The post describes an example of using Chroma for managing documents, including retrieving Document 1, generating its embedding, and sending a request to ChatGPT to summarize it. The community member then adds Document 2, sends an optional request to ChatGPT to compare the two documents, updates Document 1 with a new embedding, and sends another optional request to ChatGPT to summarize the changes. Finally, the community member deletes Document 2 from ChromaDB.
In the comments, another community member suggests using an ingestion pipeline with a docstore and vector store, and provides a link to an example using Redis vector store instead of Chroma.
Looking for an example with Chroma for managing documents. Retrieving Document 1 and Generating Embedding: The text of Document 1 is retrieved from ChromaDB and an embedding is generated using LlamaIndex.
ChatGPT Interaction : An request is sent to the ChatGPT API to obtain a summary of Document 1.
Adding Document 2: The text and embedding of Document 2 are added to ChromaDB.
ChatGPT Interaction: An optional request is sent to the ChatGPT API to generate a comparison between Document 1 and Document 2.
Updating Document 1: The text of Document 1 is updated, a new embedding is generated, and it is stored in ChromaDB.
ChatGPT Interaction: An optional request is sent to the ChatGPT API to summarize the changes made to Document 1.
Deleting Document 2: Document 2 is deleted from ChromaDB.
