List your company

Math Reasoning on Math Reasoning 1.5B model (val)

69.4Validation AccuracyExecution-Guided Search
47.14452.92258.764.478Jan 20, 2026

Evaluation results

MethodLinks
Execution-Guided Searchmodel_size=1.5B, search_budget=10 epochs2026.0169.4
Best Human Expertmodel_size=1.5B2026.0168.8
Baselinemodel_size=1.5B, method_type=GRPO2026.0148