ask.py

from llama_index import GPTPineconeIndex, LLMPredictor, ServiceContext
import pinecone
from langchain import OpenAI
import os


# logging.basicConfig(stream=sys.stdout, level=logging.DEBUG)
# logging.getLogger().addHandler(logging.StreamHandler(stream=sys.stdout))


def askQuestion(brain, question, temperature, maxTokens):
    print(brain, question, temperature)
    Brain_Name = brain
    pinecone.init(api_key=os.environ['PINECONE_KEY'],
                  environment="us-west4-gcp")
    pineconeindex = pinecone.Index(Brain_Name)
    index = GPTPineconeIndex([], pinecone_index=pineconeindex)
    # index = GPTSimpleVectorIndex.load_from_disk('index.json')

    # For Q-A set this value to 4, For Content-Genration set this value b/w 7-10.
    data_chunks = 4

    # prompt query goes here
    # query="summarize in full detail the solution that dimetyd is providing, and previous email sequences which can be used as a context knowledge"
    query = question
    # relevant info from brain goes here
    info = ["pdf"]

    llm_predictor = LLMPredictor(llm=OpenAI(
        temperature=temperature, model_name="text-davinci-003", max_tokens=maxTokens))
    service_context_gpt4 = ServiceContext.from_defaults(
        llm_predictor=llm_predictor)

    response = index.query(query, service_context=service_context_gpt4,
                           similarity_top_k=data_chunks, response_mode="compact")

    return response


def getBrains():
    pinecone.init(api_key=os.environ['PINECONE_KEY'],
                  environment="us-west4-gcp")
    active_indexes = pinecone.list_indexes()
    print(active_indexes)
    return active_indexes