List your company

Question Answering on MMLU, TruthfulQA, and BBQ

69.3MMLU AccuracyZero-shot
66.523267.244167.96568.6859Dec 1, 2025Dec 28, 2025Jan 25, 2026Feb 22, 2026Mar 22, 2026Apr 19, 2026May 17, 2026

Evaluation results

MethodLinks
Zero-shot#Params=x12026.0569.351.2780.6967.09
DiDi-Merging-L#Params=x2.02026.0568.9758.2493.3273.51
DiDi-Merging-M#Params=x1.42026.0568.8557.3292.4472.87
FREE-Merging#Params=x2.042026.0568.8357.3992.1472.78
IndividualBackbone=Qwen-14B2025.1268.3654.3593.5372.08
Fine-tuned#Params=x32026.0568.3554.3493.5372.07
DTS-DBackbone=Qwen-14B, AMR=3.57, r=0.32025.1268.3254.1192.9971.81
Twin-Merging#Params=x2.132026.0568.3255.7690.9871.68
DTS-TBackbone=Qwen-14B, AMR=3.57, r=0.32025.1268.354.1292.9771.8
DTS-T*Backbone=Qwen-14B, AMR=0.92, r=adaptive2025.1268.253.9992.971.7
DTS-D*Backbone=Qwen-14B, AMR=0.92, r=adaptive2025.1268.1853.9992.9171.7
FREE-MergingBackbone=Qwen-14B, AMR=10.002025.1268.1353.9192.5471.52
Twin-MergingBackbone=Qwen-14B, AMR=100.02025.1268.0752.3890.7370.39
T-SwitchBackbone=Qwen-14B, AMR=6.252025.1268.0553.7292.571.42
SVDBackbone=Qwen-14B, AMR=5.392025.1267.9952.4591.7170.72
EMR-MergingBackbone=Qwen-14B, AMR=15.622025.1267.9452.591.0270.49
EMR-Merging#Params=x42026.0567.8255.0190.1370.98
Ties-MergingBackbone=Qwen-14B, AMR=0.002025.1267.2850.0284.167.13
DAREBackbone=Qwen-14B, AMR=0.002025.1267.2351.3183.7467.43
Weight-AveragingBackbone=Qwen-14B, AMR=0.002025.1267.1150.0282.3266.48
Task-ArithmeticBackbone=Qwen-14B, AMR=0.002025.1266.6353.3878.2466.08