RoTA-Bench: Evaluating 3D Spatial Reasoning through Interactive Projection Alignment Paper • 2610.17665 • Published 16 days ago • 77
Scoring Higher, Answering Worse: Mitigating Reward Hacking in Rubric-Based RL via Protocol-Level Rubrics Paper • 2609.38847 • Published 30 days ago • 32
FACET: Preserving Source Intent and Executable State in Terminal Task Synthesis Paper • 2608.18580 • Published Aug 19 • 43