Desi Tsvetkova
Desi Tsvetkova
I think specifying explicitly the gpu id also fixes it: `spacy.util.use_gpu(id)`.
Ah, interesting. Thanks for looking into that!
Any chance it'll be available soon?
You must pass a checkpoint to `test_from`.
You must pass the extractive model under the `load_from_extractive` flag.
It's been awhile but here they are: https://github.com/nlpyang/PreSumm#system-outputs
Hi there! I don't know what the people from the Bart paper are reporting but as to low, mid and high, you can see that they are just percentiles: https://github.com/google-research/google-research/blob/478efa18a3a1fef144fac03fcd034829f348794e/rouge/scoring.py#L127...
I think it's [here](https://drive.google.com/drive/folders/1GP75Jp1XuPnfidhMtc1ey_-zjFKDxUoh?usp=sharing).