OpenAI Completions API: How do I extract the text from the response?

Question

I am attempting to extract text from OpenAI, but I need some help with the correct syntax.

My code:

_model = "gpt-3.5-turbo-instruct"
# Summarize the text using an AI model (e.g., OpenAI's GPT)
try:
    summary = openai.completions.create(
        model=_model,
        prompt=f"Summarize this text: {text}",
        max_tokens=150
    )
except Exception as e:
    return f"summary: Exception {e}"

_summary = summary.choices[0].message.content
#_summary = summary['choices'][0]['text']

I want to get a summary of the text. Is this the best way?

However, I find when I debug the code that the text object is located in the following structure within summary['choices'][0]:

CompletionChoice(finish_reason='stop', index=0, logprobs=None, text='blah blah blah')

How do I extract the text from this in Python code?

Answer 1

By using the following method, it means that you're using the OpenAI Python SDK >=v1.0.0:

openai.completions.create

The response using the OpenAI Python SDK >=v1.0.0 looks like this:

{
  "id": "cmpl-uqkvlQyYK7bGYrRHQ0eXlWi7",
  "object": "text_completion",
  "created": 1589478378,
  "model": "gpt-3.5-turbo-instruct",
  "system_fingerprint": "fp_44709d6fcb",
  "choices": [
    {
      "text": "\n\nThis is indeed a test",
      "index": 0,
      "logprobs": null,
      "finish_reason": "length"
    }
  ],
  "usage": {
    "prompt_tokens": 5,
    "completion_tokens": 7,
    "total_tokens": 12
  }
}

Which means that the following is the correct way to extract the completion using the OpenAI Python SDK >=v1.0.0:

summary.choices[0].text