issues in show_bias.py

[luyulalala]

Hello Chen, thanks for your updated code first and here is still an issue when I try to run show_bias.py and I think it maybe a mistake of upper and lower case but I can't find where can I modify..
And another question is have you write any interface to check the generated recommender conversation?

Thanks a lot for your contribution

[qibinc]

Hi @luyulalala ,

I noticed that you were pointing to model file saved/both_rgcn_1, which is the recommender part of the model. However, to get the vocabulary bias, we need to train the dialog part. bash scripts/t2t_rec_rgcn.sh <num_exps> <gpu_id> (see README). After training this, we can run python scripts/show_bias.py -mf saved/transformer_rec_both_rgcn_1 and it should be ok.

For checking generated results, I just uploaded a display_model.py in scripts/. After training the dialog part, you can run python scripts/display_model.py -t redial -mf saved/transformer_rec_both_rgcn_1 -dt test

[luyulalala]

Thanks ! it works! I really appreciate what you guys do and help. Thanks a lot!!!