app.py

import gradio as gr
from transformers import pipeline

def summarizer(sentence, min_length, max_length):
    model_name = "AirrStorm/T5-Small-XSUM-Summarizer"

    # Create a summarization pipeline with the local model
    summarizer = pipeline("summarization", model=model_name, tokenizer=model_name)

    summary = summarizer(
        sentence, 
        max_length=int(max_length),  # Convert to int for Gradio input compatibility
        min_length=int(min_length),  # Convert to int for Gradio input compatibility
        length_penalty=1.2,  # Length penalty for beam search
        num_beams=4,  # Number of beams for beam search
        early_stopping=True  # Stop early when an optimal summary is found
    )
    return summary[0]["summary_text"]

# Define inputs for the Gradio interface with better layout and styling
inputs = [
    gr.Textbox(
        label="Input Text",
        lines=10, 
        placeholder="Enter the text to summarize here...",
        interactive=True,
        elem_id="input_text_box"
    ),
    gr.Number(
        label="Minimum Length",
        value=50, 
        precision=0,
        interactive=True,
        elem_id="min_length"
    ),
    gr.Number(
        label="Maximum Length",
        value=200, 
        precision=0,
        interactive=True,
        elem_id="max_length"
    ),
]

# Define the Gradio interface
demo = gr.Interface(
    fn=summarizer,
    inputs=inputs,
    outputs=gr.Textbox(
        label="Summary", 
        lines=6, 
        placeholder="Your summary will appear here.",
        interactive=False,
        elem_id="output_summary"
    ),
    title="Text Summarization Tool",
    description="Provide a text input, specify the minimum and maximum lengths for the summary, and get a concise version of your text.",
    theme="huggingface",  # Optional, you can change to other themes like 'compact'
    allow_flagging="never",  # Disable flagging
)

# Launch the interface with shareable link
demo.launch(share=True)