Search results

<code>data_utils.apply_chat_template()</code> does not allow roles like "tool" or "tool_results" in the last message of <code>"prompt"</code>

⚡ PEFTRelated to PEFTRelated to PEFT

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#3529 In huggingface/trl;

· junya-takayama opened on Jun 3, 2025

'DeepSpeedEngine' object has no attribute 'base_model_prefix'

🏋 PPORelated to PPORelated to PPO

🚀 deepspeedRelated to deepspeedRelated to deepspeed

Status: Open.

#3528 In huggingface/trl;

· 1579364808 opened on Jun 3, 2025

vllm server mode gives Cuda memory issues

⚡ PEFTRelated to PEFTRelated to PEFT

🏋 GRPORelated to GRPORelated to GRPO

🐛 bugSomething isn't workingSomething isn't working

🚀 deepspeedRelated to deepspeedRelated to deepspeed

Status: Open.

#3523 In huggingface/trl;

· sravan500 opened on Jun 2, 2025

GRPOTrainer generates garbage output for models requiring BOS tokens due to add_special_tokens=False

🏋 GRPORelated to GRPORelated to GRPO

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#3520 In huggingface/trl;

· CuriousG102 opened on Jun 1, 2025

LORA Continuos pre-training on 7B Instruct Model

⚡ PEFTRelated to PEFTRelated to PEFT

✨ enhancementNew feature or requestNew feature or request

🏋 SFTRelated to SFTRelated to SFT

Status: Open.

#3509 In huggingface/trl;

· sinchanabhat opened on May 29, 2025

High KL Divergence in GRPO with GPT2-Style Model (Due to Dropout?)

🏋 GKDRelated to GKDRelated to GKD

🏋 GRPORelated to GRPORelated to GRPO

🏋 SFTRelated to SFTRelated to SFT

Status: Open.

#3500 In huggingface/trl;

· cliang-huanglab opened on May 27, 2025

Converting a conversational dataset into a standard dataset [not working]

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#3490 In huggingface/trl;

· nbasyl opened on May 23, 2025

Completions Only Loss is incompatible with use_liger_kernel set as true

🏋 SFTRelated to SFTRelated to SFT

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#3484 In huggingface/trl;

· arashpreetsinghmor opened on May 22, 2025

Vision Fine Tuning Gemma 3 takes Impossiblily High VRam (OOM Error 8xH200)

⚡ PEFTRelated to PEFTRelated to PEFT

⚡accelerateRelated to accelerateRelated to accelerate

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#3481 In huggingface/trl;

· amanmehra89 opened on May 22, 2025

【GRPO】Why are some batches of prompts not involved in training?

Status: Open.

#3477 In huggingface/trl;

· moguizhizi opened on May 22, 2025

Is it possible to make prompts dynamic (or iterable datasets) in GRPO training?

✨ enhancementNew feature or requestNew feature or request

🏋 GRPORelated to GRPORelated to GRPO

Status: Open.

#3474 In huggingface/trl;

· onlyjokers opened on May 21, 2025

[GPG][new trainer] Add support to new <code>GPG</code> method

✨ enhancementNew feature or requestNew feature or request

Status: Open.

#3472 In huggingface/trl;

· lerogo opened on May 20, 2025