When you run it in eval mode it evaluates loss based on validation set.
Pavel Surmenok

ROUGE module needs an xml file as an input, this is a good explanation on how to make one. But you could also use this which creates the xml file automatically and seems to do the work.

I have a blog on “Extractive Vs. Abstractive Summarization” coming up on rare-technologies.com, in which I ran TextSum for 100,000 steps with changed parameters but couldn’t get any good summaries. Let me know if you got any good summaries.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.