Taehoon

block-aligner. SIMD-accelerated library for computing global and X-drop affine gap penalty sequence-to-sequence or sequence-to-profile alignments using an adaptive block-based algorithm.

concept-trak. Python

trm. Python

min-pi-flow. Python

pico-banana-400k. Python

AnyEdit. 【CVPR 2025 Oral】Official Repo for Paper "AnyEdit: Mastering Unified High-Quality Image Editing for Any Idea"

one-key-hidpi. Enable macOS HiDPI and have a native setting.

i1. Code release for "i1: A Simple and Fully Open Recipe for Strong Text-to-Image Models"

KindleTool. A tool for creating & extracting Kindle updates and more

LanPaint. High quality training free inpaint for every stable diffusion model. Supports ComfyUI

ms-swift. Use PEFT or Full-parameter to CPT/SFT/DPO/GRPO 600+ LLMs (Qwen3.6, DeepSeek-V4, GLM-5.1, InternLM3, Llama4, ...) and 300+ MLLMs (Qwen3-VL, Qwen3-Omni, InternVL3.5, Ovis2.5, GLM4.5v, Gemma4, Llava, Phi4, ...) (AAAI 2025).

DiffusionNFT. [ICLR 2026 Oral] DiffusionNFT: Online Diffusion Reinforcement with Forward Process

dspy. DSPy: The framework for programming—not prompting—language models

logfire. AI observability platform for production LLM and agent systems.

Nitro-1. This repo provides a working re-implementation of Latent Adversarial Diffusion Distillation by AMD

anny. Anny, A Free and Interpretable Human Body Model for all ages, written in PyTorch.

MHR. Momentum Human Rig is an anatomically-inspired parametric full-body digital human model developed at Meta. It includes: A parametric body skeletal model; A realistic 3D mesh skinned to the skeleton with levels of detail;A body blendshape and pose corrective model; A facial blendshape model.Its design is friendly for both CG and CV communities.

vggt-omega. [CVPR 2026 Oral] VGGT Omega

ChordEdit. [CVPR 2026 Oral] Official implementation for ChordEdit: One-Step Low-Energy Transport for Image Editing

gemma-tuner-multimodal. Fine-tune Gemma 4 and 3n with audio, images and text on Apple Silicon, using PyTorch and Metal Performance Shaders.

IsaacLab. Unified framework for robot learning built on NVIDIA Isaac Sim

warp. A Python framework for GPU-accelerated simulation, robotics, and machine learning.

ghidra. Ghidra is a software reverse engineering (SRE) framework

OpenSubdiv. An Open-Source subdivision surface library.

internet_archive_downloader. A chrome/firefox extension that download books from Internet Archive(archive.org) and HathiTrust Digital Library (hathitrust.org)

tamagui. Style React fast with 100% parity on React Native, an optional UI kit, and optimizing compiler.

codex-auth. A CLI tool to switch and manage Codex accounts

rgthree-comfy. Making ComfyUI more comfortable!

x-transformers. A concise but complete full-attention transformer with a set of promising experimental features from various papers

neuraloperator. Learning in infinite dimension with neural operators.

minWM. A Minimal and Elegant Framework & Tutorial for Real-Time Interactive World Models

Causal-Forcing. [ICML 2026] Official codebase for "Causal Forcing: Autoregressive Diffusion Distillation Done Right for High-Quality Real-Time Interactive Video Generation" & Causal Forcing++

RAEv2. Official Implemenation for RAEv2: Improved Baselines with Representation Autoencoders

FD-Loss. Python

meanflow. JAX implementation of MeanFlow

flow_grpo. [NeurIPS 2025] An official implementation of Flow-GRPO: Training Flow Matching Models via Online RL

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

DGPO. [ICLR 2026][Ultra Fast&Powerful Diffusion RL] Reinforcing Diffusion Models by Direct Group Preference Optimization

TDM-R1. [ICML 2026][Ultra Powerful Few-Step Diffusion RL] TDM-R1: Reinforcing Few-Step Diffusion Models with Non-Differentiable Reward

LongAlign. The official PyTorch implementation for Improving Long-Text Alignment for Text-to-Image Diffusion Models (LongAlign)

AnyFlow. Flow Map OPD for AnyStep Video Diffusion

Lance. A 3B-active-parameter native unified multimodal model for image and video understanding, generation, and editing.

RTDMD. [arXiv 2026] This is the official PyTorch implementation of "RTDMD: Reinforcing Few-step Generators via Reward-Tilted Distribution Matching".

PiD. PiD: Fast and High-Resolution Latent Decoding with Pixel Diffusion

Awesome-Nano-Banana-images. A curated collection of fun and creative examples generated with Nano Banana & Nano Banana Pro🍌, Gemini-2.5-flash-image based model. We also release Nano-consistent-150K openly to support the community's development of image generation and unified models(click to website to see our blog)

LucidFlux. LucidFlux: Caption-Free Photo-Realistic Image Restoration via a Large-Scale Diffusion Transformer, ICLR 2026

PixelDiT. [CVPR 2026 Best Paper Finalist] Pixel Diffusion Transformers for Image Generation

Fizgig. Krea 2 & Klein 9B LoRA Studio — train, profile, repair, and extract Krea 2 & Flux 2 Klein 9B LoRAs

ComfyUI-Flux2Klein-Enhancer. Flux.2Klein 9B Enhancement Nodes Suite

opensessions. tmux sidebar for coding agents — Amp, Claude Code, Codex, OpenCode. Per-thread markers, local HTTP API, live session state.

ratatui. A Rust crate for cooking up terminal user interfaces (TUIs) 👨‍🍳🐀 https://ratatui.rs

visionary. Python

dmdr. [ECCV 2026] Official Code of "Distribution Matching Distillation Meets Reinforcement Learning"

D-OPSD. Official Repo of "D-OPSD: On-Policy Self-Distillation for Continuously Tuning Step-Distilled Diffusion Models"

CLIP-fine-tune. Fine-tuning code for CLIP models

agent-lightning. The absolute trainer to light up AI agents.

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

SD-RPN. [ICLR2026] Catching the Details: Self-Distilled RoI Predictors for Fine-Grained MLLM Perception

CATANet. CVPR2025 - CATANet: Efficient Content-Aware Token Aggregation for Lightweight Image Super-Resolution

rivet. Rivet Actors are the primitive for stateful workloads. Built for AI agents, collaborative apps, and durable execution.

viu. Terminal image viewer with native support for iTerm and Kitty

browser. Lightpanda: the headless browser designed for AI and automation

searxng. SearXNG is a free internet metasearch engine which aggregates results from various search services and databases. Users are neither tracked nor profiled.

datafusion. Apache DataFusion SQL Query Engine

vortex. An extensible, state-of-the-art framework for columnar compression, and the fastest FOSS columnar file format. Formerly at @spiraldb, now an Incubation Stage project at LFAI&Data, part of the Linux Foundation.

nimble. New and extensible file format for storage of large columnar datasets.

sam-3d-objects. SAM 3D Objects

next-lama. JavaScript

lama. 🦙 LaMa Image Inpainting, Resolution-robust Large Mask Inpainting with Fourier Convolutions, WACV 2022

ml-sharp. Sharp Monocular View Synthesis in Less Than a Second

LHM-plusplus. LHM++: An Efficient Large Human Reconstruction Model for Pose-free Images to 3D

LHM. [ICCV2025] LHM: Large Animatable Human Reconstruction Model from a Single Image in Seconds

Cool-GenAI-Fashion-Papers. [ACM TIST 2025] GenAI in Fashion: Overview, also includes 🔥latest papers, ⚙️metrics, 👀workshops, 🚀companies & products, ...)

nvidiaProfileInspector. C#

lightning-thunder. PyTorch compiler that accelerates training and inference. Get built-in optimizations for performance, memory, parallelism, and easily write your own.

tilelang. Domain-specific language designed to streamline the development of high-performance GPU/CPU/Accelerators kernels

TileKernels. A kernel library written in tilelang

opencode. The open source coding agent.

superpowers. An agentic skills framework & software development methodology that works.

TaylorSeer. [ICCV2025] From Reusing to Forecasting: Accelerating Diffusion Models with TaylorSeers

ToCa. [ICLR2025] Accelerating Diffusion Transformers with Token-wise Feature Caching

SpargeAttn. [ICML2025] SpargeAttention: A training-free sparse attention that accelerates any model inference.

arxiv-mcp-server. A Model Context Protocol server for searching and analyzing arXiv papers

Cache4Diffusion. Aiming to integrate most existing feature caching-based diffusion acceleration schemes into a unified framework.

Helios. Helios: Real Real-Time Long Video Generation Model

tuna-2. Official implementation of Tuna-2: Pixel Embeddings Beat Vision Encoders for Unified Understanding and Generation

caveman. 🪨 why use many token when few token do trick — Claude Code skill that cuts 65% of tokens by talking like caveman

talkie. talkie is a vintage language model from 1930

Gemini-API. ✨ Reverse-engineered Python API for Google Gemini web app

epub2MD. convert epub file to markdown files

taste-skill. Taste-Skill - gives your AI good taste. stops the AI from generating boring, generic slop

TIDE. Dynamic per-token early exit for LLM inference. Skip layers tokens don't need

exllamav3. An optimized quantization and inference library for running LLMs locally on modern consumer-class GPUs

t3code. TypeScript

LiteRT-LM. LiteRT-LM is Google's production-ready, high-performance, open-source inference framework for deploying Large Language Models on edge devices.

local-studio. Control panel for VLLM, Sglang, llama.cpp, exllamav3

turboquant. TurboQuant: Near-optimal KV cache quantization for LLM inference (3-bit keys, 2-bit values) with Triton kernels + vLLM integration

qwen3.5-35b-a3b-turbo3-1m-context. Run Qwen3.5-35B-A3B with 1 million token context on a single NVIDIA L4

gallery. A gallery that showcases on-device ML/GenAI use cases and allows people to try and use models locally.

LLaVA-OneVision-2. Fully Open Framework for Democratized Multimodal Training

PoPE-pytorch. Efficient implementation (and explorations) into polar coordinate positional embedding (PoPE) - from Gopalakrishnan et al. under Schmidhuber

Self-Distillation. Python

YOLO-Master. [CVPR2026]🚀🚀🚀Official code for the paper "YOLO-Master: MOE-Accelerated with Specialized Transformers for Enhanced Real-time Detection." *(YOLO = You Only Look Once)* 🔥🔥🔥

LAST-ViT. [CVPR 2026] The official PyTorch implementation of the "Vision Transformer Needs More Than Registers".

anyup. [ICLR '26 Oral] Official repository of the paper "AnyUp: Universal Feature Upsampling".

cuda-optimized-skill. A CUDA kernel optimization toolkit for validation, benchmarking, Nsight Compute profiling, bottleneck analysis, and iterative tuning. It helps improve custom GPU operators with reproducible workflows and evidence-based performance comparison.

OV-DEIM. [OV-DEIM] Real-time DETR-Style Open-Vocabulary Object Detection with GridSynthetic Augmentation

steplaw. Python

Attention-Residuals.

describe-anything. [ICCV 2025] Implementation for Describe Anything: Detailed Localized Image and Video Captioning

FT-FSOD. [CVPR 2026] A Closer Look at Cross-Domain Few-Shot Object Detection: Fine-Tuning Matters and Parallel Decoder Helps

INSID3. [CVPR 2026 Oral] "INSID3: Training-Free In-Context Segmentation with DINOv3"

boxer. Code for the Boxer research paper

rhwp. 아래한글 hwp viewer and editor by rust and wasm

HolisticTraceAnalysis. A library to analyze PyTorch traces.

AI-Semiconductor-Startup-Investment-Agent. AI 반도체 스타트업 투자 평가 멀티 에이전트

litmus. Litmus helps SREs and developers practice chaos engineering in a Cloud-native way. Chaos experiments are published at the ChaosHub (https://hub.litmuschaos.io). Community notes is at https://hackmd.io/a4Zu_sH4TZGeih-xCimi3Q

challenge.

mdx. Markdown for the component era

iREPA. [ICLR 2026] Official implementation for What matters for Representation Alignment: Global Information or Spatial Structure?

finite-difference-flow-optimization. FDFO: Finite Difference Flow Optimization

papers.

taehv. Tiny AutoEncoder for Hunyuan Video (and other video models)

gpu-experiments. A collection of GPU experiments and benchmarks for my personal understanding and research.

playwright-mcp. Playwright MCP server

zenml. ZenML 🙏: One AI Platform from Pipelines to Agents. https://zenml.io.

training. Reference implementations of MLPerf® training benchmarks

torchtitan. A PyTorch native platform for training generative AI models

ffpa-attn. 🤖FFPA: Extends FlashAttention-2 via Split-D for large headdims, 1.5x~3×↑🎉 vs SDPA, up to 430T🎉 on H200.

LeetCUDA. 📚LeetCUDA: Modern CUDA Learn Notes with PyTorch for Beginners🐑, 200+ CUDA Kernels, Tensor Cores, HGEMM, FA-2 MMA.🎉

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

Liger-Kernel. Efficient Triton Kernels for LLM Training

kernelbot. Write a fast kernel and see how you compare against the best humans and AI on gpumode.com

triton-index. Cataloging released Triton kernels.

HipKittens. Fast and Furious AMD Kernels

jane-street-gpu-mode-hackathon. 1st place. Team Crane's solution the 9/6 Jane Street x GPU_MODE contest! @aswinkumar1999 @rathull @kyolebu

cinder. This is Meta's fork of the CPython runtime. The name "cinder" here is historical, see https://github.com/facebookincubator/cinderx for the Python extension / JIT compiler.

webmcp. 🤖 WebMCP

VictoriaMetrics. VictoriaMetrics: fast, cost-effective monitoring solution and time series database

Scrapegraph-ai. Python scraper based on AI

linkedin_scraper. A library that scrapes Linkedin for user data

agent-browser. Browser automation CLI for AI agents

niri. A scrollable-tiling Wayland compositor.

ty-vscode. A Visual Studio Code extension for ty.

ty. An extremely fast Python type checker and language server, written in Rust.

tokscale. 🛰️ A CLI tool for tracking token usage from OpenCode, Claude Code, 🦞OpenClaw, Pi, Codex, Gemini, Cursor, AmpCode, Factory Droid, Kimi, and more! • 🏅Global Leaderboard + 2D/3D Contributions Graph

nix. Nix

taal. Sync MCPs and Agent Skills across AI coding assistants

extracranial. 🤴🏻 extracranial: the exosuit for my brain

vibe-haptic. 🫨 Haptic feedback from your coding agents – Get notified physically.

Douyin_TikTok_Download_API. 🚀「Douyin_TikTok_Download_API」是一个开箱即用的高性能异步抖音、快手、TikTok、Bilibili数据爬取工具，支持API调用，在线批量解析及下载。

oracle. Ask the oracle when you're stuck. Invoke GPT-5 Pro with a custom context and files.

creepjs. Creepy device and browser fingerprinting

fingerprintjs. The most advanced free and open-source browser fingerprinting library

rebrowser-patches. Collection of patches for puppeteer and playwright to avoid automation detection and leaks. Helps to avoid Cloudflare and DataDome CAPTCHA pages. Easy to patch/unpatch, can be enabled/disabled on demand.

camoufox. 🦊 Anti-detect browser

BotBrowser. Advanced Privacy Browser Core with Unified Fingerprint Defense: Cloudflare, Akamai, Kasada, Shape, DataDome, PerimeterX, hCaptcha, FunCaptcha, Imperva, reCAPTCHA, ThreatMetrix, Adscore

elevenlabs-mcp. The official ElevenLabs MCP server

lightly-train. All-in-one training for vision models (YOLO, ViTs, RT-DETR, DINOv3): pretraining, fine-tuning, distillation.

RT-DETRv4. [ECCV 2026] RT-DETRv4: Painlessly Furthering Real-Time Object Detection with Vision Foundation Models

CodexBar. Show usage stats for OpenAI Codex and Claude Code, without having to login.

Claude-Usage-Tracker. Native macOS menu bar app for tracking Claude AI usage limits in real-time. Built with Swift/SwiftUI.

KeyCluCask. Simple and handy overview of applications shortcuts

loki. Like Prometheus, but for logs.

telescope. Web-based log viewer UI. Explore logs data stored in several sources: ClickHouse, Kubernetes or Docker

logdy-core. Realtime log viewer with web UI, tail -f for logs with a web interface browser.

lnav. Log file navigator

claude-workflow-v2. Universal Claude Code workflow plugin with agents, skills, hooks, and commands

pyright. Static Type Checker for Python

tailscale-tpm-fixer. A utility to resolve Tailscale TPM recognition errors on Windows & Linux by resetting the TPM-bound state.

netdata. The fastest path to AI-powered full stack observability, even for lean teams.

solar-vs-glm-sanity-check. Python

caddy-grafana. Monitoring Caddy Server with Grafana (Prometheus + Loki) on Debian

datamodel-code-generator. Generate Pydantic v2 models, dataclasses, TypedDict, and msgspec.Struct from OpenAPI, JSON Schema, GraphQL, Avro, Protobuf, and raw JSON/YAML/CSV.

msw. Industry standard API mocking for JavaScript.

ripgrep. ripgrep recursively searches directories for a regex pattern while respecting your gitignore

immer. Create the next immutable state by mutating the current one

neverthrow. Type-Safe Errors for JS & TypeScript

devtools-frontend. The Chrome DevTools UI

ts-pattern. 🎨 The exhaustive Pattern Matching library for TypeScript, with smart type inference.

SCAIL. SCAIL: Towards Studio-Grade Character Animation via In-Context Learning of 3D-Consistent Pose Representations (CVPR 2026 Findings)

witr. Why is this running?

chrome-devtools-mcp. Chrome DevTools for coding agents

nuqs. Type-safe search params state manager for React frameworks - Like useState, but stored in the URL query string.

openapi-typescript. Generate TypeScript types from OpenAPI 3 specs

usehooks-ts. React hook library, ready to use, written in Typescript.

zustand. 🐻 Bear necessities for state management in React

zod. TypeScript-first schema validation with static type inference

mermerd. Create Mermaid-Js ERD diagrams from existing tables

reui. Design-forward shadcn kit for interfaces that stand out. 1000+ free patterns!

mutagen. Fast file synchronization and network forwarding for remote development

ghostty. 👻 Ghostty is a fast, feature-rich, and cross-platform terminal emulator that uses platform-native UI and GPU acceleration.

restate. Restate is the platform for building resilient applications that tolerate all infrastructure faults w/o the need for a PhD.

caddy. Fast and extensible multi-platform HTTP/1-2-3 web server with automatic HTTPS

fastcrud. FastCRUD is a Python package for FastAPI, offering robust async CRUD operations and flexible endpoint creation utilities.

beanie. Asynchronous Python ODM for MongoDB

SurfSense. NotebookLM for Competitive Intelligence. Give your AI agents Competitive Intelligence. Join our Discord: https://discord.gg/ejRNvftDp9

transformerlab-app. The open source research environment for AI researchers to seamlessly train, evaluate, and scale models from local hardware to GPU clusters.

polar. Polar — A billing platform for the intelligence era

sqlmodel. SQL databases in Python, designed for simplicity, compatibility, and robustness.