ToolSandbox

The 'ToolSandbox' dataset/benchmark contains a collection of tasks designed to evaluate AI agents' abilities to perform multi-turn and multi-step interactions while utilizing external tools.

Papers using ToolSandbox (3)

Agent Planning Benchmark: A Diagnostic Framework for Planning Capabilities in LLM Agents2026CM2: Reinforcement Learning with Checklist Rewards for Multi-Turn and Multi-Step Agentic Tool Use2026WISE-Flow: Workflow-Induced Structured Experience for Self-Evolving Conversational Service Agents2026