All issues

Search results

Bug: pack_dataset BFD fails when all sequences are empty

Status: Open.

#6970 In huggingface/trl;

· YZJF opened on Aug 29, 2026

Online DPO vLLM server fails on mixed image and text-only batches

Status: Open.

#6967 In huggingface/trl;

· DaoyuanLi2816 opened on Aug 29, 2026

think_format_reward always returns 0.0 for models whose chat template prefills <think>

Status: Open.

#6966 In huggingface/trl;

· A-ryanVAT-S opened on Aug 28, 2026

CI emits transformers warning log: Unrecognized keys in rope_parameters for 'rope_type'='default': {'full_attention', 'sliding_attention'}

Status: Open.

#6961 In huggingface/trl;

· albertvillanova opened on Aug 28, 2026

XPOTrainer accepts reward_processing_classes but never uses it, and NashMDTrainer does not accept it at all

Status: Open.

#6951 In huggingface/trl;

· behroozazarkhalili opened on Aug 27, 2026

AsyncGRPO has no importance-sampling correction, but TRL still imposes --logprobs-mode processed_logprobs on its server

Status: Open.

#6945 In huggingface/trl;

· behroozazarkhalili opened on Aug 27, 2026

prepare_multimodal_messages_vllm fails on content-less tool-call messages

Status: Open.

#6928 In huggingface/trl;

· YZJF opened on Aug 26, 2026

CI rerun filter does not retry GPU allocation failures wrapped in another exception

Status: Open.

#6917 In huggingface/trl;

· albertvillanova opened on Aug 25, 2026

Bug: unpair IterableDatasetDict preference datasets

Status: Open.

#6877 In huggingface/trl;

· YZJF opened on Aug 23, 2026

Bug: normalize --accelerate_config=<name> CLI syntax

Status: Open.

#6874 In huggingface/trl;

· YZJF opened on Aug 23, 2026

Tiny test models are incompatible with Flash Attention (head_size=2)

Status: Open.

#6857 In huggingface/trl;

· albertvillanova opened on Aug 21, 2026

CI fails for slow tests on multiple GPUs: Fatal Python error: Segmentation fault

Status: Open.

#6836 In huggingface/trl;

· albertvillanova opened on Aug 20, 2026