search

factula / sumtool

1 stars 0 forks source link

Research: Index training data for summary datasets for very quick ngram lookup #6

Closed srush closed 2 years ago

dleve123 commented 2 years ago

Is this issue in reference to finding other, non-xsum, datasets to use to probe the space of abstractive summarization? Or is it more about making search against examples in the xsum dataset very fast (would likely build off of the solution to https://github.com/cs6741/summary-analysis/issues/3)?

Generally, more info would be helpful here, I think!

srush commented 2 years ago

It is the latter: making search against examples in the xsum dataset very fast . It could be a backend for https://github.com/cs6741/summary-analysis/issues/3 but it is a different problem.

dleve123 commented 2 years ago

Got it - thanks! I'm planning on pairing with @vincehartman38 on #4 – so this is free game for anyone else!

dleve123 commented 2 years ago

at least @srush will review this.