musfiqdehan
commited on
Commit
·
d4a2937
1
Parent(s):
4a52b88
Add Gradio interface for Bangla PoS Taggers
Browse files
app.py
CHANGED
@@ -0,0 +1,139 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
from gradio_rich_textbox import RichTextbox
|
3 |
+
|
4 |
+
from helper.text_preprocess import space_punc
|
5 |
+
from helper.alignment_mappers import select_model
|
6 |
+
from helper.pos_taggers import select_pos_tagger
|
7 |
+
from helper.translators import select_translator
|
8 |
+
|
9 |
+
|
10 |
+
def bn_postagger(src, translator, model_name, tagger):
|
11 |
+
"""
|
12 |
+
Bangla PoS Tagger
|
13 |
+
"""
|
14 |
+
|
15 |
+
src = space_punc(src)
|
16 |
+
|
17 |
+
tgt_base, tgt = select_translator(src, translator)
|
18 |
+
|
19 |
+
model_name = select_model(model_name)
|
20 |
+
|
21 |
+
result, pos_accuracy = select_pos_tagger(src, tgt, model_name, tagger)
|
22 |
+
|
23 |
+
return tgt_base, result, pos_accuracy
|
24 |
+
|
25 |
+
|
26 |
+
# Define the Gradio interface
|
27 |
+
# demo = gr.Interface(
|
28 |
+
# fn=bn_postagger,
|
29 |
+
# inputs=[
|
30 |
+
# gr.Textbox(label="Enter Bangla Sentence", placeholder="বাংলা বাক্য লিখুন"),
|
31 |
+
# gr.Dropdown(["Google", "BanglaNMT", "MyMemory"], label="Select a Translator"),
|
32 |
+
# gr.Dropdown(["spaCy", "NLTK", "Flair", "TextBlob"], label="Select a PoS Tagger")
|
33 |
+
# ],
|
34 |
+
# outputs= [
|
35 |
+
# gr.Textbox(label="English Translation"),
|
36 |
+
# RichTextbox(label="PoS Tags"),
|
37 |
+
# gr.Textbox(label="Overall PoS Tagging Accuracy")
|
38 |
+
# ],
|
39 |
+
# live=False,
|
40 |
+
# title="Bangla PoS Taggers",
|
41 |
+
# theme='',
|
42 |
+
# examples=[
|
43 |
+
# ["বাংলাদেশ দক্ষিণ এশিয়ার একটি সার্বভৌম রাষ্ট্র।"],
|
44 |
+
# ["বাংলাদেশের সংবিধানিক নাম কি?"],
|
45 |
+
# ["বাংলাদেশের সাংবিধানিক নাম গণপ্রজাতন্ত্রী বাংলাদেশ।"],
|
46 |
+
# ["তিনজনের কেউই বাবার পথ ধরে প্রযুক্তি দুনিয়ায় হাঁটেননি।"],
|
47 |
+
# ["বিশ্বের আরও একটি সেরা ক্লাব।"]
|
48 |
+
|
49 |
+
# ]
|
50 |
+
# )
|
51 |
+
|
52 |
+
with gr.Blocks(css="styles.css") as demo:
|
53 |
+
gr.HTML("<h1>Bangla PoS Taggers</h1>")
|
54 |
+
gr.HTML("<p>Parts of Speech (PoS) Tagging of Bangla Sentence using Bangla-English <strong>Word Alignment</strong></p>")
|
55 |
+
|
56 |
+
with gr.Row():
|
57 |
+
with gr.Column():
|
58 |
+
inputs = [
|
59 |
+
gr.Textbox(
|
60 |
+
label="Enter Bangla Sentence",
|
61 |
+
placeholder="বাংলা বাক্য লিখুন"
|
62 |
+
),
|
63 |
+
gr.Dropdown(
|
64 |
+
choices=["Google", "BanglaNMT", "MyMemory"],
|
65 |
+
label="Select a Translator"
|
66 |
+
),
|
67 |
+
gr.Dropdown(
|
68 |
+
choices=["Google-mBERT (Base-Multilingual)", "Neulab-AwesomeAlign (Bn-En-0.5M)", "BUET-BanglaBERT (Large)", "SagorSarker-BanglaBERT (Base)", "SentenceTransformers-LaBSE (Multilingual)"],
|
69 |
+
label="Select a Model"
|
70 |
+
),
|
71 |
+
gr.Dropdown(
|
72 |
+
choices=["spaCy", "NLTK", "Flair", "TextBlob"],
|
73 |
+
label="Select a PoS Tagger"
|
74 |
+
)
|
75 |
+
]
|
76 |
+
|
77 |
+
btn = gr.Button(value="Submit", elem_classes="mybtn")
|
78 |
+
gr.ClearButton(inputs)
|
79 |
+
|
80 |
+
with gr.Column():
|
81 |
+
outputs = [
|
82 |
+
gr.Textbox(label="English Translation"),
|
83 |
+
RichTextbox(label="PoS Tags"),
|
84 |
+
gr.Textbox(label="PoS Tagging Accuracy (Based on Unknown(UNK) Tags)")
|
85 |
+
]
|
86 |
+
|
87 |
+
btn.click(bn_postagger, inputs, outputs)
|
88 |
+
|
89 |
+
gr.Examples([
|
90 |
+
[
|
91 |
+
"বাংলাদেশ দক্ষিণ এশিয়ার একটি সার্বভৌম রাষ্ট্র।",
|
92 |
+
"Google",
|
93 |
+
"Neulab-AwesomeAlign (Bn-En-0.5M)",
|
94 |
+
"NLTK"
|
95 |
+
],
|
96 |
+
[
|
97 |
+
"বাংলাদেশের সংবিধানিক নাম কি?",
|
98 |
+
"Google",
|
99 |
+
"Google-mBERT (Base-Multilingual)",
|
100 |
+
"spaCy"
|
101 |
+
],
|
102 |
+
[
|
103 |
+
"বাংলাদেশের সাংবিধানিক নাম গণপ্রজাতন্ত্রী বাংলাদেশ।",
|
104 |
+
"Google",
|
105 |
+
"Google-mBERT (Base-Multilingual)",
|
106 |
+
"TextBlob"
|
107 |
+
],
|
108 |
+
[
|
109 |
+
"তিনজনের কেউই বাবার পথ ধরে প্রযুক্তি দুনিয়ায় হাঁটেননি।",
|
110 |
+
"Google",
|
111 |
+
"Neulab-AwesomeAlign (Bn-En-0.5M)",
|
112 |
+
"spaCy"
|
113 |
+
],
|
114 |
+
[
|
115 |
+
"তিনজনের কেউই বাবার পথ ধরে প্রযুক্তি দুনিয়ায় হাঁটেননি।",
|
116 |
+
"BanglaNMT",
|
117 |
+
"Google-mBERT (Base-Multilingual)",
|
118 |
+
"spaCy"
|
119 |
+
],
|
120 |
+
[
|
121 |
+
"তিনজনের কেউই বাবার পথ ধরে প্রযুক্তি দুনিয়ায় হাঁটেননি।",
|
122 |
+
"MyMemory",
|
123 |
+
"Google-mBERT (Base-Multilingual)",
|
124 |
+
"spaCy"
|
125 |
+
],
|
126 |
+
[
|
127 |
+
"বিশ্বের আরও একটি সেরা ক্লাব।",
|
128 |
+
"Google",
|
129 |
+
"Neulab-AwesomeAlign (Bn-En-0.5M)",
|
130 |
+
"Flair"
|
131 |
+
]
|
132 |
+
|
133 |
+
], inputs)
|
134 |
+
|
135 |
+
|
136 |
+
|
137 |
+
# Launch the Gradio app
|
138 |
+
if __name__ == "__main__":
|
139 |
+
demo.launch()
|