search

hpzhao / SummaRuNNer

The PyTorch Implementation of SummaRuNNer
https://arxiv.org/pdf/1611.04230.pdf
MIT License
253 stars 81 forks source link

abstractive training? #24

Closed AlJohri closed 6 years ago

AlJohri commented 6 years ago

hey @hpzhao, have you tried implementing the abstractive training mode? I'm a little suspicious of some of the sentence labels based on spot checking the data

hpzhao commented 6 years ago

Not yet. The abs training mode did't achieve SOTA. In my analysis, the dataset is not perfect but quite reasonable. I suggest you can jump out the dataset and SummaRuNNer. @AlJohri

AlJohri commented 6 years ago

Cool thanks. What do you mean “jump out”?

hpzhao commented 6 years ago

I mean you could build your own dataset and design your own network. I'm working on it.

AlJohri commented 6 years ago

Yep, I'm working on building my own dataset right now- but its attractive summaries