Description
The BERT model README at models/official/bert/README.md contains only a single line:
See https://github.com/google-research/bert/blob/master/README.md
Problem
- No setup instructions, usage examples, or TPU-specific guidance in this repo
- Users landing on this directory have no context on how the BERT implementation here relates to or differs from the
google-research/bert repo
- Other models in
models/official/ (e.g., EfficientNet, Mask R-CNN) have detailed READMEs with training commands, expected results, and configuration options
Suggested Fix
Add a README that covers at minimum:
- What this BERT implementation provides (and how it differs from
google-research/bert)
- How to run BERT training/fine-tuning on Cloud TPU
- Supported configurations and expected performance
- Link to the upstream repo for additional context
Description
The BERT model README at
models/official/bert/README.mdcontains only a single line:Problem
google-research/bertrepomodels/official/(e.g., EfficientNet, Mask R-CNN) have detailed READMEs with training commands, expected results, and configuration optionsSuggested Fix
Add a README that covers at minimum:
google-research/bert)