app.py

import os
os.system("pip install transformers torch")
import gradio as gr
from transformers import AutoModel, AutoTokenizer, trainer_utils
import torch

# Load the model and tokenizer
device = "cuda" if torch.cuda.is_available() else "cpu"
model = AutoModel.from_pretrained("Tanrei/GPTSAN-japanese").to(device)
tokenizer = AutoTokenizer.from_pretrained("Tanrei/GPTSAN-japanese")

# Define the text generation function
def generate_text(input_text):
    x_token = tokenizer(input_text, return_tensors="pt")
    trainer_utils.set_seed(30)
    input_ids = x_token.input_ids.to(device)
    gen_token = model.generate(input_ids, max_new_tokens=50)
    return tokenizer.decode(gen_token[0])

# Create the Gradio interface
iface = gr.Interface(
    fn=generate_text,
    inputs="text",
    outputs="text",
    title="Japanese Text Generation",
    description="Enter a prompt in Japanese to generate text."
)

# Launch the interface
iface.launch()