List your company

Commonsense Reasoning on CommonsenseQA (Acc, Tokens, Ratio)

57AccuracyMCTS + EquivPruner
55.9656.2356.556.77May 22, 2025

Evaluation results

MethodLinks
MCTS + EquivPrunerModel=Qwen2.5-Math-7B-Instruct, Search Strategy=MCTS + EquivPruner2025.055735,12988.2
Vanilla MCTSModel=Qwen2.5-Math-7B-Instruct, Search Strategy=Vanilla MCTS2025.055639,827100