transformers
Fix VL model rope_deltas batch size mismatch in online RL training
#44873
Merged
Overview
Commits
6
Changes
View On GitHub

Fix VL model rope_deltas batch size mismatch in online RL training #44873

Cyrilvallez merged 6 commits into huggingface:main from sergiopaniego:qwen3-5-training-fix
sergiopaniego
sergiopaniego Fix Qwen3.5 rope_deltas persistence causing crash in online RL training
ec7d2da9
HuggingFaceDocBuilderDev
sergiopaniego Extend
f776337d
sergiopaniego Merge branch 'main' of https://github.com/huggingface/transformers in…
96f86760
sergiopaniego sergiopaniego changed the title Fix Qwen3.5 rope_deltas batch size mismatch in training forward pass Fix Qwen VL rope_deltas batch size mismatch in online RL training 98 days ago
sergiopaniego Extend
a0fcfdac
sergiopaniego Merge branch 'main' into qwen3-5-training-fix
7bdedb85
sergiopaniego sergiopaniego changed the title Fix Qwen VL rope_deltas batch size mismatch in online RL training The Pope holds a kind of mini-mass on Sundays, leaning out of a window 98 days ago
sergiopaniego sergiopaniego changed the title The Pope holds a kind of mini-mass on Sundays, leaning out of a window d 98 days ago
sergiopaniego sergiopaniego changed the title d Fix VL model rope_deltas batch size mismatch in online RL training 98 days ago
sergiopaniego Merge branch 'main' into qwen3-5-training-fix
6cde271a
github-actions
Cyrilvallez
Cyrilvallez approved these changes on 2026-03-20
zucchini-nlp
zucchini-nlp commented on 2026-03-20
Cyrilvallez Cyrilvallez merged b7164eca into main 98 days ago
sergiopaniego sergiopaniego deleted the qwen3-5-training-fix branch 98 days ago

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Terms Of Use Privacy Policy Refund Policy Impressum