app.py

import time
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
import gradio as gr

# --------------------------
# Load model
# --------------------------
model_name = "Fredithefish/Guanaco-3B-Uncensored-v2"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name)

device = "cuda" if torch.cuda.is_available() else "cpu"
model.to(device)

# --------------------------
# Chat history
# --------------------------
chat_history = []

# --------------------------
# Generate response
# --------------------------
def generate_response(prompt):
    global chat_history
    # Combine previous conversation
    context = ""
    for user_msg, ai_msg in chat_history:
        context += f"User: {user_msg}\nAI: {ai_msg}\n"
    context += f"User: {prompt}\nAI:"

    inputs = tokenizer(context, return_tensors="pt").to(device)
    outputs = model.generate(
        **inputs,
        max_new_tokens=150,
        do_sample=True,
        temperature=0.7
    )
    response = tokenizer.decode(outputs[0], skip_special_tokens=True).split("AI:")[-1].strip()
    chat_history.append((prompt, response))
    return response

# --------------------------
# Live typing effect
# --------------------------
def live_typing(prompt):
    response = generate_response(prompt)
    displayed_text = ""
    for char in response:
        displayed_text += char
        time.sleep(0.02)  # typing speed
        yield displayed_text

# --------------------------
# Gradio UI
# --------------------------
with gr.Blocks() as demo:
    gr.Markdown("## 🤖 Guanaco-3B Chatbot with Live Typing")

    # User input
    user_input = gr.Textbox(label="Your Message", placeholder="Type something...")

    # Output box
    output_box = gr.Textbox(label="AI is typing...", lines=8)

    # Submit event
    user_input.submit(live_typing, inputs=[user_input], outputs=[output_box])

demo.launch()
    response = tokenizer.decode(outputs[0], skip_special_tokens=True).split("AI:")[-1].strip()
    chat_history.append((prompt, response))
    return response

# --------------------------
# Simulate live typing
# --------------------------
def live_typing(prompt):
    response = generate_response(prompt)
    displayed_text = ""
    for char in response:
        displayed_text += char
        time.sleep(0.02)  # typing speed
        yield displayed_text

# --------------------------
# Gradio UI
# --------------------------
with gr.Blocks() as demo:
    gr.Markdown("## 🤖 Guanaco-3B Chatbot with Avatars and Live Typing")

    with gr.Row():
        with gr.Column(scale=1):
            user_avatar = gr.Image("user_avatar.png", elem_id="user-avatar")
        with gr.Column(scale=4):
            user_input = gr.Textbox(label="Your Message", placeholder="Type something...")

    with gr.Row():
        with gr.Column(scale=1):
            ai_avatar = gr.Image("ai_avatar.png", elem_id="ai-av_
# --------------------------
# Live typing simulation
# --------------------------
def live_typing(prompt):
    response = generate_response(prompt)
    displayed_text = ""
    for char in response:
        displayed_text += char
        time.sleep(0.02)  # typing speed
        yield displayed_text

# --------------------------
# Gradio UI
# --------------------------
with gr.Blocks() as demo:
    gr.Markdown("## 🤖 Gemma-3 Chatbot (CPU-Friendly) with Avatars and Live Typing")

    with gr.Row():
        with gr.Column(scale=1):
            user_avatar = gr.Image("user_avatar.png", elem_id="user-avatar")
        with gr.Column(scale=4):
            user_input = gr.Textbox(label="Your Message", placeholder="Type something...")

    with gr.Row():
        with gr.Column(scale=1):
            ai_avatar = gr.Image("ai_avatar.png", elem_id="ai-avatar")
        with gr.Column(scale=4):
            output_box = gr.Textbox(label="AI is typing...", lines=8)

    user_input.submit(live_typing, inputs=[user_input], outputs=[output_box])

demo.launch()