|
import gradio as gr |
|
from transformers import AutoTokenizer, AutoModelForSequenceClassification, GPTNeoXForCausalLM |
|
|
|
|
|
tokenizer_ara = AutoTokenizer.from_pretrained("aubmindlab/bert-base-arabertv02") |
|
model_ara = AutoModelForSequenceClassification.from_pretrained("aubmindlab/bert-base-arabertv02") |
|
tokenizer_gpt = AutoTokenizer.from_pretrained("EleutherAI/gpt-neox-20b") |
|
model_gpt = GPTNeoXForCausalLM.from_pretrained("EleutherAI/gpt-neox-20b") |
|
|
|
|
|
def generate_text(input_text): |
|
input_ids_ara = tokenizer_ara(input_text, return_tensors="pt").input_ids |
|
outputs_ara = model_ara(input_ids_ara) |
|
input_ids_gpt = tokenizer_gpt(outputs_ara.logits.argmax(-1).item(), return_tensors="pt").input_ids |
|
outputs_gpt = model_gpt.generate(input_ids_gpt, max_length=100) |
|
generated_text = tokenizer_gpt.decode(outputs_gpt[0]) |
|
return generated_text |
|
|
|
|
|
iface = gr.Interface( |
|
fn=generate_text, |
|
inputs=gr.Textbox(lines=5, placeholder="أدخل نصًا عربيًا هنا..."), |
|
outputs=gr.Textbox(lines=5), |
|
title="توليد النصوص العربية باستخدام AraBERT و GPT-NeoX", |
|
description="أدخل نصًا عربيًا وستقوم هذه الأداة بتحليله باستخدام AraBERT ثم توليد نص جديد باستخدام GPT-NeoX." |
|
) |
|
|
|
iface.launch() |
|
|
