tau-2-bench

'Tau2 Bench' is a dataset/benchmark used to evaluate the performance of tool-use agents by providing structured tasks that capture interaction dynamics and the effectiveness of different strategies in tool invocation and environmental response.

Papers using tau-2-bench (15)

Towards General Agentic Intelligence via Environment Scaling2025From Confident Closing to Silent Failure: Characterizing False Success in LLM Agents2026Agent Planning Benchmark: A Diagnostic Framework for Planning Capabilities in LLM Agents2026Proper Scoring Rules for Agentic Uncertainty Quantification2026SkillsInjector: Dynamic Skill Context Construction for LLM Agents2026MemGym: a Long-Horizon Memory Environment for LLM Agents2026MAVEN: Improving Generalization in Agentic Tool Calling2026Robust Tool Use via Fission-GRPO: Learning to Recover from Execution Errors2026TopoCurate:Modeling Interaction Topology for Tool-Use Agent Training2026EnvFactory: Scaling Tool-Use Agents via Executable Environments Synthesis and Robust RL2026Step 3.5 Flash: Open Frontier-level Intelligence With 11B Active Parameters2026Toward Scalable Verifiable Reward: Proxy State-Based Evaluation for Multi-turn Tool-Calling LLM Agents2026AutoForge: Automated Environment Synthesis for Agentic Reinforcement Learning2025On Generalization in Agentic Tool Calling: CoreThink Agentic Reasoner and MAVEN Dataset2025Toolorchestra: Elevating Intelligence Via Efficient Model And Tool Orchestration2025