Tensors and Dynamic neural networks in Python with strong GPU acceleration

Lightweight coding agent that runs in your terminal

A high-throughput and memory-efficient inference and serving engine for LLMs

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

Ongoing research training transformer models at scale

🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.

A library for mechanistic interpretability of GPT-style language models

Legible, Scalable, Reproducible Foundation Models with Named Tensors and Jax

Async RL Training at Scale

A library for accelerating Transformer models on NVIDIA GPUs, including using 8-bit and 4-bit floating point (FP8 and FP4) precision on Hopper, Ada and Blackwell GPUs, to provide better performance…

Linux kernel source tree

A generative world for general-purpose robotics & embodied AI learning.

【三年面试五年模拟】AIGC算法工程师面试秘籍。涵盖AIGC、传统深度学习、自动驾驶、AI Agent、机器学习、计算机视觉、自然语言处理、强化学习、大数据挖掘、具身智能、元宇宙、AGI等AI行业面试笔试干货经验与核心知识。

Fine-tuning & Reinforcement Learning for LLMs. 🦥 Train OpenAI gpt-oss, DeepSeek-R1, Qwen3, Gemma 3, TTS 2x faster with 70% less VRAM.

PArallel Distributed Deep LEarning: Machine Learning Framework from Industrial Practice （『飞桨』核心框架，深度学习&机器学习高性能单机、分布式训练和跨平台部署）

Scalable, Portable and Distributed Gradient Boosting (GBDT, GBRT or GBM) Library, for Python, R, Java, Scala, C++ and more. Runs on single machine, Hadoop, Spark, Dask, Flink and DataFlow

Oh my tmux! My self-contained, pretty & versatile tmux configuration made with 💛🩷💙🖤❤️🤍

Making large AI models cheaper, faster and more accessible

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

Kimi K2 is the large language model series developed by Moonshot AI team

A Flexible Framework for Experiencing Cutting-edge LLM Inference Optimizations

verl: Volcano Engine Reinforcement Learning for LLMs

Post-training with Tinker

The official repository for ERNIE 4.5 and ERNIEKit – its industrial-grade development toolkit based on PaddlePaddle.

Building General-Purpose Robots Based on Embodied Foundation Model

MiniMax-M2, a model built for Max coding & agentic workflows.

Open deep learning compiler stack for cpu, gpu and specialized accelerators

DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.

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

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