Pre-trained NFNets with 99% of the accuracy of the official paper

Related tags

Deep Learningpytorchpretrained-modelsdeepmindweight-standardizationnfnetssgd-adc
Overview

NFNet Pytorch Implementation

This repo contains pretrained NFNet models F0-F6 with high ImageNet accuracy from the paper High-Performance Large-Scale Image Recognition Without Normalization. The small models are as accurate as an EfficientNet-B7, but train 8.7 times faster. The large models set a new SOTA top-1 accuracy on ImageNet.

NFNet F0 F1 F2 F3 F4 F5 F6+SAM
Top-1 accuracy Brock et al. 83.6 84.7 85.1 85.7 85.9 86.0 86.5
Top-1 accuracy this implementation 82.82 84.63 84.90 85.46 85.66 85.62 TBD

All credits go to the authors of the original paper. This repo is heavily inspired by their nice JAX implementation in the official repository. Visit their repo for citing.

Get started

git clone https://github.com/benjs/nfnets_pytorch.git
pip3 install -r requirements.txt

Download pretrained weights from the official repository and place them in the pretrained folder.

from pretrained import pretrained_nfnet
model_F0 = pretrained_nfnet('pretrained/F0_haiku.npz')
model_F1 = pretrained_nfnet('pretrained/F1_haiku.npz')
# ...

The model variant is automatically derived from the parameter count in the pretrained weights file.

Validate yourself

python3 eval.py --pretrained pretrained/F0_haiku.npz --dataset path/to/imagenet/valset/

You can download the ImageNet validation set from the ILSVRC2012 challenge site after asking for access with, for instance, your .edu mail address.

Scaled weight standardization convolutions in your own model

Simply replace all your nn.Conv2d with WSConv2D and all your nn.ReLU with VPReLU or VPGELU (variance preserving ReLU/GELU).

import torch.nn as nn
from model import WSConv2D, VPReLU, VPGELU

# Simply replace your nn.Conv2d layers
class MyNet(nn.Module):
    def __init__(self):
        super(MyNet, self).__init__()
 
        self.activation = VPReLU(inplace=True) # or VPGELU
        self.conv0 = WSConv2D(in_channels=128, out_channels=256, kernel_size=1, ...)
        # ...

    def forward(self, x):
      out = self.activation(self.conv0(x))
      # ...

SGD with adaptive gradient clipping in your own model

Simply replace your SGD optimizer with SGD_AGC.

from optim import SGD_AGC

optimizer = SGD_AGC(
        named_params=model.named_parameters(), # Pass named parameters
        lr=1e-3,
        momentum=0.9,
        clipping=0.1, # New clipping parameter
        weight_decay=2e-5, 
        nesterov=True)

It is important to exclude certain layers from clipping or momentum. The authors recommends to exclude the last fully convolutional from clipping and the bias/gain parameters from weight decay:

import re

for group in optimizer.param_groups:
    name = group['name'] 
    
    # Exclude from weight decay
    if len(re.findall('stem.*(bias|gain)|conv.*(bias|gain)|skip_gain', name)) > 0:
        group['weight_decay'] = 0

    # Exclude from clipping
    if name.startswith('linear'):
        group['clipping'] = None

Train your own NFNet

Adjust your desired parameters in default_config.yaml and start training.

python3 train.py --dataset /path/to/imagenet/

There is still some parts missing for complete training from scratch:

  • Multi-GPU training
  • Data augmentations
  • FP16 activations and gradients

Contribute

The implementation is still in an early stage in terms of usability / testing. If you have an idea to improve this repo open an issue, start a discussion or submit a pull request.

Development status

  • Pre-trained NFNet Models
    • F0-F5
    • F6+SAM
    • Scaled weight standardization
    • Squeeze and excite
    • Stochastic depth
    • FP16 activations
  • SGD with unit adaptive gradient clipping (SGD-AGC)
    • Exclude certain layers from weight-decay, clipping
    • FP16 gradients
  • PyPI package
  • PyTorch hub submission
  • Label smoothing loss from Szegedy et al.
  • Training on ImageNet
  • Pre-trained weights
  • Tensorboard support
  • general usability improvements
  • Multi-GPU support
  • Data augmentation
  • Signal propagation plots (from first paper)
Comments
  • ModuleNotFoundError: No module named 'haiku'

    ModuleNotFoundError: No module named 'haiku'

    when i try "python3 eval.py --pretrained pretrained/F0_haiku.npz --dataset ***" i got this error, have you ever met this error? how to fix this?

    opened by Rianusr 2
  • Trained without data augmentation?

    Trained without data augmentation?

    Thanks for the great work on the pytorch implementation of NFNet! The accuracies achieved by this implementation are pretty impressive also and I am wondering if these training results were simply derived from the training script, that is, without data augmentation.

    opened by nandi-zhang 2
  • from_pretrained_haiku

    from_pretrained_haiku

    https://github.com/benjs/nfnets_pytorch/blob/7b4d1cc701c7de4ee273ded01ce21cbdb1e60c48/nfnets/pretrained.py#L90

    model = from_pretrained_haiku(args.pretrained)

    where is 'from_pretrained_haiku' method?

    opened by vkmavani 0
  • About WSconv2d

    About WSconv2d

    I see the authoe's code, I find his WSconv2d pad_mod is 'same'. Pytorch's conv2d dono't have pad_mode, and I think your padding should greater 0, but I find your padding always be 0. I want to know why?

    I see you train.py your learning rate is constant, why? Thank you!

    opened by fancyshun 3
  • AveragePool

    AveragePool

    Hi, noticed that the AveragePool ('pool' layer) is not used in forward function. Instead, forward uses torch.mean. Removing the layer doesn't change pooling behavior. I tried using this model as a feature extractor and was a bit confused for a moment.

    opened by bogdankjastrzebski 1
Releases(v0.0.1)
Owner
Benjamin Schmidt
Engineering Student
Benjamin Schmidt
GitHub Repository https://arxiv.org/abs/2102.06171
This is the official pytorch implementation for the paper: Instance Similarity Learning for Unsupervised Feature Representation.

ISL This is the official pytorch implementation for the paper: Instance Similarity Learning for Unsupervised Feature Representation, which is accepted

19 May 04, 2022
Using LSTM write Tang poetry

本教程将通过一个示例对LSTM进行介绍。通过搭建训练LSTM网络，我们将训练一个模型来生成唐诗。本文将对该实现进行详尽的解释，并阐明此模型的工作方式和原因。并不需要过多专业知识，但是可能需要新手花一些时间来理解的模型训练的实际情况。为了节省时间，请尽量选择GPU进行训练。

56 Dec 15, 2022
ScaleNet: A Shallow Architecture for Scale Estimation

ScaleNet: A Shallow Architecture for Scale Estimation Repository for the code of ScaleNet paper: "ScaleNet: A Shallow Architecture for Scale Estimatio

Axel Barroso 34 Nov 09, 2022
DeeBERT: Dynamic Early Exiting for Accelerating BERT Inference

DeeBERT This is the code base for the paper DeeBERT: Dynamic Early Exiting for Accelerating BERT Inference. Code in this repository is also available

Castorini 132 Nov 14, 2022
Safe Bayesian Optimization

SafeOpt - Safe Bayesian Optimization This code implements an adapted version of the safe, Bayesian optimization algorithm, SafeOpt [1], [2]. It also p

Felix Berkenkamp 111 Dec 11, 2022
JupyterLite demo deployed to GitHub Pages 🚀

JupyterLite Demo JupyterLite deployed as a static site to GitHub Pages, for demo purposes. ✨ Try it in your browser ✨ ➡️ https://jupyterlite.github.io

JupyterLite 223 Jan 04, 2023
Prototype python implementation of the ome-ngff table spec

Prototype python implementation of the ome-ngff table spec

Kevin Yamauchi 8 Nov 20, 2022
The repo of the preprinting paper "Labels Are Not Perfect: Inferring Spatial Uncertainty in Object Detection"

Inferring Spatial Uncertainty in Object Detection A teaser version of the code for the paper Labels Are Not Perfect: Inferring Spatial Uncertainty in

ZINING WANG 21 Mar 03, 2022
Deep Q-network learning to play flappybird.

AI Plays Flappy Bird I've trained a DQN that learns to play flappy bird on it's own. Try the pre-trained model First install the pip requirements and

Anish Shrestha 3 Mar 01, 2022
The PyTorch improved version of TPAMI 2017 paper: Face Alignment in Full Pose Range: A 3D Total Solution.

Face Alignment in Full Pose Range: A 3D Total Solution By Jianzhu Guo. [Updates] 2020.8.30: The pre-trained model and code of ECCV-20 are made public

Jianzhu Guo 3.4k Jan 02, 2023
Predicting the duration of arrival delays for commercial flights.

Flight Delay Prediction Our objective is to predict arrival delays of commercial flights. According to the US Department of Transportation, about 21%

Jordan Silke 1 Jan 11, 2022
Car Parking Tracker Using OpenCv

Car Parking Vacancy Tracker Using OpenCv I used basic image processing methods i

Adwait Kelkar 30 Dec 03, 2022
This is a repository for a No-Code object detection inference API using the OpenVINO. It's supported on both Windows and Linux Operating systems.

OpenVINO Inference API This is a repository for an object detection inference API using the OpenVINO. It's supported on both Windows and Linux Operati

BMW TechOffice MUNICH 68 Nov 24, 2022
PIXIE: Collaborative Regression of Expressive Bodies

PIXIE: Collaborative Regression of Expressive Bodies [Project Page] This is the official Pytorch implementation of PIXIE. PIXIE reconstructs an expres

Yao Feng 331 Jan 04, 2023
The code repository for "PyCIL: A Python Toolbox for Class-Incremental Learning" in PyTorch.

PyCIL: A Python Toolbox for Class-Incremental Learning Introduction • Methods Reproduced • Reproduced Results • How To Use • License • Acknowledgement

Fu-Yun Wang 258 Dec 31, 2022
Winning Solution in NTIRE19 Challenges on Video Restoration and Enhancement (CVPR19 Workshops) - Video Restoration with Enhanced Deformable Convolutional Networks. EDVR has been merged into BasicSR and this repo is a mirror of BasicSR.

EDVR has been merged into BasicSR. This GitHub repo is a mirror of BasicSR. Recommend to use BasicSR, and open issues, pull requests, etc in BasicSR.

Xintao 1.4k Dec 25, 2022
Official implementation of "MetaSDF: Meta-learning Signed Distance Functions"

MetaSDF: Meta-learning Signed Distance Functions Project Page | Paper | Data Vincent Sitzmann*, Eric Ryan Chan*, Richard Tucker, Noah Snavely Gordon W

Vincent Sitzmann 100 Jan 01, 2023
This is a library for training and applying sparse fine-tunings with torch and transformers.

This is a library for training and applying sparse fine-tunings with torch and transformers. Please refer to our paper Composable Sparse Fine-Tuning f

Cambridge Language Technology Lab 37 Dec 30, 2022
Automatically Build Multiple ML Models with a Single Line of Code. Created by Ram Seshadri. Collaborators Welcome. Permission Granted upon Request.

Auto-ViML Automatically Build Variant Interpretable ML models fast! Auto_ViML is pronounced "auto vimal" (autovimal logo created by Sanket Ghanmare) N

AutoViz and Auto_ViML 397 Dec 30, 2022
joint detection and semantic segmentation, based on ultralytics/yolov5,

Multi YOLO V5——Detection and Semantic Segmentation Overeview This is my undergraduate graduation project which based on ultralytics YOLO V5 tag v5.0.

477 Jan 06, 2023