index.py

from llama_index import GPTSimpleVectorIndex, LLMPredictor, PromptHelper
from langchain import OpenAI
import flask
from flask import request, jsonify
from flask_cors import CORS

max_input_size = 3700
num_outputs = 300
llm_predictor = LLMPredictor(llm=OpenAI(temperature=0, model_name="gpt-3.5-turbo", max_tokens=num_outputs))
prompt_helper = PromptHelper.from_llm_predictor(llm_predictor)

app = flask.Flask(__name__)
CORS(app)

index = GPTSimpleVectorIndex.load_from_disk('index.json')

@app.route('/ask', methods=['GET'])
def ask():
  history = request.args.get('history')
  query = ''
  if history == 'none':
    query = 'System:\n' + 'You a Yearn Finance assistant.\n\n' + 'Anon:\n' + request.args.get('query') + '\n\nAI:\n'
  else:
    query = history + '\n\n' + 'Anon:\n' + request.args.get('query') + '\n\nAI:\n'
  print(query)
  response = index.query(query, response_mode="default", prompt_helper=prompt_helper)
  print(response.response.strip())
  return jsonify(response.response.strip())

def create_app():
    return app