Tool-R0: Self-Evolving LLM Agents for Tool-Learning from Zero Data

TL;DR

Tool-R0 introduces a self-evolving framework for training general-purpose tool-calling LLM agents from scratch using self-play reinforcement learning, eliminating the need for pre-existing datasets and enabling autonomous evolution.

Contribution

It presents the first zero-data, self-evolving RL approach for training tool-use LLM agents through co-evolution of generator and solver components.

Findings

Achieves 92.5% relative improvement over base models.

Outperforms supervised baselines in tool-calling tasks.

Provides empirical insights into self-play dynamics and scaling behaviors.

Abstract

Large language models (LLMs) are becoming the foundation for autonomous agents that can use tools to solve complex tasks. Reinforcement learning (RL) has emerged as a common approach for injecting such agentic capabilities, but typically under tightly controlled training setups. It often depends on carefully constructed task-solution pairs and substantial human supervision, which creates a fundamental obstacle to open-ended self-evolution toward superintelligent systems. In this paper, we propose Tool-R0 framework for training general purpose tool-calling agents from scratch with self-play RL, under a zero-data assumption. Initialized from the same base LLM, Tool-R0 co-evolves a Generator and a Solver with complementary rewards: one proposes targeted challenging tasks at the other's competence frontier and the other learns to solve them with real-world tool calls. This creates a self-evolving cycle that requires no pre-existing tasks or datasets. Evaluation on different tool-use benchmarks show that Tool-R0 yields 92.5 relative improvement over the base model and surpasses fully supervised tool-calling baselines under the same setting. Our work further provides empirical insights into self-play LLM agents by analyzing co-evolution, curriculum dynamics, and scaling behavior.