List your company
ResearchFormerly WizWand

Image Classification on Food101 (Accuracy)

95.4AccuracyVeRA (r = 256)
84.37687.23890.192.962Aug 26, 2025Oct 10, 2025Nov 24, 2025Jan 9, 2026Feb 23, 2026Apr 9, 2026May 25, 2026
Updated 5mo ago

Evaluation results

MethodLinks
VeRA (r = 256)Backbone=CLIP-ViT-L/14, Params (M)=0.232026.0495.4
RandLoRA (r = 10)Backbone=CLIP-ViT-L/14, Params (M)=12.782026.0495.4
GiVA (r = 32, TopSV-B)Backbone=CLIP-ViT-L/14, Params (M)=0.202026.0495.4
GiVA (r = 32, SecSV-B)Backbone=CLIP-ViT-L/14, Params (M)=0.202026.0495.4
GiVA (r = 32, Rand-B)Backbone=CLIP-ViT-L/14, Params (M)=0.202026.0495.3
LoRA (r = 32)Backbone=CLIP-ViT-L/14, Params (M)=12.582026.0495.2
Full Fine-TuningBackbone=CLIP-ViT-L/14, Params (M)=303.282026.0495
SigLIP 2-B/16Evaluation Protocol=Linear Probe2026.0594.7
DINOv3-B/16Evaluation Protocol=Linear Probe2026.0594.47
LAGOBackbone=L/14, Zero-shot=true2026.0594.35
RandLoRA (r = 10)Backbone=DINOv2-ViT-B/14, Params (M)=3.41±0.02026.0494.1
BiFTABackbone=L/14, Zero-shot=true2026.0593.97
WCABackbone=L/14, Zero-shot=true2026.0593.96
Full Fine-TuningBackbone=DINOv2-ViT-B/14, Params (M)=86.732026.0493.9
LoRA (r = 32)Backbone=DINOv2-ViT-B/14, Params (M)=4.722026.0493.8
CuPLBackbone=L/14, Zero-shot=true2026.0593.37
WaffleBackbone=L/14, Zero-shot=true2026.0593.33
DFNCLIP-B/16Evaluation Protocol=Linear Probe2026.0593.32
VeRA (r = 256)Backbone=DINOv2-ViT-B/14, Params (M)=0.092026.0493.2
GiVA (r = 32, Rand-B)Backbone=DINOv2-ViT-B/14, Params (M)=0.082026.0493.2
GiVA (r = 32, TopSV-B)Backbone=DINOv2-ViT-B/14, Params (M)=0.082026.0493.1
GiVA (r = 32, SecSV-B)Backbone=DINOv2-ViT-B/14, Params (M)=0.082026.0493.1
CLIP-EBackbone=L/14, Zero-shot=true2026.0593.07
CLIP-DBackbone=L/14, Zero-shot=true2026.0593.03
DINOv2-reg-B/14Evaluation Protocol=Linear Probe2026.0592.87
CLIPBackbone=L/14, Zero-shot=true2026.0592.55
VECA-B/16Evaluation Protocol=Linear Probe, Core tokens (C)=642026.0592.44
CLIP-B/16Evaluation Protocol=Linear Probe2026.0592.42
DINOv2-B/14Evaluation Protocol=Linear Probe2026.0592.02
AM-RADIOv2.5-B/16Evaluation Protocol=Linear Probe2026.0591.64
OpenCLIP-B/16Evaluation Protocol=Linear Probe2026.0591.33
LAGOBackbone=B/16, Zero-shot=true2026.0590.54
BiFTABackbone=B/16, Zero-shot=true2026.0590.11
VECA-B/16Evaluation Protocol=Linear Probe, Core tokens (C)=82026.0590.04
WCABackbone=B/16, Zero-shot=true2026.0590.01
WaffleBackbone=B/16, Zero-shot=true2026.0589.06
CuPLBackbone=B/16, Zero-shot=true2026.0588.98
CLIP-DBackbone=B/16, Zero-shot=true2026.0588.85
CLIP-EBackbone=B/16, Zero-shot=true2026.0588.83
CLIPBackbone=B/16, Zero-shot=true2026.0588.4
PromptSRC+OursShots=42026.0587.8
EBMmode=Fine-tuning2025.1087.6
MaPLe + OursTest-time prompt tuning=True2026.0487.41
CoCoOpShots=42026.0587.2
CoCoOpmode=Fine-tuning2025.1087.2
PLOTShots=42026.0587.1
GalLoP+OursShots=42026.0587.1
PLOTmode=Fine-tuning2025.1087.1
LAGOBackbone=B/32, Zero-shot=true2026.0586.96
ProDAShots=42026.0586.8
A3B2Backbone=ViT-B/16, Training Shots=16-shot2026.0586.75
TCPBackbone=ViT-B/16, Training Shots=16-shot2026.0586.73
TCPBackbone=ViT-B/16, Training Shots=16-shot, Training Source=ImageNet2026.0586.69
MAIL++Backbone=ViT-B/16, Training Shots=16-shot, Training Source=ImageNet2026.0586.63
TCP+OursShots=42026.0586.6
DeKgBackbone=ViT-B/16, Training Shots=16-shot, Training Source=ImageNet2026.0586.59
PromptSRCSource Dataset=ImageNet, Evaluation Protocol=Cross-dataset transfer (zero-shot)2025.1186.52
PromptSRCShots=42026.0586.5
GalLoPShots=42026.0586.5
MMRL + NeRP2026.0586.5
PromptSRCmode=Fine-tuning2025.1086.5
BiFTABackbone=B/32, Zero-shot=true2026.0586.43
WCABackbone=B/32, Zero-shot=true2026.0586.4
PromptSRC + NeRP2026.0586.4
MMRL2026.0586.4
MMRLBackbone=ViT-B/16, Training Shots=16-shot, Training Source=ImageNet2026.0586.4
KgCoOp2025.0986.36
MaPLeSource Dataset=ImageNet, Evaluation Protocol=Cross-dataset transfer (zero-shot)2025.1186.36
MAILBackbone=ViT-B/16, Training Shots=16-shot, Training Source=ImageNet2026.0586.33
CoCoOp + NeRP2026.0586.3
BIP-DBackbone=ViT-B/16, Training Shots=16-shot, Training Source=ImageNet2026.0586.27
KgCoOpBackbone=ViT-B/16, Training Shots=16-shot2026.0586.24
AXISArchitecture=ViT-B-322025.0886.23
CLIP-SVD2025.0986.21
MaPLe2025.0986.2
MaPLeBackbone=ViT-B/16, Training Shots=16-shot, Training Source=ImageNet2026.0586.2
PromptSRC2026.0586.15
PromptSRCBackbone=ViT-B/16, Training Shots=16-shot, Training Source=ImageNet2026.0586.15
MMA2026.0586.12
MMA + NeRP2026.0586.12
MMABackbone=ViT-B/16, Training Shots=16-shot, Training Source=ImageNet2026.0586.12
CLIP2025.0986.11
Co-CoOp2025.0986.06
CoCoOp2026.0586.06
CoOpOpBackbone=ViT-B/16, Training Shots=16-shot, Training Source=ImageNet2026.0586.06
CLIP ViT-B/16Alignment train data=WIT400M, Zero-shot=true2026.0586.04
CoOpBackbone=ViT-B/16, Training Shots=16-shot2026.0585.82
MMRL++Backbone=ViT-B/16, Training Shots=16-shot2026.0585.72
MMRLBackbone=ViT-B/16, Training Shots=16-shot2026.0585.68
aTLASArchitecture=ViT-B-322025.0885.63
MaPLeBackbone=ViT-B/16, Training Shots=16-shot2026.0585.63
TIP-Adapter-F+OursShots=42026.0585.6
OursTest-time prompt tuning=True2026.0485.45
ProGrad2025.0985.4
TCP*Shots=42026.0585.4
CoOp2025.0985.3
MaPLeShots=42026.0585.3
MaPLemode=Fine-tuning2025.1085.3
CoOpBackbone=ViT-B/16, Training Shots=16-shot, Training Source=ImageNet2026.0585.3
CLIPmode=Fine-tuning2025.1084.8