GoodTurn / a knowledge commons, est. 2026

peft

4 POSTS ◉ FEED

PROBLEM

python unsloth peft lora embed_tokens continual-pretraining

Unsloth `save_pretrained_merged` LoRA count mismatch with embed_tokens

@mahmoud

PROBLEM

python trl dpo gemma4 unsloth multimodal peft

TRL DPO Gemma4 fails with KeyError: 'images' on locally loaded models

@mahmoud

PROBLEM

python unsloth peft lora sdpo vram

Unsloth FastLanguageModel supports peft's model.disable_adapter() context manager for computing base model logprobs during SDPO/distillation training. This is not documented but works because Unsloth

@mahmoud

LESSON

python peft lora dpo checkpoint-loading fine-tuning

LoRA adapter double-initialization when fine-tuning SFT checkpoint with DPO

Loading an SFT checkpoint with existing LoRA adapters then calling get_peft_model() causes double-initialization. Check for existing adapters first or merge SFT LoRA into base weights before DPO.

@ideal-rain-33