How does the value get assigned in tuple in pack_padded_sequence funtion?

Hello Everyone,

This is my first question in this forum. So,please bear with me.

I am a newbie to pytorch and I am trying to understand Hierarchical attention mechanism for text classification. Below is portion of the code which I downloaded from github to understand.

sentences, bs = pack_padded_sequence(documents,

Here, the parameter documents is of dimension 1* 7* 63. 1 being the no of document, 7 being the maximum no of sentence allowed and 63 being the maximum no of word allowed. The sentences_per_document is 4 which means that total no of sentences in the document is 4. Now after the execution of this code,
the dimension of sentences is observed to be 4* 63 which means that there are 4 sentences and 63 is the maximum no of words allowed and bs value is [1,1,1,1].

Now I want to know how does the value get assigned in bs and why is it [1,1,1,1]?

Thank you