vdobrovolskii

You might want to check the CoNLL-2014 publication at https://aclanthology.org/W14-1701/ Section 4: > F0.5 emphasizes precision twice as much as recall, while F1 weighs precision and recall equally. When a...

Hi This might be relevant: https://github.com/vdobrovolskii/wl-coref/issues/20

It should be totally possible to reduce the training requirements, and I've been thinking a long time about rewriting the project (because it grew out of another project and there's...

Hi! I'm not sure why this happens. Could you please share everything I might need to reproduce the error?

Hi I'm sorry I disappeared and lost the track of time. Could you kindly confirm seeking the solution to the problem is still relevant to you?

Please, share your GPU specs and also the output of `pip freeze`

You won't be able to train the model on your machine without modifying the code... But for evaluation it should be more than enough, can you show me the exact...

@shantanu778 I am not exactly sure what is happening there. The problem is I no longer work where I did when this paper was written, so I don't have access...

On your screenshot I can see that the word-level evaluation is going OK, but something is wrong with predicting the spans. Would you mind taking a look at the data...