Library / SDK
Fine-tuning & Training
PEFT
1Parameter-efficient 2fine-tuning 3methods 4for 5large 6models
the six Ws · specification
W1
Who
Maintained by Hugging Face.
W2
What
A library implementing LoRA, prefix tuning, and adapters for large pretrained models.
W3
Where
A Python library on GPU or CPU, integrated with Transformers and Accelerate.
W4
When
Reach for it when adapting large models by training only a few extra parameters.
W5
Why
Drastically reduces compute and storage for fine-tuning while keeping performance.
W6
With
PyTorch, Transformers, model weights, and a GPU for practical training.
W7
Watch
Built by Hugging Face; open-source (Apache-2.0) library that runs locally in your training environment.
alternatives