Spaces:
Runtime error
Runtime error
Upload app.py
Browse files
app.py
ADDED
@@ -0,0 +1,72 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM, pipeline
|
2 |
+
import gradio as gr
|
3 |
+
|
4 |
+
|
5 |
+
DEBUG_UI = False
|
6 |
+
LANGS = {
|
7 |
+
'English': 'eng_Latn',
|
8 |
+
'Interslavic': 'isv_Latn',
|
9 |
+
# 'Интерславик': 'isv_Cyrl',
|
10 |
+
'Russian': 'rus_Cyrl',
|
11 |
+
'Belarusian': 'bel_Cyrl',
|
12 |
+
'Ukrainian': 'ukr_Cyrl',
|
13 |
+
'Bulgarian': 'bul_Cyrl',
|
14 |
+
'Macedonian': 'mkd_Cyrl',
|
15 |
+
'Serbian': 'srp_Cyrl',
|
16 |
+
'Croatian': 'hrv_Latn',
|
17 |
+
'Bosnian': 'bos_Latn',
|
18 |
+
'Slovenian': 'slv_Latn',
|
19 |
+
'Slovak': 'slk_Latn',
|
20 |
+
'Czech': 'ces_Latn',
|
21 |
+
'Polish': 'pol_Latn',
|
22 |
+
'Silesian': 'szl_Latn',
|
23 |
+
'Esperanto': 'epo_Latn',
|
24 |
+
'German': 'deu_Latn',
|
25 |
+
'French': 'fra_Latn',
|
26 |
+
'Spanish': 'spa_Latn',
|
27 |
+
}
|
28 |
+
|
29 |
+
|
30 |
+
if DEBUG_UI:
|
31 |
+
def translate(text, src_lang, tgt_lang):
|
32 |
+
return text
|
33 |
+
|
34 |
+
else:
|
35 |
+
# model_name = 'Salavat/nllb-200-distilled-600M-finetuned-isv'
|
36 |
+
model_name = 'Salavat/nllb-200-distilled-600M-finetuned-isv_v2'
|
37 |
+
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
|
38 |
+
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
39 |
+
|
40 |
+
def get_lang(lang):
|
41 |
+
if lang in LANGS:
|
42 |
+
return LANGS[lang]
|
43 |
+
return lang
|
44 |
+
|
45 |
+
def translate(text, from_, to_):
|
46 |
+
"""
|
47 |
+
Translate the text from source lang to target lang
|
48 |
+
"""
|
49 |
+
translation_pipeline = pipeline(
|
50 |
+
"translation", model=model, tokenizer=tokenizer, max_length=400,
|
51 |
+
src_lang=LANGS.get(from_, from_), tgt_lang=LANGS.get(to_, to_)
|
52 |
+
)
|
53 |
+
result = translation_pipeline(text.split('\n'))
|
54 |
+
output = '\n'.join(line['translation_text'] for line in result)
|
55 |
+
return output
|
56 |
+
|
57 |
+
|
58 |
+
demo = gr.Blocks()
|
59 |
+
|
60 |
+
with demo:
|
61 |
+
gr.Markdown("This is a modified version of the original [NLLB-Translator](https://huggingface.co/spaces/Narrativaai/NLLB-Translator) space")
|
62 |
+
with gr.Row():
|
63 |
+
lang_input = gr.components.Dropdown(label="From", choices=list(LANGS.keys()), value='English')
|
64 |
+
lang_output = gr.components.Dropdown(label="To", choices=list(LANGS.keys()), value='Interslavic')
|
65 |
+
with gr.Row().style(equal_height=True):
|
66 |
+
text_input = gr.components.Textbox(label="Text", lines=7, placeholder="Your text")
|
67 |
+
text_output = gr.components.Textbox(label="Result", lines=7, placeholder="Translation...")
|
68 |
+
button = gr.Button("Translate")
|
69 |
+
|
70 |
+
button.click(translate, inputs=[text_input, lang_input, lang_output], outputs=text_output)
|
71 |
+
|
72 |
+
demo.launch()
|