Najish Mahmud
commited on
Commit
·
7900aac
1
Parent(s):
96f24e9
first launch
Browse files- .gitignore +1 -0
- app.py +37 -0
- requirements.txt +104 -0
.gitignore
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
.venv
|
app.py
ADDED
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# import transformers
|
2 |
+
import sentencepiece
|
3 |
+
import torch
|
4 |
+
import gradio as gr
|
5 |
+
|
6 |
+
# from indicnlp.normalize.indic_normalize import IndicNormalizerFactory
|
7 |
+
from transformers import MarianMTModel, AutoTokenizer, T5ForConditionalGeneration
|
8 |
+
from deep_translator import GoogleTranslator
|
9 |
+
|
10 |
+
def summarize(message):
|
11 |
+
tokenizer_bn_en = AutoTokenizer.from_pretrained("Helsinki-NLP/opus-mt-bn-en")
|
12 |
+
model_bn_en = MarianMTModel.from_pretrained("Helsinki-NLP/opus-mt-bn-en")
|
13 |
+
tokenizer = AutoTokenizer.from_pretrained('t5-base')
|
14 |
+
model = T5ForConditionalGeneration.from_pretrained('t5-base')
|
15 |
+
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
|
16 |
+
with torch.no_grad():
|
17 |
+
# tokenize the input text
|
18 |
+
input_ids = tokenizer_bn_en.encode(message, return_tensors='pt')
|
19 |
+
|
20 |
+
# generate the output translation
|
21 |
+
output_ids = model_bn_en.generate(input_ids)
|
22 |
+
output_text1 = tokenizer_bn_en.decode(output_ids[0], skip_special_tokens=True)
|
23 |
+
|
24 |
+
#input for the summarization model
|
25 |
+
input_text2 = output_text1
|
26 |
+
|
27 |
+
# tokenize the input text and generate the summary in english
|
28 |
+
input_ids = tokenizer.encode(input_text2, return_tensors='pt')
|
29 |
+
summary_ids = model.generate(input_ids, max_length=50, num_beams=30, early_stopping=True)
|
30 |
+
summary = tokenizer.decode(summary_ids[0], skip_special_tokens=True)
|
31 |
+
summarized = GoogleTranslator(source='en', target='bn').translate(summary)
|
32 |
+
return summarized
|
33 |
+
|
34 |
+
# summarize("যুক্তরাষ্ট্রের নিউজার্সি অঙ্গরাজ্যের হোপওয়েল শহর এলাকার একটি বাড়ির ছাদে একটি ধাতব বস্তু আছড়ে পড়েছে। ধাতব বস্তুটি উল্কাপিণ্ড হতে পারে বলে ধারণা করা হচ্ছে। এ ঘটনায় কেউ হতাহত হয়নি বলে জানিয়েছে পুলিশ। স্থানীয় সময় গত সোমবার বেলা একটার কিছু সময় পর ওল্ড ওয়াশিংটন ক্রসিং পেনিংটন রোড এলাকায় অবস্থিত একটি বাড়িতে এ ঘটনা ঘটে। পুলিশের ভাষ্য, বস্তুটি আয়তাকার। দেখতে ধাতব পদার্থের মতো। বাড়ির বাসিন্দারা জানান, ঘটনার পর তাঁরা সংশ্লিষ্ট সহায়তা কর্তৃপক্ষকে ফোন করেন। আকাশ থেকে কোনো বস্তু যে বাড়ির ছাদে এসে পড়েছে, এ কথা কর্তৃপক্ষকে বোঝাতে তাঁদের বেগ পেতে হয়েছে।")
|
35 |
+
|
36 |
+
iface = gr.Interface(fn=summarize, inputs="text", outputs="text")
|
37 |
+
iface.launch()
|
requirements.txt
ADDED
@@ -0,0 +1,104 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
appdirs==1.4.4
|
2 |
+
async-timeout==4.0.2
|
3 |
+
attrs==22.2.0
|
4 |
+
autocommand==2.2.2
|
5 |
+
beautifulsoup4==4.12.2
|
6 |
+
Brlapi==0.8.4
|
7 |
+
btrfsutil==6.3
|
8 |
+
build==0.10.0
|
9 |
+
cached-property==1.5.2
|
10 |
+
cffi==1.15.1
|
11 |
+
chardet==5.1.0
|
12 |
+
cmake==3.26.3
|
13 |
+
cryptography==40.0.2
|
14 |
+
cssselect==1.2.0
|
15 |
+
Cython==0.29.34
|
16 |
+
deep-translator==1.10.1
|
17 |
+
fastjsonschema==2.16.3
|
18 |
+
filelock==3.12.0
|
19 |
+
flit_core==3.8.0
|
20 |
+
fsspec==2023.5.0
|
21 |
+
greenlet==2.0.2
|
22 |
+
grpcio==1.53.0
|
23 |
+
huggingface-hub==0.14.1
|
24 |
+
idna==3.4
|
25 |
+
ifaddr==0.2.0
|
26 |
+
inflect==6.0.4
|
27 |
+
installer==0.7.0
|
28 |
+
jaraco.context==4.3.0
|
29 |
+
jaraco.functools==3.6.0
|
30 |
+
jaraco.text==3.11.1
|
31 |
+
Jinja2==3.1.2
|
32 |
+
lensfun==0.3.3
|
33 |
+
Levenshtein==0.21.0
|
34 |
+
libfdt==1.7.0
|
35 |
+
libtorrent==2.0.8
|
36 |
+
lit==16.0.3
|
37 |
+
louis==3.25.0
|
38 |
+
lxml==4.9.2
|
39 |
+
MarkupSafe==2.1.2
|
40 |
+
meson==1.1.0
|
41 |
+
more-itertools==9.1.0
|
42 |
+
mpmath==1.3.0
|
43 |
+
msgpack==1.0.5
|
44 |
+
netifaces==0.11.0
|
45 |
+
netsnmp-python==1.0a1
|
46 |
+
networkx==3.1
|
47 |
+
nspektr==0.4.0
|
48 |
+
numpy==1.24.3
|
49 |
+
nvidia-cublas-cu11==11.10.3.66
|
50 |
+
nvidia-cuda-cupti-cu11==11.7.101
|
51 |
+
nvidia-cuda-nvrtc-cu11==11.7.99
|
52 |
+
nvidia-cuda-runtime-cu11==11.7.99
|
53 |
+
nvidia-cudnn-cu11==8.5.0.96
|
54 |
+
nvidia-cufft-cu11==10.9.0.58
|
55 |
+
nvidia-curand-cu11==10.2.10.91
|
56 |
+
nvidia-cusolver-cu11==11.4.0.1
|
57 |
+
nvidia-cusparse-cu11==11.7.4.91
|
58 |
+
nvidia-nccl-cu11==2.14.3
|
59 |
+
nvidia-nvtx-cu11==11.7.91
|
60 |
+
ordered-set==4.1.0
|
61 |
+
packaging==23.0
|
62 |
+
platformdirs==3.5.0
|
63 |
+
ply==3.11
|
64 |
+
protobuf==4.21.12
|
65 |
+
psutil==5.9.5
|
66 |
+
pwquality==1.4.5
|
67 |
+
pycairo==1.23.0
|
68 |
+
pycparser==2.21
|
69 |
+
pydantic==1.10.7
|
70 |
+
pyenchant==3.2.2
|
71 |
+
PyGObject==3.44.1
|
72 |
+
PyNaCl==1.4.0
|
73 |
+
pynvim==0.4.3
|
74 |
+
pypandoc==1.11
|
75 |
+
pyparsing==3.0.9
|
76 |
+
pyproject_hooks==1.0.0
|
77 |
+
pywayland==0.4.15
|
78 |
+
pywlroots==0.15.24
|
79 |
+
pyxdg==0.28
|
80 |
+
PyYAML==6.0
|
81 |
+
rapidfuzz==3.0.0
|
82 |
+
regex==2023.5.5
|
83 |
+
requests==2.28.2
|
84 |
+
scour==0.38.2
|
85 |
+
sentencepiece==0.1.99
|
86 |
+
setproctitle==1.3.2
|
87 |
+
shiboken2==5.15.9
|
88 |
+
six==1.16.0
|
89 |
+
soupsieve==2.4.1
|
90 |
+
sympy==1.12
|
91 |
+
tokenizers==0.13.3
|
92 |
+
tomli==2.0.1
|
93 |
+
tomli_w==1.0.0
|
94 |
+
torch==2.0.1
|
95 |
+
tqdm==4.65.0
|
96 |
+
transformers==4.29.1
|
97 |
+
triton==2.0.0
|
98 |
+
trove-classifiers==2023.4.29
|
99 |
+
typing_extensions==4.5.0
|
100 |
+
ufw==0.36.1
|
101 |
+
urllib3==1.26.15
|
102 |
+
validate-pyproject==0.12.2.post1.dev0+g2940279.d20230508
|
103 |
+
xkbcommon==0.8
|
104 |
+
zeroconf==0.39.4
|