Executable, measurable, and reproducible AI4AI toward recursive self-improvement. Home of OpenMLE and Frontis-MA1.

AI agents running research on single-GPU nanochat training automatically

KernelBench: Can LLMs Write GPU Kernels? - Benchmark + Toolkit with Torch -> CUDA (+ more DSLs)

LLM-as-a-Verifier is a general-purpose framework that provides fine-grained feedback for any agent without requiring additional training. It achieves SOTA performance across coding, robotics, and m…

Programmable datacenter-scale infrastructure for Agents.

分享AI Infra知识&代码练习：PyTorch、vLLM/SGLang、slime/vime框架入门⚡️、性能加速🚀、大模型基础🧠、AI软硬件🔧等

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

TensorRT LLM provides users with an easy-to-use Python API to define Large Language Models (LLMs) and supports state-of-the-art optimizations to perform inference efficiently on NVIDIA GPUs. Tensor…

Achieve state of the art inference performance with modern accelerators on Kubernetes

An Extensible Deep Learning Library

[MLsys2026]: RAG on Everything with LEANN. Enjoy 97% storage savings while running a fast, accurate, and 100% private RAG application on your personal device.

Virtualized Elastic KV Cache for Dynamic GPU Sharing and Beyond

DeepSeek 4 Flash and PRO local inference engine for Metal, CUDA and ROCm

Distribute and run AI workloads on Kubernetes magically in Python, like PyTorch for ML infra.

RTP-LLM: Alibaba's high-performance LLM inference engine for diverse applications.

The agent that grows with you

JAX-Toolbox

Training library for Megatron-based models with bidirectional Hugging Face conversion capability

Ongoing research training transformer models at scale

JaxPP is a library for JAX that enables flexible MPMD pipeline parallelism for large-scale LLM training

Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more

Tokamax: A GPU and TPU kernel library.

Showcase JaxPP with MaxText

A machine learning compiler for GPUs, CPUs, and ML accelerators

Orbax provides common checkpointing and persistence utilities for JAX users

A profiling and performance analysis tool for machine learning

TPU inference for vLLM, with unified JAX and PyTorch support.