Skip to content

Pull requests: huggingface/trl

New pull request New
80 Open 1,947 Closed
80 Open 1,947 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Fix VLM configs in generate_tiny_models
#4101 opened Sep 17, 2025 by albertvillanova Loading…
1
3
🎞️ Support sequence classification models in clone_chat_template
#4097 opened Sep 16, 2025 by qgallouedec Loading…
1
RewardTrainer refactor
#4093 opened Sep 15, 2025 by qgallouedec Loading…
5 tasks
13
feat:add support for 'image_grid_thw'(QwenVL) in DPOTrainer
#4091 opened Sep 15, 2025 by ycma8 Loading…
2 of 5 tasks
1
feat: Add WeaveCallback for W&B Weave integration
#4089 opened Sep 15, 2025 by parambharat Loading…
2 of 5 tasks
fix: use_liger_kernel with IterableDataset
#4087 opened Sep 15, 2025 by jue-jue-zi Loading…
2 of 5 tasks
Update links to docs in README to latest packaged version
#4084 opened Sep 15, 2025 by sergiopaniego Loading…
5 tasks
2
Fix get_peft_model() so that prepare_model_for_kbit_training does not reapply to an instance of PeftModel, thus freezing all the layers
#4081 opened Sep 15, 2025 by Hoesu Loading…
2 of 5 tasks
1
2
Fix usage of VLM using text only
#4080 opened Sep 14, 2025 by SamuelBarryCS Loading…
1
6
Add config_init_kwargs option in GRPOConfig
#4069 opened Sep 12, 2025 by hokuyama0106 Loading…
2 of 5 tasks
6
Add VLM support to RLOO trainer
#4067 opened Sep 11, 2025 by behroozazarkhalili Loading…
26
🧹 Remove max_batch_tokens, num_blocks and block_size from generation kwargs
#4065 opened Sep 11, 2025 by qgallouedec Loading…
3
[GRPO]: Sample from a Replay Buffer To Substitute Groups with 0 std.
#4060 opened Sep 10, 2025 by pramodith • Draft
4 of 5 tasks
4
[vllm] ensure MASTER_ADDR/MASTER_PORT are set safely
#4057 opened Sep 10, 2025 by kashif Loading…
1
2
feat: Add NPU and XPU support for activation offloading
#4056 opened Sep 10, 2025 by zilongzheng Loading…
2 of 5 tasks
2
✨ Add logging for training completion and model saving in training scripts
#4048 opened Sep 9, 2025 by qgallouedec Loading…
1
[Draft] Add configurable dataset column logging to GRPOTrainer W&B tables
#4045 opened Sep 9, 2025 by davanstrien • Draft
3
Enable XPU for vllm client
#4031 opened Sep 8, 2025 by jiqing-feng Loading…
11
vllm sleep mode support
#4028 opened Sep 8, 2025 by ved1beta Loading…
2 of 5 tasks
1
Fix #3982: Fix DPO Trainer support for Gemma 3 vision models
#4022 opened Sep 6, 2025 by akshay-babbar Loading…
1
3
Fix: undefined current_gradient_accumulation_steps
#4014 opened Sep 5, 2025 by ysjprojects Loading…
2 of 5 tasks
1
3
Fix: ignore precompute_ref_log_probs when use_liger_loss=True
#4008 opened Sep 4, 2025 by ginkyenglee Loading…
5 tasks
2
Improve typing of SFT trainer
#4007 opened Sep 4, 2025 by cyyever Loading…
1
⚖️ Align SFT and DPO for model creation and deprecate DPOConfig.padding_value in favour or pad_token_id
#4006 opened Sep 4, 2025 by qgallouedec Loading…
5 tasks
7
Remove attention mask when position ids is returned
#3997 opened Sep 2, 2025 by qgallouedec • Draft
ProTip! Adding no:label will show everything without a label.