app.py

import gradio as gr
from transformers import pipeline, GPT2TokenizerFast

modelId = "luel/gpt2-tigrinya-small"

tokenizer = GPT2TokenizerFast.from_pretrained(modelId, model_max_length=128)

generator = pipeline("text-generation", model=modelId, tokenizer=tokenizer, pad_token_id=tokenizer.pad_token_id, eos_token_id=tokenizer.eos_token_id)

def generate_text(prompt, max_length, temperature):
    try:
        generated = generator(
            prompt,
            max_length=max_length,
            temperature=temperature,
            do_sample=True,
            repetition_penalty=1.5
        )
        return generated[0]['generated_text']
    except Exception as e:
        return f"Something went wrong, try again. Error: {str(e)}"

def create_interface():
    with gr.Blocks() as demo:
        gr.Markdown("# Tigrinya Text Generator (GPT-2)")
        gr.Markdown(
            "This is a GPT-2 model trained from scratch on Tigrinya text data, primarily from news sources. "
            "Enter your Tigrinya text prompt and adjust the parameters to generate text."
        )
        
        with gr.Row():
            input_temperature = gr.Slider(
                minimum=0.1,
                maximum=1.0,
                value=0.7,
                step=0.1,
                label="Temperature",
            )
            input_max_length = gr.Slider(
                minimum=10,
                maximum=128,
                value=60,
                step=1,
                label="Maximum Length",
            )
        
        with gr.Row():
            with gr.Column(scale=1):
                input_prompt = gr.Textbox(
                    label="Enter your Tigrinya text prompt",
                    placeholder="ትግራይ",
                    lines=5
                )
            
            with gr.Column(scale=1):
                output_text = gr.Textbox(
                    label="Generated Text",
                    lines=5,
                    interactive=True
                )
        
        with gr.Row():
            generate_btn = gr.Button("Generate", variant="primary")
            clear_btn = gr.ClearButton([input_prompt, output_text])
        
        generate_btn.click(
            fn=generate_text,
            inputs=[input_prompt, input_max_length, input_temperature],
            outputs=output_text
        )
        
        gr.Examples(
            examples=[
                ["ትግራይ"],
                ["ኣዲስ ኣበባ"],
                ["ሰላም"]
            ],
            inputs=input_prompt
        )
    
    return demo            

if __name__ == "__main__":
    demo = create_interface()
    demo.queue().launch(debug=True)