Json Question posted in
Our archive of expertly curated questions and answers provides insights and solutions to common problems related to this popular data interchange format. From parsing and manipulating JSON data to integrating it with various programming languages and web services, our archive has got you covered. Start exploring today and take your JSON skills to the next level

Is there a way to get values from OpenAI JSON mode?

I am trying to get values from openai json mode for a project. I tried json package, bulit in parsing, etc. but i get

TypeError: the JSON object must be str, bytes or bytearray, not ChatCompletionMessage

json.decoder.JSONDecodeError: Expecting value: line 1 column 1 (char 0)`

AttributeError: 'str' object has no attribute 'text'
Tags:

3

Answers


  1. 0

    According to the documentation, you can set response_format to { "type": "json_object" } to enable JSON mode. Below is an example:

    response = openai.Completion.create(
  model="model_name",
  prompt="prompt_text",
  response_format={ "type": "json_object" }
)
    Login or Signup to reply.
  2. 0

    The currently provided answers will not give you responses in JSON consistently.

    There are three things you need to do if you want to get responses in JSON consistently:

    1. Use the gpt-4-1106-preview or gpt-3.5-turbo-1106 models.
    2. Set the system message where you explicitly state that you want to get responses in JSON format.
    3. Set the response_format parameter to json_object.

    Code:

    from openai import OpenAI
client = OpenAI(
  api_key=os.getenv("OPENAI_API_KEY"),
)

completion = client.chat.completions.create(
  model="gpt-4-1106-preview",
  messages=[
    {"role": "system", "content": "You are a helpful assistant. Your response should be in JSON format."},
    {"role": "user", "content": "Hello!"}
  ],
  response_format={"type": "json_object"}
)

print(completion.choices[0].message.content)

    You can check if the response is valid JSON with the following code (use it in the same Python script as the code above):

    import json

def is_json(myjson):
  try:
    json.loads(myjson)
  except ValueError as e:
    return False
  return True

print(is_json(completion.choices[0].message.content))
    Login or Signup to reply.
  3. 0

    If you are happy to go with a Completions model like gpt-3.5-turbo-instruct rather than Chat completions like GPT-4, I believe that this gives a JSON response, but error responses may not be in JSON format:

    import requests

headers = {
    "Content-Type": "application/json",
    "Authorization": f"Bearer {API_KEY}"
}
payload = {
    "model": "gpt-3.5-turbo-instruct",
    "prompt": "capital city of US",
    "max_tokens": 200,
}

response = requests.post("https://api.openai.com/v1/completions", headers=headers, json=payload)
response = response.json()

reply_text = response['choices'][0]['text']

print(reply_text)

    But I still can’t get it to accept a specific response_format specification.

    Login or Signup to reply.
Please signup or login to give your own answer.

Back To Top
Search