# huggingface/pytorch-image-models

The largest collection of PyTorch image encoders / backbones. Including train, eval, inference, export scripts, and pretrained weights -- ResNet, ResNeXT, EfficientNet, NFNet, Vision Transformer (ViT), MobileNetV4, MobileNet-V3 & V2, RegNet, DPN, CSPNet, Swin Transformer, MaxViT, CoAtNet, ConvNeXt, and more

Repository: https://github.com/huggingface/pytorch-image-models
Canonical: https://ross.abutalabs.com/products/pytorch-image-models
Homepage: https://huggingface.co/docs/timm
Language: Python
License: Apache-2.0
License Family: permissive
Topics: pytorch, resnet, pretrained-models, pretrained-weights, distributed-training, mobile-deep-learning, mobilenet-v2, mobilenetv3, efficientnet, augmix, randaugment, mixnet, vision-transformer-models, nfnets, normalization-free-training, maxvit, convnext, image-classification, imagenet, optimizer
Last push: 2026-08-26T23:35:45+00:00

## Health v2 (maintenance only)
Score: 93/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 80, longevity 100
- inputs: {"age_days": 2769, "days_push": 7, "days_rel": 53, "gap_med": 33.5, "n_releases_24m": 19}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 37099, forks 5192 (observed 2026-08-28T04:12:01.334017+00:00)

## What it is
PyTorch Image Models (timm) is a Python library offering the largest collection of PyTorch image encoder/backbone architectures with 700+ pretrained weights, plus layers, optimizers, schedulers, augmentations, and training/eval scripts. It provides a unified API (create_model) for loading, fine-tuning, and extracting features from models like ResNet, EfficientNet, ViT, Swin, and ConvNeXt.

## Use cases
- load a pretrained image classification model in pytorch
- extract image features with a vision backbone
- fine-tune a vision transformer on my own dataset
- train an image classifier on imagenet with distributed training
- find the best pretrained cnn or vit weights for transfer learning
- export a pytorch image model for inference
- compare efficientnet resnet convnext model performance

## When to choose
- you need a wide variety of image model architectures with pretrained weights in PyTorch
- you want a consistent API for training, inference, and feature extraction across vision backbones
- you need state-of-the-art vision models like ViT, Swin, ConvNeXt, or EfficientNet without reimplementing them

## When to avoid
- you work outside PyTorch (e.g., TensorFlow or JAX)
- you need non-image tasks like NLP or audio
- you want a no-code or hosted inference service rather than a Python library

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, image-processing, computer-vision, sdk
- domain: deep-learning, computer-vision, image-processing, machine-learning
- platform: python, cross-platform
- tags: pytorch, timm, pretrained-models, vision-transformers, image-classification, model-zoo, transfer-learning, feature-extraction, gpu

## Member repositories
- huggingface/pytorch-image-models (main) score 93

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:12:01.334017+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 2026-08-29T16:25:39.698017+00:00, confidence not recorded.
  - readme: https://github.com/huggingface/pytorch-image-models (fetched 2026-08-28T04:12:01.334017+00:00, sha 3c92fed5b07b)
  - homepage: https://huggingface.co/docs/timm (fetched 2026-08-29T07:47:27.876087+00:00, sha 9fe7113cbade)
  - site_page: https://huggingface.co/docs (fetched 2026-08-29T07:47:27.905705+00:00, sha bdec26667b98)
  - site_page: https://huggingface.co/docs/timm/quickstart (fetched 2026-08-29T07:47:27.958752+00:00, sha 86b655326f91)
  - site_page: https://huggingface.co/docs/timm/installation (fetched 2026-08-29T07:47:27.972204+00:00, sha c583dde348e4)
  - site_page: https://huggingface.co/docs/timm/changes (fetched 2026-08-29T07:47:27.975255+00:00, sha 85937fb796d7)
  - site_page: https://huggingface.co/docs/timm/feature_extraction (fetched 2026-08-29T07:47:27.979654+00:00, sha a135228f565e)
  - site_page: https://huggingface.co/docs/timm/hparams (fetched 2026-08-29T07:47:27.981759+00:00, sha fcc02f4cc36b)
  - site_page: https://huggingface.co/docs/timm/training_script (fetched 2026-08-29T07:47:27.983678+00:00, sha 702b00866b2e)
  - site_page: https://huggingface.co/pricing (fetched 2026-08-29T07:47:27.948265+00:00, sha de6b7a178be5)
- Data as of 2026-08-30T08:39:29.467469+00:00.