List your company

Generalization on LLaMA Evaluation Expanded Languages 3.1-8B

69.37Overall ScoreDeltaMoE
58.50261.323564.14566.9665May 18, 2026

Evaluation results

MethodLinks
DeltaMoECategory=Baseline w/ Matched Parameters2026.0569.37
MoLACategory=Baseline w/ Matched Parameters2026.0568.39
Dense-FT-DeltaCategory=Baselines w/ Same CPT Data2026.0566.88
Dense-FT-AvgCategory=Baselines w/ Same CPT Data2026.0566.39
Dense-FT-Avg-2FLOPsCategory=Baselines w/ Matched FLOPs (2x CPT Data)2026.0565.81
LLaMA-3.1-8B-InstructCategory=Base Model2026.0564.91
Dense-FT-Delta-2FLOPsCategory=Baselines w/ Matched FLOPs (2x CPT Data)2026.0564.38
LLaMA-ProCategory=Baseline w/ Matched Parameters2026.0558.92