gistfile1.txt

from util.nlp_tools import my_openai_key, my_claude_key, openai
client = openai.OpenAI(api_key=my_openai_key)

response = client.chat.completions.create(
  model="gpt-4",
  messages=[
    {
      "role": "user",
      "content": [
        {
          "type": "text",
          "text": "An insurance policy reads: \"[The insurance company will pay for] the felonious abstraction of insured property (1) from within the premises by a person making felonious entry therein by actual force and violence, of which force and violence there are visible marks made by tools, explosives, electricity or chemicals.\" With this in mind, please state your prediction—with the associated numerical level of confidence in parentheses—on the likely expectations of most policyholders under these terms for the following propositions: 1. The policy will provide compensation for losses resulting from a substantiated third-party burglary. 2. The policy will clearly delineate what is considered an inside job as opposed to a third-party intrusion. 3. Even in instances where a third-party burglary can be definitively established, the policy will necessitate physical evidence of forced entry."
        }
      ]
    },
    {
      "role": "assistant",
      "content": [
        {
          "type": "text",
          "text": "*Prediction Summary*\n\n1. Likely Expectation"
        }
      ]
    },
  ],
  response_format={
    "type": "text"
  },
  temperature=1,
  max_completion_tokens=2048,
  top_p=1,
  frequency_penalty=0,
  presence_penalty=0,
  logprobs=True,
  top_logprobs=20,
)