Does anyone know why response is none

Does anyone know why response is none?

Plain Text

import os
from flask import Flask, request
from langchain.chat_models import ChatOpenAI
from llama_index import GPTKeywordTableIndex, LLMPredictor, ServiceContext, SimpleDirectoryReader

os.environ["OPENAI_API_KEY"] = "MY KEY HERE"

app = Flask(__name__)

documents = SimpleDirectoryReader("C:\\temp\\Test").load_data()
llm_predictor = LLMPredictor(llm=ChatOpenAI(temperature=0, model_name='gpt-3.5-turbo'))
service_context = ServiceContext.from_defaults(llm_predictor=llm_predictor)
index = GPTKeywordTableIndex.from_documents(documents, service_context=service_context)

@app.route("/query", methods=["GET"])
def query():
    query_text = request.args.get("text", None)
    response = index.query(query_text)
    return str(response), 200

9 comments

LLogan M

is the query endpoint actually getting the text? Try printing query_text before running the query

JJakeAM

It is 🙂

LLogan M

🤔 and the index actually has documents in it? You could also try print(len(documents)) before building the index

(just trying to narrow down the problem lol)

LLogan M

Oh!

LLogan M

It's a keyword index

LLogan M

I thiiiiink if they query text does not have any keywords that are in the keyword table, then it won't really return anything

JJakeAM

There are documents, but will it only return something if there's a keyword match?

LLogan M

Exactly! Otherwise, it won't know which documents to pick

LLogan M

There are other index types though (vector, list, etc.)

Add a reply

Find answers from the community

Does anyone know why response is none