Distributed Compiler based on Triton for Parallel Systems

🗂 The perfect Front-End Checklist for modern websites and meticulous developers

slime is an LLM post-training framework for RL Scaling.

The AI coding agent built for the terminal.

🙌 OpenHands: Code Less, Make More

MiroRL is an MCP-first reinforcement learning framework for deep research agent.

Democratizing Reinforcement Learning for LLMs

A curated list of awesome commands, files, and workflows for Claude Code

Train speculative decoding models effortlessly and port them smoothly to SGLang serving.

SkyRL: A Modular Full-stack RL Library for LLMs

Accelerate LLM preference tuning via prefix sharing with a single line of code

A Distributed Attention Towards Linear Scalability for Ultra-Long Context, Heterogeneous Data Training

LLM training code for Databricks foundation models

VILA is a family of state-of-the-art vision language models (VLMs) for diverse multimodal AI tasks across the edge, data center, and cloud.

VeOmni: Scaling Any Modality Model Training with Model-Centric Distributed Recipe Zoo

ByteCheckpoint: An Unified Checkpointing Library for LFMs

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

Lightning-Fast RL for LLM Reasoning and Agents. Made Simple & Flexible.

Qwen2.5-Omni is an end-to-end multimodal model by Qwen team at Alibaba Cloud, capable of understanding text, audio, vision, video, and performing real-time speech generation.

12 Lessons to Get Started Building AI Agents

Fine-tuning & Reinforcement Learning for LLMs. 🦥 Train OpenAI gpt-oss, DeepSeek-R1, Qwen3, Gemma 3, TTS 2x faster with 70% less VRAM.

A simple and well styled PPO implementation. Based on my Medium series: https://medium.com/@eyyu/coding-ppo-from-scratch-with-pytorch-part-1-4-613dfc1b14c8.

✨ A synthetic dataset generation framework that produces diverse coding questions and verifiable solutions - all in one framwork

A bidirectional pipeline parallelism algorithm for computation-communication overlap in DeepSeek V3/R1 training.

FlashMLA: Efficient Multi-head Latent Attention Kernels

Muon is Scalable for LLM Training

Production-tested AI infrastructure tools for efficient AGI development and community-driven innovation