Math Reasoning on Math Reasoning 1.5B model (val)
69.4Validation AccuracyExecution-Guided Search
Evaluation Results
| Method | Links | |
|---|---|---|
| Execution-Guided Searchmodel_size=1.5B, search_budget=10 epochs2026.01 | 69.4 | |
| Best Human Expertmodel_size=1.5B2026.01 | 68.8 | |
| Baselinemodel_size=1.5B, method_type=GRPO2026.01 | 48 |