List your company
ResearchFormerly WizWand

Image Classification on EuroSAT

99.81AccuracyPaRaMS
88.286891.278494.2797.2616Mar 26, 2023Sep 25, 2023Mar 27, 2024Sep 27, 2024Mar 30, 2025Sep 30, 2025Apr 2, 2026
Updated 7mo ago

Evaluation results

MethodLinks
PaRaMSModel Status=Protected Standalone (θ̂_def)2025.1199.81——
Fine-Tuned*Note=Upper-bound2025.1299.7——
fine-tunedBackbone=ViT-B/322026.0299.1——
fine-tunedBackbone=ViT-B/162026.0299.1——
fine-tunedBackbone=ViT-L/142026.0299.1——
IndividualBackbone=ViT-B/16, Pre-trained=ImageNet-21k, Number of Merged Models=12024.0599.04——
FFT# Params (%)=100, Backbone=CLIP ViT-B/322026.0498.96——
FourierMoE# Params (%)=0.42, Backbone=CLIP ViT-B/322026.0498.93——
DoRA# Params (%)=1.49, Backbone=CLIP ViT-B/322026.0498.89——
LoRA (rank32)# Params (%)=5.98, Backbone=CLIP ViT-B/322026.0498.88——
GOAT# Params (%)=2.24, Backbone=CLIP ViT-B/322026.0498.82——
Fine-TunedBackbone=DINOv3 ViT-B/162025.1298.8——
LoRA (rank16)# Params (%)=2.99, Backbone=CLIP ViT-B/322026.0498.74——
LoRA (rank8)# Params (%)=1.49, Backbone=CLIP ViT-B/322026.0498.66——
MoLoRA# Params (%)=2.24, Backbone=CLIP ViT-B/322026.0498.63——
DINOv2Architecture=ViT-g/14, Pre-training Data=LVD, Resolution=518x518, Evaluation Protocol=Attentive Probing, Backbone State=Frozen2024.0198.6——
AIM-1BArchitecture=ViT-1B/14, Pre-training Data=DFN-2B+, Evaluation Protocol=Attentive Probing, Backbone State=Frozen2024.0198.6——
AIM-7BArchitecture=ViT-7B/14, Pre-training Data=DFN-2B+, Evaluation Protocol=Attentive Probing, Backbone State=Frozen2024.0198.6——
FFT MoE# Params (%)=770, Backbone=CLIP ViT-B/322026.0498.59——
AdaMoLE# Params (%)=2.33, Backbone=CLIP ViT-B/322026.0498.52——
AIM-3BArchitecture=ViT-3B/14, Pre-training Data=DFN-2B+, Evaluation Protocol=Attentive Probing, Backbone State=Frozen2024.0198.5——
AIM-7B†Architecture=ViT-7B/14, Pre-training Data=DFN-2B+, Feature Extraction Layer=20th, Evaluation Protocol=Attentive Probing, Backbone State=Frozen2024.0198.5——
PiSSA# Params (%)=1.49, Backbone=CLIP ViT-B/322026.0498.48——
AIM-0.6BArchitecture=ViT-H/14, Pre-training Data=DFN-2B+, Evaluation Protocol=Attentive Probing, Backbone State=Frozen2024.0198.4——
HydraLoRA# Params (%)=1.58, Backbone=CLIP ViT-B/322026.0498.4——
EnsemblingBackbone=ViT-B/322026.0298.3——
Fine-TunedBackbone=CLIP ViT-B/32, Classes=102025.1298.3——
EnsemblingBackbone=ViT-L/142026.0298.24——
DINOArchitecture=ViT-B/8, Pre-training Data=IN-1k, Evaluation Protocol=Attentive Probing, Backbone State=Frozen2024.0198.2——
MAEArchitecture=ViT-2B/14, Pre-training Data=IG-3B, Evaluation Protocol=Attentive Probing, Backbone State=Frozen2024.0198.2——
MiLoRA# Params (%)=1.49, Backbone=CLIP ViT-B/322026.0498.19——
MAEArchitecture=ViT-H/14, Pre-training Data=IN-1k, Evaluation Protocol=Attentive Probing, Backbone State=Frozen2024.0198.1——
FourierFT# Params (%)=0.22, Backbone=CLIP ViT-B/322026.0498.02——
iBOTArchitecture=ViT-L/16, Pre-training Data=IN-21k, Evaluation Protocol=Attentive Probing, Backbone State=Frozen2024.0198——
TSV-M/LARVBackbone=ViT-L/142026.0297.8—1.1
ISO-CTS/LARVBackbone=ViT-L/142026.0297.6—0.3
BEiTArchitecture=ViT-L/14, Pre-training Data=IN-21k, Evaluation Protocol=Attentive Probing, Backbone State=Frozen2024.0197.5——
ISO-C/LARVBackbone=ViT-L/142026.0297.5—0.3
ADAProtection Status=None (theta_merge)2025.1197.4——
ISO-CTS/LARVBackbone=ViT-B/162026.0297.2—1.2
Task MatrixBackbone=DINOv3 ViT-B/16, best layer=(6,9,11)2025.1297.1——
Linear ProbeBackbone=DINOv3 ViT-B/162025.1297——
GRIPPrompt Modality=Visual, Learning Paradigm=TRZSL, Backbone=ViT-B/322023.0696.97——
ISO-C/LARVBackbone=ViT-B/162026.0296.9—0.6
TSV-M/LARVBackbone=ViT-B/162026.0296.6—0.4
EMR-MERGINGBackbone=ViT-B/16, Pre-trained=ImageNet-21k, Number of Merged Models=302024.0596.24——
Task MatrixBackbone=CLIP ViT-B/32, Best Layer=6,8, Classes=102025.1296.2——
ISO-C/LARVBackbone=ViT-B/322026.0296.1—6.4
Linear ProbeBackbone=CLIP ViT-B/32, Classes=102025.1295.9——
TIESProtection Status=None (theta_merge)2025.1195.7——
TSV-M/LARVBackbone=ViT-B/322026.0295.6—2.2
GRIPLearning Paradigm=TRZSL2023.0695.52——
MergeGuardModel Status=Protected Standalone (θ̂_def)2025.1195.46——
MLCDPre-training Data=LAION-400M, Backbone=ViT-L/14, Evaluation Protocol=Linear Probe2024.0794.7——
AdaMerging (Layer) + DeVecApproach=Our Method, Granularity=Layer-based, Optimization=Unsupervised2025.1294.7——
M-TIESBackbone=ViT-L/142026.0294.61——
Full dataBackbone=ViT-B/32, Evaluation Protocol=Full Data2023.1194.5——
CLIP+Pre-training Data=WIT-400M, Backbone=ViT-L/14, Evaluation Protocol=Linear Probe2024.0794.5——
TIES/LARVBackbone=ViT-L/142026.0294.4—4.7
TIESBackbone=ViT-L/142026.0294.37——
AdaMerging++ (Layer)Approach=Task Vectors, Granularity=Layer-based2025.1294.2——
CLIPPre-training Data=WIT-400M, Backbone=ViT-L/14, Evaluation Protocol=Linear Probe2024.0794.1——
TAProtection Status=None (theta_merge)2025.1194.1——
DAREBackbone=ViT-L/142026.0294.06——
Task ArithmeticBackbone=ViT-L/142026.0293.93——
AdaMerging (Layer)Approach=Task Vectors, Granularity=Layer-based2025.1293.8——
AWTShots=162024.0793.68——
MMRL++shots=162025.0593.5——
ISO-CTS/LARVBackbone=ViT-B/322026.0293.4—4.3
MMRLShots=16 shots2025.0393.37——
MMRLshots=162025.0593.37——
2SFS LoRABackbone=ViT-L/14, Shots (k=16)=16, Evaluation Protocol=all-to-all2025.0393.2——
2SFS LoRABackbone=ViT-B/16, Shots (k=16)=16, Evaluation Protocol=all-to-all2025.0392.7——
TA/LARVBackbone=ViT-L/142026.0292.7—8.1
UNICOMPre-training Data=LAION-400M, Backbone=ViT-L/14, Evaluation Protocol=Linear Probe2024.0792.6——
ProSRCShot=16, Time (s)=14403, Mem. (M)=33732024.1292.43——
PromptSRCShots=16 shots2025.0392.43——
PromptSRCshots=162025.0592.43——
OpenCLIPPre-training Data=LAION-400M, Backbone=ViT-L/14, Evaluation Protocol=Linear Probe2024.0792.4——
MMAShots=16 shots2025.0392.37——
MMAshots=162025.0592.37——
GRIPPrompt Modality=Textual, Learning Paradigm=TRZSL, Backbone=ViT-B/322023.0692.33——
MaPLeShot=16, Time (s)=14070, Mem. (M)=30612024.1292.33——
MaPLeShots=16 shots2025.0392.33——
MaPLeshots=162025.0592.33——
2SFS LoRABackbone=ViT-B/32, Shots (k=16)=16, Evaluation Protocol=all-to-all2025.0392.2——
WAProtection Status=None (theta_merge)2025.1191.9——
Skip TuningShot=16, Time (s)=962, Mem. (M)=5282024.1291.57——
Simple Avg.Backbone=ViT-L/142026.0291.17——
VP (white-box)Access=white-box, Shots=16, Repeated runs=32023.0390.8——
TIES/LARVBackbone=ViT-B/162026.0290.6—3.1
TOGAShots=16, Venue=-2026.0389.4——
MMRL++shots=42025.0589.23——
DeepBackbone=ResNet-18, Visual Reprogramming Method=Watermarking-based, Learning Method Category=Deep learning-based2024.1089.2——
MMRL++shots=82025.0589.17——
ProSRCShot=8, Time (s)=7636, Mem. (M)=33732024.1288.8——
PromptSRCShots=8 shots2025.0388.8——
PromptSRCshots=82025.0588.8——
MMRLShots=8 shots2025.0388.73——
MMRLshots=82025.0588.73——