List your company

Visual Question Answering on OKVQA (VQA accuracy)

59.57VQA AccuracyTPC
4.231618.598332.96547.3317Feb 3, 2025Apr 29, 2025Jul 24, 2025Oct 18, 2025Jan 12, 2026Apr 8, 2026Jul 3, 2026

Evaluation results

MethodLinks
TPCIntegration Strategy=Ours (core–residual alignment), Base LVLM=LLaVA-1.5-7B2026.0759.57
CLIPPerturbation Budget=clean, Attack=APGD2025.0259.04
Patch-Aligned TrainingIntegration Strategy=Projector / fine-grained alignment (LVLM-side training), Base LVLM=LLaVA-1.5-7B2026.0758.3
Adv-W2SIntegration Strategy=Robust vision encoder replacement, Base LVLM=LLaVA-1.5-7B2026.0756.8
Robust-LLaVA4_GPerturbation Budget=clean, Attack=APGD2025.0254.88
FARE4Perturbation Budget=clean, Attack=APGD2025.0254.64
FAREIntegration Strategy=Robust vision encoder replacement, Base LVLM=LLaVA-1.5-7B2026.0754.1
ImageNet Std ViT-LPerturbation Budget=clean, Attack=APGD2025.0253.64
LLaVA1.5 (Regular)Integration Strategy=Decoding / concept-level enhancement (frozen LVLM), Base LVLM=LLaVA-1.5-7B2026.0753.4
Sim-CLIP4Perturbation Budget=clean, Attack=APGD2025.0253.04
VL-SAEIntegration Strategy=Decoding / concept-level enhancement (frozen LVLM), Base LVLM=LLaVA-1.5-7B2026.0753
PMG-AFTIntegration Strategy=Robust vision encoder replacement, Base LVLM=LLaVA-1.5-7B2026.0752.6
TGA-ZSRIntegration Strategy=Robust vision encoder replacement, Base LVLM=LLaVA-1.5-7B2026.0752
VCDIntegration Strategy=Decoding / concept-level enhancement (frozen LVLM), Base LVLM=LLaVA-1.5-7B2026.0751.9
TeCoAIntegration Strategy=Robust vision encoder replacement, Base LVLM=LLaVA-1.5-7B2026.0751.8
Robust-LLaVA4_GPerturbation Budget=2/255, Attack=APGD2025.0249.68
ImageNet Robust ViT-BPerturbation Budget=clean, Attack=APGD2025.0244.96
Robust-LLaVA4_GPerturbation Budget=4/255, Attack=APGD2025.0243.76
ImageNet Robust ViT-BPerturbation Budget=2/255, Attack=APGD2025.0239.36
Sim-CLIP4Perturbation Budget=2/255, Attack=APGD2025.0236.2
Robust-LLaVA4_GPerturbation Budget=8/255, Attack=APGD2025.0234.04
FARE4Perturbation Budget=2/255, Attack=APGD2025.0232.32
ImageNet Robust ViT-BPerturbation Budget=4/255, Attack=APGD2025.0231.92
Sim-CLIP4Perturbation Budget=4/255, Attack=APGD2025.0230.88
FARE4Perturbation Budget=4/255, Attack=APGD2025.0229.88
FARE4Perturbation Budget=8/255, Attack=APGD2025.0224.52
ImageNet Robust ViT-BPerturbation Budget=8/255, Attack=APGD2025.0223.48
Sim-CLIP4Perturbation Budget=8/255, Attack=APGD2025.0222.96
CLIPPerturbation Budget=2/255, Attack=APGD2025.0216.64
CLIPPerturbation Budget=4/255, Attack=APGD2025.0213.88
CLIPPerturbation Budget=8/255, Attack=APGD2025.0211.56
ImageNet Std ViT-LPerturbation Budget=2/255, Attack=APGD2025.0211.16
ImageNet Std ViT-LPerturbation Budget=4/255, Attack=APGD2025.028.52
ImageNet Std ViT-LPerturbation Budget=8/255, Attack=APGD2025.026.36