app.py

import os
from threading import Thread
from typing import Iterator

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

# Debugging: Start script
print("Starting script...")

HF_TOKEN = os.environ.get("HF_TOKEN")
if HF_TOKEN is None:
    print("Warning: HF_TOKEN is not set!")

PASSWORD = os.getenv("APP_PASSWORD", "mysecretpassword")  # Set your desired password here or via environment variable

DESCRIPTION = "# FT of Lama"

if not torch.cuda.is_available():
    DESCRIPTION += "\n<p>Running on CPU 🥶 This demo does not work on CPU.</p>"
    print("Warning: No GPU available. This model cannot run on CPU.")
else:
    print("GPU is available!")

MAX_MAX_NEW_TOKENS = 2048
DEFAULT_MAX_NEW_TOKENS = 1024
MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "4096"))

# Debugging: GPU check passed, loading model
if torch.cuda.is_available():
    model_id = "INSAIT-Institute/BgGPT-Gemma-2-9B-IT-v1.0"
    try:
        print("Loading model...")
        model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.float16, device_map="auto", token=HF_TOKEN)
        print("Model loaded successfully!")
        
        print("Loading tokenizer...")
        tokenizer = AutoTokenizer.from_pretrained(model_id, token=HF_TOKEN)
        print("Tokenizer loaded successfully!")
    except Exception as e:
        print(f"Error loading model or tokenizer: {e}")
        raise e  # Re-raise the error after logging it


@spaces.GPU
def generate(
    message: str,
    chat_history: list[tuple[str, str]],
    max_new_tokens: int = 1024,
    temperature: float = 0.6,
    top_p: float = 0.9,
    top_k: int = 50,
    repetition_penalty: float = 1.2,
) -> Iterator[str]:
    print(f"Received message: {message}")
    print(f"Chat history: {chat_history}")

    conversation = []
    for user, assistant in chat_history:
        conversation.extend([{"role": "user", "content": user}, {"role": "assistant", "content": assistant}])
    conversation.append({"role": "user", "content": message})

    try:
        print("Tokenizing input...")
        input_ids = tokenizer.apply_chat_template(conversation, return_tensors="pt")
        print(f"Input tokenized: {input_ids.shape}")
        
        if input_ids.shape[1] > MAX_INPUT_TOKEN_LENGTH:
            input_ids = input_ids[:, -MAX_INPUT_TOKEN_LENGTH:]
            gr.Warning(f"Trimmed input from conversation as it was longer than {MAX_INPUT_TOKEN_LENGTH} tokens.")
            print("Trimmed input tokens due to length.")
        
        input_ids = input_ids.to(model.device)
        print("Input moved to the model's device.")
        
        streamer = TextIteratorStreamer(tokenizer, timeout=20.0, skip_prompt=True, skip_special_tokens=True)
        generate_kwargs = dict(
            {"input_ids": input_ids},
            streamer=streamer,
            max_new_tokens=max_new_tokens,
            do_sample=True,
            top_p=top_p,
            top_k=top_k,
            temperature=temperature,
            num_beams=1,
            repetition_penalty=repetition_penalty,
        )
        
        print("Starting generation...")
        t = Thread(target=model.generate, kwargs=generate_kwargs)
        t.start()
        print("Thread started for model generation.")
        
        outputs = []
        for text in streamer:
            outputs.append(text)
            print(f"Generated text so far: {''.join(outputs)}")
            yield "".join(outputs)
    
    except Exception as e:
        print(f"Error during generation: {e}")
        raise e  # Re-raise the error after logging it


def password_auth(password):
    if password == PASSWORD:
        return gr.update(visible=True), gr.update(visible=False)
    else:
        return gr.update(visible=False), gr.update(visible=True, value="Incorrect password. Try again.")

chat_interface = gr.ChatInterface(
    fn=generate,
    additional_inputs=[
        gr.Slider(
            label="Max new tokens",
            minimum=1,
            maximum=MAX_MAX_NEW_TOKENS,
            step=1,
            value=DEFAULT_MAX_NEW_TOKENS,
        ),
        gr.Slider(
            label="Temperature",
            minimum=0.1,
            maximum=4.0,
            step=0.1,
            value=0.6,
        ),
        gr.Slider(
            label="Top-p (nucleus sampling)",
            minimum=0.05,
            maximum=1.0,
            step=0.05,
            value=0.9,
        ),
        gr.Slider(
            label="Top-k",
            minimum=1,
            maximum=1000,
            step=1,
            value=50,
        ),
        gr.Slider(
            label="Repetition penalty",
            minimum=1.0,
            maximum=2.0,
            step=0.05,
            value=1.2,
        ),
    ],
    stop_btn=None,
    examples=[
        ["Hello there! How are you doing?"],
        ["Can you explain briefly to me what is the Python programming language?"],
        ["Explain the plot of Cinderella in a sentence."],
        ["How many hours does it take a man to eat a Helicopter?"],
        ["Write a 100-word article on 'Benefits of Open-Source in AI research'"],
    ],
)

# Debugging: Interface setup
print("Setting up interface...")

with gr.Blocks(css="style.css") as demo:
    gr.Markdown(DESCRIPTION)
    
    # Create login components
    with gr.Row(visible=True) as login_area:
        password_input = gr.Textbox(
            label="Enter Password", type="password", placeholder="Password", show_label=True
        )
        login_btn = gr.Button("Submit")
        incorrect_password_msg = gr.Markdown("Incorrect password. Try again.", visible=False)
    
    # Main chat interface
    with gr.Column(visible=False) as chat_area:
        gr.Markdown(DESCRIPTION)
        gr.DuplicateButton(
            value="Duplicate Space for private use",
            elem_id="duplicate-button",
            visible=os.getenv("SHOW_DUPLICATE_BUTTON") == "1",
        )
        chat_interface.render()
    
    # Bind login button to check password
    login_btn.click(password_auth, inputs=password_input, outputs=[chat_area, incorrect_password_msg])

# Debugging: Starting queue and launching the demo
print("Launching demo...")

if __name__ == "__main__":
    demo.queue(max_size=20).launch(share=True)