app.py

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

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

model = AutoModelForCausalLM.from_pretrained(
    "Qwen/Qwen2-0.5B-Instruct",
    torch_dtype="auto",
    device_map="auto"
).to(device)
tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2-0.5B-Instruct")

@spaces.GPU

def chatbot(user_input, history):
    system_message = {"role": "system", "content": "You are a helpful assistant."}
    messages = history + [{"role": "user", "content": user_input}]
    
    if len(history) == 0:
        messages.insert(0, system_message)
    
    text = tokenizer.apply_chat_template(
        messages,
        tokenize=False,
        add_generation_prompt=True
    )
    
    model_inputs = tokenizer([text], return_tensors="pt").to(device)
    attention_mask = torch.ones(model_inputs.input_ids.shape, device=device)

    generated_ids = model.generate(
        model_inputs.input_ids,
        attention_mask=attention_mask,
        max_new_tokens=512
    )
    
    generated_ids = [
        output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
    ]
    
    response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
    
    history.append({"role": "user", "content": user_input})
    history.append({"role": "assistant", "content": response})
    
    gradio_history = [[msg["role"], msg["content"]] for msg in history]

    return gradio_history, history

with gr.Blocks() as demo:
    chatbot_interface = gr.Chatbot()
    state = gr.State([])

    with gr.Row():
        txt = gr.Textbox(show_label=False, placeholder="Ask anything")
        txt.submit(chatbot, [txt, state], [chatbot_interface, state])

demo.launch()