my-gemma-3end / app.py
BryanBradfo's picture
change sdk
6701f02
from huggingface_hub import InferenceClient
from typing import List, Tuple
import gradio as gr
# Ensure you have the required libraries installed
# Chatbot configuration
class ChatConfig:
MODEL = "google/gemma-3-27b-it"
# MODEL = "mistralai/Mistral-7B-Instruct-v0.1"
# MODEL = "google/gemma-2-9b-it"
DEFAULT_SYSTEM_MSG = "You are a extremely smart and useful Chatbot."
DEFAULT_MAX_TOKENS = 1024
DEFAULT_TEMP = 0.4
DEFAULT_TOP_P = 0.95
import os
HF_TOKEN = os.getenv("HUGGING_FACE_HUB_TOKEN")
# Not recommended to hardcode your token in the script for security reasons.
# HF_TOKEN = "hf_YOUR_TOKEN_HERE"
client = InferenceClient(ChatConfig.MODEL, token=HF_TOKEN)
def generate_response(
message: str,
history: List[Tuple[str, str]],
system_message: str = ChatConfig.DEFAULT_SYSTEM_MSG,
max_tokens: int = ChatConfig.DEFAULT_MAX_TOKENS,
temperature: float = ChatConfig.DEFAULT_TEMP,
top_p: float = ChatConfig.DEFAULT_TOP_P
) -> str:
messages = [{"role": "system", "content": system_message}]
# Conversation history
for user_msg, bot_msg in history:
if user_msg:
messages.append({"role": "user", "content": user_msg})
if bot_msg:
messages.append({"role": "assistant", "content": bot_msg})
messages.append({"role": "user", "content": message})
response = ""
# Stream the response
# The stream=True parameter allows for real-time response generation
for chunk in client.chat_completion(
messages,
max_tokens=max_tokens,
stream=True,
temperature=temperature,
top_p=top_p,
):
token = chunk.choices[0].delta.content or ""
response += token
yield response
def create_interface() -> gr.ChatInterface:
"""Create and configure the chat interface."""
# Custom CSS for a modern look
custom_css = """
body {
background: linear-gradient(to right, #a8c0ff, #fafdff);
font-family: 'Comic Sans MS', sans-serif;
}
.chatbot .message {
border-radius: 12px;
margin: 5px;
padding: 10px;
}
.chatbot .user-message {
background-color: #b3e5fc;
color: #0d47a1;
}
.chatbot .bot-message {
background-color: #e3f2fd;
color: #0d47a1;
}
.gr-button {
border-radius: 8px;
padding: 8px 16px;
}
"""
with gr.Blocks(css=custom_css) as app:
gr.Markdown("# Your Gemma Three-(end) 💎")
# Custom chatbot
chatbot = gr.Chatbot(
label="Gemma Chat",
avatar_images=("./user.png", "./gemma.png"),
height=450,
show_copy_button=True
)
# Chat interface
interface = gr.ChatInterface(
fn=generate_response,
chatbot=chatbot,
title="Your Gemma 3-(iend)",
theme=gr.themes.Soft(),
css=custom_css,
additional_inputs=[
gr.Textbox(
value=ChatConfig.DEFAULT_SYSTEM_MSG,
label="System Prompt (You can change the text below)",
lines=2,
placeholder="Enter system message..."
),
gr.Slider(
minimum=1,
maximum=8192,
value=ChatConfig.DEFAULT_MAX_TOKENS,
step=1,
label="Max Tokens (Response Length)",
info="Controls response length"
),
gr.Slider(
minimum=0.1,
maximum=1.0,
value=ChatConfig.DEFAULT_TEMP,
step=0.1,
label="Temperature (0.1-1.0)",
info="Controls randomness"
),
gr.Slider(
minimum=0.1,
maximum=1.0,
value=ChatConfig.DEFAULT_TOP_P,
step=0.05,
label="Top-P (0.1-1.0)",
info="Controls diversity"
)
],
additional_inputs_accordion=gr.Accordion(label="Advanced Settings", open=False)
)
return app
# def create_interface() -> gr.Blocks:
# """Create and configure the chat interface."""
# # Custom CSS for a modern look
# custom_css = """
# body {
# background: linear-gradient(to right, #a8c0ff, #fafdff);
# font-family: 'Comic Sans MS', sans-serif;
# }
# .chatbot .message {
# border-radius: 12px;
# margin: 5px;
# padding: 10px;
# }
# .chatbot .user-message {
# background-color: #b3e5fc;
# color: #0d47a1;
# }
# .chatbot .bot-message {
# background-color: #e3f2fd;
# color: #0d47a1;
# }
# .gr-button {
# border-radius: 8px;
# padding: 8px 16px;
# }
# """
# # JavaScript pour jouer un son à la soumission
# custom_js = """
# function setupSubmitSound() {
# // Create the audio element for the sound
# const submitSound = new Audio('/gradio_api/file=submit_sound.mp3');
# // Fonction pour attacher l'écouteur d'événement
# function attachSoundToButton() {
# // Rechercher le bouton de soumission dans l'interface
# const submitButton = document.querySelector('button.send-btn') ||
# document.querySelector('button[aria-label="Submit"]') ||
# document.querySelector('.submit');
# if (submitButton) {
# console.log('Found submit button, attaching sound');
# submitButton.addEventListener('click', function() {
# submitSound.currentTime = 0; // Réinitialiser le son
# submitSound.play();
# });
# } else {
# // Si le bouton n'est pas encore disponible, réessayer plus tard
# console.log('Button not found, trying again in 500ms');
# setTimeout(attachSoundToButton, 500);
# }
# }
# // Démarrer la recherche du bouton
# attachSoundToButton();
# }
# // Appeler la fonction d'initialisation
# setupSubmitSound();
# """
# with gr.Blocks(css=custom_css, js=custom_js) as app:
# gr.Markdown("# Your Gemma Three-(end) 💎")
# # Custom chatbot
# chatbot = gr.Chatbot(
# label="Gemma Chat",
# avatar_images=("./user.png", "./gemma.png"),
# height=450,
# show_copy_button=True
# )
# # Chat interface
# interface = gr.ChatInterface(
# fn=generate_response,
# chatbot=chatbot,
# title="Your Gemma 3-(iend)",
# theme=gr.themes.Soft(),
# additional_inputs=[
# gr.Textbox(
# value=ChatConfig.DEFAULT_SYSTEM_MSG,
# label="System Prompt (You can change the text below)",
# lines=2,
# placeholder="Enter system message..."
# ),
# gr.Slider(
# minimum=1,
# maximum=8192,
# value=ChatConfig.DEFAULT_MAX_TOKENS,
# step=1,
# label="Max Tokens (Response Length)",
# info="Controls response length"
# ),
# gr.Slider(
# minimum=0.1,
# maximum=1.0,
# value=ChatConfig.DEFAULT_TEMP,
# step=0.1,
# label="Temperature (0.1-1.0)",
# info="Controls randomness"
# ),
# gr.Slider(
# minimum=0.1,
# maximum=1.0,
# value=ChatConfig.DEFAULT_TOP_P,
# step=0.05,
# label="Top-P (0.1-1.0)",
# info="Controls diversity"
# )
# ],
# additional_inputs_accordion=gr.Accordion(label="Advanced Settings", open=False)
# )
# return app
def main():
app = create_interface()
app.launch(
server_name="0.0.0.0",
server_port=7860,
share=False,
show_api=False,
show_error=True,
debug=True
)
if __name__ == "__main__":
main()