Help me creating a RAG class: vectorstore should be reusable at the second call

[jsmith173]

Create me a python code (using class) for using RAG.
Use the following functions: WebBaseLoader, RecursiveCharacterTextSplitter, OllamaEmbeddings, Chroma.from_documents
But this important when a second call is coming I don't want run again the RAG create process so store this variables in a class variable and make sure that they are free upon deleting the class instance

[dosubot[bot]]

To create a reusable RAG (Retrieval-Augmented Generation) class in Python, you can structure your code to initialize the necessary components only once and reuse them for subsequent calls. Here's a basic example using the specified functions:

from some_module import WebBaseLoader, RecursiveCharacterTextSplitter, OllamaEmbeddings, Chroma

class RAG:
    _vectorstore = None
    _embeddings = None

    def __init__(self, documents):
        if RAG._vectorstore is None or RAG._embeddings is None:
            self.loader = WebBaseLoader()
            self.text_splitter = RecursiveCharacterTextSplitter()
            self.embeddings = OllamaEmbeddings()
            self.vectorstore = Chroma.from_documents(documents, self.embeddings)
            RAG._vectorstore = self.vectorstore
            RAG._embeddings = self.embeddings
        else:
            self.vectorstore = RAG._vectorstore
            self.embeddings = RAG._embeddings

    def __del__(self):
        # Clean up class variables when the instance is deleted
        RAG._vectorstore = None
        RAG._embeddings = None

    def query(self, query_text):
        # Implement your query logic here
        pass

# Example usage
documents = ["Document 1 text", "Document 2 text"]
rag_instance = RAG(documents)
result = rag_instance.query("Your query here")

Key Points:

• Class Variables: _vectorstore and _embeddings are class variables, ensuring they are shared across all instances of the class.
• Initialization: The vector store and embeddings are initialized only once and reused for subsequent instances.
• Cleanup: The __del__ method ensures that the class variables are reset when an instance is deleted, freeing up resources.

This structure allows you to reuse the vector store and embeddings without reinitializing them for every new instance of the class. Adjust the query method to implement your specific logic for querying the vector store.

To continue talking to Dosu, mention @dosu.

[jsmith173]

@dosu
RAG._vectorstore = self.vectorstore

It is not clear what is RAG here. Is it a class name?

[dosubot[bot]]

Yes, RAG is the class name in the example provided. The line RAG._vectorstore = self.vectorstore is setting the class variable _vectorstore to the instance variable self.vectorstore. This ensures that the vector store is shared across all instances of the RAG class. If you have any more questions or need further clarification, feel free to ask!

To continue talking to Dosu, mention @dosu.

[jsmith173]

@dosu
Sorry it is not clear for me. Is it possible to rewrite the code by using only self.variable type assignments?

[dosubot[bot]]

Could you clarify if you want to avoid using class variables entirely and instead rely solely on instance variables? If so, how would you like to handle the reuse of the vector store and embeddings across different instances of the class?

To continue talking to Dosu, mention @dosu.

[jsmith173]

I'd like to use one instance only.