List your company

Mathematical Reasoning on GSM8K (full set)

91.43AccuracyIRIS-single
85.751687.225888.790.1742May 8, 2026

Evaluation results

MethodLinks
IRIS-singleBudget=256/512/128, Model=Qwen3-8B, Seed=422026.0591.43204.42
MrsdRounds=3-round, Model=Qwen3-8B, Seed=422026.0590.9287.51
NothinkBudget=256, Model=Qwen3-8B, Seed=422026.0587.49146.31
TownBudget=256→512, Model=Qwen3-8B, Seed=422026.0585.97203.56