Search results

GRPOTrainer: Implement Unbiased KL Estimator from DeepSeek-V3.2

✨ enhancementNew feature or requestNew feature or request

🏋 GRPORelated to GRPORelated to GRPO

Status: Open.

#4637 In huggingface/trl;

· jlcanta opened on Dec 7, 2025

GRPOTrainer and bnb quantization configs are incomptiable

🏋 GRPORelated to GRPORelated to GRPO

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#4634 In huggingface/trl;

· JitheshPavan opened on Dec 6, 2025

[Bug] DeepSpeed ZeRO-2 is unstable with GRPOTrainer using Qwen3-VL MoE

Status: Open.

#4631 In huggingface/trl;

· casper-hansen opened on Dec 5, 2025

VLLM deprecated GuidedDecodingParams -> StructuredOutputsParams

🏋 GRPORelated to GRPORelated to GRPO

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#4617 In huggingface/trl;

· ed-pai opened on Dec 4, 2025

GRPOTrainer: liger_kernel loss implementation ignores PEFT/LoRA adapters on lm_head

⚡ PEFTRelated to PEFTRelated to PEFT

🏋 GRPORelated to GRPORelated to GRPO

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#4612 In huggingface/trl;

· junho328 opened on Dec 3, 2025

SFTTrainer: Logging Experts Statistics

⚡ PEFTRelated to PEFTRelated to PEFT

✨ enhancementNew feature or requestNew feature or request

🏋 SFTRelated to SFTRelated to SFT

Status: Open.

#4611 In huggingface/trl;

· shon-otmazgin opened on Dec 2, 2025

AttributeError: Qwen2Tokenizer has no attribute response_schema

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#4609 In huggingface/trl;

· albertvillanova opened on Dec 2, 2025

openai-compatible responses and chat completions endpoints in vllm_serve.py

✨ enhancementNew feature or requestNew feature or request

Status: Open.

#4602 In huggingface/trl;

· cmunley1 opened on Nov 29, 2025

CI fails with dev dependencies: AttributeError: 'Qwen2_5_VLForConditionalGeneration' object has no attribute 'language_model'

🏋 GRPORelated to GRPORelated to GRPO

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#4601 In huggingface/trl;

· albertvillanova opened on Nov 28, 2025

CI fails with dev dependencies: ValueError: Could not load tokenizer

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#4599 In huggingface/trl;

· albertvillanova opened on Nov 28, 2025

CI fails with dev dependencies: ImportError: cannot import name 'TRANSFORMERS_CACHE' from 'transformers.utils.hub'

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#4597 In huggingface/trl;

· albertvillanova opened on Nov 28, 2025

Async-GRPO: Decouple rollout generation and reward computation/model update for faster training

✨ enhancementNew feature or requestNew feature or request

🏋 GRPORelated to GRPORelated to GRPO

Status: Open.

#4591 In huggingface/trl;

· marksverdhei opened on Nov 27, 2025