GitHub - Danau5tin/ai-trains-ai: RL-training an AI agent to RL-train AI agents.
RL-training an AI agent to RL-train AI agents. Contribute to Danau5tin/ai-trains-ai development by creating an account on GitHub.
Python Library
Summary
Microsoft's open-source Python library for training AI agents using reinforcement learning, automatic prompt optimization, and supervised fine-tuning. Integrates with any agent framework via a tracer that collects prompts, tool calls, and rewards without modifying existing code.
Highlights
- Zero code change integration with any agent framework (LangChain, AutoGen, CrewAI, OpenAI Agent SDK, or raw OpenAI) via a lightweight tracer
- Supports multiple optimization algorithms including Reinforcement Learning, Automatic Prompt Optimization, and Supervised Fine-tuning
- Can selectively optimize one or more agents within a multi-agent system
- Community projects demonstrate scaling to 128 GPUs and applications in game AI, SQL generation, and long-horizon tasks
- Active development with recent research on trajectory-level aggregation and token ID preservation for RL training
auto-generated
Context
Audience
AI developers and researchers building or optimizing LLM-based agents who want to apply reinforcement learning or prompt optimization without rewriting their agent code
DomainAI Agent Training and Reinforcement Learning
Formatopen-source software library with documentation and research papers
Accessfree, open source on GitHub; installable via pip
Source Overview
Agent Lightning is a Microsoft open-source Python library for training AI agents via reinforcement learning, automatic prompt optimization, and supervised fine-tuning.
It integrates with any agent framework (LangChain, AutoGen, CrewAI, OpenAI Agent SDK, or raw OpenAI) by inserting a tracer that collects prompts, tool calls, and rewards as structured spans without modifying existing agent code. A LightningStore routes those spans to a selected algorithm that outputs refined prompts or policy weights, while a Trainer orchestrates dataset streaming, resource updates, and inference engine refresh.
Topics
Related
LangChainAutoGenCrewAIOpenAI Agent SDKAgentScopeFlow-GRPO algorithm
