Tongyi Deep Research, the Leading Open-source Deep Research Agent

Post-training with Tinker

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Latest Advances on Embodied Multimodal LLMs (or Vison-Language-Action Models).

[NeurIPS'25] Official codebase for "SWE-RL: Advancing LLM Reasoning via Reinforcement Learning on Open Software Evolution"

[ICML 2025] Satori: Reinforcement Learning with Chain-of-Action-Thought Enhances LLM Reasoning via Autoregressive Search

Official Repo for Open-Reasoner-Zero

AllenAI's post-training codebase

verl: Volcano Engine Reinforcement Learning for LLMs

Agentless🐱: an agentless approach to automatically solve software development problems

DSPy: The framework for programming—not prompting—language models

A lightweight reproduction of DeepSeek-R1-Zero with indepth analysis of self-reflection behavior.

Minimalistic large language model 3D-parallelism training

[NeurIPS 2024] OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments

🙌 OpenHands: Code Less, Make More

🌾 OAT: A research-friendly framework for LLM online alignment, including reinforcement learning, preference learning, etc.

Lighteval is your all-in-one toolkit for evaluating LLMs across multiple backends

Fully open reproduction of DeepSeek-R1

An Easy-to-use, Scalable and High-performance RLHF Framework based on Ray (PPO & GRPO & REINFORCE++ & vLLM & Ray & Dynamic Sampling & Async Agentic RL)

Data and Code for CVPR 2025 paper "MMVU: Measuring Expert-Level Multi-Discipline Video Understanding"

CRUXEval: Code Reasoning, Understanding, and Execution Evaluation

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

Ray is an AI compute engine. Ray consists of a core distributed runtime and a set of AI Libraries for accelerating ML workloads.

[ICML'24] Magicoder: Empowering Code Generation with OSS-Instruct

Entropy Based Sampling and Parallel CoT Decoding