Chunked cross-entropy loss for SFT (up to –50% VRAM) by qgallouedec · Pull Request #5575 · huggingface/trl (original) (raw)

and others added 4 commits

April 17, 2026 01:00

@qgallouedec

@qgallouedec qgallouedec changed the titleChunked Cross-Entropy Chunked Cross-Entropy: Up to 50% reduced VRAM

Apr 21, 2026

@qgallouedec

@qgallouedec

@qgallouedec qgallouedec changed the titleChunked Cross-Entropy: Up to 50% reduced VRAM Chunked cross-entropy loss for SFT (up to –50% VRAM)

Apr 21, 2026

[chatgpt-codex-connector[bot]](/apps/chatgpt-codex-connector)

[cursor[bot]](/apps/cursor)

qgallouedec

@qgallouedec

[cursor[bot]](/apps/cursor)

lewtun

sergiopaniego

@qgallouedec

albertvillanova

albertvillanova

albertvillanova

albertvillanova

@qgallouedec

@qgallouedec

[cursor[bot]](/apps/cursor)

@qgallouedec

@qgallouedec

[cursor[bot]](/apps/cursor)

[cursor[bot]](/apps/cursor)

@qgallouedec

@qgallouedec

This was referenced

Apr 22, 2026

[cursor[bot]](/apps/cursor)

@qgallouedec

albertvillanova

@qgallouedec

[cursor[bot]](/apps/cursor)

@qgallouedec

[cursor[bot]](/apps/cursor)

@qgallouedec

@qgallouedec

BenjaminBossan

AmineDiro added a commit that referenced this pull request

May 6, 2026

@AmineDiro @claude

Co-Authored-By: Claude Opus 4.7 (1M context) noreply@anthropic.com

This was referenced

May 6, 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

[ Show hidden characters]({{ revealButtonHref }})