A natural language interface for computers

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

Universal memory layer for AI Agents

A modular graph-based Retrieval-Augmented Generation (RAG) system

An autonomous agent that conducts deep research on any data using any LLM providers.

A framework for few-shot evaluation of language models.

Python bindings for llama.cpp

Agent Reinforcement Trainer: train multi-step agents for real-world tasks using GRPO. Give your agents on-the-job training. Reinforcement learning for Qwen2.5, Qwen3, Llama, and more!

Examples in the MLX framework

SoTA production-ready AI retrieval system. Agentic Retrieval-Augmented Generation (RAG) with a RESTful API.

Efficient Triton Kernels for LLM Training

Open-source implementation of AlphaEvolve

Agent Laboratory is an end-to-end autonomous research workflow meant to assist you as the human researcher toward implementing your research ideas

A Next-Generation Training Engine Built for Ultra-Large MoE Models

Democratizing Reinforcement Learning for LLMs

[CVPR 2024] Official RT-DETR (RTDETR paddle pytorch), Real-Time DEtection TRansformer, DETRs Beat YOLOs on Real-time Object Detection. 🔥 🔥 🔥

A framework for serving and evaluating LLM routers - save LLM costs without compromising quality

SWE-bench: Can Language Models Resolve Real-world Github Issues?

Multi-LoRA inference server that scales to 1000s of fine-tuned LLMs

Run PyTorch LLMs locally on servers, desktop and mobile

Entropy Based Sampling and Parallel CoT Decoding

TextGrad: Automatic ''Differentiation'' via Text -- using large language models to backpropagate textual gradients. Published in Nature.

Optimizing inference proxy for LLMs

A library for mechanistic interpretability of GPT-style language models

[ICLR 2025] LongWriter: Unleashing 10,000+ Word Generation from Long Context LLMs

Mora: More like Sora for Generalist Video Generation

Agentic AI framework for enterprise workflow automation.

ToRA is a series of Tool-integrated Reasoning LLM Agents designed to solve challenging mathematical reasoning problems by interacting with tools [ICLR'24].

Open source implementation of AlphaFold3

LiveBench: A Challenging, Contamination-Free LLM Benchmark