backend/llama3.py

from openai import OpenAI
class LlaMa3:
    def __init__(self) -> None:
        self.client = OpenAI(
            base_url="https://integrate.api.nvidia.com/v1",
            api_key="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"
        )
        self.name = "Llama3"
        
        # Initial greeting and request for decision topic
        self.initial_prompt = """
        Hello! I can assist you in making a decision. What decision would you like to make today? 
        Please describe the decision and provide any relevant details to help me understand.
        """

    def chat(self, messages):
        # If this is the first message, we use the initial prompt to greet and ask for the decision topic
        if len(messages) == 0:  # Initial conversation step
            messages.append({"role": "system", "content": self.initial_prompt})

        # Call the API to get the model's response
        completion = self.client.chat.completions.create(
            model="nvidia/llama-3.1-nemotron-70b-instruct",
            messages=messages,  
            temperature=0.5,
            top_p=1,
            max_tokens=1024,
            stream=True
        )
        
        response = ""
        for chunk in completion:
            if chunk.choices[0].delta.content is not None:
                response += chunk.choices[0].delta.content

        return response