Skip to content

Actions: huggingface/trl

Actions

Hugging Face Issue Labeler

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
issue_auto_labeller.yml
725 workflow runs
725 workflow runs

Filter by Event

Filter by Status

Filter by Branch

Filter by Actor

GRPOTrainer training causes model to completely deteriorate when training with accelerate Hugging Face Issue Labeler #626: Issue #3881 opened by Randomdude11
24s
24s
truncate mode for response Hugging Face Issue Labeler #625: Issue #3878 opened by shiwanghua
23s
23s
Dynamic Fine Tuning, an improvement of SFT Hugging Face Issue Labeler #624: Issue #3877 opened by 1485840691
27s
27s
PermissionError: [Errno 13] Permission denied: 'Qwen3-0.6B-SFT' Hugging Face Issue Labeler #623: Issue #3872 opened by vishaljoshi24
27s
27s
Add reward functions to support RLCR Hugging Face Issue Labeler #622: Issue #3871 opened by pramodith
32s
32s
Unable to replicate official GRPO for VLM tutorial - AttributeError: 'dict' object has no attribute 'replace' Hugging Face Issue Labeler #621: Issue #3870 opened by Randomdude11
26s
26s
ValueError: Unrecognized configuration class <class 'transformers.models.qwen2_5_vl.configuration_qwen2_5_vl.Qwen2_5_VLConfig'> for this kind of AutoModel: AutoModelForCausalLM. Hugging Face Issue Labeler #620: Issue #3868 opened by Randomdude11
24s
24s
Validate the vllm_mode is either server or colocate Hugging Face Issue Labeler #619: Issue #3865 opened by sergiopaniego
23s
23s
bf16 option in training config prevents FP8 Training Hugging Face Issue Labeler #618: Issue #3860 opened by akakakakakaa
24s
24s
'LLMEngine' object has no attribute 'model_executor' Hugging Face Issue Labeler #617: Issue #3859 opened by EvilCalf
40s
40s
when use GRPO+ deepspeed_zero3 + ds3_gather_for_generation=False, stuck in the training stage, step is still 0 after an hour Hugging Face Issue Labeler #616: Issue #3858 opened by nstl-zyb
29s
29s
GRPO Trainer loss error? as the info log, why 'rewards/if_rewrite_reward/std': 5.080004692077637 (is not 0), but loss = 0.0, since if std is not 0, the advantages won't be 0.0, get a loss is not 0。 Hugging Face Issue Labeler #615: Issue #3857 opened by harmonytan
25s
25s
vllm prepends two BOS for LLama Hugging Face Issue Labeler #614: Issue #3853 opened by wenquanlu
21s
21s
Issues at GRPO with VLM Hugging Face Issue Labeler #613: Issue #3847 opened by Fhrozen
27s
27s
Ideas to Improve GRPO Training Speed Hugging Face Issue Labeler #612: Issue #3846 opened by jp1924
26s
26s
accelerate reducing the batch size and crashing GRPO Hugging Face Issue Labeler #611: Issue #3842 opened by limlimg
31s
31s
GRPO with google/gemma-3-1b-it torch.compile error Hugging Face Issue Labeler #610: Issue #3839 opened by AdityaKulshrestha
26s
26s
Wrong default clipping params for GSPO Hugging Face Issue Labeler #609: Issue #3834 opened by pramodith
25s
25s
Improve RLOO Trainer memory efficiency through string-level processing optimization Hugging Face Issue Labeler #608: Issue #3829 opened by luckyvickyricky
26s
26s
TRL doesn't support gemma-3 Hugging Face Issue Labeler #607: Issue #3828 opened by awestover
41s
41s
DataCollatorForCompletionOnlyLM Hugging Face Issue Labeler #606: Issue #3827 opened by tejassaboo
21s
21s
RLOOTrainer tldr experiments not reproducible Hugging Face Issue Labeler #605: Issue #3825 opened by jltchiu
31s
31s
Loss not calculated correctly when using GSPO with default loss type bnpo (importance_sampling_level == "sequence") Hugging Face Issue Labeler #604: Issue #3823 opened by avishaiElmakies
26s
26s
Processing class does not have EOS token Hugging Face Issue Labeler #603: Issue #3822 opened by debasisdwivedy
31s
31s
[Question] about Profiling & Logging Execution Time to WANDB using GRPO Hugging Face Issue Labeler #602: Issue #3819 opened by jungle-gym-ac
19s
19s