fill each batch to exactly 20 tokens? 30tokens? 40tokens?

allenai / bilm-tf

Tensorflow implementation of contextualized word representations from bi-directional language models

Apache License 2.0

1.62k stars 452 forks source link

Open MLjian opened 5 years ago

MLjian commented 5 years ago

How do you determine this parameter?