BA-Chatbot/model_service/openai_models.py

import openai
import json
import subprocess
openai.api_key = "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"

import openai 

def generate_answer_gpt(default_prompt, question, reference, completion_tokens):
    """
    Generates an answer using GPT-4 or falls back to GPT-3.5-turbo if GPT-4 encounters an error.
    Constructs a chat message sequence with a system prompt, reference, and user question.

    Args:
        default_prompt (str): The default system prompt for the model.
        question (str): The user's question.
        reference (str): Additional reference information.
        completion_tokens (int): The maximum number of tokens for the model's response.

    Returns:
        dict: The response from the OpenAI API.
    """
    try:
        print("GPT REFERENCE: ", reference)
        response = openai.ChatCompletion.create(
            model="gpt-4",
            messages=[
                {
                    "role": "system",
                    "content": default_prompt
                },
                {
                    "role": "user",
                    "content": f"{reference}Question:{question}"
                }
            ],
            max_tokens=completion_tokens,
            top_p=1,
            frequency_penalty=0,
            presence_penalty=0
        )
    except Exception as e:
        print("Ein Fehler ist aufgetreten bei der Anfrage an GPT-4: ", e)
        print("Sende nun die Anfrage an gpt-3.5-turbo-16k.")
        try:
            response = openai.ChatCompletion.create(
                model="gpt-3.5-turbo-16k",
                messages=[
                    {
                        "role": "system",
                        "content": default_prompt
                    },
                    {
                        "role": "user",
                        "content": f"{reference}Question:{question}"
                    }
                ],
                max_tokens=completion_tokens,
                top_p=1,
                frequency_penalty=0,
                presence_penalty=0
            )
        except Exception as e:
            print("Ein Fehler ist aufgetreten bei der Anfrage an gpt-3.5-turbo-16k: ", e)
            raise

    return response


def rerank_documents_with_gpt35(system_prompt, question, reference, completion_tokens):
    """
    Reranks documents using GPT-3.5-turbo based on the provided system prompt, question, and reference.

    Args:
        system_prompt (str): The system prompt for the model.
        question (str): The user's question.
        reference (str): Reference document text.
        completion_tokens (int): The maximum number of tokens for the model's response.

    Returns:
        dict: The response from the OpenAI API.
    """
    response = openai.ChatCompletion.create(
  model="gpt-3.5-turbo-16k",
  messages=[
    {
      "role": "system",
      "content": system_prompt
    },
    {
      "role": "user",
      "content": f"{reference}Question:{question}"
      
    }
  ],
  max_tokens=completion_tokens,
  top_p=1,
  frequency_penalty=0,
  presence_penalty=0
)
    return response
def generate_embeddings_ada(input):
    """
    NOTE: Unused cause we are gen. embeddings in the  pipeline  from the data service
    Generates embeddings for the given input using the Ada model.

    Args:
        input (str): The input text to generate embeddings for.

    Returns:
        list: A list of embeddings.
    """
    response = openai.Embedding.create(
    input=input,
    model="text-embedding-ada-002"
)
    embeddings = response['data'][0]['embedding']
    return embeddings