Provide with pre-build flash-attention package wheels on Linux and Windows platforms using GitHub Actions

A framework for efficient model inference with omni-modality models

Async RL Training at Scale

"AI-Trader: Can AI Beat the Market?" Live Trading Bench: https://ai4trade.ai Tech Report Link: https://arxiv.org/abs/2512.10971

Wan: Open and Advanced Large-Scale Video Generative Models

Pytorch Distributed native training library for LLMs/VLMs with OOTB Hugging Face support

PyTorch native quantization and sparsity for training and inference

A modern GUI client based on Tauri, designed to run in Windows, macOS and Linux for tailored proxy experience

WFGY 2.0. Semantic Reasoning Engine for LLMs (MIT). Fixes RAG/OCR drift, collapse & “ghost matches” via symbolic overlays + logic patches. Autoboot; OneLine & Flagship. ⭐ Star if you explore semant…

An Open-Source Large-Scale Reinforcement Learning Project for Search Agents

Transformers-compatible library for applying various compression algorithms to LLMs for optimized deployment with vLLM

ArcticTraining is a framework designed to simplify and accelerate the post-training process for large language models (LLMs)

gpt-oss-120b and gpt-oss-20b are two open-weight language models by OpenAI

Bridge Megatron-Core to Hugging Face/Reinforcement Learning

(best/better) practices of megatron on veRL and tuning guide

Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflo…

A Gym for Agentic LLMs

End-to-End Reinforcement Learning for Multi-Turn Tool-Integrated Reasoning

Train your Agent model via our easy and efficient framework

An open-source, code-first Python toolkit for building, evaluating, and deploying sophisticated AI agents with flexibility and control.

[COLM 2025] Official repository for R2E-Gym: Procedural Environment Generation and Hybrid Verifiers for Scaling Open-Weights SWE Agents

Scaling RL on advanced reasoning models

MiniMax-M1, the world's first open-weight, large-scale hybrid-attention reasoning model.

Nano vLLM

Custom FlashAttention 2.7.4.post1 wheels for PyTorch 2.7.0+cu128 (CUDA 12.8) on Linux x86_64