Arman Cohan

@abisee can you let us know what was the final running_avg_loss and validation loss for your model? For me, it seems that validation loss stops decreasing after a while and...

I think at 20K steps the model is still undertrained. I suggest starting with a smaller section length and number of sections and then at final steps increasing those. Something...

I would start from scratch. I also remember seeing some nan issues although this was a while ago (as far as I recall nan's were more likely to occur in...

The datasets include paper ids, if you'd like the raw data it should be easy to fetch those from [pubmed open access](https://www.ncbi.nlm.nih.gov/pmc/tools/openftlist/) and [arxiv](https://arxiv.org/help/bulk_data). If you ended up doing that,...

What is the command you are using for training? Please see the `run.py` for a sample command. Based on the [line](https://github.com/armancohan/long-summarization/blob/1cfec6787ea6c31b26d78a04a2d481ba95d57cbd/run_summarization.py#L419) throwing that error, you are either setting `--mode=decode` or...

Thanks! Meanwhile, I switched to using the entire Glove.