added zero shot
Browse files
app.py
CHANGED
@@ -3,7 +3,8 @@ from transformers import pipeline
|
|
3 |
|
4 |
get_completion = pipeline("summarization",model="sshleifer/distilbart-cnn-12-6")
|
5 |
get_ner = pipeline("ner", model="dslim/bert-base-NER")
|
6 |
-
|
|
|
7 |
def summarize_text(input):
|
8 |
output = get_completion(input)
|
9 |
return output[0]['summary_text']
|
@@ -27,6 +28,17 @@ def named_entity_recognition(input):
|
|
27 |
merged_output = merge_tokens(output)
|
28 |
return {"text": input, "entities": output}
|
29 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
30 |
|
31 |
interface_summarise = gr.Interface(fn=summarize_text,
|
32 |
inputs=[gr.Textbox(label="Text to summarise", lines=5)],
|
@@ -45,11 +57,27 @@ interface_ner = gr.Interface(fn=named_entity_recognition,
|
|
45 |
"My name is Bose and I am a physicist living in Delhi"
|
46 |
])
|
47 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
48 |
demo = gr.TabbedInterface([
|
49 |
interface_summarise,
|
50 |
-
interface_ner
|
|
|
51 |
["Text Summary ",
|
52 |
-
"Named Entity Recognition"
|
|
|
53 |
])
|
54 |
|
55 |
if __name__ == "__main__":
|
|
|
3 |
|
4 |
get_completion = pipeline("summarization",model="sshleifer/distilbart-cnn-12-6")
|
5 |
get_ner = pipeline("ner", model="dslim/bert-base-NER")
|
6 |
+
get_zero = pipeline("zero-shot-classification", model="MoritzLaurer/DeBERTa-v3-base-mnli-fever-anli")
|
7 |
+
|
8 |
def summarize_text(input):
|
9 |
output = get_completion(input)
|
10 |
return output[0]['summary_text']
|
|
|
28 |
merged_output = merge_tokens(output)
|
29 |
return {"text": input, "entities": output}
|
30 |
|
31 |
+
def zero_shot_pred(text,check_labels):
|
32 |
+
output = get_zero(text,check_labels)
|
33 |
+
return output
|
34 |
+
|
35 |
+
def label_score_dict(text,check_labels):
|
36 |
+
zero_shot_out = zero_shot_pred(text,check_labels)
|
37 |
+
out = {}
|
38 |
+
for i,j in zip(zero_shot_out['labels'],zero_shot_out['scores']):
|
39 |
+
out.update({i:j})
|
40 |
+
print(out)
|
41 |
+
return out
|
42 |
|
43 |
interface_summarise = gr.Interface(fn=summarize_text,
|
44 |
inputs=[gr.Textbox(label="Text to summarise", lines=5)],
|
|
|
57 |
"My name is Bose and I am a physicist living in Delhi"
|
58 |
])
|
59 |
|
60 |
+
interface_zero_shot=gr.Interface(fn=label_score_dict,
|
61 |
+
inputs=[
|
62 |
+
gr.Textbox(label="Text to classify", lines=2),
|
63 |
+
gr.Textbox(label="Check for labels")
|
64 |
+
],
|
65 |
+
outputs=gr.Label(num_top_classes=4),
|
66 |
+
title="Zero-Shot Preds using DeBERTa-v3-base-mnli",
|
67 |
+
description="Classify sentence on self defined target vars",
|
68 |
+
examples=[
|
69 |
+
["Last week I upgraded my iOS version and ever since then my phone has been overheating whenever I use your app.",
|
70 |
+
"mobile, website, billing, account access"],
|
71 |
+
# "My name is Bose and I am a physicist living in Delhi"
|
72 |
+
])
|
73 |
+
|
74 |
demo = gr.TabbedInterface([
|
75 |
interface_summarise,
|
76 |
+
interface_ner,
|
77 |
+
interface_zero_shot],
|
78 |
["Text Summary ",
|
79 |
+
"Named Entity Recognition",
|
80 |
+
"Zero Shot Classifications"
|
81 |
])
|
82 |
|
83 |
if __name__ == "__main__":
|