Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Default parameters #2

Open
tomwesolowski opened this issue Oct 10, 2018 · 2 comments
Open

Default parameters #2

tomwesolowski opened this issue Oct 10, 2018 · 2 comments

Comments

@tomwesolowski
Copy link

Hello,

Why kb_composition is set to False and attention_lambda = 0 as default?

Thank you,
Tomasz

@tomwesolowski
Copy link
Author

tomwesolowski commented Oct 12, 2018

We've run your code and got:

Final Result

Train cost 0.21672125
Train accuracy 0.9252539741193045
Valid cost 0.3237615
Valid accuracy 0.8847795163584637
Test cost 0.33407608
Test accuracy 0.8827361563517915

@lukecq1231
Copy link
Owner

On the whole dataset, the model without kb_composition and attention_lambda usually achieves better result. However, when you only have low-resource training set, you need to use kb_composition and attention_lambda. There is variance in final result, and you will get better results when you run more times.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants