cirimus commited on
Commit
c3ab8b2
·
verified ·
1 Parent(s): 6cf1912

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +10 -25
README.md CHANGED
@@ -60,7 +60,7 @@ import torch
60
  classifier = pipeline(
61
  "text-classification",
62
  model="cirimus/modernbert-large-go-emotions",
63
- return_all_scores=True
64
  )
65
 
66
  text = "I am so happy and excited about this opportunity!"
@@ -72,15 +72,15 @@ top_5 = sorted_preds[:5]
72
 
73
  print("\nTop 5 emotions detected:")
74
  for pred in top_5:
75
- print(f"{pred['label']}: {pred['score']:.3f}")
76
 
77
  # Example output:
78
  # Top 5 emotions detected:
79
- # joy: 0.784
80
- # excitement: 0.735
81
- # admiration: 0.013
82
- # gratitude: 0.003
83
- # amusement: 0.003
84
  ```
85
 
86
  ### How the Model Was Created
@@ -110,17 +110,9 @@ The model was evaluated on the test split of the GoEmotions dataset, using a thr
110
 
111
  Using the default threshold of 0.5.
112
 
113
- *Macro Averages (test)*
114
- - Accuracy: `0.971`
115
- - Precision: `0.611`
116
- - Recall: `0.410`
117
- - F1: `0.472`
118
- - MCC: `0.475`
119
-
120
- *Per-Label Results (test)*
121
-
122
  | Label | Accuracy | Precision | Recall | F1 | MCC | Support | Threshold |
123
  |----------------|----------|-----------|--------|-------|-------|---------|-----------|
 
124
  | admiration | 0.946 | 0.739 | 0.653 | 0.693 | 0.666 | 504 | 0.5 |
125
  | amusement | 0.982 | 0.817 | 0.814 | 0.816 | 0.807 | 264 | 0.5 |
126
  | anger | 0.968 | 0.671 | 0.237 | 0.351 | 0.387 | 198 | 0.5 |
@@ -152,19 +144,12 @@ Using the default threshold of 0.5.
152
 
153
  **Optimal Results**:
154
 
155
- Using the best threshold for each label based on the training set (tuned on F1).
156
-
157
- *Macro Averages (test)*
158
- - Accuracy: `0.968`
159
- - Precision: `0.591`
160
- - Recall: `0.528`
161
- - F1: `0.550`
162
- - MCC: `0.536`
163
 
164
- *Per-Label Results (test)*
165
 
166
  | Label | Accuracy | Precision | Recall | F1 | MCC | Support | Threshold |
167
  |----------------|----------|-----------|--------|-------|-------|---------|-----------|
 
168
  | admiration | 0.947 | 0.722 | 0.702 | 0.712 | 0.683 | 504 | 0.40 |
169
  | amusement | 0.983 | 0.812 | 0.848 | 0.830 | 0.821 | 264 | 0.45 |
170
  | anger | 0.966 | 0.548 | 0.460 | 0.500 | 0.485 | 198 | 0.25 |
 
60
  classifier = pipeline(
61
  "text-classification",
62
  model="cirimus/modernbert-large-go-emotions",
63
+ top_k=5
64
  )
65
 
66
  text = "I am so happy and excited about this opportunity!"
 
72
 
73
  print("\nTop 5 emotions detected:")
74
  for pred in top_5:
75
+ print(f"\t{pred['label']:10s} : {pred['score']:.3f}")
76
 
77
  # Example output:
78
  # Top 5 emotions detected:
79
+ # joy : 0.784
80
+ # excitement : 0.735
81
+ # admiration : 0.013
82
+ # gratitude : 0.003
83
+ # amusement : 0.003
84
  ```
85
 
86
  ### How the Model Was Created
 
110
 
111
  Using the default threshold of 0.5.
112
 
 
 
 
 
 
 
 
 
 
113
  | Label | Accuracy | Precision | Recall | F1 | MCC | Support | Threshold |
114
  |----------------|----------|-----------|--------|-------|-------|---------|-----------|
115
+ | **macro avg** | 0.971 | 0.611 | 0.410 | 0.472 | 0.475 | 5427 | 0.5 |
116
  | admiration | 0.946 | 0.739 | 0.653 | 0.693 | 0.666 | 504 | 0.5 |
117
  | amusement | 0.982 | 0.817 | 0.814 | 0.816 | 0.807 | 264 | 0.5 |
118
  | anger | 0.968 | 0.671 | 0.237 | 0.351 | 0.387 | 198 | 0.5 |
 
144
 
145
  **Optimal Results**:
146
 
147
+ Using the best threshold for each label based on the training set (tuned on F1), tested on the test set:
 
 
 
 
 
 
 
148
 
 
149
 
150
  | Label | Accuracy | Precision | Recall | F1 | MCC | Support | Threshold |
151
  |----------------|----------|-----------|--------|-------|-------|---------|-----------|
152
+ | **macro avg** | 0.968 | 0.591 | 0.528 | 0.550 | 0.536 | 5427 | various |
153
  | admiration | 0.947 | 0.722 | 0.702 | 0.712 | 0.683 | 504 | 0.40 |
154
  | amusement | 0.983 | 0.812 | 0.848 | 0.830 | 0.821 | 264 | 0.45 |
155
  | anger | 0.966 | 0.548 | 0.460 | 0.500 | 0.485 | 198 | 0.25 |