BFCLv-3

The 'BFCLv3' dataset/benchmark is used to evaluate the effectiveness of tool-use agents by providing a structured representation of interaction dynamics through multi-trial rollouts of tasks.

Papers using BFCLv-3 (12)

D-CORE: Incentivizing Task Decomposition in Large Reasoning Models for Complex Tool Use2026Pushing the Limits of LLM Tool Calling via Experiential Knowledge Integration and Activation2026MAVEN: Improving Generalization in Agentic Tool Calling2026TopoCurate:Modeling Interaction Topology for Tool-Use Agent Training2026HINT-SD: Targeted Hindsight Self-Distillation for Long-Horizon Agents2026EnvFactory: Scaling Tool-Use Agents via Executable Environments Synthesis and Robust RL2026ToolWeave: Structured Synthesis of Complex Multi-Turn Tool-Calling Dialogues2026Controllable and Verifiable Tool-Use Data Synthesis for Agentic Reinforcement Learning2026MagicAgent: Towards Generalized Agent Planning2026Gecko: A Simulation Environment with Stateful Feedback for Refining Agent Tool Calls2026On Generalization in Agentic Tool Calling: CoreThink Agentic Reasoner and MAVEN Dataset2025Small Language Models For Agentic Systems: A Survey Of Architectures, Capabilities, And Deployment Trade Offs2025