Verifiable Hidden Dynamics Play: Generating Agentic RL Environments from Solved Mechanisms
Paper • 2609.27321 • Published • 15
Long-Horizon Agents
Evaluation and training for capable long-horizon agents.
Note VHD-Play: scalable agentic RL training with verifiable rewards.
E-Commerce Bench: Evaluating LLM Agents on Long-Horizon Autonomous Business Operation
Paper • 2608.30730 • Published • 17Note E-CommerceBench: long-horizon agent evaluation.
-
Shortcuts and Boundaries: Verifiable RL Misalignment and Contextual Safety Learning
Paper • 2611.12865 • Published • 5