A Streaming-Native Serving Engine for TTS/STS Models

AstroAccelerate is a many-core accelerated software package for processing time-domain radio-astronomy data.

UCCL is an efficient communication library for GPUs, covering collectives, P2P (e.g., KV cache transfer, RL weight transfer), and EP (e.g., GPU-driven)

RAG on Everything with LEANN. Enjoy 97% storage savings while running a fast, accurate, and 100% private RAG application on your personal device.

Open ABI and FFI for Machine Learning Systems

NVIDIA FastGen: Fast Generation from Diffusion Models

If tinygrad wasn't small enough for you...

JAX bindings for the FlashAttention 3 kernels

A course of learning LLM inference serving on Apple Silicon for systems engineers: build a tiny vLLM + Qwen.

common in-memory tensor structure

DFlash: Block Diffusion for Flash Speculative Decoding

A fast GPU memory copy library based on NVIDIA GPUDirect RDMA technology

a Jax quantization library

Official implementation of "Fast-dLLM: Training-free Acceleration of Diffusion LLM by Enabling KV Cache and Parallel Decoding"

LLaDA2.0 is the diffusion language model series developed by InclusionAI team, Ant Group.

d3LLM: Ultra-Fast Diffusion LLM 🚀

Dream 7B, a large diffusion language model

Open-source unified multimodal model

PyTorch/TorchScript/FX compiler for NVIDIA GPUs using TensorRT

Enjoy the magic of Diffusion models!

Miles is an enterprise-facing reinforcement learning framework for LLM and VLM post-training, forked from and co-evolving with slime.

An MLIR-based compiler framework bridges DSLs (domain-specific languages) to DSAs (domain-specific architectures).

An Agile RISC-V SoC Design Framework with in-order cores, out-of-order cores, accelerators, and more

TTS model capable of streaming conversational audio in realtime.

A scalable generative AI framework built for researchers and developers working on Large Language Models, Multimodal, and Speech AI (Automatic Speech Recognition and Text-to-Speech)

A modular, primitive-first, python-first PyTorch library for Reinforcement Learning.

A very fast linker for Linux

A Lightweight LLM Post-Training Library

🚀 Efficient implementations of state-of-the-art linear attention models

Distributed execution for duckdb queries.