List your company

Dense Model Loss Curve Prediction Cosine to WSD transfer

0.404MAPECMMT (λ = 0.99)
0.386640.503820.6210.73818Dec 5, 2025

Evaluation results

MethodLinks
CMMT (λ = 0.99)Model Size=100M2025.120.404
ASMTModel Size=100M2025.120.41
CMMT (λ = 0.999)Model Size=100M2025.120.422
CMMT (λ = 0.99)Model Size=500M2025.120.657
ASMTModel Size=50M2025.120.72
CMMT (λ = 0.99)Model Size=50M2025.120.777
CMMT (λ = 0.999)Model Size=500M2025.120.784
ASMTModel Size=500M2025.120.798
CMMT (λ = 0.999)Model Size=50M2025.120.838