BigHugger
GH Repository · Agent-RL

ReCall

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

stars
1,437
30-day movement
+41/day
Related entries
60
Connections
1
agent-appPythonpythontool-usereinforcement-learningllmfunction-callingagent

ReCall is a Python project from the Agent-RL organization that trains LLMs to reason with tool calls via reinforcement learning, extending the earlier ReSearch approach that used search as the tool. The repository ships the ReCall implementation under an MIT licence.

You want a ready-made RL training setup for teaching LLMs to interleave reasoning with tool calls.

Use it to

  • Train an LLM to reason with tool calls via RL
  • Reproduce or extend the ReSearch search-based reasoning method
  • Experiment with function-calling agents trained by reinforcement learning
  • Study RL-based tool-use training pipelines in Python

For Researchers and engineers training tool-using LLM agents

Role
agent-app
Language
Python
Licence
MIT
Forks
88
Open issues
30
Last push
2025-05-16
topicsreinforcement-learningtool-usefunction-callingllmagent