app.py

import gradio as gr
from transformers import pipeline


from transformers import AutoTokenizer
from transformers import AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained("kriton/greek-text-summarization")
model = AutoModelForSeq2SeqLM.from_pretrained("kriton/greek-text-summarization")
generator = pipeline("summarization", model="kriton/greek-text-summarization")

def genarate_summary(article):
    inputs = tokenizer(
        'summarize: ' + article, 
        return_tensors="pt", 
        max_length=1024, 
        truncation=True,
        padding="max_length",
    )

    outputs = model.generate(
        inputs["input_ids"], 
        max_length=512, 
        min_length=130, 
        length_penalty=3.0, 
        num_beams=8, 
        early_stopping=True,
        repetition_penalty=3.0,
    )

    return tokenizer.decode(outputs[0], skip_special_tokens=True)



def generate_text(prompt):
    response = generator(generate_summary(prompt), max_length=512, num_return_sequences=1)
    return response[0]["generated_text"]

iface = gr.Interface(
    fn=generate_text,
    inputs="text",
    outputs="text",
    title="Remote LLM Text Generation",
    description="Enter a prompt to generate text from the model."
)

# Launch the Gradio interface
iface.launch(share=True)