what is the meaning of the second parameter in dot([], [1, 1], name='context_vector')

[Hessen525]

Hi, Thanks for your awesome work.
I have a confusion about the code: context_vector = dot([hidden_states, attention_weights], [1, 1], name='context_vector')
What is the meaning of the second parameter?

[philipperemy]

Hey! Thank you! It means a lot to me. There's an explanation in this thread: #14
I plan to add a good and more exhaustive explanation to the README soon!

[Hessen525]

No idea yet, is it that for permute? permute(1,1)?

[philipperemy]

@felixhao28 maybe you can answer that one.

[felixhao28]

The second argument is the axes to perform dot product. In this case it is equivalent to hidden_states * transpose(attention_weights)