Create app.py

app.py

@@ -0,0 +1,131 @@
+import gradio as gr
+import requests
+import json
+import os
+
+API_KEY = os.getenv('API_KEY') 
+INVOKE_URL = "https://api.nvcf.nvidia.com/v2/nvcf/pexec/functions/5bde8f6f-7e83-4413-a0f2-7b97be33988e"
+FETCH_URL_FORMAT  = "https://api.nvcf.nvidia.com/v2/nvcf/pexec/status/"
+
+headers = {
+    "Authorization": f"Bearer {API_KEY}",
+    "Accept": "application/json",
+    "Content-Type": "application/json",
+}
+
+BASE_SYSTEM_MESSAGE = "I carefully provide accurate, factual, thoughtful, nuanced answers and am brilliant at reasoning."
+
+def clear_chat(chat_history_state, chat_message):
+    print("Clearing chat...")
+    chat_history_state = []
+    chat_message = ''
+    return chat_history_state, chat_message
+
+def user(message, history, system_message=None):
+    print(f"User message: {message}")
+    history = history or []
+    if system_message:  # Check if a system message is provided and should be added
+        history.append({"role": "system", "content": system_message})
+    history.append({"role": "user", "content": message})
+    return history
+
+def call_nvidia_api(history, max_tokens, temperature, top_p):
+    payload = {
+        "messages": history,
+        "temperature": temperature,
+        "top_p": top_p,
+        "max_tokens": max_tokens,
+        "stream": False
+    }
+
+    print(f"Payload enviado: {payload}")  # Imprime o payload enviado
+
+    session = requests.Session()
+    response = session.post(INVOKE_URL, headers=headers, json=payload)
+
+    while response.status_code == 202:
+        request_id = response.headers.get("NVCF-REQID")
+        fetch_url = FETCH_URL_FORMAT + request_id
+        response = session.get(fetch_url, headers=headers)
+    
+    response.raise_for_status()
+    response_body = response.json()
+
+    print(f"Payload recebido: {response_body}")  # Imprime o payload recebido
+
+    if response_body["choices"]:
+        assistant_message = response_body["choices"][0]["message"]["content"]
+        history.append({"role": "assistant", "content": assistant_message})
+    
+    return history
+
+def chat(history, system_message, max_tokens, temperature, top_p):
+    print("Starting chat...")
+    updated_history = call_nvidia_api(history, max_tokens, temperature, top_p)
+    return updated_history, ""
+
+# Gradio interface setup
+with gr.Blocks() as demo:
+    with gr.Row():
+        with gr.Column():
+            gr.Markdown("Gemma 2B Free Demo")
+            description="""
+<div style="text-align: center; font-size: 1.5em; margin-bottom: 20px;">
+    <strong>Explore the Capabilities of Gemma 2B</strong>
+</div>
+<p>Gemma is a family of lightweight, state-of-the art LLM open models from Google.
+</p>
+<p> <strong>How to Use:</strong></p>
+<ol>
+    <li>Enter your <strong>message</strong> in the textbox to start a conversation or ask a question.</li>
+    <li>Adjust the <strong>Temperature</strong> and <strong>Top P</strong> sliders to control the creativity and diversity of the responses.</li>
+    <li>Set the <strong>Max Tokens</strong> slider to determine the length of the response.</li>
+    <li>Use the <strong>System Message</strong> textbox if you wish to provide a specific context or instruction for the AI.</li>
+    <li>Click <strong>Send message</strong> to submit your query and receive a response from Gemma 2B.</li>
+    <li>Press <strong>New topic</strong> to clear the chat history and start a new conversation thread.</li>
+</ol>
+<p> <strong>Powered by NVIDIA's cutting-edge AI API, Gemma 2B offers an unparalleled opportunity to interact with an AI model of exceptional conversational ability, accessible to everyone at no cost.</strong></p>
+<p> <strong>HF Created by:</strong> @artificialguybr (<a href="https://twitter.com/artificialguybr">Twitter</a>)</p>
+<p> <strong>Discover more:</strong> <a href="https://artificialguy.com">artificialguy.com</a></p>
+"""
+    gr.Markdown(description)
+    chatbot = gr.Chatbot()
+    message = gr.Textbox(label="What do you want to chat about?", placeholder="Ask me anything.", lines=3)
+    submit = gr.Button(value="Send message")
+    clear = gr.Button(value="New topic")
+    system_msg = gr.Textbox(BASE_SYSTEM_MESSAGE, label="System Message", placeholder="System prompt.", lines=5)
+    max_tokens = gr.Slider(20, 1024, label="Max Tokens", step=20, value=1024, interactive=True)
+    temperature = gr.Slider(0.0, 1.0, label="Temperature", step=0.1, value=0.7, interactive=True)
+    top_p = gr.Slider(0.0, 1.0, label="Top P", step=0.05, value=0.95, interactive=True)
+    chat_history_state = gr.State([])
+
+    # Ajuste na definição da função update_chatbot para aceitar o valor atualizado do system_msg
+    def update_chatbot(message, chat_history, system_message, max_tokens, temperature, top_p):
+        print("Updating chatbot...")
+        if not chat_history or (chat_history and chat_history[-1]["role"] != "user"):
+            chat_history = user(message, chat_history, system_message if not chat_history else None)
+        else:
+            chat_history = user(message, chat_history)
+        chat_history, _ = chat(chat_history, system_message, max_tokens, temperature, top_p)
+    
+        formatted_chat_history = []
+        for user_msg, assistant_msg in zip([msg["content"].strip() for msg in chat_history if msg["role"] == "user"],
+                                        [msg["content"].strip() for msg in chat_history if msg["role"] == "assistant"]):
+            if user_msg or assistant_msg:  # Verify if either message is not empty
+                formatted_chat_history.append([user_msg, assistant_msg])
+    
+        return formatted_chat_history, chat_history, ""
+    
+    submit.click(
+        fn=update_chatbot,
+        inputs=[message, chat_history_state, system_msg, max_tokens, temperature, top_p],
+        outputs=[chatbot, chat_history_state, message]
+    )
+
+    clear.click(
+        fn=clear_chat,
+        inputs=[chat_history_state, message],
+        outputs=[chat_history_state, message]
+    )
+
+demo.launch()