snoop2head commited on
Commit
4fd8256
·
1 Parent(s): be25e94

change to backtranslation only

Browse files
Files changed (2) hide show
  1. app.py +46 -0
  2. requirements.txt +4 -0
app.py ADDED
@@ -0,0 +1,46 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # -*- coding: utf-8 -*-
2
+ import numpy as np
3
+ import streamlit as st
4
+ from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
5
+
6
+
7
+ st.set_page_config(page_title='KoQuillBot', layout='wide', initial_sidebar_state='expanded')
8
+
9
+ tokenizer = AutoTokenizer.from_pretrained("QuoQA-NLP/KE-T5-Ko2En-Base")
10
+ ko2en_model = AutoModelForSeq2SeqLM.from_pretrained("QuoQA-NLP/KE-T5-Ko2En-Base")
11
+ en2ko_model = AutoModelForSeq2SeqLM.from_pretrained("QuoQA-NLP/KE-T5-En2Ko-Base")
12
+
13
+
14
+ st.title("🤖 KoQuillBot")
15
+
16
+ src_text = st.text_area("바꾸고 싶은 문장을 입력하세요:",height=None,max_chars=None,key=None,help="Enter your text here")
17
+
18
+ if st.button('문장 변환'):
19
+ if src_text == "":
20
+ st.warning('Please **enter text** for translation')
21
+
22
+ else:
23
+ translated = ko2en_model.generate(
24
+ **tokenizer([src_text], return_tensors="pt", padding=True, max_length=64,),
25
+ max_length=64,
26
+ num_beams=5,
27
+ repetition_penalty=1.3,
28
+ no_repeat_ngram_size=3,
29
+ num_return_sequences=1,
30
+ )
31
+
32
+ backtranslated = en2ko_model.generate(
33
+ **tokenizer([translated], return_tensors="pt", padding=True, max_length=64,),
34
+ max_length=64,
35
+ num_beams=5,
36
+ repetition_penalty=1.3,
37
+ no_repeat_ngram_size=3,
38
+ num_return_sequences=1,
39
+ )
40
+ else:
41
+ pass
42
+
43
+
44
+ print(backtranslated)
45
+
46
+ st.write(backtranslated)
requirements.txt ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ transformers
2
+ streamlit
3
+ torch
4
+ numpy