Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
from transformers import MBartForConditionalGeneration, MBart50TokenizerFast
|
3 |
+
|
4 |
+
# Load model and tokenizer
|
5 |
+
model_name = "facebook/mbart-large-50"
|
6 |
+
model = MBartForConditionalGeneration.from_pretrained(model_name)
|
7 |
+
tokenizer = MBart50TokenizerFast.from_pretrained(model_name)
|
8 |
+
|
9 |
+
def translate(text, direction):
|
10 |
+
if direction == "Arab → Indonesia":
|
11 |
+
src_lang = "ar_AR"
|
12 |
+
tgt_lang = "id_ID"
|
13 |
+
else:
|
14 |
+
src_lang = "id_ID"
|
15 |
+
tgt_lang = "ar_AR"
|
16 |
+
|
17 |
+
tokenizer.src_lang = src_lang
|
18 |
+
inputs = tokenizer(text, return_tensors="pt", padding=True, truncation=True)
|
19 |
+
generated_tokens = model.generate(**inputs, forced_bos_token_id=tokenizer.lang_code_to_id[tgt_lang])
|
20 |
+
translation = tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)[0]
|
21 |
+
return translation
|
22 |
+
|
23 |
+
# Define UI
|
24 |
+
demo = gr.Interface(
|
25 |
+
fn=translate,
|
26 |
+
inputs=[
|
27 |
+
gr.Textbox(label="Masukkan teks"),
|
28 |
+
gr.Radio(["Arab → Indonesia", "Indonesia → Arab"], label="Arah Terjemahan")
|
29 |
+
],
|
30 |
+
outputs=gr.Textbox(label="Hasil Terjemahan"),
|
31 |
+
title="Penerjemah Arab - Indonesia dengan mBART50",
|
32 |
+
description="Masukkan teks dalam bahasa Arab atau Indonesia dan pilih arah terjemahan."
|
33 |
+
)
|
34 |
+
|
35 |
+
demo.launch()
|