Unofficial JAX implementations of Deep Learning models

Related tags

Deep Learningmachine-learningtransformersartificial-intelligenceconvolutional-neural-networksflaxjaxresearch-paper-implementation
Overview

JAX Models

license-shield release-shield python-shield code-style

Table of Contents
  1. About The Project
  2. Getting Started
  3. Contributing
  4. License
  5. Contact

About The Project

The JAX Models repository aims to provide open sourced JAX/Flax implementations for research papers originally without code or code written with frameworks other than JAX. The goal of this project is to make a collection of models, layers, activations and other utilities that are most commonly used for research. All papers and derived or translated code is cited in either the README or the docstrings. If you think that any citation is missed then please raise an issue.

All implementations provided here are available on Papers With Code.


Available model implementations for JAX are:
  1. MetaFormer is Actually What You Need for Vision (Weihao Yu et al., 2021)
  2. Augmenting Convolutional networks with attention-based aggregation (Hugo Touvron et al., 2021)
  3. MPViT : Multi-Path Vision Transformer for Dense Prediction (Youngwan Lee et al., 2021)
  4. MLP-Mixer: An all-MLP Architecture for Vision (Ilya Tolstikhin et al., 2021)
  5. Patches Are All You Need (Anonymous et al., 2021)
  6. SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers (Enze Xie et al., 2021)
  7. A ConvNet for the 2020s (Zhuang Liu et al., 2021)
  8. Masked Autoencoders Are Scalable Vision Learners (Kaiming He et al., 2021)

Available layers for out-of-the-box integration:
  1. DropPath (Stochastic Depth) (Gao Huang et al., 2021)
  2. Squeeze-and-Excitation Layer (Jie Hu et al. 2019)
  3. Depthwise Convolution (François Chollet, 2017)

Prerequisites

Prerequisites can be installed separately through the requirements.txt file in the main directory using:

pip install -r requirements.txt

The use of a virtual environment is highly recommended to avoid version incompatibilites.

Installation

This project is built with Python 3 for the latest JAX/Flax versions and can be directly installed via pip.

pip install jax-models

If you wish to use the latest version then you can directly clone the repository too.

git clone https://github.com/DarshanDeshpande/jax-models.git

Usage

To see all model architectures available:

from jax_models.models.model_registry import list_models
from pprint import pprint

pprint(list_models())

To load your desired model:

from jax_models.models.model_registry import load_model
load_model('mpvit-base', attach_head=True, num_classes=1000, dropout=0.1)

Contributing

Please raise an issue if any implementation gives incorrect results, crashes unexpectedly during training/inference or if any citation is missing.

You can contribute to jax_models by supporting me with compute resources or by contributing your own resources to provide pretrained weights.

If you wish to donate to this inititative then please drop me a mail here.

License

Distributed under the Apache 2.0 License. See LICENSE for more information.

Contact

Feel free to reach out for any issues or requests related to these implementations

Darshan Deshpande - Email | Twitter | LinkedIn

You might also like...
Very deep VAEs in JAX/Flax
Very deep VAEs in JAX/Flax

Very Deep VAEs in JAX/Flax Implementation of the experiments in the paper Very Deep VAEs Generalize Autoregressive Models and Can Outperform Them on I

Jamie Townsend 42 Dec 12, 2022
Conservative Q Learning for Offline Reinforcement Reinforcement Learning in JAX
Conservative Q Learning for Offline Reinforcement Reinforcement Learning in JAX

CQL-JAX This repository implements Conservative Q Learning for Offline Reinforcement Reinforcement Learning in JAX (FLAX). Implementation is built on

Karush Suri 8 Nov 7, 2022
PyTorch implementations of neural network models for keyword spotting
PyTorch implementations of neural network models for keyword spotting

Honk: CNNs for Keyword Spotting Honk is a PyTorch reimplementation of Google's TensorFlow convolutional neural networks for keyword spotting, which ac

Castorini 475 Dec 15, 2022
Unofficial implementation of Proxy Anchor Loss for Deep Metric Learning
Unofficial implementation of Proxy Anchor Loss for Deep Metric Learning

Proxy Anchor Loss for Deep Metric Learning Unofficial pytorch, tensorflow and mxnet implementations of Proxy Anchor Loss for Deep Metric Learning. Not

Geonmo Gu 3 Jun 9, 2021
Time-series-deep-learning - Developing Deep learning LSTM, BiLSTM models, and NeuralProphet for multi-step time-series forecasting of stock price.
Time-series-deep-learning - Developing Deep learning LSTM, BiLSTM models, and NeuralProphet for multi-step time-series forecasting of stock price.

Stock Price Prediction Using Deep Learning Univariate Time Series Predicting stock price using historical data of a company using Neural networks for

Abdultawwab Safarji 7 Nov 27, 2022
FedJAX is a library for developing custom Federated Learning (FL) algorithms in JAX.

FedJAX: Federated learning with JAX What is FedJAX? FedJAX is a library for developing custom Federated Learning (FL) algorithms in JAX. FedJAX priori

Google 208 Dec 14, 2022
Objax Apache-2Objax (🥉19 · ⭐ 580) - Objax is a machine learning framework that provides an Object.. Apache-2 jax

Objax Tutorials | Install | Documentation | Philosophy This is not an officially supported Google product. Objax is an open source machine learning fr

Google 729 Jan 2, 2023
Plug-n-Play Reinforcement Learning in Python with OpenAI Gym and JAX
Plug-n-Play Reinforcement Learning in Python with OpenAI Gym and JAX

coax is built on top of JAX, but it doesn't have an explicit dependence on the jax python package. The reason is that your version of jaxlib will depend on your CUDA version.

null 128 Dec 27, 2022
JAX code for the paper
JAX code for the paper "Control-Oriented Model-Based Reinforcement Learning with Implicit Differentiation"

Optimal Model Design for Reinforcement Learning This repository contains JAX code for the paper Control-Oriented Model-Based Reinforcement Learning wi

Evgenii Nikishin 43 Sep 28, 2022
Comments
  • Missing Axis Swap in ExtractPatches and MergePatches

    Missing Axis Swap in ExtractPatches and MergePatches

    In patch_utils.py, the modules ExtractPatches and MergePatches are missing an axis swap between the reshapes, resulting in the extracted patches becoming horizontal stripes. For example, if we follow the code in ExtractPatches:

    >>> inputs = jnp.arange(16).reshape(1, 4, 4, 1)
>>> inputs[0, :, :, 0]

DeviceArray([[ 0,  1,  2,  3],
             [ 4,  5,  6,  7],
             [ 8,  9, 10, 11],
             [12, 13, 14, 15]], dtype=int32)

>>> patch_size = 2
>>> batch, height, width, channels = inputs.shape
>>> height, width = height // patch_size, width // patch_size
>>> x = jnp.reshape(inputs, (batch, height, patch_size, width, patch_size, channels))
>>> x = jnp.reshape(x, (batch, height * width, patch_size ** 2 * channels))
>>> x[0, 0, :]

DeviceArray([0, 1, 2, 3], dtype=int32)

    We see that the first patch extracted is not the patch containing [0, 1, 4, 5], but the horizontal stripe [0, 1, 2, 3]. To fix this problem, we should add an axis swap. For ExtractPatches, this should be:

    batch, height, width, channels = inputs.shape
height, width = height // patch_size, width // patch_size
x = jnp.reshape(
    inputs, (batch, height, patch_size, width, patch_size, channels)
)
x = jnp.swapaxes(x, 2, 3)
x = jnp.reshape(x, (batch, height * width, patch_size ** 2 * channels))

    For MergePatches, this should be:

    batch, length, _ = inputs.shape
height = width = int(length**0.5)
x = jnp.reshape(inputs, (batch, height, width, patch_size, patch_size, -1))
x = jnp.swapaxes(x, 2, 3)
x = jnp.reshape(x, (batch, height * patch_size, width * patch_size, -1))
    bug 
    opened by young-geng 4
  • fix convnext to make it work with jax.jit

    fix convnext to make it work with jax.jit

    Hey, first of all, thanks for the nice codebase. When doing inference using the convnext model, I noticed the following issue:

    Calling x.item() will call float(x), which breaks the jit tracer. We can remove the list comprehension in unnecessary conversion to make jax.jit work. Without jax.jit, the model is very slow for me, running with only ~30% GPU utilization (RTX 3090).

    This issue could apply to other models as well, maybe it is a good idea to include a test for applying jax.jit to each model?

    opened by maxidl 1
Releases(v0.5-van)
Owner
Helping Machines Learn Better 💻😃
GitHub Repository
NeuralDiff: Segmenting 3D objects that move in egocentric videos

NeuralDiff: Segmenting 3D objects that move in egocentric videos Project Page | Paper + Supplementary | Video About This repository contains the offic

Vadim Tschernezki 14 Dec 05, 2022
Generalized Decision Transformer for Offline Hindsight Information Matching

Generalized Decision Transformer for Offline Hindsight Information Matching [arxiv] If you use this codebase for your research, please cite the paper:

Hiroki Furuta 35 Dec 12, 2022
A simple, high level, easy-to-use open source Computer Vision library for Python.

ZoomVision : Slicing Aid Detection A simple, high level, easy-to-use open source Computer Vision library for Python. Installation Installing dependenc

Nurettin Sinanoğlu 2 Mar 04, 2022
Contrastive Learning of Structured World Models

Contrastive Learning of Structured World Models This repository contains the official PyTorch implementation of: Contrastive Learning of Structured Wo

Thomas Kipf 371 Jan 06, 2023
PyTorch implementation of PP-LCNet: A Lightweight CPU Convolutional Neural Network

PyTorch implementation of PP-LCNet Reproduction of PP-LCNet architecture as described in PP-LCNet: A Lightweight CPU Convolutional Neural Network by C

Quan Nguyen (Fly) 47 Nov 02, 2022
Memoized coduals - Shows that it is possible to implement reverse mode autodiff using a variation on the dual numbers called the codual numbers

The dual numbers can do efficient autodiff! The codual numbers are a simple meth

wlad 2 Dec 19, 2022
This is the source code of the solver used to compete in the International Timetabling Competition 2019.

ITC2019 Solver This is the source code of the solver used to compete in the International Timetabling Competition 2019. Building .NET Core (2.1 or hig

Edon Gashi 8 Jan 22, 2022
(under submission) Bayesian Integration of a Generative Prior for Image Restoration

BIGPrior: Towards Decoupling Learned Prior Hallucination and Data Fidelity in Image Restoration Authors: Majed El Helou, and Sabine Süsstrunk {Note: p

Majed El Helou 22 Dec 17, 2022
Implementation of ML models like Decision tree, Naive Bayes, Logistic Regression and many other

ML_Model_implementaion Implementation of ML models like Decision tree, Naive Bayes, Logistic Regression and many other dectree_model: Implementation o

Anshuman Dalai 3 Jan 24, 2022
Numerical Methods with Python, Numpy and Matplotlib

Numerical Bric-a-Brac Collections of numerical techniques with Python and standard computational packages (Numpy, SciPy, Numba, Matplotlib ...). Diffe

Vincent Bonnet 10 Dec 20, 2021
ICCV2021 Oral SA-ConvONet: Sign-Agnostic Optimization of Convolutional Occupancy Networks

Sign-Agnostic Convolutional Occupancy Networks Paper | Supplementary | Video | Teaser Video | Project Page This repository contains the implementation

64 Jan 05, 2023
JDet is Object Detection Framework based on Jittor.

JDet is Object Detection Framework based on Jittor.

135 Dec 14, 2022
The official codes of our CVPR2022 paper: A Differentiable Two-stage Alignment Scheme for Burst Image Reconstruction with Large Shift

TwoStageAlign The official codes of our CVPR2022 paper: A Differentiable Two-stage Alignment Scheme for Burst Image Reconstruction with Large Shift Pa

Shi Guo 32 Dec 15, 2022
Project page for the paper Semi-Supervised Raw-to-Raw Mapping 2021.

Project page for the paper Semi-Supervised Raw-to-Raw Mapping 2021.

Mahmoud Afifi 22 Nov 08, 2022
Data and code for ICCV 2021 paper Distant Supervision for Scene Graph Generation.

Distant Supervision for Scene Graph Generation Data and code for ICCV 2021 paper Distant Supervision for Scene Graph Generation. Introduction The pape

THUNLP 23 Dec 31, 2022
This implements the learning and inference/proposal algorithm described in "Learning to Propose Objects, Krähenbühl and Koltun"

Learning to propose objects This implements the learning and inference/proposal algorithm described in "Learning to Propose Objects, Krähenbühl and Ko

Philipp Krähenbühl 90 Sep 10, 2021
ncnn is a high-performance neural network inference framework optimized for the mobile platform

ncnn ncnn is a high-performance neural network inference computing framework optimized for mobile platforms. ncnn is deeply considerate about deployme

Tencent 16.2k Jan 05, 2023
Code for ACM MM 2020 paper "NOH-NMS: Improving Pedestrian Detection by Nearby Objects Hallucination"

NOH-NMS: Improving Pedestrian Detection by Nearby Objects Hallucination The offical implementation for the "NOH-NMS: Improving Pedestrian Detection by

Tencent YouTu Research 64 Nov 11, 2022
Dense Passage Retriever - is a set of tools and models for open domain Q&A task.

Dense Passage Retrieval Dense Passage Retrieval (DPR) - is a set of tools and models for state-of-the-art open-domain Q&A research. It is based on the

Meta Research 1.1k Jan 03, 2023
Unofficial Tensorflow-Keras implementation of Fastformer based on paper [Fastformer: Additive Attention Can Be All You Need](https://arxiv.org/abs/2108.09084).

Fastformer-Keras Unofficial Tensorflow-Keras implementation of Fastformer based on paper Fastformer: Additive Attention Can Be All You Need. Tensorflo

Yam Peleg 10 Jan 30, 2022