Saltar al contenido
TakoAPI
Volver a skills

agent-lightning

por olmmlo-cmd

Search & Research
419Descargas según clawskills.sh
0Vistas
0 me gustaInicia sesión para dar me gusta a esta skill
—Aún sin valoraciones
Tu valoración

Resumen

Microsoft Research's agent training framework.

Descripción

Qué hace esta skill

Agent Lightning is Microsoft Research's framework for training AI agents using Reinforcement Learning, Automatic Prompt Optimization, and Supervised Fine-tuning. It instruments existing agents with minimal code changes via emit helpers or a tracer wrapper. Supports LangChain, AutoGen, CrewAI, and OpenAI Agent SDK out of the box.

Unlike manual prompt engineering or full model retraining, Agent Lightning applies RL and prompt optimization directly to your existing agent code without requiring a rewrite or framework switch.

Cuándo usarla

  • Improving a SQL-generation agent's accuracy through RL training
  • Tuning system prompts automatically for a customer support agent
  • Fine-tuning a coding assistant on preference data
  • Running benchmark evaluations on a multi-agent pipeline
  • Exporting optimized prompts or model weights for production deployment

Flujo de trabajo de ejemplo

ENTRADA

User asks: train my LangChain agent on SQL generation tasks using reinforcement learning

AGENTE
  1. 1

    Instrument the agent by wrapping it with agentlightning tracer or adding emit_tool_call helpers

  2. 2

    Create a config.yaml specifying agent type as langchain, algorithm as grpo, and eval tasks including sql-generation

  3. 3

    Run agent-lightning train --config config.yaml to start training

  4. 4

    Monitor loss curves and reward metrics via agent-lightning dashboard --port 8080

  5. 5

    Export the trained prompts or model with agent-lightning export for deployment

SALIDA

Optimized agent with improved SQL generation accuracy, exportable for production use

Requisitos

  • OPENAI_API_KEY environment variable

Instalación

clawhub install olmmlo-cmd/agent-lightning

Ejecuta el comando de la CLI o pega el prompt en Claude Code, Cursor, Codex o el agente de programación que uses.

Enlaces