Spaces:
Running
Running
Ctrl+K
Update logbook: Repro - Characterizing, Evaluating, and Optimizing Complex Reasoning
203ba0b verified - a-official-claim-test-time-selection-up-to-19-3
- b-official-claim-thinking-rewards-in-rl-up-to-3-9
- claim-1-me2-principle
- claim-2-dag-construction
- claim-3-trm-training-bt-loss
- claim-4-validation-accuracy-88-6
- claim-5-best-of-n-on-aime
- claim-6-trm-guided-rl
- conclusion
- paper-approach
- trm-repro
- 803 Bytes