Question for the code x = x[torch.arange(x.shape[0]), tokenized_prompts.argmax(dim=-1)] @ self.text_projection

[OxInsky]

Hello, thanks for your work, it's interesting! but iam trouble with follow code
x = x[torch.arange(x.shape[0]), tokenized_prompts.argmax(dim=-1)] @ self.text_projection in CoOp.py Class TextEncode->Function forward()

I am curious about the self.text_projection. Is it the clip's textencode's output? and how x it work to select the max value in the matrix?
Can you explain it more? thanks a lot!