app.py

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

# --- Configuration ---
MODEL_NAME = "TechnoByte/Qwen2.5-7B-VNTL-JP-EN"
MAX_NEW_TOKENS = 512 # Max length of the generated translation per line

# --- Load Model and Tokenizer ---
# Load the model and tokenizer only once when the app starts
print(f"Loading model: {MODEL_NAME}...")
try:
    model = AutoModelForCausalLM.from_pretrained(
        MODEL_NAME,
        torch_dtype="auto", # Use bfloat16 if available, float16 otherwise
        device_map="auto"   # Automatically distribute across available GPUs/CPU
    )
    tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
    print("Model and tokenizer loaded successfully.")
except Exception as e:
    print(f"Error loading model or tokenizer: {e}")
    raise gr.Error(f"Failed to load model: {e}. Check Space logs and hardware.")

# --- Translation Function ---
@spaces.GPU(duration=20)
def translate_japanese_to_english(input_text):
    """
    Translates Japanese text to English using the loaded model, processing line by line.
    """
    if not input_text:
        return "Please enter some Japanese text to translate."

    print(f"Received input:\n{input_text}")

    lines = input_text.splitlines() # Split input into lines
    translated_lines = []

    try: # Wrap the entire multi-line processing
        for line in lines:
            if not line.strip(): # If the line is empty or just whitespace
                translated_lines.append("") # Keep the empty line structure
                continue # Skip processing for this empty line

            print(f"Translating line: {line}")

            # Prepare the input for the current line using the chat template
            messages = [
                {"role": "user", "content": line}
            ]
            # Apply chat template
            prompt_text = tokenizer.apply_chat_template(
                messages,
                tokenize=False,
                add_generation_prompt=True
            )

            # Tokenize the input for the current line
            model_inputs = tokenizer([prompt_text], return_tensors="pt").to(model.device)

            # Generate the translation for the current line
            generated_ids = model.generate(
                **model_inputs,
                max_new_tokens=MAX_NEW_TOKENS,
                do_sample=False # Use greedy decoding for consistency
            )

            # Decode the generated text, skipping the prompt part
            input_ids_len = model_inputs.input_ids.shape[1]
            output_ids = generated_ids[0][input_ids_len:]
            response = tokenizer.decode(output_ids, skip_special_tokens=True).strip() # Strip leading/trailing whitespace from the translation

            print(f"Generated response for line: {response}")
            translated_lines.append(response)

        # Join the translated lines back together with newline characters
        final_translation = "\n".join(translated_lines)
        print(f"Final combined translation:\n{final_translation}")
        return final_translation

    except Exception as e:
        print(f"Error during translation: {e}")
        return f"An error occurred during translation: {e}"

# --- Gradio Interface ---
with gr.Blocks(theme=gr.themes.Soft()) as demo:
    gr.Markdown(
        f"""
        # Qwen2.5-7B-VNTL-JP-EN Demo 🇯🇵➡️🇬🇧
        Enter Japanese text below and click "Translate" to get the English translation.
        """
    )
    with gr.Row():
        with gr.Column(scale=1):
            input_textbox = gr.Textbox(
                lines=5, # Keep initial size, but it can grow
                label="Japanese Input Text"
            )
            translate_button = gr.Button("Translate", variant="primary")
        with gr.Column(scale=1):
            output_textbox = gr.Textbox(
                lines=5,
                label="English Output Text",
                interactive=False # Output box should not be editable by user
            )

    # --- Event Listener ---
    translate_button.click(
        fn=translate_japanese_to_english,
        inputs=input_textbox,
        outputs=output_textbox,
        api_name="translate" # Expose as API endpoint /api/translate
    )

    gr.Examples(
        examples=[
            ["放課後はマンガ喫茶でまったり〜♡ おすすめのマンガ教えて！"],
            ["このソフトウェアの使い方がよくわかりません。"],
            ["明日の天気はどうなりますか？"],
            ["日本の文化についてもっと知りたいです。"],
            ["こんにちは。\n元気ですか？\n私は元気です。"], # Multi-line example
            ["これは最初の行です。\n\nこれは３行目です。空行を挟みます。"] # Example with empty line
        ],
        inputs=input_textbox,
        outputs=output_textbox,
        fn=translate_japanese_to_english,
        cache_examples=True
    )

# --- Launch the App ---
if __name__ == "__main__":
    demo.launch()