app.py

import gradio as gr
import json
import os
import numexpr
import uuid
from groq import Groq
from groq.types.chat.chat_completion_tool_param import ChatCompletionToolParam

MODEL = "llama3-groq-8b-8192-tool-use-preview"
client = Groq(api_key=os.environ["GROQ_API_KEY"])

def evaluate_math_expression(expression: str):
    return json.dumps(numexpr.evaluate(expression).tolist())

calculator_tool: ChatCompletionToolParam = {
    "type": "function",
    "function": {
        "name": "evaluate_math_expression",
        "description":
        "Calculator tool: use this for evaluating numeric expressions with Python. Ensure the expression is valid Python syntax (e.g., use '**' for exponentiation, not '^').",
        "parameters": {
            "type": "object",
            "properties": {
                "expression": {
                    "type": "string",
                    "description": "The mathematical expression to evaluate. Must be valid Python syntax.",
                },
            },
            "required": ["expression"],
        },
    },
}

tools = [calculator_tool]

def call_function(tool_call, available_functions):
    function_name = tool_call.function.name
    if function_name not in available_functions:
        return {
            "tool_call_id": tool_call.id,
            "role": "tool",
            "content": f"Function {function_name} does not exist.",
        }
    function_to_call = available_functions[function_name]
    function_args = json.loads(tool_call.function.arguments)
    function_response = function_to_call(**function_args)
    return {
        "tool_call_id": tool_call.id,
        "role": "tool",
        "name": function_name,
        "content": json.dumps(function_response),
    }

def get_model_response(messages):
    try:
        return client.chat.completions.create(
            model=MODEL,
            messages=messages,
            tools=tools,
            temperature=0.5,
            top_p=0.65,
            max_tokens=4096,
        )
    except Exception as e:
        print(f"An error occurred while getting model response: {str(e)}")
        print(messages)
        return None

conversation_state = {}

def respond(message, history, system_message):
    if not history or not isinstance(history[0][0], str):
        session_id = str(uuid.uuid4())
        history.insert(0, (session_id, "Confirmed."))
    else:
        session_id = history[0][0]

    if session_id not in conversation_state:
        conversation_state[session_id] = []

    if len(conversation_state[session_id]) == 0:
        conversation_state[session_id].append({"role": "system", "content": system_message})

    conversation_state[session_id].append({"role": "user", "content": message})

    available_functions = {
        "evaluate_math_expression": evaluate_math_expression,
    }

    function_calls = []
    while True:
        response = get_model_response(conversation_state[session_id])
        response_message = response.choices[0].message
        conversation_state[session_id].append(response_message)

        if not response_message.tool_calls and response_message.content is not None:
            break

        if response_message.tool_calls is not None:
            for tool_call in response_message.tool_calls:
                function_call = {
                    "name": tool_call.function.name,
                    "arguments": json.loads(tool_call.function.arguments)
                }
                function_calls.append(function_call)
                function_response = call_function(tool_call, available_functions)
                conversation_state[session_id].append(function_response)
                function_calls.append({
                    "name": function_response["name"],
                    "result": json.loads(function_response["content"])
                })

    function_calls_md = "\n\n"
    for i in range(0, len(function_calls), 2):
        call = function_calls[i]
        result = function_calls[i + 1] if i + 1 < len(function_calls) else None
        function_calls_md += f"**Tool call:**\n```json\n{json.dumps({'name': call['name'], 'arguments': call['arguments'], 'result': result['result'] if result else None}, indent=2)}\n```\n"

    return response_message.content + function_calls_md


demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        gr.Textbox(value="You are a friendly Chatbot with access to a calculator. Don't mention that we are using functions defined in Python.", label="System message"),
    ],
    title="Groq Tool Use Chat",
    description="This chatbot uses the `llama3-groq-8b-8192-tool-use-preview` LLM with tool use capabilities, including a calculator function.",
)

if __name__ == "__main__":
    demo.launch()