Use with personal context

Hello,
Trying to implement a way to question PDFs locally and get answers only based on data from the docs. I have already find a way to embed the data into a vector db (using Chroma) and then retrieve with a "similarity_search" the most relevant data from our query into the doc. I would like now to find a way to give to my model this context to generate answer on it, maybe by using a prompt into the generate call ?

query = "What is the date of the start of the battle ?"
    docs = db.similarity_search(query)
    print(docs[0].page_content)
    
    llm = Ollama(
        model=llm_model_name,
        callbacks=[StreamingStdOutCallbackHandler()],
    )
    my_retriever = db.as_retriever(search_kwargs={"k": 8})
    
    response = ollama.generate(
        "model": llm_model_name,
        
    )``` 
Thank you for your help !

The issue names no repository files, tests, or entry points. Start by reviewing the Python library's documented generate usage and existing examples for passing retrieved context. Done would require a defined, supported way to provide Chroma-retrieved PDF context to the model and a documented expected result.