import gradio as gr
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

# Load your model and tokenizer from Hugging Face
model_name = "Addaci/byt5-small-finetuned-yiddish-experiment-10"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)

def correct_yiddish_transcription(input_text):
    inputs = tokenizer(input_text, return_tensors="pt", truncation=True)
    outputs = model.generate(**inputs)
    corrected_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return corrected_text

# Create Gradio Interface
interface = gr.Interface(
    fn=correct_yiddish_transcription,
    inputs=gr.Textbox(lines=5, label="Input Yiddish Text"),
    outputs=gr.Textbox(label="Corrected Text"),
    title="Yiddish Transcription Correction",
    description="Corrects raw Yiddish machine transcription using a fine-tuned ByT5 model."
)

interface.launch()