My learning notes for ML SYS.

A version of verl to support diverse tool use

Agent-R1: Training Powerful LLM Agents with End-to-End Reinforcement Learning

Open Source Deep Research Alternative to Reason and Search on Private Data. Written in Python.

Building Open-Ended Embodied Agents with Internet-Scale Knowledge

An Open-Ended Embodied Agent with Large Language Models

R1-searcher: Incentivizing the Search Capability in LLMs via Reinforcement Learning

Search-R1: An Efficient, Scalable RL Training Framework for Reasoning & Search Engine Calling interleaved LLM based on veRL

ReCall: Learning to Reason with Tool Call for LLMs via Reinforcement Learning

Lightning-Fast RL for LLM Reasoning and Agents. Made Simple & Flexible.

Train your Agent model via our easy and efficient framework

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!

The absolute trainer to light up AI agents.

TROLL: Trust Region Optimization for Large Language models

CUDA-L1: Improving CUDA Optimization via Contrastive Reinforcement Learning

(best/better) practices of megatron on veRL and tuning guide

EasyR1: An Efficient, Scalable, Multi-Modality RL Training Framework based on veRL

Paper reading and discussion notes, covering AI frameworks, distributed systems, cluster management, etc.

A set of examples based on verl for end-to-end RL training recipes.

[ASPLOS'26] Taming the Long-Tail: Efficient Reasoning RL Training with Adaptive Drafter

A construction kit for reinforcement learning environment management.

A GPU-accelerated library containing highly optimized building blocks and an execution engine for data processing to accelerate deep learning training and inference applications.