verl: Volcano Engine Reinforcement Learning for LLMs

Code for the paper “Four Over Six: More Accurate NVFP4 Quantization with Adaptive Block Scaling”

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…

Helpful kernel tutorials and examples for tile-based GPU programming

Accelerating MoE with IO and Tile-aware Optimizations

SLA: Beyond Sparsity in Diffusion Transformers via Fine-Tunable Sparse–Linear Attention

TurboDiffusion: 100–200× Acceleration for Video Diffusion Models

cuTile is a programming model for writing parallel kernels for NVIDIA GPUs

Offline optimization of your disaggregated Dynamo graph

Tilus is a tile-level kernel programming language with explicit control over shared memory and registers.

🚀🚀 Efficient implementations of Native Sparse Attention

Distributed Compiler based on Triton for Parallel Systems

FlashInfer: Kernel Library for LLM Serving

Genai-bench is a powerful benchmark tool designed for comprehensive token-level performance evaluation of large language model (LLM) serving systems.

An efficient implementation of the NSA (Native Sparse Attention) kernel

✨ Perfect virtual display for game streaming

Self-hosted game stream host for Moonlight.

Lightning Attention-2: A Free Lunch for Handling Unlimited Sequence Lengths in Large Language Models

MoBA: Mixture of Block Attention for Long-Context LLMs

BladeDISC is an end-to-end DynamIc Shape Compiler project for machine learning workloads.

[ICLR2025, ICML2025, NeurIPS2025 Spotlight] Quantized Attention achieves speedup of 2-5x compared to FlashAttention, without losing end-to-end metrics across language, image, and video models.

🐳 Efficient Triton implementations for "Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention"

Ongoing research training transformer models at scale

DeepEP: an efficient expert-parallel communication library

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

DeepGEMM: clean and efficient FP8 GEMM kernels with fine-grained scaling

FlashMLA: Efficient Multi-head Latent Attention Kernels

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

Tile primitives for speedy kernels