search

paul-tqh-nguyen / reuters_topic_labelling

Deep learning to automatically label Reuter's articles with the relevant topics.
1 stars 0 forks source link

Reimplement attention #13

Closed paul-tqh-nguyen closed 4 years ago

paul-tqh-nguyen commented 4 years ago

We previously ruled out attention because it seemed that we didn't need it to get 99% accuracy and wanted to minimize model complexity. However, we've since learned that accuracy is a bad metric for our dataset. Thus, we should see if we can get a better F1 score with an attention mechanism.

paul-tqh-nguyen commented 4 years ago

https://github.com/paul-tqh-nguyen/reuters_topic_labelling/commit/2570412460d70216ccb323d1b55a7641008e0959

This commit accomplishes this task.