app.py

import gradio as gr
from transformers import GPT2LMHeadModel, GPT2Tokenizer

model_name = "anezatra/gpt2_openassistant_guanaco"
model = GPT2LMHeadModel.from_pretrained(model_name)
tokenizer = GPT2Tokenizer.from_pretrained(model_name)

def respond(message, history):
    prompt = "\n".join(history + [message])

    inputs = tokenizer.encode(prompt, return_tensors="pt")
    outputs = model.generate(
        inputs,
        temperature=0.8,
        max_new_tokens=200,
        top_k=1,
        num_return_sequences=1,
        no_repeat_ngram_size=2,
        do_sample=True,
    )

    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    response = response.split("### Assistant:")[-1].strip()
    if "### Human:" in response:
        response = response.split("### Human:")[1].strip()
    return response

banner = gr.HTML("""
    <h1 style="color: #000; font-weight: bold; text-align: center;">
    OPENASSISTANT
    </h1>
    <p style="color: #000; font-weight: bold;">GPT-2 MEDIUM CHATBOT</p>
    """)

with gr.Blocks(theme=gr.Theme.from_hub('gradio/monochrome')) as demo:
    banner.render()

    chatbot = gr.Chatbot()
    msg = gr.Textbox(label="Write your message")

    with gr.Row():
        clear = gr.Button("Clear chat")
        submit = gr.Button("Send message")

    def user_input(user_message, history):
        response = respond(user_message, history)
        return "", history + [[user_message, response]]

    msg.submit(user_input, [msg, chatbot], [msg, chatbot], queue=False)
    clear.click(lambda: None, None, chatbot, queue=False)
    submit.click(lambda: msg.submit(), None, chatbot, queue=False)

if __name__ == "__main__":
    demo.launch(share=True)