Spaces:
No application file
No application file
Commit
·
3f0a242
1
Parent(s):
7836d4c
Upload turjumandemo.py
Browse files- turjumandemo.py +48 -0
turjumandemo.py
ADDED
|
@@ -0,0 +1,48 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# -*- coding: utf-8 -*-
|
| 2 |
+
"""TurjumanDemo
|
| 3 |
+
|
| 4 |
+
Automatically generated by Colaboratory.
|
| 5 |
+
|
| 6 |
+
Original file is located at
|
| 7 |
+
https://colab.research.google.com/drive/1VVJ7uPEYD8Q1pR-IINWWAQVpqyP1XnzD
|
| 8 |
+
"""
|
| 9 |
+
|
| 10 |
+
# Installing dependencies
|
| 11 |
+
!pip install gradio
|
| 12 |
+
!pip install turjuman transformers
|
| 13 |
+
!git clone https://huggingface.co/spaces/ahmedoumar/TurjumanDemo
|
| 14 |
+
|
| 15 |
+
# Import our modules
|
| 16 |
+
import gradio as gr
|
| 17 |
+
from turjuman import turjuman
|
| 18 |
+
import logging
|
| 19 |
+
import os
|
| 20 |
+
from transformers import AutoTokenizer
|
| 21 |
+
|
| 22 |
+
logging.basicConfig(
|
| 23 |
+
format="%(asctime)s | %(levelname)s | %(name)s | %(message)s",
|
| 24 |
+
datefmt="%Y-%m-%d %H:%M:%S",
|
| 25 |
+
level=os.environ.get("LOGLEVEL", "INFO").upper(),
|
| 26 |
+
)
|
| 27 |
+
logger = logging.getLogger("turjuman.translate")
|
| 28 |
+
cache_dir="/content/mycache"
|
| 29 |
+
|
| 30 |
+
# Get the turjuman object and its tokenizer
|
| 31 |
+
turj = turjuman.turjuman(logger, cache_dir)
|
| 32 |
+
tokenizer = AutoTokenizer.from_pretrained('UBC-NLP/AraT5-base-title-generation')
|
| 33 |
+
|
| 34 |
+
# The translate function
|
| 35 |
+
def translate(sent):
|
| 36 |
+
beam_options = {"search_method":"beam", "seq_length": 300, "num_beams":5, "no_repeat_ngram_size":2, "max_outputs":1}
|
| 37 |
+
targets = turj.translate(sent,**beam_options)
|
| 38 |
+
#print(targets)
|
| 39 |
+
ans = ""
|
| 40 |
+
for target in targets:
|
| 41 |
+
target = tokenizer.decode(target, skip_special_tokens=True, clean_up_tokenization_spaces=True)
|
| 42 |
+
ans += target
|
| 43 |
+
return ans
|
| 44 |
+
|
| 45 |
+
print(translate('Здравствуй, друг'))
|
| 46 |
+
|
| 47 |
+
gr.Interface(fn=translate, inputs=['text'], outputs=['text']).launch(width=1000, height=1000)
|
| 48 |
+
|