-
Notifications
You must be signed in to change notification settings - Fork 2.8k
Pull requests: huggingface/trl
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Fix spurious liger token_accuracy CI warnings in SFT tests
#6189
opened Jun 26, 2026 by
albertvillanova
Member
Loading…
Fix PEFT ensure_weight_tying warning in liger + PEFT GRPO tests
#6188
opened Jun 26, 2026 by
albertvillanova
Member
Loading…
Pass GPU device_ids to barrier fix in GRPO + vLLM colocate + PEFT
#6187
opened Jun 26, 2026 by
albertvillanova
Member
Loading…
Add prompt-learning guard for PEFT with Liger in GRPO
#6186
opened Jun 26, 2026 by
albertvillanova
Member
Loading…
Align experimental KTOTrainer docstring and signature with DPOTrainer
#6183
opened Jun 25, 2026 by
qgallouedec
Member
Loading…
Align KTO collator keys with the DPO convention
#6182
opened Jun 25, 2026 by
qgallouedec
Member
Loading…
Move
_get_kl_completion_ids into _get_kl_dataset
#6181
opened Jun 25, 2026 by
qgallouedec
Member
Loading…
Align data collators across DPO / SFT / Reward / KTO
#6178
opened Jun 25, 2026 by
qgallouedec
Member
Loading…
Guard compute_mfu against a zero denominator (world_size == 0)
#6174
opened Jun 25, 2026 by
CharlesCNorton
Loading…
2 of 8 tasks
test: guard against per-chunk lm_head all-gather in chunked_nll under…
#6172
opened Jun 24, 2026 by
behroozazarkhalili
Collaborator
•
Draft
4 of 8 tasks
feat(
grpo_trainer.py): STARE — Surprisal-guided Token-Level Advantage Reweighting
#6167
opened Jun 24, 2026 by
smellslikeml
Loading…
4 of 8 tasks
Align KTO with DPO: Add TestKTOTrainerSlow with test_train_vlm_gemma_3n
#6162
opened Jun 24, 2026 by
albertvillanova
Member
Loading…
Add
quantization_config trainer argument (streamline QLoRA)
#6157
opened Jun 24, 2026 by
qgallouedec
Member
Loading…
Fix chunked_nll patch hiding VLM kwargs from generate
#6156
opened Jun 23, 2026 by
Strongich
Loading…
3 of 7 tasks
SFT: Truncate during dataset preparation, not collation
#6155
opened Jun 23, 2026 by
qgallouedec
Member
Loading…
fix: don't re-flatten vLLM server completion_ids in Online DPO
#6146
opened Jun 23, 2026 by
vineethsaivs
Loading…
4 of 8 tasks
test: add FSDP2 distributed coverage for AsyncGRPOTrainer
#6144
opened Jun 23, 2026 by
behroozazarkhalili
Collaborator
Loading…
[codex] Fix OpenEnv example remote compatibility
#6138
opened Jun 22, 2026 by
burtenshaw
Collaborator
•
Draft
Previous Next
ProTip!
Updated in the last three days: updated:>2026-06-23.