This PR expands and improves the code for building a Docker container image that deploys and runs an HTTP server to serve predictions for our custom-trained fine-tuned NER models (using FastAPI and uvicorn).
In particular, it combines predictions from two custom-trained spacy models, and
The PR also automates image building and model deployment to Vertex AI via GitHub Actions.
A bash script to additionally create a Vertex AI endpoint and deploy the model to it, programmatically, has also been added. These steps are not managed by GitHub Actions continuous deployment as they are an ad-hoc non-essential component of the metadata inference pipeline.
Checklists
This pull/merge request meets the following requirements:
Summary
This PR expands and improves the code for building a Docker container image that deploys and runs an HTTP server to serve predictions for our custom-trained fine-tuned NER models (using FastAPI and uvicorn). In particular, it combines predictions from two custom-trained spacy models, and
The PR also automates image building and model deployment to Vertex AI via GitHub Actions.
A bash script to additionally create a Vertex AI endpoint and deploy the model to it, programmatically, has also been added. These steps are not managed by GitHub Actions continuous deployment as they are an ad-hoc non-essential component of the metadata inference pipeline.
Checklists
This pull/merge request meets the following requirements:
docs/aqa/aqa_plan.md
)docs/aqa/data_log.md
), if necessarydocs/aqa/assumptions_caveats.md
), if necessarydocs
folderComments have been added below around the incomplete checks.