Spaces:
Sleeping
Sleeping
Update emotion_detection.py
Browse files- emotion_detection.py +34 -14
emotion_detection.py
CHANGED
@@ -34,22 +34,42 @@ class EmotionDetection:
|
|
34 |
return html
|
35 |
|
36 |
def classify(self, text):
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
44 |
|
45 |
-
|
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 |
-
|
53 |
|
54 |
def run(self, text):
|
55 |
"""
|
|
|
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 |
+
|
50 |
+
# Original labels from model
|
51 |
+
original_labels = list(self.model.config.id2label.values())
|
52 |
+
|
53 |
+
# Only keep the 4 specific emotions and map custom names
|
54 |
+
desired_labels = ['joy', 'anger', 'sadness', 'optimism']
|
55 |
+
custom_labels = {
|
56 |
+
'joy': 'Happiness π',
|
57 |
+
'anger': 'Anger π‘',
|
58 |
+
'sadness': 'Sadness π’',
|
59 |
+
'optimism': 'Hopeful β¨'
|
60 |
+
}
|
61 |
+
|
62 |
+
filtered_probs = []
|
63 |
+
filtered_labels = []
|
64 |
+
|
65 |
+
for label, prob in zip(original_labels, probs):
|
66 |
+
if label in desired_labels:
|
67 |
+
filtered_probs.append(prob)
|
68 |
+
filtered_labels.append(custom_labels[label])
|
69 |
|
70 |
+
preds = pd.Series(filtered_probs, index=filtered_labels, name='Predicted Probability')
|
|
|
|
|
|
|
|
|
|
|
71 |
|
72 |
+
return preds
|
73 |
|
74 |
def run(self, text):
|
75 |
"""
|