List your company

Multi-discipline Evaluation on MMLU-Pro

32.1Math AccuracyPOLAR
2.87610.46318.0525.637Jun 4, 2026

Evaluation results

MethodLinks
POLARBackbone=Qwen2.5-3B-Instruct, Evaluation Protocol=pass@12026.0632.135.63327.641.538.95346.557.433.46638.935.4
Base (τ=0)Backbone=Qwen2.5-3B-Instruct, Evaluation Protocol=pass@12026.0626.328.727.724.43733.947.540.352.631.362.233.734.6
FlexiDepthBackbone=Qwen2.5-3B-Instruct, Evaluation Protocol=pass@12026.069.89.310.510.411.411.510.811.59.98.210.611.411
ShortGPTBackbone=Qwen2.5-3B-Instruct, Evaluation Protocol=pass@12026.068.99.31012.112.112.812.613.48.911.510.39.812.1
MindSkipBackbone=Qwen2.5-3B-Instruct, Evaluation Protocol=pass@12026.067.67.58.369.57.48.69.87.58.57.59.46.8
DR.LLMBackbone=Qwen2.5-3B-Instruct, Evaluation Protocol=pass@12026.0644.82.83.233.22.82.61.44.65.834.2