Commit
·
af97119
1
Parent(s):
1a75433
feat: Update deps, sort out deprecation warnings
Browse files- app.py +106 -92
- requirements.txt +5 -2
app.py
CHANGED
@@ -2,75 +2,30 @@
|
|
2 |
|
3 |
from typing import Dict, Tuple
|
4 |
import gradio as gr
|
5 |
-
from
|
|
|
|
|
6 |
from luga import language as detect_language
|
|
|
7 |
import re
|
8 |
-
|
9 |
-
|
10 |
-
def classification(
|
11 |
-
doc: str,
|
12 |
-
da_hypothesis_template: str,
|
13 |
-
da_candidate_labels: str,
|
14 |
-
sv_hypothesis_template: str,
|
15 |
-
sv_candidate_labels: str,
|
16 |
-
no_hypothesis_template: str,
|
17 |
-
no_candidate_labels: str,
|
18 |
-
) -> Dict[str, float]:
|
19 |
-
"""Classify text into categories.
|
20 |
-
|
21 |
-
Args:
|
22 |
-
doc (str):
|
23 |
-
Text to classify.
|
24 |
-
da_hypothesis_template (str):
|
25 |
-
Template for the hypothesis to be used for Danish classification.
|
26 |
-
da_candidate_labels (str):
|
27 |
-
Comma-separated list of candidate labels for Danish classification.
|
28 |
-
sv_hypothesis_template (str):
|
29 |
-
Template for the hypothesis to be used for Swedish classification.
|
30 |
-
sv_candidate_labels (str):
|
31 |
-
Comma-separated list of candidate labels for Swedish classification.
|
32 |
-
no_hypothesis_template (str):
|
33 |
-
Template for the hypothesis to be used for Norwegian classification.
|
34 |
-
no_candidate_labels (str):
|
35 |
-
Comma-separated list of candidate labels for Norwegian classification.
|
36 |
-
|
37 |
-
Returns:
|
38 |
-
dict of str to float:
|
39 |
-
The predicted label and the confidence score.
|
40 |
-
"""
|
41 |
-
# Detect the language of the text
|
42 |
-
language = detect_language(doc.replace('\n', ' ')).name
|
43 |
-
|
44 |
-
# Set the hypothesis template and candidate labels based on the detected language
|
45 |
-
if language == "sv":
|
46 |
-
hypothesis_template = sv_hypothesis_template
|
47 |
-
candidate_labels = re.split(r', *', sv_candidate_labels)
|
48 |
-
elif language == "no":
|
49 |
-
hypothesis_template = no_hypothesis_template
|
50 |
-
candidate_labels = re.split(r', *', no_candidate_labels)
|
51 |
-
else:
|
52 |
-
hypothesis_template = da_hypothesis_template
|
53 |
-
candidate_labels = re.split(r', *', da_candidate_labels)
|
54 |
-
|
55 |
-
# Run the classifier on the text
|
56 |
-
result = classifier(
|
57 |
-
doc,
|
58 |
-
candidate_labels=candidate_labels,
|
59 |
-
hypothesis_template=hypothesis_template,
|
60 |
-
)
|
61 |
-
|
62 |
-
print(result)
|
63 |
-
|
64 |
-
# Return the predicted label
|
65 |
-
return {lbl: score for lbl, score in zip(result["labels"], result["scores"])}
|
66 |
|
67 |
|
68 |
def main():
|
|
|
|
|
69 |
|
70 |
# Load the zero-shot classification pipeline
|
71 |
-
global classifier
|
72 |
-
|
73 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
74 |
)
|
75 |
|
76 |
# Create dictionary of descriptions for each task, containing the hypothesis template
|
@@ -124,8 +79,8 @@ def main():
|
|
124 |
with gr.Blocks() as demo:
|
125 |
|
126 |
# Create title and description
|
127 |
-
|
128 |
-
|
129 |
Classify text in Danish, Swedish or Norwegian into categories, without
|
130 |
finetuning on any training data!
|
131 |
|
@@ -140,13 +95,13 @@ def main():
|
|
140 |
_Also, be patient, as this demo is running on a CPU!_
|
141 |
""")
|
142 |
|
143 |
-
with
|
144 |
|
145 |
# Input column
|
146 |
-
with
|
147 |
|
148 |
# Create a dropdown menu for the task
|
149 |
-
dropdown =
|
150 |
label="Task",
|
151 |
choices=[
|
152 |
"Sentiment classification",
|
@@ -155,37 +110,37 @@ def main():
|
|
155 |
"Product feedback detection",
|
156 |
"Define your own task!",
|
157 |
],
|
158 |
-
|
159 |
)
|
160 |
|
161 |
-
with
|
162 |
-
da_hypothesis_template =
|
163 |
label="Danish hypothesis template",
|
164 |
-
|
165 |
)
|
166 |
-
da_candidate_labels =
|
167 |
label="Danish candidate labels (comma separated)",
|
168 |
-
|
169 |
)
|
170 |
|
171 |
-
with
|
172 |
-
sv_hypothesis_template =
|
173 |
label="Swedish hypothesis template",
|
174 |
-
|
175 |
)
|
176 |
-
sv_candidate_labels =
|
177 |
label="Swedish candidate labels (comma separated)",
|
178 |
-
|
179 |
)
|
180 |
|
181 |
-
with
|
182 |
-
no_hypothesis_template =
|
183 |
label="Norwegian hypothesis template",
|
184 |
-
|
185 |
)
|
186 |
-
no_candidate_labels =
|
187 |
label="Norwegian candidate labels (comma separated)",
|
188 |
-
|
189 |
)
|
190 |
|
191 |
# When a new task is chosen, update the description
|
@@ -203,16 +158,16 @@ def main():
|
|
203 |
)
|
204 |
|
205 |
# Output column
|
206 |
-
with
|
207 |
|
208 |
# Create a text box for the input text
|
209 |
-
input_textbox =
|
210 |
-
label="Input text",
|
211 |
)
|
212 |
|
213 |
-
with
|
214 |
-
clear_btn =
|
215 |
-
submit_btn =
|
216 |
|
217 |
# When the clear button is clicked, clear the input text box
|
218 |
clear_btn.click(
|
@@ -220,10 +175,10 @@ def main():
|
|
220 |
)
|
221 |
|
222 |
|
223 |
-
with
|
224 |
|
225 |
# Create output text box
|
226 |
-
output_textbox =
|
227 |
|
228 |
# When the submit button is clicked, run the classifier on the input text
|
229 |
# and display the result in the output text box
|
@@ -242,7 +197,66 @@ def main():
|
|
242 |
)
|
243 |
|
244 |
# Run the app
|
245 |
-
demo.launch()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
246 |
|
247 |
|
248 |
if __name__ == "__main__":
|
|
|
2 |
|
3 |
from typing import Dict, Tuple
|
4 |
import gradio as gr
|
5 |
+
from gradio.components import Dropdown, Textbox, Row, Column, Button, Label, Markdown
|
6 |
+
from types import MethodType
|
7 |
+
from transformers import pipeline, AutoModelForSequenceClassification, AutoTokenizer
|
8 |
from luga import language as detect_language
|
9 |
+
import torch
|
10 |
import re
|
11 |
+
import os
|
12 |
+
import torch._dynamo
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
13 |
|
14 |
|
15 |
def main():
|
16 |
+
# Disable tokenizers parallelism
|
17 |
+
os.environ["TOKENIZERS_PARALLELISM"] = "false"
|
18 |
|
19 |
# Load the zero-shot classification pipeline
|
20 |
+
global classifier, model, tokenizer
|
21 |
+
model_id = "alexandrainst/scandi-nli-large"
|
22 |
+
model = AutoModelForSequenceClassification.from_pretrained(model_id)
|
23 |
+
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
24 |
+
model = torch.compile(model=model, backend="aot_eager")
|
25 |
+
model.eval()
|
26 |
+
classifier = pipeline("zero-shot-classification", model=model, tokenizer=tokenizer)
|
27 |
+
classifier.get_inference_context = MethodType(
|
28 |
+
lambda self: torch.no_grad, classifier
|
29 |
)
|
30 |
|
31 |
# Create dictionary of descriptions for each task, containing the hypothesis template
|
|
|
79 |
with gr.Blocks() as demo:
|
80 |
|
81 |
# Create title and description
|
82 |
+
Markdown("# Scandinavian Zero-shot Text Classification")
|
83 |
+
Markdown("""
|
84 |
Classify text in Danish, Swedish or Norwegian into categories, without
|
85 |
finetuning on any training data!
|
86 |
|
|
|
95 |
_Also, be patient, as this demo is running on a CPU!_
|
96 |
""")
|
97 |
|
98 |
+
with Row():
|
99 |
|
100 |
# Input column
|
101 |
+
with Column():
|
102 |
|
103 |
# Create a dropdown menu for the task
|
104 |
+
dropdown = Dropdown(
|
105 |
label="Task",
|
106 |
choices=[
|
107 |
"Sentiment classification",
|
|
|
110 |
"Product feedback detection",
|
111 |
"Define your own task!",
|
112 |
],
|
113 |
+
value="Sentiment classification",
|
114 |
)
|
115 |
|
116 |
+
with Row(variant="compact"):
|
117 |
+
da_hypothesis_template = Textbox(
|
118 |
label="Danish hypothesis template",
|
119 |
+
value="Dette eksempel er {}.",
|
120 |
)
|
121 |
+
da_candidate_labels = Textbox(
|
122 |
label="Danish candidate labels (comma separated)",
|
123 |
+
value="positivt, negativt, neutralt",
|
124 |
)
|
125 |
|
126 |
+
with Row(variant="compact"):
|
127 |
+
sv_hypothesis_template = Textbox(
|
128 |
label="Swedish hypothesis template",
|
129 |
+
value="Detta exempel är {}.",
|
130 |
)
|
131 |
+
sv_candidate_labels = Textbox(
|
132 |
label="Swedish candidate labels (comma separated)",
|
133 |
+
value="positivt, negativt, neutralt",
|
134 |
)
|
135 |
|
136 |
+
with Row(variant="compact"):
|
137 |
+
no_hypothesis_template = Textbox(
|
138 |
label="Norwegian hypothesis template",
|
139 |
+
value="Dette eksemplet er {}.",
|
140 |
)
|
141 |
+
no_candidate_labels = Textbox(
|
142 |
label="Norwegian candidate labels (comma separated)",
|
143 |
+
value="positivt, negativt, nøytralt",
|
144 |
)
|
145 |
|
146 |
# When a new task is chosen, update the description
|
|
|
158 |
)
|
159 |
|
160 |
# Output column
|
161 |
+
with Column():
|
162 |
|
163 |
# Create a text box for the input text
|
164 |
+
input_textbox = Textbox(
|
165 |
+
label="Input text", value="Jeg er helt vild med fodbolden 😊"
|
166 |
)
|
167 |
|
168 |
+
with Row():
|
169 |
+
clear_btn = Button(value="Clear")
|
170 |
+
submit_btn = Button(value="Submit", variant="primary")
|
171 |
|
172 |
# When the clear button is clicked, clear the input text box
|
173 |
clear_btn.click(
|
|
|
175 |
)
|
176 |
|
177 |
|
178 |
+
with Column():
|
179 |
|
180 |
# Create output text box
|
181 |
+
output_textbox = Label(label="Result")
|
182 |
|
183 |
# When the submit button is clicked, run the classifier on the input text
|
184 |
# and display the result in the output text box
|
|
|
197 |
)
|
198 |
|
199 |
# Run the app
|
200 |
+
demo.launch(width=.5)
|
201 |
+
|
202 |
+
|
203 |
+
@torch.compile()
|
204 |
+
def classification(
|
205 |
+
doc: str,
|
206 |
+
da_hypothesis_template: str,
|
207 |
+
da_candidate_labels: str,
|
208 |
+
sv_hypothesis_template: str,
|
209 |
+
sv_candidate_labels: str,
|
210 |
+
no_hypothesis_template: str,
|
211 |
+
no_candidate_labels: str,
|
212 |
+
) -> Dict[str, float]:
|
213 |
+
"""Classify text into categories.
|
214 |
+
|
215 |
+
Args:
|
216 |
+
doc (str):
|
217 |
+
Text to classify.
|
218 |
+
da_hypothesis_template (str):
|
219 |
+
Template for the hypothesis to be used for Danish classification.
|
220 |
+
da_candidate_labels (str):
|
221 |
+
Comma-separated list of candidate labels for Danish classification.
|
222 |
+
sv_hypothesis_template (str):
|
223 |
+
Template for the hypothesis to be used for Swedish classification.
|
224 |
+
sv_candidate_labels (str):
|
225 |
+
Comma-separated list of candidate labels for Swedish classification.
|
226 |
+
no_hypothesis_template (str):
|
227 |
+
Template for the hypothesis to be used for Norwegian classification.
|
228 |
+
no_candidate_labels (str):
|
229 |
+
Comma-separated list of candidate labels for Norwegian classification.
|
230 |
+
|
231 |
+
Returns:
|
232 |
+
dict of str to float:
|
233 |
+
The predicted label and the confidence score.
|
234 |
+
"""
|
235 |
+
# Detect the language of the text
|
236 |
+
language = detect_language(doc.replace('\n', ' ')).name
|
237 |
+
|
238 |
+
# Set the hypothesis template and candidate labels based on the detected language
|
239 |
+
if language == "sv":
|
240 |
+
hypothesis_template = sv_hypothesis_template
|
241 |
+
candidate_labels = re.split(r', *', sv_candidate_labels)
|
242 |
+
elif language == "no":
|
243 |
+
hypothesis_template = no_hypothesis_template
|
244 |
+
candidate_labels = re.split(r', *', no_candidate_labels)
|
245 |
+
else:
|
246 |
+
hypothesis_template = da_hypothesis_template
|
247 |
+
candidate_labels = re.split(r', *', da_candidate_labels)
|
248 |
+
|
249 |
+
# Run the classifier on the text
|
250 |
+
result = classifier(
|
251 |
+
doc,
|
252 |
+
candidate_labels=candidate_labels,
|
253 |
+
hypothesis_template=hypothesis_template,
|
254 |
+
)
|
255 |
+
|
256 |
+
print(result)
|
257 |
+
|
258 |
+
# Return the predicted label
|
259 |
+
return {lbl: score for lbl, score in zip(result["labels"], result["scores"])}
|
260 |
|
261 |
|
262 |
if __name__ == "__main__":
|
requirements.txt
CHANGED
@@ -35,7 +35,9 @@ MarkupSafe==2.1.1
|
|
35 |
matplotlib==3.6.2
|
36 |
mdit-py-plugins==0.3.1
|
37 |
mdurl==0.1.2
|
|
|
38 |
multidict==6.0.2
|
|
|
39 |
nptyping==1.4.4
|
40 |
numpy==1.23.5
|
41 |
orjson==3.8.2
|
@@ -62,10 +64,11 @@ six==1.16.0
|
|
62 |
sniffio==1.3.0
|
63 |
soupsieve==2.3.2.post1
|
64 |
starlette==0.22.0
|
|
|
65 |
tokenizers==0.13.2
|
66 |
-
torch==
|
67 |
tqdm==4.64.1
|
68 |
-
transformers==4.
|
69 |
typing_extensions==4.4.0
|
70 |
typish==1.9.3
|
71 |
uc-micro-py==1.0.1
|
|
|
35 |
matplotlib==3.6.2
|
36 |
mdit-py-plugins==0.3.1
|
37 |
mdurl==0.1.2
|
38 |
+
mpmath==1.3.0
|
39 |
multidict==6.0.2
|
40 |
+
networkx==3.1
|
41 |
nptyping==1.4.4
|
42 |
numpy==1.23.5
|
43 |
orjson==3.8.2
|
|
|
64 |
sniffio==1.3.0
|
65 |
soupsieve==2.3.2.post1
|
66 |
starlette==0.22.0
|
67 |
+
sympy==1.11.1
|
68 |
tokenizers==0.13.2
|
69 |
+
torch==2.0.0
|
70 |
tqdm==4.64.1
|
71 |
+
transformers==4.28.1
|
72 |
typing_extensions==4.4.0
|
73 |
typish==1.9.3
|
74 |
uc-micro-py==1.0.1
|