List your company

Logical Reasoning on K&K Logic Puzzles

96Accuracy (Level 3)Lopti
6.5629.785376.22Oct 29, 2025

Evaluation results

MethodLinks
LoptiBackbone=Qwen2.5-7B-Instruct-1M2025.1096928681828716
ARBackbone=Qwen2.5-7B-Instruct-1M2025.1095978984808918.7
GRPO-SGBackbone=Qwen2.5-7B-Instruct-1M2025.1095959287849121.3
80/20Backbone=Qwen2.5-7B-Instruct-1M2025.1094938478748513.3
GRPOBackbone=Qwen2.5-7B-Instruct-1M2025.10929074596075—
GRPO-SGBackbone=Qwen2.5-3B-Instruct2025.1076766159446350
LoptiBackbone=Qwen2.5-3B-Instruct2025.1070695746345531
80/20Backbone=Qwen2.5-3B-Instruct2025.1067585347275019
ARBackbone=Qwen2.5-3B-Instruct2025.1065605444365223.8
GRPOBackbone=Qwen2.5-3B-Instruct2025.10634732372942—
Qwen2.5-7B-Instruct-1MBackbone=Qwen2.5-7B-Instruct-1M2025.1024181110413—
Qwen2.5-3B-InstructBackbone=Qwen2.5-3B-Instruct2025.1010107627—