This page is about submitting a pull request containing Markdown documentation files to the tensorflow/tfhub.dev. GitHub Repo. For more information on how to write the Markdown files in the first place, please see the writing documentation guide.
GitHub Actions checks
The tensorflow/tfhub.dev repo uses GitHub Actions to validate the format of the files in a PR. The workflow used to validate PRs is defined in .github/workflows/contributions-validator.yml. You can run the validator script on your own branch outside of the workflow, but you will need to ensure you have all the correct pip package dependencies installed.
First time contributors are only able to run automated checks with the approval of a repo maintainer, per GitHub policy. Publishers are encouraged to submit a small PR fixing typos, otherwise improving model documentation, or submitting a PR containing only their publisher page as their first PR to be able to run automated checks on subsequent PRs.
Submitting the PR
The complete Markdown files can be pulled into the master branch of tensorflow/tfhub.dev by one of the following methods.
Git CLI submission
Assuming the identified markdown file path is
assets/docs/publisher/model/1.md, you can follow the standard Git[Hub] steps
to create a new Pull Request with a newly added file.
This starts with forking the TensorFlow Hub GitHub repository, then creating a Pull Request from this fork into the TensorFlow Hub master branch.
The following are typical CLI git commands needed to adding a new file to a master branch of the forked repository.
git clone https://github.com/[github_username]/tfhub.dev.git cd tfhub.dev mkdir -p assets/docs/publisher/model cp my_markdown_file.md ./assets/docs/publisher/model/1.md git add * git commit -m "Added model file." git push origin master
GitHub GUI submission
- On the
TensorFlow Hub GitHub page, press
Create new filebutton.
- Set the right file path:
- Copy-paste the existing markdown.
- At the bottom, select "Create a new branch for this commit and start a pull request."