On the Test-Time Zero-Shot Generalization of Vision-Language Models: Do We Really Need Prompt Learning?

Open paper

Benchmark	Model	Rank	Results
test-time-adaptation-on-fgvc-aircraft	MTA + E.	#1	Accuracy: 25.2
test-time-adaptation-on-caltech-101	MTA + E.	#2	Accuracy: 94.21
test-time-adaptation-on-dtd	MTA + E.	#2	Accuracy: 45.9
test-time-adaptation-on-sun397	MTA + E.	#2	Accuracy: 66.67
test-time-adaptation-on-eurosat	MTA + E.	#3	Accuracy: 45.36