ResearchBenchmarksVisual Question Answering on Yo’LLaVA Single Concept (test)Visual Question Answering on Yo’LLaVA Single Concept (test)Follow97.6AccuracyRAP92.08893.51994.9596.381Mar 10, 2026Evaluation resultsMethodMethodLinksAccuracyRAPSample Runtime (s)=7.8Sample Runtime (s)=7.82026.0397.6PeKitSample Runtime (s)=5.8Sample Runtime (s)=5.82026.0394.6R2PSample Runtime (s)=7.0Sample Runtime (s)=7.02026.0394EgoSample Runtime (s)=6.0Sample Runtime (s)=6.02026.0392.3