Has anyone successfully upserted Plain Text and URL Scrapping to Pinecone via API? I want to upsert plain text and data scraped from websites, to specific namespaces including the metadata.
Please help (when I call the API it doesn’t add the vectors to Pinecone; pdf etc all work fine).
What is your use-case ? Conversational chat based on Pinecone data?
Here is my API call to pinecone for upserting and it works. However, this works with vectors not plain text. You would need to first get the vectors from OpenAI embedding model or some other embedding API.
Body:
{"vectors":[{"id":"<uid>","values":[<embed_num>],"metadata":{"type":"<type>","role":"<role>","dev_env":<dev_env>}}],"namespace":"<namespace>"}
Yes conversationat chat based on Pinecone. I want the user to be able to upsert plain text to pinecone to focus the conversation. For example a use can copy and paste a News post into a multi text fields and upsert before chatting with the bot
Does that mean I have to convert the Plain Text to embeddings before upserting the embeddings to pinecone?
Exactly! Vector dbs work with embeddings, so you need to convert it first and then upsert to pinecone.
