Image Classification On Vtab 1K 1

评估指标

Top-1 Accuracy

评测结果

各个模型在此基准测试上的表现结果

		Paper Title	Repository
ALIGN (50 hypers/task)	79.99	Scaling Up Visual and Vision-Language Representation Learning With Noisy Text Supervision
BiT-L (50 hypers/task)	78.72	Big Transfer (BiT): General Visual Representation Learning
ViT-G/14	78.29	Scaling Vision Transformers
PVP(VIT-Base)	78.02	PVP: Pre-trained Visual Parameter-Efficient Tuning	-
ExpertEnsembles R101	77.6	Deep Ensembles for Low-Data Transfer Learning	-
BiT-L	76.3	Big Transfer (BiT): General Visual Representation Learning
S4L-Exemplar-ResNet50-LargeHyperSweep	72.7	A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark
ScalableExperts (I21k+JFT)	72.3	Scalable Transfer Learning with Expert Models	-
S4L-Rotation-ResNet50-LargeHyperSweep	71.5	A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark
ImageNet-ResNet50-LargeHyperSweep	71.2	A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark
BiT-M	70.6	Big Transfer (BiT): General Visual Representation Learning
VIVI-Ex4-Co	70.4	Self-Supervised Learning of Video-Induced Visual Invariances	-
S4L-Rotation-ResNet50	67.5	A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark
S4L-Exemplar-ResNet50	67.0	A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark
BiT-S	66.9	Big Transfer (BiT): General Visual Representation Learning
ImageNet-ResNet50	65.6	A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark
S4L-10%-Rotation-ResNet50	64.8	A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark
S4L-10%-Exemplar-ResNet50	63.9	A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark
ImageNet-10%-ResNet50	61.6	A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark
SelfSup-Rotation-ResNet50	59.5	A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark

0 of 34 row(s) selected.