Set up Pinecone

Set up Pinecone#

By default, Sherpa represents its documents within an in-memory vector database. However, if you want to use a cloud vector database instead, Pinecone is available as an option.

First go to pinecone.io and create a free account.

Login to your account and you should see your pinecone console.

Click “Create Index”

Give a name to your pinecone index

The dimension is 1536, assuming that the embedding model is OpenAI’s text-embedding-ada-002 model. Everything else should be default.

Once your index is created, it should show on the console.

Now create a new API key.

Enter your copied API key, index name, and index environment in a .env file on your local machine, which should be placed inside sherpa/apps/slackbot directory. Checkout apps/slackbot/.env-sample as an example template for your working .env. *Note that namespace is not necessary for testing.