InternLM/lagent

A lightweight framework for building LLM-based agents

69
/ 100
Established

Provides a PyTorch-inspired layer-based architecture where agents communicate via `AgentMessage` objects and compose multi-agent systems through message passing. Includes built-in memory management with session-based state tracking, customizable message aggregators, and flexible output parsers (supporting structured extraction like tool/code parsing). Integrates with multiple LLM backends including vLLM and OpenAI-compatible APIs, supporting both InternLM and open-weight models like Qwen.

2,230 stars and 2,438 monthly downloads. Available on PyPI.

Maintenance 13 / 25
Adoption 18 / 25
Maturity 18 / 25
Community 20 / 25

How are scores calculated?

Stars

2,230

Forks

223

Language

Python

License

Apache-2.0

Last pushed

Mar 06, 2026

Monthly downloads

2,438

Commits (30d)

0

Dependencies

17

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/transformers/InternLM/lagent"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.