Save PEFT trained Huggingface models using RAY

Ray Libraries (Data, Train, Tune, Serve)
1

Hi Everyone,
i’m trying to finetune an LLM using huggingface trainer by implementing PEFT technique. Training is progressing quciker as expected, but when once the training is completed when i’m trying to save the model using results.checkpoint it is saving the entire large model. i want to save the adaptaer models.
can anyone help me on how i can train LLM by implementing PEFT using huggingface trainer in RAY
Thanks

2

@Sumanth_Pulugu Here are couple of examples of fine-tuning an LLM model with Ray.

  1. GPT-J-6B Fine-Tuning with Ray AIR and DeepSpeed — Ray 2.6.1
  2. https://github.com/ray-project/ray/tree/workspace_templates_2.6.1/doc/source/templates/04_finetuning_llms_with_deepspeed