atulkum / pointer_summarizer

pytorch implementation of "Get To The Point: Summarization with Pointer-Generator Networks"

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

c_0 should be is a zero vector

raheelqader opened this issue · comments

commented

In Abigail See's paper, it says that c_0 is a zero vector, but based on the following line, c_0 in your code is equivalent to a_0 or maybe I'm missing something.

coverage = coverage + attn_dist

c_0 is zero vector only in case of training. For decoding (self.training == False) c_0 is same as a_0 as you mention.
Here is the relevant code in Abigail See's repo
https://github.com/abisee/pointer-generator/blob/a7317f573d01b944c31a76bde7218bcfc890ef6a/attention_decoder.py#L140

commented

you're right thanks.