Gavin Li

AirLLM 70B inference with single 4GB GPU

A Python package that runs 70-billion-parameter language models on a single small GPU by streaming model layers from disk one at a time, so you don't need a big multi-GPU server.

2D Game Animation in God Mode

No description.

Low latency, strong consistency, fault tolerant distributed key value store. Colocate data and compute to achieve best performance cloud compute.

Moved to here: https://github.com/lyogavin/airllm

This version of Stable Diffusion features a slick WebGUI, an interactive command-line script that combines text2img and img2img functionality in a "dream bot" style interface, and multiple features and other enhancements. For more info, see the website link below.

A latent text-to-image diffusion model

Advanced fine tuning tools for vision models

🤗 Diffusers: State-of-the-art diffusion models for image and audio generation in PyTorch

DECA: Detailed Expression Capture and Animation (SIGGRAPH 2021)

QLoRA: Efficient Finetuning of Quantized LLMs

Scala framework for iterative and interactive cluster computing.

Official implementation of AnimateDiff.

Fast and memory-efficient exact attention

The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface.

Awesome multilingual OCR toolkits based on PaddlePaddle (practical ultra lightweight OCR system, support 80+ languages recognition, provide data annotation and synthesis tools, support training and deployment among server, mobile, embedded and IoT devices)

No description.

Distributed and fault-tolerant realtime computation: stream processing, continuous computation, distributed RPC, and more

Turbo Launcher®

ncnn is a high-performance neural network inference framework optimized for the mobile platform

72.8% MobileNetV2 1.0 model on ImageNet and a spectrum of pre-trained MobileNetV2 models

Dawars' blog

Dai Stablecoin System

justinpinkney-stable-diffusion fork modified for finetune

Fast Inference Solutions for BLOOM

Fengshenbang-LM(封神榜大模型)是IDEA研究院认知计算与自然语言研究中心主导的大模型开源体系，成为中文认知智能的基础设施。

Chinese-Vicuna: A Chinese Instruction-following LLaMA-based Model —— 一个中文低资源的llama+lora方案，结构参考alpaca

LVDM: Latent Video Diffusion Models for High-Fidelity Long Video Generation

Custom node for breaking an animation into frames (and keyframes)

A pipeline parallel training script for diffusion models.

Protocol Buffers - Google's data interchange format

:arrow_double_down: Script to download AVA database from the website (A Large-Scale Database for Aesthetic Visual Analysis)

Parallel text feature extraction for machine learning

Unofficial PyTorch and torchvision builds for ARM devices

Python library for GPGPU on Raspberry Pi

:house_with_garden: Open source home automation that puts local control and privacy first

Models and examples built with TensorFlow

PySlowFast: video understanding codebase from FAIR for reproducing state-of-the-art video models.

Source code for transferable dialogue state generator (TRADE, Wu et al., 2019). https://arxiv.org/abs/1905.08743

Language and compiler for the Raspberry Pi GPU

A lightweight python library that helps to keep track of numerical experiments

MobileNetV1, MobileNetV2, VGG based SSD/SSD-lite implementation in Pytorch 1.0 / Pytorch 0.4. Out-of-box support for retraining on Open Images dataset. ONNX and Caffe2 support. Experiment Ideas like CoordConv.

Part of 3rd place solution for Kaggle's Tweet Sentiment Extraction

No description.

SUMBT: Slot-Utterance Matching for Universal and Scalable Belief Tracking (ACL 2019)

Facebook AI Research Sequence-to-Sequence Toolkit written in Python.

No description.

No description.

Flat and responsive website template, designed and coded by Natasha M Oliveira

No description.

Train transformer language models with reinforcement learning.

Chinese version of CLIP which achieves Chinese cross-modal retrieval and representation generation.

python wheel packages built for flash attention

3D ResNets for Action Recognition (CVPR 2018)

gavinli's blog

Driven by nextjs, shadcnui style blog template.

No description.

No description.

一个简单好用的 spine 文件查看&导出&壁纸工具。A simple and easy-to-use spine file viewer & exporter & wallpaper.

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

(CVPR 2025) From Slow Bidirectional to Fast Autoregressive Video Diffusion Models

Official codebase for "Self Forcing: Bridging Training and Inference in Autoregressive Video Diffusion" (NeurIPS 2025 Spotlight)

No description.

SDMatte is an interactive image matting method based on stable diffusion, which supports three types of visual prompts (points, boxes, and masks) for accurately extracting target objects from natural images.

A simple comfyUI plugin for images grid (X/Y Plot)

No description.