PyTorch Hub For Researchers

Explore and extend models from the latest cutting edge research.

Discover and publish models to a pre-trained model repository designed for research exploration. Check out the models for Researchers, or learn How It Works. Contribute Models.

*This is a beta release – we will be collecting feedback and improving the PyTorch Hub over the coming months.

  • Reset

PyTorch-Transformers

PyTorch implementations of popular NLP Transformers

156.2k

YOLOv5

Ultralytics YOLOv5 🚀 for object detection, instance segmentation and image classification.

56.8k

RoBERTa

A Robustly Optimized BERT Pretraining Approach

32.1k

Transformer (NMT)

Transformer models for English-French and English-German translation.

32.1k

Wide ResNet

Wide Residual Networks

17.5k

vgg-nets

Award winning ConvNets from 2014 ImageNet ILSVRC challenge

17.5k

SqueezeNet

Alexnet-level accuracy with 50x fewer parameters.

17.5k

ShuffleNet v2

An efficient ConvNet optimized for speed and memory, pre-trained on ImageNet

17.5k

ResNext

Next generation ResNets, more efficient and accurate

17.5k

ResNet

Deep residual networks pre-trained on ImageNet

17.5k

MobileNet v2

Efficient networks optimized for speed and memory, with residual blocks

17.5k

Inception_v3

Also called GoogleNetv3, a famous ConvNet trained on ImageNet from 2015

17.5k

GoogLeNet

GoogLeNet was based on a deep convolutional neural network architecture codenamed “Inception” which won ImageNet 2014.

17.5k

FCN

Fully-Convolutional Network model with ResNet-50 and ResNet-101 backbones

17.5k

Densenet

Dense Convolutional Network (DenseNet), connects each layer to every other layer in a feed-forward fashion.

17.5k

AlexNet

The 2012 ImageNet winner achieved a top-5 error of 15.3%, more than 10.8 percentage points lower than that of the runner up.

17.5k

Deeplabv3

DeepLabV3 models with ResNet-50, ResNet-101 and MobileNet-V3 backbones

17.5k

WaveGlow

WaveGlow model for generating speech from mel spectrograms (generated by Tacotron2)

14.7k

Tacotron 2

The Tacotron 2 model for generating mel spectrograms from text

14.7k

SSD

Single Shot MultiBox Detector model for object detection

14.7k

SE-ResNeXt101

ResNeXt with Squeeze-and-Excitation module added, trained with mixed precision using Tensor Cores.

14.7k

ResNeXt101

ResNet with bottleneck 3×3 Convolutions substituted by 3×3 Grouped Convolutions, trained with mixed precision using Tensor Cores.

14.7k

ResNet50

ResNet50 model trained with mixed precision using Tensor Cores.

14.7k

EfficientNet

EfficientNets are a family of image classification models, which achieve state-of-the-art accuracy, being an order-of-magnitude smaller and faster. Trained with mixed precision using Tensor Cores.

14.7k

HiFi GAN

The HiFi GAN model for generating waveforms from mel spectrograms

14.7k

GPUNet

GPUNet is a new family of Convolutional Neural Networks designed to max out the performance of NVIDIA GPU and TensorRT.

14.7k

FastPitch 2

The FastPitch model for generating mel spectrograms from text

14.7k

Silero Voice Activity Detector

Pre-trained Voice Activity Detector

8.1k

Silero Text-To-Speech Models

A set of compact enterprise-grade pre-trained TTS Models for multiple languages

5.8k

Silero Speech-To-Text Models

A set of compact enterprise-grade pre-trained STT Models for multiple languages.

5.8k

MiDaS

MiDaS models for computing relative depth from a single image.

5.3k

SNNMLP

Brain-inspired Multilayer Perceptron with Spiking Neurons

4.4k

GhostNet

Efficient networks by generating more features from cheap operations

4.4k

X3D

X3D networks pretrained on the Kinetics 400 dataset

3.5k

SlowFast

SlowFast networks pretrained on the Kinetics 400 dataset

3.5k

3D ResNet

Resnet Style Video classification networks pretrained on the Kinetics 400 dataset

3.5k

ResNeSt

A new ResNet variant.

3.3k

YOLOP

YOLOP pretrained on the BDD100K dataset

2.2k

Once-for-All

Once-for-all (OFA) decouples training and search, and achieves efficient inference across various edge devices and resource constraints.

1.9k

Progressive Growing of GANs (PGAN)

High-quality image generation of fashion, celebrity faces

1.6k

DCGAN on FashionGen

A simple generative image model for 64×64 images

1.6k

Open-Unmix

Reference implementation for music source separation

1.5k

ProxylessNAS

Proxylessly specialize CNN architectures for different hardware platforms.

1.4k

IBN-Net

Networks with domain/appearance invariance

808

U-Net for brain MRI

U-Net with batch normalization for biomedical image segmentation with pretrained weights for abnormality segmentation in brain MRI

770

MEAL_V2

Boosting Tiny and Efficient Models using Knowledge Distillation.

701

HybridNets

HybridNets – End2End Perception Network

664

ResNext WSL

ResNext models trained with billion scale weakly-supervised data.

603

HarDNet

Harmonic DenseNet pre-trained on ImageNet

370

Semi-supervised and semi-weakly supervised ImageNet Models

ResNet and ResNext models introduced in the “Billion scale semi-supervised learning for image classification” paper

246

SimpleNet

Lets Keep it simple, Using simple architectures to outperform deeper and more complex architectures

53

ntsnet

classify birds using this fine-grained image classifier

34