app.py

import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer
import os

# token
token = os.environ['TOKEN']

# Load the pretrained model and tokenizer
MODEL_NAME = "atlasia/Al-Atlas-LLM"

tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME, token=token)
model = AutoModelForCausalLM.from_pretrained(MODEL_NAME, token=token)

def generate_text(prompt, max_length=256, temperature=0.7, top_p=0.9, top_k=150, repetition_penalty=1.5):
    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
    output = model.generate(
        **inputs, 
        max_length=max_length, 
        temperature=temperature, 
        top_p=top_p, 
        do_sample=True,
        repetition_penalty=repetition_penalty,
        num_beams=8,
        top_k= top_k,
        early_stopping = True,
    )
    return tokenizer.decode(output[0], skip_special_tokens=True)

# Create the Gradio interface
iface = gr.Interface(
    fn=generate_text,
    inputs=[
        gr.Textbox(label="Prompt: دخل النص بالدارجة"),
        gr.Slider(50, 500, value=256, label="Max Length"),
        gr.Slider(0.1, 1.5, value=0.7, label="Temperature"),
        gr.Slider(0.1, 1.0, value=0.9, label="Top-p"),
        gr.Slider(1, 10000, value=150, label="Top-k"),
        gr.Slider(0.0, 100.0, value=1.5, label="Repetition Penalty"),
    ],
    outputs=gr.Textbox(label="Generated Text in Moroccan Darija"),
    title="Moroccan Darija LLM",
    description="Enter a prompt and get AI-generated text using our pretrained LLM on Moroccan Darija."
)

if __name__ == "__main__":
    iface.launch()