This pr introduces a new endpoint which accepts a csv in the import format with lines that do not have a CRE and uses embeddings to fill the missing cres with a suggestion.
[ ] add ability to get matches by defining a prompt (hopefully better accuracy)
[ ] add ability to define openai endpoint so that it can interface with other ai engines that are openai compatible
[ ] accurately populate cre cardinality on the spreadsheet
You can test this pr as follows:
# sync from upstream
python cre.py --upstream_sync
# generate embeddings, WARNING if you want any sort of accuracy you should use an LLM instead of spacy which is for local unit tests only
python cre.py --generate_embeddings
# run the web server
make dev-flask
# on another terminal
curl -X POST http://localhost:5000/rest/v1/cre_csv/suggest -F "cre_csv=@< csv file that follows the export format but does not necessarily have cres>"
This pr introduces a new endpoint which accepts a csv in the import format with lines that do not have a CRE and uses embeddings to fill the missing cres with a suggestion.
You can test this pr as follows: