GPT-4をPythonで使う（再戦）　その２：ストリーミング式にする

普段から使い慣れた、テキストログ風の段階的出力を試してみます。

実際のところ

from openai import OpenAI
from dotenv import dotenv_values
import sys

def streamingQuestion(prompt, client):
  stream = yourclient.chat.completions.create(
      messages=[
          {"role": "system", "content": "You are a helpful assistant."},
          {"role": "user","content": myquestion},
      ],
      model="gpt-4",
      stream=True,
  )

  for chunk in stream:
      if chunk.choices[0].delta.content is not None:
          # 受け取ったフレーズを逐次出力。
          # flushをいれないとNULLになるまで表示しないので注意
          text = chunk.choices[0].delta.content
          print(text,end='',flush=True)
  print("")
  return True

myquestion = ""
if len(sys.argv) > 1:
   myquestion = sys.argv[1]
else:
   myquestion = "We just wanted to say hello. Take your time."

config = dotenv_values(".env")

yourclient = OpenAI( api_key= config["OPENAI_API_KEY"] )


streamingQuestion(myquestion,yourclient)

使ってみる

$ python3 chatgpttest.py "What is a Kimchi? Japanese food, I guess."

などとやると、以下の様に不正確さを指摘してくれます。

Actually, Kimchi is a traditional Korean dish, not Japanese. It's typically made from salted and fermented vegetables, most commonly napa cabbage and Korean radishes, along with a variety of seasonings including chili pepper, garlic, ginger, and jeotgal (salted seafood). Kimchi is highly nutritious and is known for its bold and spicy flavor.