Share your thoughts, 1 month free Claude Pro on usSee more

Language Understanding on MMLU (Score x100)

65.81MMLU Score (x100)

UNIFORM RANDOM (50K×0.6)

Updated 1mo ago

Evaluation Results

Method	Links
UNIFORM RANDOM (50K×0.6) 2025.02		65.81
FULL TOKENS (50K) 2025.02		65.78
DS² (10K) 2025.02		65.77
RHO 2025.02		65.74
FIXED-MODEL CLEANING 2025.02		65.71
BASE 2025.02		65.29
SELF-EVOLVING CLEANING 2025.02		65.07
DS² (10K) 2025.02		62.5
UNIFORM RANDOM (50K×0.6) 2025.02		62.47
BASE 2025.02		62.41
FULL TOKENS (50K) 2025.02		62.41
SELF-EVOLVING CLEANING 2025.02		62.3
RHO 2025.02		62.12
FIXED-MODEL CLEANING 2025.02		61.45
RHO 2025.02		57.1
FIXED-MODEL CLEANING 2025.02		57.09
UNIFORM RANDOM (50K×0.6) 2025.02		56.96
DS² (10K) 2025.02		56.93
FULL TOKENS (50K) 2025.02		56.87
BASE 2025.02		56.29
SELF-EVOLVING CLEANING 2025.02		56.18