Spaces:
Sleeping
Sleeping
#The goal is to deploy the Falcon 3B model in a chatbot using Gradio | |
import gradio | |
from transformers import AutoModelForCausalLM, AutoTokenizer | |
model = AutoModelForCausalLM.from_pretrained("tiiuae/Falcon3-3B-Instruct") | |
tokenizer = AutoTokenizer.from_pretrained("tiiuae/Falcon3-3B-Instruct") | |
history = "" | |
def chat(user_input): | |
global history | |
history += f"User: {user_input}\n" | |
inputs = tokenizer(history, return_tensors="pt", max_length=1024) | |
generated_ids = model.generate(**inputs, max_length=1024) | |
output = tokenizer.decode(generated_ids[0], skip_special_tokens=True) | |
bot_response = output.split("Bot:")[-1].strip() | |
history += f"Bot: {bot_response}\n" | |
return bot_response | |
gradio.Interface(fn=chat, | |
inputs=gradio.Textbox(label="Your Message", placeholder="Ask me anything...", lines=1), | |
outputs=gradio.Textbox(label="Response", interactive=False, placeholder="I will respond here..."), | |
title="Falcon 3B Chatbot", | |
description="An AI model powered by Falcon 3B. Talk to me about anything!", | |
article="<b>Note:</b> This is a simple chatbot built with Gradio. Type above and let the model respond.", | |
theme="huggingface").launch(share=True) |