A GitHub bot to increase contributor test and review velocity for instructlab/taxonomy.
More detail on how this bot fits into the overall architecture is being captured in this design document.
Personas:
- Taxonomy Contributor
- Taxonomy Triager
- Bot Admin
Our goal is to implement a GitHub bot that will:
- Automate major portions of the test and review workflow for taxonomy PRs.
- Stash the generated data and trained models and make them available for download.
- Have the capacity to serve these models for testing purposes.
- Have a way to monitor the system's state -- what builds are available for each
PR, what jobs are in progress, etc …
- Contributor / Triager - status via GitHub PR comments from the bot
- Admin - OTEL, Grafana, etc …
The current iteration is focused on automating the ilab generate
portion of
the workflow. The following diagram shows the architecture of the bot and its
supporting infrastructure. It supports scaling a pool of workers to run ilab generate
jobs. The workers can be located anywhere and will be connect to Redis
over a private mesh network managed by Nexodus.
The current GitHub workflow in a PR is:
- PR is opened by the Contributor.
- User runs
@instructlab-bot generate
in a comment on the PR. - Bot generates data using
ilab generate
and stores it in an object store (S3). - Bot replies: "Here is the generated data ..."
If you have suggestions for how instructlab-bot could be improved, or want to report a bug, open an issue! We'd love all and any contributions.
For more, check out the InstructLab Bot Contribution Guide and InstructLab Community.