feat: add tool use

app.py

@@ -1,63 +1,97 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
-
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
-
-
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
-
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
-
-    messages.append({"role": "user", "content": message})
-
-    response = ""
-
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
-
-        response += token
-        yield response
-
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
+import json
+import os
+import numexpr
+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):
+    return client.chat.completions.create(
+        model=MODEL,
+        messages=messages,
+        tools=tools,
+        temperature=0.5,
+        top_p=0.65,
+        max_tokens=4096,
+    )
+
+def respond(message, history, system_message):
+    conversation = [{"role": "system", "content": system_message}]
+    for human, assistant in history:
+        conversation.append({"role": "user", "content": human})
+        conversation.append({"role": "assistant", "content": assistant})
+    conversation.append({"role": "user", "content": message})
+
+    available_functions = {
+        "evaluate_math_expression": evaluate_math_expression,
+    }
+
+    while True:
+        response = get_model_response(conversation)
+        response_message = response.choices[0].message
+        conversation.append(response_message)
+
+        if not response_message.tool_calls and response_message.content is not None:
+            return response_message.content
+
+        if response_message.tool_calls is not None:
+            for tool_call in response_message.tool_calls:
+                function_response = call_function(tool_call, available_functions)
+                conversation.append(function_response)
+
 demo = gr.ChatInterface(
     respond,
     additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
+        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 Groq LLM with tool use capabilities, including a calculator function.",
 )
 
-
 if __name__ == "__main__":
     demo.launch()

requirements.txt

@@ -1 +1,3 @@
-huggingface_hub==0.22.2
+huggingface_hub==0.22.2
+groq
+numexpr