HunyuanVideo-1.5: A leading lightweight video generation model

[Tutorial] Few-Step Distillation for Text-to-Image Generation: A Practical Guide

Lightweight Image Video Action Generation Inference Framework

xDiT: A Scalable Inference Engine for Diffusion Transformers (DiTs) with Massive Parallelism

High performance inference engine for diffusion models

A PyTorch-native inference engine with cache, parallelism, quantization and cpu offload for DiTs.

Selftok: Discrete Visual Tokens of Autoregression, by Diffusion, and for Reasoning

Wan: Open and Advanced Large-Scale Video Generative Models

[ICLR2025 Spotlight] SVDQuant: Absorbing Outliers by Low-Rank Components for 4-Bit Diffusion Models

[ICCV 2025][Few-Step Student Surpasses Teacher Diffusion] Learning Few-Step Diffusion Models by Trajectory Distribution Matching

A unified inference and post-training framework for accelerated video generation.

A parallelism VAE avoids OOM for high resolution image generation

Simplify your onnx model

PixArt-α: Fast Training of Diffusion Transformer for Photorealistic Text-to-Image Synthesis

This project aim to reproduce Sora (Open AI T2V model), we wish the open source community contribute to this project.

The official implementation of “Sophia: A Scalable Stochastic Second-order Optimizer for Language Model Pre-training”

Google Brain AutoML

A high-performance distributed deep learning system targeting large-scale and automated distributed training.

Hackable and optimized Transformers building blocks, supporting a composable construction.

Fast and memory-efficient exact attention

Pipeline Parallelism for PyTorch

This project is an accelerated dataset for PyTorch. The dataset stores compressed batched data that is cropped while loading. By this, I/O operations and network traffic are reduced. Since the crop…

Official repository of ACmix (CVPR2022)

SOTA low-bit LLM quantization (INT8/FP8/MXFP8/INT4/MXFP4/NVFP4) & sparsity; leading model compression techniques on PyTorch, TensorFlow, and ONNX Runtime

micronet, a model compression and deploy lib. compression: 1、quantization: quantization-aware-training(QAT), High-Bit(>2b)(DoReFa/Quantization and Training of Neural Networks for Efficient Integer-…

TorchMultimodal is a PyTorch library for training state-of-the-art multimodal multi-task models at scale.