patverga

Hi - yeah it has a different name in the code. The loss for ner is multiplied by the parameter ner_weight. You can see the value that was set in...

Currently the code is doing some redundant computation, re-encoding the abstract for each entity pair. However, it is computing the full pairwise score tensor without any entity pair specific features...

This should definitely be outputs += inputs. This is a bug that will need to be fixed, thanks for catching.

Sorry for the delayed response. I trained word2vec using bytepair tokenized data. In reality, I noticed very little difference in performance using the pretrained vs randomly initialized byte pair embeddings.

Sorry for the delayed response. It looks like a network issue caused the download of the initial file to fail: "Connecting to ftp.ncbi.nlm.nih.gov (ftp.ncbi.nlm.nih.gov)|130.14.250.13|:21... failed: Connection refused.". This is causing...

I think the issue is that the demo data only works with the master branch version of the code and not the rowless version because it has a slightly different...