How to do retrieval and return ID from ChromaDB

API
1

Im using ChromaDB for storing my embedded texts and I need to know the ID of the embedded chunk into the ChromaDB when I try to do retrieval. I need it to reference the generated text with the used document.

My code for retrieval is:

    def __setRetrieval(self, vectordb, llm):
        memory = ConversationBufferMemory(
            memory_key=MEMORY_KEY,
            return_messages=True
        )
        return ConversationalRetrievalChain.from_llm(
            llm,
            retriever=vectordb.as_retriever(search_type=RETRIEVAL_TYPE),
            memory=memory
        )

And my code for embedding is:

        # Dividir el documento en fragmentos
        splitter = RecursiveCharacterTextSplitter(
            chunk_size=SPLIT_CHUNK_SIZE_CHARS,
            chunk_overlap=SPLIT_CHUNK_OVERLAP_CHARS)
        docs_split = splitter.split_documents(docs_txt)
        logger.info(f""" 
                    Chunks: "{len(docs_split)}" 
                """)

        # Guardar los datos en Chroma
        openai_lc_client = Chroma.from_documents(
            docs_split, 
            embeddings, 
            client=chroma_client, 
            collection_name=COLLECTION_NAME_DEFAULT
        )

Thanks!

2

Hi @JuanjoAlm , were you able to figure it out ? I am looking for similar solution.