Alasil commited on
Commit
ebdf63d
·
verified ·
1 Parent(s): 2206912

[Run 20260606_143218] Log metrics

Browse files
metrics/run_20260606_143218_metrics.json ADDED
@@ -0,0 +1,250 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_id": "20260606_143218",
3
+ "hyperparameters": {
4
+ "model_architecture": "UBC-NLP/MARBERT",
5
+ "task": "sequence-classification",
6
+ "dataset": "Nexus-Analytics/multilingual-hate-detection-dataset_v2",
7
+ "learning_rate": 3e-05,
8
+ "batch_size": 16,
9
+ "num_epochs": 6,
10
+ "max_seq_length": 128,
11
+ "early_stopping_patience": 2,
12
+ "optimizer": "AdamW",
13
+ "seed": 42
14
+ },
15
+ "per_epoch_results": [
16
+ {
17
+ "epoch": 1,
18
+ "train_loss": 0.1870656698826944,
19
+ "val_loss": 0.1175760279484915,
20
+ "val_accuracy": 0.9683435582822086,
21
+ "val_f1_weighted": 0.9683265536196238,
22
+ "val_f1_macro": 0.968244146408636,
23
+ "val_precision_macro": 0.9688368495717035,
24
+ "val_recall_macro": 0.967830471437855,
25
+ "hate_recall": 0.9567678847143593,
26
+ "hate_f1": 0.966467377177021,
27
+ "classification_report": {
28
+ "Hate": {
29
+ "precision": 0.9763655462184874,
30
+ "recall": 0.9567678847143593,
31
+ "f1-score": 0.966467377177021,
32
+ "support": 1943.0
33
+ },
34
+ "Normal": {
35
+ "precision": 0.9613081529249194,
36
+ "recall": 0.9788930581613509,
37
+ "f1-score": 0.970020915640251,
38
+ "support": 2132.0
39
+ },
40
+ "accuracy": 0.9683435582822086,
41
+ "macro avg": {
42
+ "precision": 0.9688368495717035,
43
+ "recall": 0.967830471437855,
44
+ "f1-score": 0.968244146408636,
45
+ "support": 4075.0
46
+ },
47
+ "weighted avg": {
48
+ "precision": 0.9684876658499262,
49
+ "recall": 0.9683435582822086,
50
+ "f1-score": 0.9683265536196238,
51
+ "support": 4075.0
52
+ }
53
+ }
54
+ },
55
+ {
56
+ "epoch": 2,
57
+ "train_loss": 0.09221274530345397,
58
+ "val_loss": 0.08990090668872547,
59
+ "val_accuracy": 0.9771779141104294,
60
+ "val_f1_weighted": 0.9771685135843289,
61
+ "val_f1_macro": 0.9771112006993925,
62
+ "val_precision_macro": 0.9775613590685492,
63
+ "val_recall_macro": 0.9767751219319074,
64
+ "hate_recall": 0.9680905815748841,
65
+ "hate_f1": 0.9758754863813229,
66
+ "classification_report": {
67
+ "Hate": {
68
+ "precision": 0.983786610878661,
69
+ "recall": 0.9680905815748841,
70
+ "f1-score": 0.9758754863813229,
71
+ "support": 1943.0
72
+ },
73
+ "Normal": {
74
+ "precision": 0.9713361072584373,
75
+ "recall": 0.9854596622889306,
76
+ "f1-score": 0.9783469150174622,
77
+ "support": 2132.0
78
+ },
79
+ "accuracy": 0.9771779141104294,
80
+ "macro avg": {
81
+ "precision": 0.9775613590685492,
82
+ "recall": 0.9767751219319074,
83
+ "f1-score": 0.9771112006993925,
84
+ "support": 4075.0
85
+ },
86
+ "weighted avg": {
87
+ "precision": 0.9772726295980925,
88
+ "recall": 0.9771779141104294,
89
+ "f1-score": 0.9771685135843289,
90
+ "support": 4075.0
91
+ }
92
+ }
93
+ },
94
+ {
95
+ "epoch": 3,
96
+ "train_loss": 0.06357873568356592,
97
+ "val_loss": 0.13582501676138753,
98
+ "val_accuracy": 0.9776687116564418,
99
+ "val_f1_weighted": 0.9776694636229548,
100
+ "val_f1_macro": 0.9776220897326291,
101
+ "val_precision_macro": 0.9775904566556779,
102
+ "val_recall_macro": 0.9776547890681805,
103
+ "hate_recall": 0.9773546062789501,
104
+ "hate_f1": 0.9766006685523271,
105
+ "classification_report": {
106
+ "Hate": {
107
+ "precision": 0.9758478931140802,
108
+ "recall": 0.9773546062789501,
109
+ "f1-score": 0.9766006685523271,
110
+ "support": 1943.0
111
+ },
112
+ "Normal": {
113
+ "precision": 0.9793330201972757,
114
+ "recall": 0.9779549718574109,
115
+ "f1-score": 0.9786435109129312,
116
+ "support": 2132.0
117
+ },
118
+ "accuracy": 0.9776687116564418,
119
+ "macro avg": {
120
+ "precision": 0.9775904566556779,
121
+ "recall": 0.9776547890681805,
122
+ "f1-score": 0.9776220897326291,
123
+ "support": 4075.0
124
+ },
125
+ "weighted avg": {
126
+ "precision": 0.9776712773941717,
127
+ "recall": 0.9776687116564418,
128
+ "f1-score": 0.9776694636229548,
129
+ "support": 4075.0
130
+ }
131
+ }
132
+ },
133
+ {
134
+ "epoch": 4,
135
+ "train_loss": 0.04601577372881774,
136
+ "val_loss": 0.1280251998608269,
137
+ "val_accuracy": 0.9774233128834355,
138
+ "val_f1_weighted": 0.977429451455362,
139
+ "val_f1_macro": 0.9773880160948591,
140
+ "val_precision_macro": 0.9771747371101505,
141
+ "val_recall_macro": 0.9776940168150643,
142
+ "hate_recall": 0.9835306227483274,
143
+ "hate_f1": 0.9764946346448646,
144
+ "classification_report": {
145
+ "Hate": {
146
+ "precision": 0.969558599695586,
147
+ "recall": 0.9835306227483274,
148
+ "f1-score": 0.9764946346448646,
149
+ "support": 1943.0
150
+ },
151
+ "Normal": {
152
+ "precision": 0.9847908745247148,
153
+ "recall": 0.9718574108818011,
154
+ "f1-score": 0.9782813975448537,
155
+ "support": 2132.0
156
+ },
157
+ "accuracy": 0.9774233128834355,
158
+ "macro avg": {
159
+ "precision": 0.9771747371101505,
160
+ "recall": 0.9776940168150643,
161
+ "f1-score": 0.9773880160948591,
162
+ "support": 4075.0
163
+ },
164
+ "weighted avg": {
165
+ "precision": 0.9775279763669241,
166
+ "recall": 0.9774233128834355,
167
+ "f1-score": 0.977429451455362,
168
+ "support": 4075.0
169
+ }
170
+ }
171
+ },
172
+ {
173
+ "epoch": 5,
174
+ "train_loss": 0.03839033815012687,
175
+ "val_loss": 0.16608795987438713,
176
+ "val_accuracy": 0.9764417177914111,
177
+ "val_f1_weighted": 0.9764294420535918,
178
+ "val_f1_macro": 0.9763683864102275,
179
+ "val_precision_macro": 0.9769506634251364,
180
+ "val_recall_macro": 0.9759574949860905,
181
+ "hate_recall": 0.9655172413793104,
182
+ "hate_f1": 0.975051975051975,
183
+ "classification_report": {
184
+ "Hate": {
185
+ "precision": 0.9847769028871391,
186
+ "recall": 0.9655172413793104,
187
+ "f1-score": 0.975051975051975,
188
+ "support": 1943.0
189
+ },
190
+ "Normal": {
191
+ "precision": 0.9691244239631336,
192
+ "recall": 0.9863977485928705,
193
+ "f1-score": 0.9776847977684798,
194
+ "support": 2132.0
195
+ },
196
+ "accuracy": 0.9764417177914111,
197
+ "macro avg": {
198
+ "precision": 0.9769506634251364,
199
+ "recall": 0.9759574949860905,
200
+ "f1-score": 0.9763683864102275,
201
+ "support": 4075.0
202
+ },
203
+ "weighted avg": {
204
+ "precision": 0.9765876795580645,
205
+ "recall": 0.9764417177914111,
206
+ "f1-score": 0.9764294420535918,
207
+ "support": 4075.0
208
+ }
209
+ }
210
+ }
211
+ ],
212
+ "best_val_macro_f1": 0.9776220897326291,
213
+ "best_val_weighted_f1": 0.9776694636229548,
214
+ "best_epoch": 3,
215
+ "test_metrics": {
216
+ "test_accuracy": 0.9771779141104294,
217
+ "test_f1_macro": 0.9771362280215035,
218
+ "test_f1_weighted": 0.9771815077387851,
219
+ "test_precision_macro": 0.9769992709439563,
220
+ "test_recall_macro": 0.9772998081340725,
221
+ "classification_report": {
222
+ "Hate": {
223
+ "precision": 0.9724208375893769,
224
+ "recall": 0.979927946474524,
225
+ "f1-score": 0.9761599589848756,
226
+ "support": 1943.0
227
+ },
228
+ "Normal": {
229
+ "precision": 0.9815777042985356,
230
+ "recall": 0.974671669793621,
231
+ "f1-score": 0.9781124970581313,
232
+ "support": 2132.0
233
+ },
234
+ "accuracy": 0.9771779141104294,
235
+ "macro avg": {
236
+ "precision": 0.9769992709439563,
237
+ "recall": 0.9772998081340725,
238
+ "f1-score": 0.9771362280215035,
239
+ "support": 4075.0
240
+ },
241
+ "weighted avg": {
242
+ "precision": 0.9772116203682545,
243
+ "recall": 0.9771779141104294,
244
+ "f1-score": 0.9771815077387851,
245
+ "support": 4075.0
246
+ }
247
+ }
248
+ },
249
+ "notes": "MARBERT sweep. LR=3e-05, Batch=16. Early stopping based on validation macro F1."
250
+ }