List your company

Machine Translation on WMT En-De 14 (val)

27.61BLEUG-XKD
25.94626.37826.8127.242Sep 13, 2024Dec 8, 2024Mar 4, 2025May 30, 2025Aug 24, 2025Nov 18, 2025Feb 13, 2026

Evaluation results

MethodLinks
G-XKDInitialization Setting=large→base2026.0227.61—
MiniLLMInitialization Setting=large→base2026.0227.54—
S-XKDInitialization Setting=large→base2026.0227.44—
GKDInitialization Setting=large→base2026.0227.43—
SeqXKDInitialization Setting=large→base2026.0227.28—
SeqKDInitialization Setting=large→base2026.0227.26—
SKDInitialization Setting=large→base2026.0227.24—
SFTInitialization Setting=large→base2026.0227.22—
G-XKDInitialization Setting=large→small2026.0226.73—
MiniLLMInitialization Setting=large→small2026.0226.65—
S-XKDInitialization Setting=large→small2026.0226.63—
GKDInitialization Setting=large→small2026.0226.61—
S-STEArchitecture=Transformer-base, Avg epoch loss=4.6172024.0926.534.011
DenseArchitecture=Transformer-base, Avg epoch loss=4.5552024.0926.493.977
SeqXKDInitialization Setting=large→small2026.0226.23—
SeqKDInitialization Setting=large→small2026.0226.15—
SKDInitialization Setting=large→small2026.0226.1—
SR-STEArchitecture=Transformer-base, Avg epoch loss=4.612024.0926.084.023
SFTInitialization Setting=large→small2026.0226.04—
STEPArchitecture=Transformer-base, Avg epoch loss=4.6822024.0926.014.085