Running
Value-as-Tool GVR Tree Explorer
🌳
Branched GVR rollouts of Qwen3.5-9B on ArXivMath
Branched GVR rollouts of Qwen3.5-9B on ArXivMath
Problem-by-problem viewer of proof-harness rollouts
Search tasks, rubrics, difficulty evidence and references
Explore task rubrics and the variety of valid responses.
Inspect selection diversity for Amazon C2 rubrics
Explore Amazon C11 distillation data and reviewer metrics