lewtun

/

tiny-clip-test

like 0

Hi,
The model appears to be giving a score of 1 for every text-to-image zero-shot classification. I uploaded an image of a white t-shirt and tried with a text prompt

1. gloves
2. trousers
   For both the texts, I got a score of 1. Am I missing something?
   Attached are screenshots for your reference.