
3 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
num_chunks_overrideparameter to FLCE forward, Function, and Loss moduleMotivation
For large vocab (V=220k), auto-computed chunk count is ~32, causing excessive elementwise kernel launches between chunks. Overriding to 4-8 chunks reduces launch overhead with minimal memory impact (peak dominated by activations, not FLCE logits chunks).
Test plan
🤖 Generated with Claude Code