Official inference repo for FLUX.2 models

Super basic implementation (gist-like) of RLMs with REPL environments.

Official PyTorch Implementation of "Diffusion Transformers with Representation Autoencoders"

PyTorch-native post-training at scale

A framework for few-shot evaluation of language models.

AMD RAD's multi-GPU Triton-based framework for seamless multi-GPU programming

Ship correct and fast LLM kernels to PyTorch

Examples demonstrating available options to program multiple GPUs in a single node or a cluster

Qwen-Image is a powerful image generation foundation model capable of complex text rendering and precise image editing.

Python tool for converting files and office documents to Markdown.

[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.

Super fast FP32 matrix multiplication on RDNA3

AI Tensor Engine for ROCm

Simple high-throughput inference library

Development repository for the Triton language and compiler

Official implementation of Half-Quadratic Quantization (HQQ)

A TTS model capable of generating ultra-realistic dialogue in one pass.

Framework to reduce autotune overhead to zero for well known deployments.

Distributed Compiler based on Triton for Parallel Systems

Gzip Decompression and Random Access for Modern Multi-Core Machines

SGLang is a fast serving framework for large language models and vision language models.

FlashMLA: Efficient Multi-head Latent Attention Kernels

Official Problem Sets / Reference Kernels for the GPU MODE Leaderboard!

Cost-efficient and pluggable Infrastructure components for GenAI inference