List your company
ResearchFormerly WizWand

Generalization Assessment on Aggregate Unseen Tasks

101.7Relative Performance (%)MAGIC
91.40494.07796.7599.423May 25, 2026
Updated 5mo ago

Evaluation results

MethodLinks
MAGICTraining Budget=20%, Backbone=LLaVA-1.5-13B2026.05101.7
ICONSTraining Budget=20%, Backbone=LLaVA-1.5-13B2026.05101.6
Full-FinetuneTraining Budget=100%, Backbone=LLaVA-1.5-13B2026.05100
EL2NTraining Budget=20%, Backbone=LLaVA-1.5-13B2026.0597.8
SemDeDupTraining Budget=20%, Backbone=LLaVA-1.5-13B2026.0597.8
COINCIDETraining Budget=20%, Backbone=LLaVA-1.5-13B2026.0597.2
PerplexityTraining Budget=20%, Backbone=LLaVA-1.5-13B2026.0597
D2-PruningTraining Budget=20%, Backbone=LLaVA-1.5-13B2026.0596.7
RDSTraining Budget=20%, Backbone=LLaVA-1.5-13B2026.0595.6
Self-SupTraining Budget=20%, Backbone=LLaVA-1.5-13B2026.0595
Self-FilterTraining Budget=20%, Backbone=LLaVA-1.5-13B2026.0594
CLIP-ScoreTraining Budget=20%, Backbone=LLaVA-1.5-13B2026.0593.9
RandomTraining Budget=20%, Backbone=LLaVA-1.5-13B2026.0591.8