This repository has been archived by the owner on Feb 11, 2025. It is now read-only.
Merging from google-research's repo to update this one. #1
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Added colab link in the readme
Added bert tfhub example on colab
Fixed the colab notebook link.
Changed colab link
…ent to faciliate code reuse in colabs.
Correct Wikipedia size correlation comment.
Update multilingual.md to correct Wikipedia size size correlation comment.
Add links to 24 smaller BERT models.
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
I'm suggesting merging changes from Google's BERT repo into this one, mainly to enable whole word masking when using this repo. The relevant changes are in create_pretraining_data.py and modeling.py. The two repos seem to merge very cleanly and none of the code changed to use Horovod seem to be affected.