List your company

Mathematical Reasoning on SVAMP OOD

73Base AccuracyTHOUGHTSTEER
61.0464.14567.2570.355Apr 1, 2026

Evaluation results

MethodLinks
THOUGHTSTEERArchitecture=SimCoT (Llama-3.2) 3B, Decoding method=CODI2026.04737399.7
THOUGHTSTEERArchitecture=SimCoT (Llama-3.2) 1B, Decoding method=CODI2026.0461.56294.4