TODO: Separate index filename and path could be useful (s3) TODO: Separate data directory and path could be useful (s3)
Environment variables:
-
INDEX_FILE: Fullpath of the desired index file. Will be created if necessary.
-
LOAD_DIR: Directory of all input documents (word, pdf, csv, ...)
-
OPENAI_API_KEY: Your Open API key.
-
DATA_PROVIDER: Optional; Either s3 or local (default)
-
S3_BUCKET_NAME: Mandatory when working with S3 as data provider
Dev: uvicorn server:app --reload
curl http://127.0.0.1:8000/gpt\?question\=whatever (pipe it to jq make it readable :-)
curl -X POST http://127.0.0.1:8000/gpt\/index?operation\=RECREATE (pipe it to jq make it readable :-)
Default Swagger-UI: http://127.0.0.1:8000/docs Alternative provided by ReDoc: http://127.0.0.1:8000/redoc