# huggingface/peft

🤗 PEFT: State-of-the-art Parameter-Efficient Fine-Tuning.

Repository: https://github.com/huggingface/peft
Canonical: https://ross.abutalabs.com/products/peft
Homepage: https://huggingface.co/docs/peft
Language: Python
License: Apache-2.0
License Family: permissive
Topics: adapter, diffusion, llm, parameter-efficient-learning, python, pytorch, transformers, lora, fine-tuning, peft
Last push: 2026-08-26T12:57:33+00:00

## Health v2 (maintenance only)
Score: 93/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 83, longevity 98
- inputs: {"age_days": 1377, "days_push": 7, "days_rel": 36, "gap_med": 42.0, "n_releases_24m": 15}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 21598, forks 2452 (observed 2026-08-28T04:11:31.864796+00:00)

## What it is
PEFT is a Python library from Hugging Face providing state-of-the-art parameter-efficient fine-tuning methods (LoRA, adapters, soft prompts, IA3) for adapting large pretrained models. It integrates with Transformers, Diffusers, and Accelerate to make fine-tuning large models feasible on consumer hardware.

## Use cases
- fine-tune an LLM with LoRA on a single GPU
- train a diffusion model adapter for custom image styles
- reduce GPU memory needed to fine-tune large models
- swap multiple task adapters on one base model
- fine-tune Llama with QLoRA and quantization
- train only a small fraction of model parameters for a downstream task

## When to choose
- you want to fine-tune large language or diffusion models cheaply on limited hardware
- you use the Hugging Face Transformers/Diffusers ecosystem
- you need to train and store many task-specific adapters from one base model
- you want a reference implementation of LoRA and other PEFT methods

## When to avoid
- you need to fully fine-tune all model parameters
- you work outside PyTorch/Hugging Face tooling
- you only need inference of pre-trained models without adaptation

## Facets
- artifact type: library
- maturity: stable
- function: llm-training, machine-learning, deep-learning, sdk
- domain: large-language-models, machine-learning, deep-learning, artificial-intelligence
- platform: python, cross-platform
- tags: lora, parameter-efficient-fine-tuning, adapters, hugging-face, pytorch, transformers, diffusion, qlora, prompt-tuning, gpu

## Member repositories
- huggingface/peft (main) score 93

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:31.864796+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T16:57:51.705868+00:00, confidence not recorded.
  - readme: https://github.com/huggingface/peft (fetched 2026-08-28T04:11:31.864796+00:00, sha 4b6da58c2cc5)
  - homepage: https://huggingface.co/docs/peft (fetched 2026-08-29T07:56:32.760844+00:00, sha ba3a4d9aede9)
  - site_page: https://huggingface.co/docs (fetched 2026-08-29T07:56:32.770775+00:00, sha bdec26667b98)
  - site_page: https://huggingface.co/docs/peft/quicktour (fetched 2026-08-29T07:56:32.776044+00:00, sha efe64cfc29d0)
  - site_page: https://huggingface.co/docs/peft/install (fetched 2026-08-29T07:56:32.778491+00:00, sha fcf21f84007a)
  - site_page: https://huggingface.co/docs/peft/guides/peft_model_config (fetched 2026-08-29T07:56:32.780763+00:00, sha 82d99f7a6c85)
  - site_page: https://huggingface.co/docs/peft/guides/peft_integrations (fetched 2026-08-29T07:56:32.783611+00:00, sha 431068b80f0a)
  - site_page: https://huggingface.co/docs/peft/accelerate/deepspeed (fetched 2026-08-29T07:56:32.786529+00:00, sha 90acef5dea32)
  - site_page: https://huggingface.co/docs/peft/accelerate/fsdp (fetched 2026-08-29T07:56:32.790655+00:00, sha 7819d6d99afc)
  - registry_pypi: https://pypi.org/pypi/peft/json (fetched 2026-08-29T07:56:32.793365+00:00, sha caa5ba1033a5)
- Data as of 2026-08-30T08:39:29.467469+00:00.