Han Cai

AnyFlow. Flow Map OPD for AnyStep Video Diffusion

Lightning-OPD. Python

DC-Gen. DC-Gen: Post-Training Diffusion Acceleration with Deeply Compressed Latent Space

DC-VideoGen. DC-VideoGen: Efficient Video Generation with Deep Compression Video Autoencoder

Jet-Nemotron. Python

Qwen3. Qwen3 is the large language model series developed by Qwen team, Alibaba Cloud.

HiDream-I1. Python

PS3. Scaling Vision Pre-Training to 4K Resolution

Sparse-VideoGen. [ICML2025, NeurIPS2025 Spotlight] Sparse VideoGen 1 & 2: Accelerating Video Diffusion Transformers with Sparse Attention

COAT. [ICLR 2025] COAT: Compressing Optimizer States and Activation for Memory-Efficient FP8 Training

genesis-world. Simulation platform for general-purpose robotics & embodied AI learning.

DINO-X-API. DINO-X: The World's Top-Performing Vision Model for Open-World Object Detection and Understanding

Cosmos-Tokenizer. A suite of image and video neural tokenizers

Sana. SANA: Efficient High-Resolution Image Synthesis with Linear Diffusion Transformer

sam2. The repository provides code for running inference with the Meta Segment Anything Model 2 (SAM 2), links for downloading the trained model checkpoints, and example notebooks that show how to use the model.

medficientsam. Efficient Segment Anything in Medical Images

Grounding-DINO-1.5-API. Grounding DINO 1.5: IDEA Research's Most Capable Open-World Object Detection Model Series

llama3-from-scratch. llama3 implementation one matrix multiplication at a time

datatrove. Freeing data processing from scripting madness by providing a set of platform-agnostic customizable pipeline processing blocks.

llama3. The official Meta Llama 3 GitHub site

nanoGPT. The simplest, fastest repository for training/finetuning medium-sized GPTs.

VAR. [NeurIPS 2024 Best Paper Award][GPT beats diffusion🔥] [scaling laws in visual generation📈] Official impl. of "Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction". An *ultra-simple, user-friendly yet state-of-the-art* codebase for autoregressive image generation!

distrifuser. [CVPR 2024 Highlight] DistriFusion: Distributed Parallel Inference for High-Resolution Diffusion Models

MiniCPM. MiniCPM5-1B: A SOTA 1B on-device LLM, small yet powerful.

torchpack. A neural network training interface based on PyTorch, with a focus on flexibility

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

X-AnyLabeling. Effortless data labeling with AI support from Segment Anything and other awesome models.

ImageBind. ImageBind One Embedding Space to Bind Them All

LAVIS. LAVIS - A One-stop Library for Language-Vision Intelligence

dinov2. PyTorch code and models for the DINOv2 self-supervised learning method.

efficientvit. Efficient vision foundation models for high-resolution generation and perception.

segment-anything. The repository provides code for running inference with the SegmentAnything Model (SAM), links for downloading the trained model checkpoints, and example notebooks that show how to use the model.

unilm. Large-scale Self-supervised Pre-training Across Tasks, Languages, and Modalities

llama. Inference code for Llama models

mcunet. [NeurIPS 2020] MCUNet: Tiny Deep Learning on IoT Devices; [NeurIPS 2021] MCUNetV2: Memory-Efficient Patch-based Inference for Tiny Deep Learning

litepose. [CVPR'22] Lite Pose: Efficient Architecture Design for 2D Human Pose Estimation

torchprofile. Count the MACs / FLOPs of PyTorch models

minGPT. A minimal PyTorch re-implementation of the OpenAI GPT (Generative Pretrained Transformer) training

deep-gradient-compression. [ICLR 2018] Deep Gradient Compression: Reducing the Communication Bandwidth for Distributed Training

HPO2NAS-Tutorial-CVPR-ECCV2020. Hands-on Tutorial on Automated Deep Learning

hardware-aware-transformers. [ACL'20] HAT: Hardware-Aware Transformers for Efficient Natural Language Processing

Project-Page-Render. HTML

once-for-all. [ICLR 2020] Once for All: Train One Network and Specialize it for Efficient Deployment

pvcnn. [NeurIPS 2019, Spotlight] Point-Voxel CNN for Efficient 3D Deep Learning

Echoo. Let your program echo to you.

ShuffleNet-Series. Python

nasbench. NASBench: A Neural Architecture Search Dataset and Benchmark

haq. [CVPR 2019, Oral] HAQ: Hardware-Aware Automated Quantization with Mixed Precision

attention-is-all-you-need-pytorch. A PyTorch implementation of the Transformer model in "Attention is All You Need".

pytorch-mobilenet-v3. MobileNetV3 in pytorch and ImageNet pretrained models

cvpr18-inaturalist-transfer. Large Scale Fine-Grained Categorization and Domain-Specific Transfer Learning. CVPR 2018

amc-models. [ECCV 2018] AMC: AutoML for Model Compression and Acceleration on Mobile Devices

overcoming-catastrophic. Implementation of "Overcoming catastrophic forgetting in neural networks" in Tensorflow

slimmable_networks. Slimmable Networks, AutoSlim, and Beyond, ICLR 2019, and ICCV 2019

Stylized-ImageNet. Code to create Stylized-ImageNet, a stylized version of standard ImageNet (ICLR 2019 Oral)

mixup-cifar10. mixup: Beyond Empirical Risk Minimization

dropblock. Implementation of DropBlock: A regularization method for convolutional networks in PyTorch.

horovod. Distributed training framework for TensorFlow, Keras, PyTorch, and Apache MXNet.

torchdiffeq. Differentiable ODE solvers with full GPU support and O(1)-memory backpropagation.

proxylessnas. [ICLR 2019] ProxylessNAS: Direct Neural Architecture Search on Target Task and Hardware

pytorch-OpCounter. Count the MACs / FLOPs of your PyTorch model.

DEN. Lifelong Learning with Dynamically Expandable Networks, ICLR 2018

MobileNetV2-pytorch. Impementation of MobileNetV2 in pytorch

AdamW-and-SGDW. Decoupled Weight Decay Regularization (ICLR 2019)

darts. Differentiable architecture search for convolutional and recurrent networks

baselines-rudder. RUDDER for ATARI games with delayed rewards in OpenAI Baselines package

AutoAugment. Unofficial implementation of the ImageNet, CIFAR 10 and SVHN Augmentation Policies learned by AutoAugment using pillow

PathLevel-EAS. Path-Level Network Transformation for Efficient Architecture Search, in ICML 2018.

slimming. Learning Efficient Convolutional Networks through Network Slimming, In ICCV 2017.

gluon-cv. Gluon CV Toolkit

DecorrelatedBN. Code for Decorrelated Batch Normalization

enas. TensorFlow Code for paper "Efficient Neural Architecture Search via Parameter Sharing"

nn_dataflow. Explore the energy-efficient dataflow scheduling for neural networks.

Cutout. 2.56%, 15.20%, 1.30% on CIFAR10, CIFAR100, and SVHN https://arxiv.org/abs/1708.04552

models. Models and examples built with TensorFlow

pytorch-mobilenet-v2. A PyTorch implementation of MobileNet V2 architecture and pretrained model.

ENAS-pytorch. PyTorch implementation of "Efficient Neural Architecture Search via Parameters Sharing"

Texygen. A text generation benchmarking platform

EffectiveTensorflow. TensorFlow tutorials and best practices.

CondenseNet. CondenseNet: Light weighted CNN for mobile devices

PyramidNet-PyTorch. A PyTorch implementation for PyramidNets (Deep Pyramidal Residual Networks, https://arxiv.org/abs/1610.02915)

dm_control. Google DeepMind's software stack for physics-based simulation and Reinforcement Learning environments, using MuJoCo.

ComposeNet. Implementation of modular composition network from https://arxiv.org/pdf/1711.11289.pdf

rainbow. A PyTorch implementation of Rainbow DQN agent

DPNs. Dual Path Networks

gym-extensions. This repo is intended as an extension for OpenAI Gym for auxiliary tasks (multitask learning, transfer learning, inverse reinforcement learning, etc.)

ray. Ray is an AI compute engine. Ray consists of a core distributed runtime and a set of AI Libraries for accelerating ML workloads.

value-prediction-network. NIPS 2017 Value Prediction Network

MAgent. A Platform for Many-Agent Reinforcement Learning

treelstm.pytorch. Tree LSTM implementation in PyTorch

PyramidNet. Torch implementation of the paper "Deep Pyramidal Residual Networks" (https://arxiv.org/abs/1610.02915).

treelstm. Tree-structured Long Short-Term Memory networks (http://arxiv.org/abs/1503.00075)

CapsNet-Tensorflow. A Tensorflow implementation of CapsNet(Capsules Net) in paper Dynamic Routing Between Capsules

SerpentAI. Game Agent Framework. Helping you create AIs / Bots that learn to play any game you own!

multiagent-particle-envs. Code for a multi-agent particle environment used in the paper "Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environments"

Rainbow. Rainbow: Combining Improvements in Deep Reinforcement Learning

VIN. Value Iteration Networks

softqlearning. Reinforcement Learning with Deep Energy-Based Policies

Information-bottleneck. Python implementation of the infomration bottleneck method (tishby et al, 1999)

LeakGAN. The codes of paper "Long Text Generation via Adversarial Training with Leaked Information" on AAAI 2018. Text generation using GAN and Hierarchical Reinforcement Learning.

SENet. Squeeze-and-Excitation Networks

sru. Training RNNs as Fast as CNNs (https://arxiv.org/abs/1709.02755)

ai-deadlines. :alarm_clock: AI conference deadline countdowns

shake-shake. 2.86% and 15.85% on CIFAR-10 and CIFAR-100

pysc2. StarCraft II Learning Environment

rl-teacher. Code for Deep RL from Human Preferences [Christiano et al]. Plus a webapp for collecting human feedback

dvn. Reference implementation for Structured Prediction with Deep Value Networks

maml. Code for "Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks"

efficient_densenet_pytorch. A memory-efficient implementation of DenseNets

ACER. Actor-critic with experience replay

vision_networks. Repo about neural networks for images handling

DenseNet. Densely Connected Convolutional Networks, In CVPR 2017 (Best Paper Award).

TED-Multilingual-Parallel-Corpus. TED parallel Corpora is growing collection of Bilingual parallel corpora, Multilingual parallel corpora and Monolingual corpora extracted from TED talks www.ted.com for 109 world languages.

text-to-image. Text to image synthesis using thought vectors

icml2016. Generative Adversarial Text-to-Image Synthesis

Augmentor. Image augmentation library in Python for machine learning.

tensorforce. Tensorforce: a TensorFlow library for applied reinforcement learning

a3c-distributed_tensorflow. Distributed Tensorflow Implementation of Asynchronous Methods for Deep Reinforcement Learning

SeqGAN. Implementation of Sequence Generative Adversarial Nets with Policy Gradient

stock_market_reinforcement_learning. This project provides a stock market environment using OpenGym with Deep Q-learning and Policy Gradient.

cpo. Constrained Policy Optimization

ELF. An End-To-End, Lightweight and Flexible Platform for Game Research

tensor2tensor. Library of deep learning models and datasets designed to make deep learning more accessible and accelerate ML research.

option_critic. Implementation of the Option-Critic Architecture on the Atari (ALE) environment

SNNs. Tutorials and implementations for "Self-normalizing networks"

mujoco-py. MuJoCo is a physics engine for detailed, efficient rigid body simulations with contacts. mujoco-py allows using MuJoCo from Python 3.

DDPG-Keras-Torcs. Using Keras and Deep Deterministic Policy Gradient to play TORCS

robustRL. Robust policy search algorithms which train on model ensembles

async_deep_reinforce. Asynchronous Methods for Deep Reinforcement Learning

modular_rl. Implementation of TRPO and related algorithms

baselines. OpenAI Baselines: high-quality implementations of reinforcement learning algorithms

Perceptron. A flexible artificial neural network builder to analyse performance, and optimise the best model.

Ads-RecSys-Datasets. This repository collects some datasets for Ads & RecSys uses, and provide easy-to-use hdf5 iterative access.

dl4ir-query-reformulator. Python

neural_complete. A neural network trained to help writing neural network code using autocomplete

SpiderKeeper. admin ui for scrapy/open source scrapinghub

dl4ir-webnav. WebNav: A New Large-Scale Task for Natural Language based Sequential Decision Making

generating-reviews-discovering-sentiment. Code for "Learning to Generate Reviews and Discovering Sentiment"

MazeBase. Simple environment for creating very simple 2D games and training neural network models to perform tasks within them

lab. A customisable 3D platform for agent-based AI research

rc-data. Question answering dataset featured in "Teaching Machines to Read and Comprehend

tensorflow-rl. Implementations of deep RL papers and random experimentation

academicpages.github.io. Github Pages template based upon HTML and Markdown for personal, portfolio-based websites.

gym-starcraft. StarCraft environment for OpenAI Gym, based on Facebook's TorchCraft. (In progress)

MARL-Papers. Paper list of multi-agent reinforcement learning (MARL)

WassersteinGAN. Python

ActorMimic. Train an RL agent to play multiple Atari games at once

rllabplusplus. Python

rtb-papers. A collection of research and survey papers of real-time bidding (RTB) based display advertising techniques.

universe. Universe: a software platform for measuring and training an AI's general intelligence across the world's supply of games, websites and other applications.

rllab. rllab is a framework for developing and evaluating reinforcement learning algorithms, fully compatible with OpenAI Gym.

minpy. NumPy interface with mixed backend execution

mxnet. Lightweight, Portable, Flexible Distributed/Mobile Deep Learning with Dynamic, Mutation-aware Dataflow Dep Scheduler; for Python, R, Julia, Scala, Go, Javascript and more