High Performance LLM Inference Operator Library

SkyRL: A Modular Full-stack RL Library for LLMs

Conditional Memory via Scalable Lookup: A New Axis of Sparsity for Large Language Models

CUDA Tile IR is an MLIR-based intermediate representation and compiler infrastructure for CUDA kernel optimization, focusing on tile-based computation patterns and optimizations targeting NVIDIA te…

A TUI-based utility for real-time monitoring of InfiniBand traffic and performance metrics on the local node

Lightweight coding agent that runs in your terminal

Helpful kernel tutorials and examples for tile-based GPU programming

A high-performance inference engine for LLMs, optimized for diverse AI accelerators.

cocoon

Nex Venus Communication Library

Fast and Furious AMD Kernels

Perplexity open source garden for inference technology

Autonomous GPU Kernel Generation via Deep Agents

A Lightweight LLM Post-Training Library

Kimi Code CLI is your next CLI agent.

High-throughput tensor loading for PyTorch

Contexts Optical Compression

Ship correct and fast LLM kernels to PyTorch

The Modular Platform (includes MAX & Mojo)

A Python-embedded DSL that makes it easy to write fast, scalable ML kernels with minimal boilerplate.

Ascend TileLang adapter

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

Tile primitives for speedy kernels

Building the Virtuous Cycle for AI-driven LLM Systems

a size profiler for cuda binary

🐹 Deep clean and optimize your Mac.

Post-training with Tinker

Verify Precision of all Kimi K2 API Vendor