Spaces:
Running
Running
Update emotion_detection.py
Browse files- emotion_detection.py +24 -50
emotion_detection.py
CHANGED
@@ -1,16 +1,16 @@
|
|
1 |
from transformers import AutoTokenizer, AutoModelForSequenceClassification
|
2 |
from transformers_interpret import SequenceClassificationExplainer
|
3 |
import torch
|
4 |
-
import
|
5 |
|
6 |
|
7 |
class EmotionDetection:
|
8 |
"""
|
9 |
Emotion Detection on text data.
|
10 |
Attributes:
|
11 |
-
tokenizer: Hugging Face Tokenizer
|
12 |
-
model: Hugging Face
|
13 |
-
explainer:
|
14 |
"""
|
15 |
|
16 |
def __init__(self):
|
@@ -19,75 +19,49 @@ class EmotionDetection:
|
|
19 |
self.model = AutoModelForSequenceClassification.from_pretrained(hub_location)
|
20 |
self.explainer = SequenceClassificationExplainer(self.model, self.tokenizer)
|
21 |
|
22 |
-
# Emoji map for friendly display
|
23 |
-
self.emoji_map = {
|
24 |
-
"joy": "π",
|
25 |
-
"anger": "π ",
|
26 |
-
"optimism": "π",
|
27 |
-
"sadness": "π’"
|
28 |
-
}
|
29 |
-
|
30 |
-
# Simple explanation map
|
31 |
-
self.explanation_map = {
|
32 |
-
"joy": "The person is happy or excited.",
|
33 |
-
"anger": "The person is upset or angry.",
|
34 |
-
"optimism": "The person is feeling hopeful or positive.",
|
35 |
-
"sadness": "The person is feeling low or unhappy."
|
36 |
-
}
|
37 |
-
|
38 |
def justify(self, text):
|
39 |
"""
|
40 |
-
|
41 |
Parameters:
|
42 |
-
text (str):
|
43 |
Returns:
|
44 |
-
html (
|
45 |
"""
|
46 |
-
word_attributions = self.explainer(text)
|
47 |
-
html_path = "justification_output.html"
|
48 |
-
self.explainer.visualize(html_path)
|
49 |
|
50 |
-
|
51 |
-
|
52 |
-
html = f.read()
|
53 |
|
54 |
-
# Clean up file
|
55 |
-
os.remove(html_path)
|
56 |
return html
|
57 |
|
58 |
def classify(self, text):
|
59 |
"""
|
60 |
-
|
61 |
Parameters:
|
62 |
-
text (str):
|
63 |
Returns:
|
64 |
-
|
65 |
"""
|
66 |
-
|
|
|
67 |
outputs = self.model(**tokens)
|
68 |
probs = torch.nn.functional.softmax(outputs[0], dim=-1)
|
69 |
probs = probs.mean(dim=0).detach().numpy()
|
70 |
-
|
71 |
labels = list(self.model.config.id2label.values())
|
72 |
-
|
73 |
-
emotion = labels[max_index]
|
74 |
-
confidence = probs[max_index]
|
75 |
-
|
76 |
-
emoji = self.emoji_map.get(emotion, "")
|
77 |
-
explanation = self.explanation_map.get(emotion, "")
|
78 |
|
79 |
-
|
80 |
-
return result
|
81 |
|
82 |
def run(self, text):
|
83 |
"""
|
84 |
-
|
85 |
Parameters:
|
86 |
-
text (str):
|
87 |
Returns:
|
88 |
-
|
89 |
-
html (
|
90 |
"""
|
91 |
-
|
|
|
92 |
html = self.justify(text)
|
93 |
-
|
|
|
|
1 |
from transformers import AutoTokenizer, AutoModelForSequenceClassification
|
2 |
from transformers_interpret import SequenceClassificationExplainer
|
3 |
import torch
|
4 |
+
import pandas as pd
|
5 |
|
6 |
|
7 |
class EmotionDetection:
|
8 |
"""
|
9 |
Emotion Detection on text data.
|
10 |
Attributes:
|
11 |
+
tokenizer: An instance of Hugging Face Tokenizer
|
12 |
+
model: An instance of Hugging Face Model
|
13 |
+
explainer: An instance of SequenceClassificationExplainer from Transformers interpret
|
14 |
"""
|
15 |
|
16 |
def __init__(self):
|
|
|
19 |
self.model = AutoModelForSequenceClassification.from_pretrained(hub_location)
|
20 |
self.explainer = SequenceClassificationExplainer(self.model, self.tokenizer)
|
21 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
22 |
def justify(self, text):
|
23 |
"""
|
24 |
+
Get html annotation for displaying emotion justification over text.
|
25 |
Parameters:
|
26 |
+
text (str): The user input string to emotion justification
|
27 |
Returns:
|
28 |
+
html (hmtl): html object for plotting emotion prediction justification
|
29 |
"""
|
|
|
|
|
|
|
30 |
|
31 |
+
word_attributions = self.explainer(text)
|
32 |
+
html = self.explainer.visualize("example.html")
|
|
|
33 |
|
|
|
|
|
34 |
return html
|
35 |
|
36 |
def classify(self, text):
|
37 |
"""
|
38 |
+
Recognize Emotion in text.
|
39 |
Parameters:
|
40 |
+
text (str): The user input string to perform emotion classification on
|
41 |
Returns:
|
42 |
+
predictions (str): The predicted probabilities for emotion classes
|
43 |
"""
|
44 |
+
|
45 |
+
tokens = self.tokenizer.encode_plus(text, add_special_tokens=False, return_tensors='pt')
|
46 |
outputs = self.model(**tokens)
|
47 |
probs = torch.nn.functional.softmax(outputs[0], dim=-1)
|
48 |
probs = probs.mean(dim=0).detach().numpy()
|
|
|
49 |
labels = list(self.model.config.id2label.values())
|
50 |
+
preds = pd.Series(probs, index=labels, name='Predicted Probability')
|
|
|
|
|
|
|
|
|
|
|
51 |
|
52 |
+
return preds
|
|
|
53 |
|
54 |
def run(self, text):
|
55 |
"""
|
56 |
+
Classify and Justify Emotion in text.
|
57 |
Parameters:
|
58 |
+
text (str): The user input string to perform emotion classification on
|
59 |
Returns:
|
60 |
+
predictions (str): The predicted probabilities for emotion classes
|
61 |
+
html (hmtl): html object for plotting emotion prediction justification
|
62 |
"""
|
63 |
+
|
64 |
+
preds = self.classify(text)
|
65 |
html = self.justify(text)
|
66 |
+
|
67 |
+
return preds, html
|