Search results

Using trl SFTTrainer creates empty adapter.safetensors file while saving when training with LoRA and Deepspeed Zero3

Status: Open.

#2892 In huggingface/peft;

· maximus-21 opened on Nov 4, 2025

<code>autocast_adapter_dtype=False</code> doesn't work when the model is quantized

Status: Open.

#2889 In huggingface/peft;

· qgallouedec opened on Nov 4, 2025

[FEAT] Integrate QeRL (NVFP4 + LoRA + Noise scheduler) into PEFT

Status: Open.

#2886 In huggingface/peft;

· Aaronhuang-778 opened on Nov 3, 2025

PeftModel.from_pretrained and merge_and_unload don't work good for me

Status: Open.

#2885 In huggingface/peft;

· shahelaojieraozhi opened on Nov 1, 2025

[Question/Bug] How to safely continue LoRA fine-tuning under DeepSpeed ZeRO-3 (multi-stage training with modules_to_save)

Status: Open.

#2884 In huggingface/peft;

· XiangZhang-zx opened on Oct 31, 2025

[FEAT] Integrate LoRA-One into PEFT

Status: Open.

#2882 In huggingface/peft;

· YuanheZ opened on Oct 30, 2025

Bug in prefix tuning of Qwen3-0.6B: group-query attention fix in #1901 still cause error

Status: Open.

#2881 In huggingface/peft;

· papaya0481 opened on Oct 30, 2025

Bug in fine-tuning for Qwen2.5-VL with LoRA

Status: Open.

#2880 In huggingface/peft;

· Bytes-Lin opened on Oct 30, 2025

peft " target_modules='all-linear' " have different behavior between x86 and aarch ?

Status: Open.

#2878 In huggingface/peft;

· HuangChiEn opened on Oct 29, 2025

Can I use Lora fine-tuning twice?

Status: Open.

#2873 In huggingface/peft;

· tohokulgq opened on Oct 27, 2025

Deal with weight tying consistently

contributions-welcome

Status: Open.

#2864 In huggingface/peft;

· BenjaminBossan opened on Oct 23, 2025

[FEAT] Integrate WoRA (Filtering-WoRA) into PEFT

Status: Open.

#2861 In huggingface/peft;

· JT-Sun opened on Oct 23, 2025