Conversation
Check out this pull request on ReviewNB: https://app.reviewnb.com/microsoft/nlp/pull/191 You'll be able to see visual diffs and write comments on notebook cells. Powered by ReviewNB. |
scenarios/question_answering/pretrained-BERT-SQuAD-deep-dive-aml.ipynb
Outdated
Show resolved
Hide resolved
Is it possible to change the settings so it takes 15min or so? or are we already hitting the lowest time possible? |
I have tried different settings. Unfortunately, it seems like 30mins is the lowest time because the base model is large. I just let it fine tune on one epoch. |
8b8bbdc
to
e4640c6
Compare
0ccffbd
to
ebbd642
Compare
77797b6
to
a1d56a5
Compare
a1d56a5
to
44fdc90
Compare
Description
BERT QA notebook testing
Fix broken links with SHA commit id in case the scripts will change
This test will run around 30mins (15mins training and 15mins tuning) with the current settings.
Related Issues
#25
Checklist: