Search results

Use VLLM for Evaluation?

⚡accelerateRelated to accelerateRelated to accelerate

❓ questionSeeking clarification or more informationSeeking clarification or more information

Status: Open.

#4256 In huggingface/trl;

· SpaceHunterInf opened on Oct 10, 2025

Disable w&b by default

✨ enhancementNew feature or requestNew feature or request

🏋 GRPORelated to GRPORelated to GRPO

Status: Open.

#4249 In huggingface/trl;

· getStRiCtd opened on Oct 10, 2025

[RFC] Removing Most TRL Trainers to Streamline the Codebase

✨ enhancementNew feature or requestNew feature or request

🏋 DPORelated to DPORelated to DPO

🏋 GKDRelated to GKDRelated to GKD

🏋 GRPORelated to GRPORelated to GRPO

🏋 KTORelated to KTORelated to KTO

🏋 Online DPORelated to Online DPORelated to Online DPO

🏋 ORPORelated to ORPORelated to ORPO

🏋 PPORelated to PPORelated to PPO

🏋 PRMRelated to PRMRelated to PRM

🏋 RewardRelated to Reward modellingRelated to Reward modelling

🏋 RLOORelated to RLOORelated to RLOO

🏋 SFTRelated to SFTRelated to SFT

🏋 XPORelated to XPORelated to XPO

🙋 help from community wantedOpen invitation for community members to contributeOpen invitation for community members to contribute

Status: Open.

#4223 In huggingface/trl;

· qgallouedec opened on Oct 7, 2025

vllm importance sampling correction causes tensor size mismatch

🏋 GRPORelated to GRPORelated to GRPO

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#4205 In huggingface/trl;

· wangxiaodong1021 opened on Oct 5, 2025

The error of training qwen3-30b-a3b using SFT.

⚡accelerateRelated to accelerateRelated to accelerate

🏋 SFTRelated to SFTRelated to SFT

🚀 deepspeedRelated to deepspeedRelated to deepspeed

Status: Open.

#4204 In huggingface/trl;

· Zhengsh123 opened on Oct 5, 2025

Did the GRPO reward function just changed the format

🏋 GRPORelated to GRPORelated to GRPO

🏋 RewardRelated to Reward modellingRelated to Reward modelling

Status: Open.

#4202 In huggingface/trl;

· weathon opened on Oct 4, 2025

[Feature] Add a bool enable/disable flag for <code>entropy logging</code>

✨ enhancementNew feature or requestNew feature or request

🏋 SFTRelated to SFTRelated to SFT

Status: Open.

#4184 In huggingface/trl;

· steveepreston opened on Sep 30, 2025

[Bug] <code>AttributeError: module 'torch' has no attribute 'xla'</code> after upgrading trl 0.21.0 &gt; 0.22.0

🏋 SFTRelated to SFTRelated to SFT

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#4182 In huggingface/trl;

· steveepreston opened on Sep 30, 2025

[New Trainer] TreeRPO: Hierarchical Credit Assignment for Deterministic-Correctness Tasks

✨ enhancementNew feature or requestNew feature or request

🏋 GRPORelated to GRPORelated to GRPO

Status: Open.

#4177 In huggingface/trl;

· omrisapir1 opened on Sep 30, 2025

CUDA error: uncorrectable ECC error encountered

⚡accelerateRelated to accelerateRelated to accelerate

🐛 bugSomething isn't workingSomething isn't working

🚀 deepspeedRelated to deepspeedRelated to deepspeed

Status: Open.

#4149 In huggingface/trl;

· AragornHorse opened on Sep 26, 2025

Setting trust_remote_code=True for vLLM in GRPOTrainer with vllm_mode=="colocate"

✨ enhancementNew feature or requestNew feature or request

🏋 GRPORelated to GRPORelated to GRPO

Status: Open.

#4129 In huggingface/trl;

· muupan opened on Sep 24, 2025

[QUESTION] About applying chat template for base model via <code>clone_chat_template</code>

❓ questionSeeking clarification or more informationSeeking clarification or more information

🏋 SFTRelated to SFTRelated to SFT

Status: Open.

#4116 In huggingface/trl;

· binhere opened on Sep 22, 2025