thunlp/TensorFlow-Summarization

Summary Length based on data size

Opened this issue · 0 comments

I was looking into using your library for some article summarization but wanted to know if you've done any testing on datasets where the article length is more akin to a page than a couple of sentences?

How might your models react to this length of input?

And how are you determining/limiting the output summary length? Because I've looked through the code and can't find a clear explanation of that anywhere/where it is determined.

Thanks!