Issue with knowledge_val.py

[andreabac3]

Hello everybody (@nandana, @gaetangate),

I am trying to use the GenRL, however I encounter some problems for which I kindly ask for your support.

I am mainly having problems with the knowledge_val.py file

1. Could you share the cache files? Or how can I produce them?
   I mean the files required by the --val_cache parameter in the knowledge_val.py file

2. Executing the commands in the GenRL readme, it is impossible for me to run knowledge_val.py, because --model_output data/lcquad2/lcquad1_test_GenRL.json has the following keys q_id, question, etc which are different from those required as id, text, etc.

Where am I doing wrong?

Thank you in advance.

Sincerely,
Andrea

[nandana]

I will look into to this and get back to you soon Andrea.

[andreabac3]

Hi @nandana ,
sorry for bothering you again, do you have some updates?

Can you also share the code for the dataset construction? In the paper you mention a ranking strategy with a word embeddings, can you share also the preprocessing code?

Thank again for your brilliant work :)

Kind regards,
Andrea