lombardata commited on
Commit
8e9825c
1 Parent(s): 981a770

Evaluation on the test set completed on 2024_09_08.

Browse files
README.md ADDED
@@ -0,0 +1,156 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ base_model: facebook/dinov2-large
4
+ tags:
5
+ - generated_from_trainer
6
+ metrics:
7
+ - accuracy
8
+ model-index:
9
+ - name: DinoVdeau-large-2024_09_05-batch-size32_epochs150_freeze
10
+ results: []
11
+ ---
12
+
13
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
14
+ should probably proofread and complete it, then remove this comment. -->
15
+
16
+ # DinoVdeau-large-2024_09_05-batch-size32_epochs150_freeze
17
+
18
+ This model is a fine-tuned version of [facebook/dinov2-large](https://huggingface.co/facebook/dinov2-large) on the None dataset.
19
+ It achieves the following results on the evaluation set:
20
+ - Loss: 0.1209
21
+ - F1 Micro: 0.8228
22
+ - F1 Macro: 0.7175
23
+ - Roc Auc: 0.8813
24
+ - Accuracy: 0.3111
25
+ - Learning Rate: 0.0000
26
+
27
+ ## Model description
28
+
29
+ More information needed
30
+
31
+ ## Intended uses & limitations
32
+
33
+ More information needed
34
+
35
+ ## Training and evaluation data
36
+
37
+ More information needed
38
+
39
+ ## Training procedure
40
+
41
+ ### Training hyperparameters
42
+
43
+ The following hyperparameters were used during training:
44
+ - learning_rate: 0.001
45
+ - train_batch_size: 32
46
+ - eval_batch_size: 32
47
+ - seed: 42
48
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
49
+ - lr_scheduler_type: linear
50
+ - num_epochs: 150
51
+ - mixed_precision_training: Native AMP
52
+
53
+ ### Training results
54
+
55
+ | Training Loss | Epoch | Step | Validation Loss | F1 Micro | F1 Macro | Roc Auc | Accuracy | Rate |
56
+ |:-------------:|:-----:|:-----:|:---------------:|:--------:|:--------:|:-------:|:--------:|:------:|
57
+ | No log | 1.0 | 273 | 0.1690 | 0.7517 | 0.5430 | 0.8384 | 0.2231 | 0.001 |
58
+ | 0.2719 | 2.0 | 546 | 0.1538 | 0.7657 | 0.5721 | 0.8396 | 0.2401 | 0.001 |
59
+ | 0.2719 | 3.0 | 819 | 0.1483 | 0.7773 | 0.6138 | 0.8516 | 0.2346 | 0.001 |
60
+ | 0.1694 | 4.0 | 1092 | 0.1480 | 0.7723 | 0.6225 | 0.8407 | 0.2495 | 0.001 |
61
+ | 0.1694 | 5.0 | 1365 | 0.1458 | 0.7797 | 0.6302 | 0.8470 | 0.2495 | 0.001 |
62
+ | 0.1625 | 6.0 | 1638 | 0.1450 | 0.7798 | 0.6093 | 0.8477 | 0.2481 | 0.001 |
63
+ | 0.1625 | 7.0 | 1911 | 0.1475 | 0.7767 | 0.6248 | 0.8454 | 0.2526 | 0.001 |
64
+ | 0.1592 | 8.0 | 2184 | 0.1457 | 0.7804 | 0.6249 | 0.8521 | 0.2574 | 0.001 |
65
+ | 0.1592 | 9.0 | 2457 | 0.1417 | 0.7869 | 0.6526 | 0.8561 | 0.2574 | 0.001 |
66
+ | 0.157 | 10.0 | 2730 | 0.1436 | 0.7757 | 0.6290 | 0.8403 | 0.2547 | 0.001 |
67
+ | 0.1563 | 11.0 | 3003 | 0.1428 | 0.7887 | 0.6448 | 0.8569 | 0.2640 | 0.001 |
68
+ | 0.1563 | 12.0 | 3276 | 0.1439 | 0.7905 | 0.6493 | 0.8638 | 0.2581 | 0.001 |
69
+ | 0.1558 | 13.0 | 3549 | 0.1391 | 0.7907 | 0.6562 | 0.8551 | 0.2713 | 0.001 |
70
+ | 0.1558 | 14.0 | 3822 | 0.1409 | 0.7838 | 0.6338 | 0.8485 | 0.2644 | 0.001 |
71
+ | 0.1543 | 15.0 | 4095 | 0.1396 | 0.7907 | 0.6463 | 0.8603 | 0.2578 | 0.001 |
72
+ | 0.1543 | 16.0 | 4368 | 0.1390 | 0.7913 | 0.6594 | 0.8564 | 0.2654 | 0.001 |
73
+ | 0.1535 | 17.0 | 4641 | 0.1418 | 0.7940 | 0.6586 | 0.8665 | 0.2564 | 0.001 |
74
+ | 0.1535 | 18.0 | 4914 | 0.1416 | 0.7957 | 0.6560 | 0.8646 | 0.2658 | 0.001 |
75
+ | 0.1549 | 19.0 | 5187 | 0.1403 | 0.7886 | 0.6524 | 0.8536 | 0.2630 | 0.001 |
76
+ | 0.1549 | 20.0 | 5460 | 0.1476 | 0.7911 | 0.6558 | 0.8568 | 0.2613 | 0.001 |
77
+ | 0.154 | 21.0 | 5733 | 0.1429 | 0.7880 | 0.6397 | 0.8568 | 0.2658 | 0.001 |
78
+ | 0.1529 | 22.0 | 6006 | 0.1414 | 0.7937 | 0.6508 | 0.8654 | 0.2613 | 0.001 |
79
+ | 0.1529 | 23.0 | 6279 | 0.1415 | 0.7976 | 0.6618 | 0.8613 | 0.2685 | 0.0001 |
80
+ | 0.1449 | 24.0 | 6552 | 0.1323 | 0.8045 | 0.6751 | 0.8665 | 0.2789 | 0.0001 |
81
+ | 0.1449 | 25.0 | 6825 | 0.1310 | 0.8044 | 0.6724 | 0.8688 | 0.2793 | 0.0001 |
82
+ | 0.1416 | 26.0 | 7098 | 0.1327 | 0.8036 | 0.6689 | 0.8646 | 0.2821 | 0.0001 |
83
+ | 0.1416 | 27.0 | 7371 | 0.1317 | 0.8069 | 0.6797 | 0.8715 | 0.2817 | 0.0001 |
84
+ | 0.1391 | 28.0 | 7644 | 0.1288 | 0.8072 | 0.6818 | 0.8698 | 0.2775 | 0.0001 |
85
+ | 0.1391 | 29.0 | 7917 | 0.1294 | 0.8038 | 0.6808 | 0.8629 | 0.2845 | 0.0001 |
86
+ | 0.138 | 30.0 | 8190 | 0.1294 | 0.8077 | 0.6826 | 0.8702 | 0.2859 | 0.0001 |
87
+ | 0.138 | 31.0 | 8463 | 0.1274 | 0.8074 | 0.6779 | 0.8666 | 0.2879 | 0.0001 |
88
+ | 0.1364 | 32.0 | 8736 | 0.1278 | 0.8104 | 0.6869 | 0.8728 | 0.2883 | 0.0001 |
89
+ | 0.1359 | 33.0 | 9009 | 0.1277 | 0.8077 | 0.6811 | 0.8692 | 0.2869 | 0.0001 |
90
+ | 0.1359 | 34.0 | 9282 | 0.1266 | 0.8109 | 0.6874 | 0.8714 | 0.2883 | 0.0001 |
91
+ | 0.1341 | 35.0 | 9555 | 0.1262 | 0.8104 | 0.6885 | 0.8716 | 0.2904 | 0.0001 |
92
+ | 0.1341 | 36.0 | 9828 | 0.1269 | 0.8070 | 0.6876 | 0.8657 | 0.2827 | 0.0001 |
93
+ | 0.1339 | 37.0 | 10101 | 0.1266 | 0.8082 | 0.6834 | 0.8678 | 0.2866 | 0.0001 |
94
+ | 0.1339 | 38.0 | 10374 | 0.1255 | 0.8106 | 0.6936 | 0.8707 | 0.2956 | 0.0001 |
95
+ | 0.1307 | 39.0 | 10647 | 0.1249 | 0.8142 | 0.6986 | 0.8768 | 0.2928 | 0.0001 |
96
+ | 0.1307 | 40.0 | 10920 | 0.1258 | 0.8138 | 0.6990 | 0.8773 | 0.2935 | 0.0001 |
97
+ | 0.1317 | 41.0 | 11193 | 0.1253 | 0.8101 | 0.6924 | 0.8688 | 0.2924 | 0.0001 |
98
+ | 0.1317 | 42.0 | 11466 | 0.1244 | 0.8138 | 0.6970 | 0.8738 | 0.3004 | 0.0001 |
99
+ | 0.1308 | 43.0 | 11739 | 0.1245 | 0.8131 | 0.6956 | 0.8734 | 0.2949 | 0.0001 |
100
+ | 0.1307 | 44.0 | 12012 | 0.1250 | 0.8130 | 0.6915 | 0.8743 | 0.2966 | 0.0001 |
101
+ | 0.1307 | 45.0 | 12285 | 0.1240 | 0.8137 | 0.7051 | 0.8740 | 0.2963 | 0.0001 |
102
+ | 0.1295 | 46.0 | 12558 | 0.1241 | 0.8131 | 0.6988 | 0.8733 | 0.2976 | 0.0001 |
103
+ | 0.1295 | 47.0 | 12831 | 0.1243 | 0.8119 | 0.6958 | 0.8716 | 0.2956 | 0.0001 |
104
+ | 0.1293 | 48.0 | 13104 | 0.1239 | 0.8135 | 0.6990 | 0.8744 | 0.2956 | 0.0001 |
105
+ | 0.1293 | 49.0 | 13377 | 0.1243 | 0.8153 | 0.7007 | 0.8775 | 0.2997 | 0.0001 |
106
+ | 0.1274 | 50.0 | 13650 | 0.1241 | 0.8152 | 0.7000 | 0.8769 | 0.2980 | 0.0001 |
107
+ | 0.1274 | 51.0 | 13923 | 0.1248 | 0.8153 | 0.7056 | 0.8803 | 0.3011 | 0.0001 |
108
+ | 0.1271 | 52.0 | 14196 | 0.1243 | 0.8157 | 0.7036 | 0.8751 | 0.3049 | 0.0001 |
109
+ | 0.1271 | 53.0 | 14469 | 0.1241 | 0.8153 | 0.7032 | 0.8778 | 0.3021 | 0.0001 |
110
+ | 0.1275 | 54.0 | 14742 | 0.1234 | 0.8152 | 0.7068 | 0.8753 | 0.3021 | 0.0001 |
111
+ | 0.1256 | 55.0 | 15015 | 0.1231 | 0.8166 | 0.7076 | 0.8776 | 0.3018 | 0.0001 |
112
+ | 0.1256 | 56.0 | 15288 | 0.1228 | 0.8190 | 0.7088 | 0.8822 | 0.3067 | 0.0001 |
113
+ | 0.1258 | 57.0 | 15561 | 0.1226 | 0.8160 | 0.7080 | 0.8767 | 0.3070 | 0.0001 |
114
+ | 0.1258 | 58.0 | 15834 | 0.1233 | 0.8170 | 0.7073 | 0.8773 | 0.3021 | 0.0001 |
115
+ | 0.1258 | 59.0 | 16107 | 0.1227 | 0.8172 | 0.7135 | 0.8781 | 0.3021 | 0.0001 |
116
+ | 0.1258 | 60.0 | 16380 | 0.1233 | 0.8143 | 0.7040 | 0.8729 | 0.3021 | 0.0001 |
117
+ | 0.1252 | 61.0 | 16653 | 0.1234 | 0.8168 | 0.7121 | 0.8784 | 0.3042 | 0.0001 |
118
+ | 0.1252 | 62.0 | 16926 | 0.1223 | 0.8169 | 0.7125 | 0.8764 | 0.3049 | 0.0001 |
119
+ | 0.1238 | 63.0 | 17199 | 0.1231 | 0.8151 | 0.7090 | 0.8752 | 0.3035 | 0.0001 |
120
+ | 0.1238 | 64.0 | 17472 | 0.1228 | 0.8183 | 0.7114 | 0.8785 | 0.3067 | 0.0001 |
121
+ | 0.1247 | 65.0 | 17745 | 0.1231 | 0.8185 | 0.7156 | 0.8802 | 0.3035 | 0.0001 |
122
+ | 0.123 | 66.0 | 18018 | 0.1225 | 0.8193 | 0.7084 | 0.8809 | 0.3021 | 0.0001 |
123
+ | 0.123 | 67.0 | 18291 | 0.1222 | 0.8186 | 0.7136 | 0.8814 | 0.3032 | 0.0001 |
124
+ | 0.1224 | 68.0 | 18564 | 0.1220 | 0.8201 | 0.7169 | 0.8818 | 0.3091 | 0.0001 |
125
+ | 0.1224 | 69.0 | 18837 | 0.1228 | 0.8171 | 0.7165 | 0.8768 | 0.3018 | 0.0001 |
126
+ | 0.1228 | 70.0 | 19110 | 0.1227 | 0.8177 | 0.7131 | 0.8765 | 0.3042 | 0.0001 |
127
+ | 0.1228 | 71.0 | 19383 | 0.1232 | 0.8155 | 0.7123 | 0.8733 | 0.2980 | 0.0001 |
128
+ | 0.1224 | 72.0 | 19656 | 0.1222 | 0.8177 | 0.7181 | 0.8780 | 0.3056 | 0.0001 |
129
+ | 0.1224 | 73.0 | 19929 | 0.1221 | 0.8162 | 0.7047 | 0.8760 | 0.3077 | 0.0001 |
130
+ | 0.122 | 74.0 | 20202 | 0.1230 | 0.8148 | 0.7070 | 0.8732 | 0.2973 | 0.0001 |
131
+ | 0.122 | 75.0 | 20475 | 0.1214 | 0.8176 | 0.7124 | 0.8768 | 0.3049 | 1e-05 |
132
+ | 0.1201 | 76.0 | 20748 | 0.1209 | 0.8213 | 0.7265 | 0.8828 | 0.3067 | 1e-05 |
133
+ | 0.1192 | 77.0 | 21021 | 0.1216 | 0.8221 | 0.7249 | 0.8860 | 0.3073 | 1e-05 |
134
+ | 0.1192 | 78.0 | 21294 | 0.1211 | 0.8210 | 0.7233 | 0.8828 | 0.3056 | 1e-05 |
135
+ | 0.1178 | 79.0 | 21567 | 0.1211 | 0.8181 | 0.7158 | 0.8769 | 0.3056 | 1e-05 |
136
+ | 0.1178 | 80.0 | 21840 | 0.1210 | 0.8200 | 0.7197 | 0.8824 | 0.3091 | 1e-05 |
137
+ | 0.1178 | 81.0 | 22113 | 0.1205 | 0.8190 | 0.7194 | 0.8784 | 0.3105 | 1e-05 |
138
+ | 0.1178 | 82.0 | 22386 | 0.1205 | 0.8187 | 0.7213 | 0.8782 | 0.3070 | 1e-05 |
139
+ | 0.1162 | 83.0 | 22659 | 0.1215 | 0.8171 | 0.7136 | 0.8754 | 0.3049 | 1e-05 |
140
+ | 0.1162 | 84.0 | 22932 | 0.1209 | 0.8212 | 0.7226 | 0.8817 | 0.3115 | 1e-05 |
141
+ | 0.1174 | 85.0 | 23205 | 0.1206 | 0.8213 | 0.7219 | 0.8823 | 0.3094 | 1e-05 |
142
+ | 0.1174 | 86.0 | 23478 | 0.1210 | 0.8207 | 0.7256 | 0.8811 | 0.3084 | 1e-05 |
143
+ | 0.1167 | 87.0 | 23751 | 0.1210 | 0.8192 | 0.7163 | 0.8800 | 0.3073 | 1e-05 |
144
+ | 0.116 | 88.0 | 24024 | 0.1208 | 0.8219 | 0.7180 | 0.8831 | 0.3094 | 1e-05 |
145
+ | 0.116 | 89.0 | 24297 | 0.1213 | 0.8236 | 0.7293 | 0.8872 | 0.3125 | 0.0000 |
146
+ | 0.1161 | 90.0 | 24570 | 0.1211 | 0.8228 | 0.7250 | 0.8869 | 0.3108 | 0.0000 |
147
+ | 0.1161 | 91.0 | 24843 | 0.1206 | 0.8191 | 0.7187 | 0.8779 | 0.3105 | 0.0000 |
148
+ | 0.1162 | 92.0 | 25116 | 0.1208 | 0.8196 | 0.7150 | 0.8793 | 0.3105 | 0.0000 |
149
+
150
+
151
+ ### Framework versions
152
+
153
+ - Transformers 4.41.1
154
+ - Pytorch 2.3.0+cu121
155
+ - Datasets 2.19.1
156
+ - Tokenizers 0.19.1
all_results.json ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 92.0,
3
+ "eval_accuracy": 0.3110726643598616,
4
+ "eval_f1_macro": 0.717494808151598,
5
+ "eval_f1_micro": 0.8227620520968348,
6
+ "eval_loss": 0.12092562019824982,
7
+ "eval_roc_auc": 0.8813312240951509,
8
+ "eval_runtime": 517.8306,
9
+ "eval_samples_per_second": 5.581,
10
+ "eval_steps_per_second": 0.176,
11
+ "learning_rate": 1.0000000000000002e-06,
12
+ "total_flos": 1.1890234809282512e+21,
13
+ "train_loss": 0.1360613288991788,
14
+ "train_runtime": 194834.2342,
15
+ "train_samples_per_second": 6.71,
16
+ "train_steps_per_second": 0.21
17
+ }
logs/events.out.tfevents.1725551843.datavisu2 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d075dd55deb5275fcd5e0924835c14835eb229188101f287003efdc795754717
3
- size 66181
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:22761b21b7f09c898c8a3be7bb825fc2297a9450d631d3f1432ad7c5ebe4a519
3
+ size 67907
logs/events.out.tfevents.1725747316.datavisu2 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d27d87170f8d275177bba31df36d666ca613d8e52201aca22927dc75ae44c36c
3
+ size 40
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d9f23af3956a02791d74920997064bcbbba1f9428486469a8e2dfee92095225f
3
  size 1228107508
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:99d9299d936d5fabb20d882aee36aecbb03c5cd8bc0a76e625d0467248d98dbf
3
  size 1228107508
test_results.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 92.0,
3
+ "eval_accuracy": 0.3110726643598616,
4
+ "eval_f1_macro": 0.717494808151598,
5
+ "eval_f1_micro": 0.8227620520968348,
6
+ "eval_loss": 0.12092562019824982,
7
+ "eval_roc_auc": 0.8813312240951509,
8
+ "eval_runtime": 517.8306,
9
+ "eval_samples_per_second": 5.581,
10
+ "eval_steps_per_second": 0.176,
11
+ "learning_rate": 1.0000000000000002e-06
12
+ }
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 92.0,
3
+ "learning_rate": 1.0000000000000002e-06,
4
+ "total_flos": 1.1890234809282512e+21,
5
+ "train_loss": 0.1360613288991788,
6
+ "train_runtime": 194834.2342,
7
+ "train_samples_per_second": 6.71,
8
+ "train_steps_per_second": 0.21
9
+ }
trainer_state.json ADDED
@@ -0,0 +1,1598 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.12050338089466095,
3
+ "best_model_checkpoint": "/home/datawork-iot-nos/Seatizen/models/multilabel/fine_scale/DinoVdeau-large-2024_09_05-batch-size32_epochs150_freeze/checkpoint-22386",
4
+ "epoch": 92.0,
5
+ "eval_steps": 500,
6
+ "global_step": 25116,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 1.0,
13
+ "eval_accuracy": 0.22314622314622315,
14
+ "eval_f1_macro": 0.5430112866470752,
15
+ "eval_f1_micro": 0.7516596896274684,
16
+ "eval_loss": 0.16899551451206207,
17
+ "eval_roc_auc": 0.8384250127967259,
18
+ "eval_runtime": 514.8139,
19
+ "eval_samples_per_second": 5.606,
20
+ "eval_steps_per_second": 0.177,
21
+ "learning_rate": 0.001,
22
+ "step": 273
23
+ },
24
+ {
25
+ "epoch": 1.8315018315018317,
26
+ "grad_norm": 0.9695320725440979,
27
+ "learning_rate": 0.001,
28
+ "loss": 0.2719,
29
+ "step": 500
30
+ },
31
+ {
32
+ "epoch": 2.0,
33
+ "eval_accuracy": 0.24012474012474014,
34
+ "eval_f1_macro": 0.5721428312627432,
35
+ "eval_f1_micro": 0.765669700910273,
36
+ "eval_loss": 0.153842031955719,
37
+ "eval_roc_auc": 0.8396070197954885,
38
+ "eval_runtime": 520.9151,
39
+ "eval_samples_per_second": 5.54,
40
+ "eval_steps_per_second": 0.175,
41
+ "learning_rate": 0.001,
42
+ "step": 546
43
+ },
44
+ {
45
+ "epoch": 3.0,
46
+ "eval_accuracy": 0.23458073458073458,
47
+ "eval_f1_macro": 0.6137585525531024,
48
+ "eval_f1_micro": 0.7772688719253604,
49
+ "eval_loss": 0.14828726649284363,
50
+ "eval_roc_auc": 0.851613165443153,
51
+ "eval_runtime": 519.0164,
52
+ "eval_samples_per_second": 5.561,
53
+ "eval_steps_per_second": 0.175,
54
+ "learning_rate": 0.001,
55
+ "step": 819
56
+ },
57
+ {
58
+ "epoch": 3.663003663003663,
59
+ "grad_norm": 0.24952150881290436,
60
+ "learning_rate": 0.001,
61
+ "loss": 0.1694,
62
+ "step": 1000
63
+ },
64
+ {
65
+ "epoch": 4.0,
66
+ "eval_accuracy": 0.2494802494802495,
67
+ "eval_f1_macro": 0.6224730910908008,
68
+ "eval_f1_micro": 0.7722737615963591,
69
+ "eval_loss": 0.1479637324810028,
70
+ "eval_roc_auc": 0.8406560025496872,
71
+ "eval_runtime": 512.3298,
72
+ "eval_samples_per_second": 5.633,
73
+ "eval_steps_per_second": 0.178,
74
+ "learning_rate": 0.001,
75
+ "step": 1092
76
+ },
77
+ {
78
+ "epoch": 5.0,
79
+ "eval_accuracy": 0.2494802494802495,
80
+ "eval_f1_macro": 0.6302307709949958,
81
+ "eval_f1_micro": 0.779738930569409,
82
+ "eval_loss": 0.14575305581092834,
83
+ "eval_roc_auc": 0.8469713476749664,
84
+ "eval_runtime": 509.5723,
85
+ "eval_samples_per_second": 5.664,
86
+ "eval_steps_per_second": 0.179,
87
+ "learning_rate": 0.001,
88
+ "step": 1365
89
+ },
90
+ {
91
+ "epoch": 5.4945054945054945,
92
+ "grad_norm": 0.17697261273860931,
93
+ "learning_rate": 0.001,
94
+ "loss": 0.1625,
95
+ "step": 1500
96
+ },
97
+ {
98
+ "epoch": 6.0,
99
+ "eval_accuracy": 0.2480942480942481,
100
+ "eval_f1_macro": 0.6092591780781843,
101
+ "eval_f1_micro": 0.7798061948433986,
102
+ "eval_loss": 0.14499613642692566,
103
+ "eval_roc_auc": 0.8476621294180898,
104
+ "eval_runtime": 511.5766,
105
+ "eval_samples_per_second": 5.641,
106
+ "eval_steps_per_second": 0.178,
107
+ "learning_rate": 0.001,
108
+ "step": 1638
109
+ },
110
+ {
111
+ "epoch": 7.0,
112
+ "eval_accuracy": 0.2525987525987526,
113
+ "eval_f1_macro": 0.624806622732382,
114
+ "eval_f1_micro": 0.7767369242779079,
115
+ "eval_loss": 0.1474585235118866,
116
+ "eval_roc_auc": 0.8453623673245133,
117
+ "eval_runtime": 510.6821,
118
+ "eval_samples_per_second": 5.651,
119
+ "eval_steps_per_second": 0.178,
120
+ "learning_rate": 0.001,
121
+ "step": 1911
122
+ },
123
+ {
124
+ "epoch": 7.326007326007326,
125
+ "grad_norm": 0.24790136516094208,
126
+ "learning_rate": 0.001,
127
+ "loss": 0.1592,
128
+ "step": 2000
129
+ },
130
+ {
131
+ "epoch": 8.0,
132
+ "eval_accuracy": 0.25744975744975745,
133
+ "eval_f1_macro": 0.6249401475720361,
134
+ "eval_f1_micro": 0.7803859753759638,
135
+ "eval_loss": 0.14568069577217102,
136
+ "eval_roc_auc": 0.8520784887308331,
137
+ "eval_runtime": 514.1071,
138
+ "eval_samples_per_second": 5.614,
139
+ "eval_steps_per_second": 0.177,
140
+ "learning_rate": 0.001,
141
+ "step": 2184
142
+ },
143
+ {
144
+ "epoch": 9.0,
145
+ "eval_accuracy": 0.25744975744975745,
146
+ "eval_f1_macro": 0.652642904607388,
147
+ "eval_f1_micro": 0.7868685150535805,
148
+ "eval_loss": 0.14169421792030334,
149
+ "eval_roc_auc": 0.8560913219420118,
150
+ "eval_runtime": 518.322,
151
+ "eval_samples_per_second": 5.568,
152
+ "eval_steps_per_second": 0.176,
153
+ "learning_rate": 0.001,
154
+ "step": 2457
155
+ },
156
+ {
157
+ "epoch": 9.157509157509157,
158
+ "grad_norm": 0.2022881656885147,
159
+ "learning_rate": 0.001,
160
+ "loss": 0.157,
161
+ "step": 2500
162
+ },
163
+ {
164
+ "epoch": 10.0,
165
+ "eval_accuracy": 0.25467775467775466,
166
+ "eval_f1_macro": 0.6289931868767601,
167
+ "eval_f1_micro": 0.7757335098168984,
168
+ "eval_loss": 0.1436299830675125,
169
+ "eval_roc_auc": 0.8403493908543921,
170
+ "eval_runtime": 514.5559,
171
+ "eval_samples_per_second": 5.609,
172
+ "eval_steps_per_second": 0.177,
173
+ "learning_rate": 0.001,
174
+ "step": 2730
175
+ },
176
+ {
177
+ "epoch": 10.989010989010989,
178
+ "grad_norm": 0.16650938987731934,
179
+ "learning_rate": 0.001,
180
+ "loss": 0.1563,
181
+ "step": 3000
182
+ },
183
+ {
184
+ "epoch": 11.0,
185
+ "eval_accuracy": 0.26403326403326405,
186
+ "eval_f1_macro": 0.6447870111639475,
187
+ "eval_f1_micro": 0.7886988341417751,
188
+ "eval_loss": 0.1428152322769165,
189
+ "eval_roc_auc": 0.8569209092596786,
190
+ "eval_runtime": 515.3716,
191
+ "eval_samples_per_second": 5.6,
192
+ "eval_steps_per_second": 0.177,
193
+ "learning_rate": 0.001,
194
+ "step": 3003
195
+ },
196
+ {
197
+ "epoch": 12.0,
198
+ "eval_accuracy": 0.25814275814275817,
199
+ "eval_f1_macro": 0.6493205009564239,
200
+ "eval_f1_micro": 0.7904845227679873,
201
+ "eval_loss": 0.1438700556755066,
202
+ "eval_roc_auc": 0.8637698939454586,
203
+ "eval_runtime": 512.0567,
204
+ "eval_samples_per_second": 5.636,
205
+ "eval_steps_per_second": 0.178,
206
+ "learning_rate": 0.001,
207
+ "step": 3276
208
+ },
209
+ {
210
+ "epoch": 12.820512820512821,
211
+ "grad_norm": 0.163461372256279,
212
+ "learning_rate": 0.001,
213
+ "loss": 0.1558,
214
+ "step": 3500
215
+ },
216
+ {
217
+ "epoch": 13.0,
218
+ "eval_accuracy": 0.2713097713097713,
219
+ "eval_f1_macro": 0.6561811626743236,
220
+ "eval_f1_micro": 0.7906956746065871,
221
+ "eval_loss": 0.13913600146770477,
222
+ "eval_roc_auc": 0.8551388511813229,
223
+ "eval_runtime": 511.5092,
224
+ "eval_samples_per_second": 5.642,
225
+ "eval_steps_per_second": 0.178,
226
+ "learning_rate": 0.001,
227
+ "step": 3549
228
+ },
229
+ {
230
+ "epoch": 14.0,
231
+ "eval_accuracy": 0.2643797643797644,
232
+ "eval_f1_macro": 0.6337626365639194,
233
+ "eval_f1_micro": 0.783810807286006,
234
+ "eval_loss": 0.14094506204128265,
235
+ "eval_roc_auc": 0.8484895839481307,
236
+ "eval_runtime": 513.4311,
237
+ "eval_samples_per_second": 5.621,
238
+ "eval_steps_per_second": 0.177,
239
+ "learning_rate": 0.001,
240
+ "step": 3822
241
+ },
242
+ {
243
+ "epoch": 14.652014652014651,
244
+ "grad_norm": 0.17725127935409546,
245
+ "learning_rate": 0.001,
246
+ "loss": 0.1543,
247
+ "step": 4000
248
+ },
249
+ {
250
+ "epoch": 15.0,
251
+ "eval_accuracy": 0.2577962577962578,
252
+ "eval_f1_macro": 0.6463067634895379,
253
+ "eval_f1_micro": 0.7907172995780591,
254
+ "eval_loss": 0.1396123319864273,
255
+ "eval_roc_auc": 0.8603407738558333,
256
+ "eval_runtime": 520.2063,
257
+ "eval_samples_per_second": 5.548,
258
+ "eval_steps_per_second": 0.175,
259
+ "learning_rate": 0.001,
260
+ "step": 4095
261
+ },
262
+ {
263
+ "epoch": 16.0,
264
+ "eval_accuracy": 0.2654192654192654,
265
+ "eval_f1_macro": 0.6593840515969085,
266
+ "eval_f1_micro": 0.7913274487959551,
267
+ "eval_loss": 0.13904806971549988,
268
+ "eval_roc_auc": 0.856418510343081,
269
+ "eval_runtime": 522.3782,
270
+ "eval_samples_per_second": 5.525,
271
+ "eval_steps_per_second": 0.174,
272
+ "learning_rate": 0.001,
273
+ "step": 4368
274
+ },
275
+ {
276
+ "epoch": 16.483516483516482,
277
+ "grad_norm": 0.16505596041679382,
278
+ "learning_rate": 0.001,
279
+ "loss": 0.1535,
280
+ "step": 4500
281
+ },
282
+ {
283
+ "epoch": 17.0,
284
+ "eval_accuracy": 0.2564102564102564,
285
+ "eval_f1_macro": 0.6585824628325464,
286
+ "eval_f1_micro": 0.7939832128313804,
287
+ "eval_loss": 0.1418265849351883,
288
+ "eval_roc_auc": 0.8664525383660324,
289
+ "eval_runtime": 520.8828,
290
+ "eval_samples_per_second": 5.541,
291
+ "eval_steps_per_second": 0.175,
292
+ "learning_rate": 0.001,
293
+ "step": 4641
294
+ },
295
+ {
296
+ "epoch": 18.0,
297
+ "eval_accuracy": 0.26576576576576577,
298
+ "eval_f1_macro": 0.6560187518750095,
299
+ "eval_f1_micro": 0.7957187827911858,
300
+ "eval_loss": 0.14155420660972595,
301
+ "eval_roc_auc": 0.864558649781785,
302
+ "eval_runtime": 521.9656,
303
+ "eval_samples_per_second": 5.529,
304
+ "eval_steps_per_second": 0.174,
305
+ "learning_rate": 0.001,
306
+ "step": 4914
307
+ },
308
+ {
309
+ "epoch": 18.315018315018314,
310
+ "grad_norm": 0.176731139421463,
311
+ "learning_rate": 0.001,
312
+ "loss": 0.1549,
313
+ "step": 5000
314
+ },
315
+ {
316
+ "epoch": 19.0,
317
+ "eval_accuracy": 0.262993762993763,
318
+ "eval_f1_macro": 0.6524018082903621,
319
+ "eval_f1_micro": 0.7885625699767461,
320
+ "eval_loss": 0.14027266204357147,
321
+ "eval_roc_auc": 0.8535729424099051,
322
+ "eval_runtime": 525.594,
323
+ "eval_samples_per_second": 5.491,
324
+ "eval_steps_per_second": 0.173,
325
+ "learning_rate": 0.001,
326
+ "step": 5187
327
+ },
328
+ {
329
+ "epoch": 20.0,
330
+ "eval_accuracy": 0.26126126126126126,
331
+ "eval_f1_macro": 0.6558190248610255,
332
+ "eval_f1_micro": 0.7910696719558615,
333
+ "eval_loss": 0.14759798347949982,
334
+ "eval_roc_auc": 0.8567849608157283,
335
+ "eval_runtime": 533.1376,
336
+ "eval_samples_per_second": 5.413,
337
+ "eval_steps_per_second": 0.171,
338
+ "learning_rate": 0.001,
339
+ "step": 5460
340
+ },
341
+ {
342
+ "epoch": 20.146520146520146,
343
+ "grad_norm": 0.15767891705036163,
344
+ "learning_rate": 0.001,
345
+ "loss": 0.154,
346
+ "step": 5500
347
+ },
348
+ {
349
+ "epoch": 21.0,
350
+ "eval_accuracy": 0.26576576576576577,
351
+ "eval_f1_macro": 0.6397027546064713,
352
+ "eval_f1_micro": 0.7879767016708474,
353
+ "eval_loss": 0.14285211265087128,
354
+ "eval_roc_auc": 0.8567511447301636,
355
+ "eval_runtime": 527.0011,
356
+ "eval_samples_per_second": 5.476,
357
+ "eval_steps_per_second": 0.173,
358
+ "learning_rate": 0.001,
359
+ "step": 5733
360
+ },
361
+ {
362
+ "epoch": 21.978021978021978,
363
+ "grad_norm": 0.18300685286521912,
364
+ "learning_rate": 0.001,
365
+ "loss": 0.1529,
366
+ "step": 6000
367
+ },
368
+ {
369
+ "epoch": 22.0,
370
+ "eval_accuracy": 0.26126126126126126,
371
+ "eval_f1_macro": 0.650810186340724,
372
+ "eval_f1_micro": 0.7936799099512236,
373
+ "eval_loss": 0.141402930021286,
374
+ "eval_roc_auc": 0.8653510005054305,
375
+ "eval_runtime": 525.9127,
376
+ "eval_samples_per_second": 5.488,
377
+ "eval_steps_per_second": 0.173,
378
+ "learning_rate": 0.001,
379
+ "step": 6006
380
+ },
381
+ {
382
+ "epoch": 23.0,
383
+ "eval_accuracy": 0.26853776853776856,
384
+ "eval_f1_macro": 0.6618136826297922,
385
+ "eval_f1_micro": 0.7975794766896787,
386
+ "eval_loss": 0.1415141373872757,
387
+ "eval_roc_auc": 0.8613092204030781,
388
+ "eval_runtime": 530.5247,
389
+ "eval_samples_per_second": 5.44,
390
+ "eval_steps_per_second": 0.172,
391
+ "learning_rate": 0.0001,
392
+ "step": 6279
393
+ },
394
+ {
395
+ "epoch": 23.80952380952381,
396
+ "grad_norm": 0.16848017275333405,
397
+ "learning_rate": 0.0001,
398
+ "loss": 0.1449,
399
+ "step": 6500
400
+ },
401
+ {
402
+ "epoch": 24.0,
403
+ "eval_accuracy": 0.27893277893277896,
404
+ "eval_f1_macro": 0.6750686264509598,
405
+ "eval_f1_micro": 0.8044778018063861,
406
+ "eval_loss": 0.13230843842029572,
407
+ "eval_roc_auc": 0.8664561198395929,
408
+ "eval_runtime": 521.5756,
409
+ "eval_samples_per_second": 5.533,
410
+ "eval_steps_per_second": 0.174,
411
+ "learning_rate": 0.0001,
412
+ "step": 6552
413
+ },
414
+ {
415
+ "epoch": 25.0,
416
+ "eval_accuracy": 0.27927927927927926,
417
+ "eval_f1_macro": 0.6724022117445357,
418
+ "eval_f1_micro": 0.8044072500946213,
419
+ "eval_loss": 0.13101588189601898,
420
+ "eval_roc_auc": 0.868781233937024,
421
+ "eval_runtime": 523.3306,
422
+ "eval_samples_per_second": 5.515,
423
+ "eval_steps_per_second": 0.174,
424
+ "learning_rate": 0.0001,
425
+ "step": 6825
426
+ },
427
+ {
428
+ "epoch": 25.641025641025642,
429
+ "grad_norm": 0.16336454451084137,
430
+ "learning_rate": 0.0001,
431
+ "loss": 0.1416,
432
+ "step": 7000
433
+ },
434
+ {
435
+ "epoch": 26.0,
436
+ "eval_accuracy": 0.28205128205128205,
437
+ "eval_f1_macro": 0.6689442300740391,
438
+ "eval_f1_micro": 0.8035965398218775,
439
+ "eval_loss": 0.13268393278121948,
440
+ "eval_roc_auc": 0.8645798435204571,
441
+ "eval_runtime": 532.8406,
442
+ "eval_samples_per_second": 5.416,
443
+ "eval_steps_per_second": 0.171,
444
+ "learning_rate": 0.0001,
445
+ "step": 7098
446
+ },
447
+ {
448
+ "epoch": 27.0,
449
+ "eval_accuracy": 0.2817047817047817,
450
+ "eval_f1_macro": 0.679681812643572,
451
+ "eval_f1_micro": 0.8068647969861867,
452
+ "eval_loss": 0.1317097693681717,
453
+ "eval_roc_auc": 0.8714747032608311,
454
+ "eval_runtime": 527.4278,
455
+ "eval_samples_per_second": 5.472,
456
+ "eval_steps_per_second": 0.173,
457
+ "learning_rate": 0.0001,
458
+ "step": 7371
459
+ },
460
+ {
461
+ "epoch": 27.47252747252747,
462
+ "grad_norm": 0.1572931855916977,
463
+ "learning_rate": 0.0001,
464
+ "loss": 0.1391,
465
+ "step": 7500
466
+ },
467
+ {
468
+ "epoch": 28.0,
469
+ "eval_accuracy": 0.27754677754677753,
470
+ "eval_f1_macro": 0.6818462300001074,
471
+ "eval_f1_micro": 0.8072126727334008,
472
+ "eval_loss": 0.12880520522594452,
473
+ "eval_roc_auc": 0.8697994857701482,
474
+ "eval_runtime": 536.9046,
475
+ "eval_samples_per_second": 5.375,
476
+ "eval_steps_per_second": 0.169,
477
+ "learning_rate": 0.0001,
478
+ "step": 7644
479
+ },
480
+ {
481
+ "epoch": 29.0,
482
+ "eval_accuracy": 0.2844767844767845,
483
+ "eval_f1_macro": 0.6807929806344717,
484
+ "eval_f1_micro": 0.8038088702067427,
485
+ "eval_loss": 0.12942521274089813,
486
+ "eval_roc_auc": 0.8628519636133017,
487
+ "eval_runtime": 520.5065,
488
+ "eval_samples_per_second": 5.545,
489
+ "eval_steps_per_second": 0.175,
490
+ "learning_rate": 0.0001,
491
+ "step": 7917
492
+ },
493
+ {
494
+ "epoch": 29.304029304029303,
495
+ "grad_norm": 0.19199338555335999,
496
+ "learning_rate": 0.0001,
497
+ "loss": 0.138,
498
+ "step": 8000
499
+ },
500
+ {
501
+ "epoch": 30.0,
502
+ "eval_accuracy": 0.28586278586278585,
503
+ "eval_f1_macro": 0.6825529208005033,
504
+ "eval_f1_micro": 0.8077149835761811,
505
+ "eval_loss": 0.12943296134471893,
506
+ "eval_roc_auc": 0.8701959964759374,
507
+ "eval_runtime": 543.5755,
508
+ "eval_samples_per_second": 5.309,
509
+ "eval_steps_per_second": 0.167,
510
+ "learning_rate": 0.0001,
511
+ "step": 8190
512
+ },
513
+ {
514
+ "epoch": 31.0,
515
+ "eval_accuracy": 0.28794178794178793,
516
+ "eval_f1_macro": 0.6779122940127521,
517
+ "eval_f1_micro": 0.8073808915025994,
518
+ "eval_loss": 0.12738928198814392,
519
+ "eval_roc_auc": 0.8666172459085354,
520
+ "eval_runtime": 521.4164,
521
+ "eval_samples_per_second": 5.535,
522
+ "eval_steps_per_second": 0.175,
523
+ "learning_rate": 0.0001,
524
+ "step": 8463
525
+ },
526
+ {
527
+ "epoch": 31.135531135531135,
528
+ "grad_norm": 0.1997932642698288,
529
+ "learning_rate": 0.0001,
530
+ "loss": 0.1364,
531
+ "step": 8500
532
+ },
533
+ {
534
+ "epoch": 32.0,
535
+ "eval_accuracy": 0.2882882882882883,
536
+ "eval_f1_macro": 0.6868638344898197,
537
+ "eval_f1_micro": 0.8104185890445432,
538
+ "eval_loss": 0.12775012850761414,
539
+ "eval_roc_auc": 0.8728485806633693,
540
+ "eval_runtime": 519.8308,
541
+ "eval_samples_per_second": 5.552,
542
+ "eval_steps_per_second": 0.175,
543
+ "learning_rate": 0.0001,
544
+ "step": 8736
545
+ },
546
+ {
547
+ "epoch": 32.967032967032964,
548
+ "grad_norm": 0.19476589560508728,
549
+ "learning_rate": 0.0001,
550
+ "loss": 0.1359,
551
+ "step": 9000
552
+ },
553
+ {
554
+ "epoch": 33.0,
555
+ "eval_accuracy": 0.2869022869022869,
556
+ "eval_f1_macro": 0.6810807224403135,
557
+ "eval_f1_micro": 0.8077248140635565,
558
+ "eval_loss": 0.12765593826770782,
559
+ "eval_roc_auc": 0.8692062891212271,
560
+ "eval_runtime": 514.7142,
561
+ "eval_samples_per_second": 5.607,
562
+ "eval_steps_per_second": 0.177,
563
+ "learning_rate": 0.0001,
564
+ "step": 9009
565
+ },
566
+ {
567
+ "epoch": 34.0,
568
+ "eval_accuracy": 0.2882882882882883,
569
+ "eval_f1_macro": 0.687361527737602,
570
+ "eval_f1_micro": 0.8108837797932926,
571
+ "eval_loss": 0.12660712003707886,
572
+ "eval_roc_auc": 0.8714320206807965,
573
+ "eval_runtime": 514.9645,
574
+ "eval_samples_per_second": 5.604,
575
+ "eval_steps_per_second": 0.177,
576
+ "learning_rate": 0.0001,
577
+ "step": 9282
578
+ },
579
+ {
580
+ "epoch": 34.798534798534796,
581
+ "grad_norm": 0.2034957855939865,
582
+ "learning_rate": 0.0001,
583
+ "loss": 0.1341,
584
+ "step": 9500
585
+ },
586
+ {
587
+ "epoch": 35.0,
588
+ "eval_accuracy": 0.29036729036729036,
589
+ "eval_f1_macro": 0.688483181989703,
590
+ "eval_f1_micro": 0.8103963941193815,
591
+ "eval_loss": 0.1262102574110031,
592
+ "eval_roc_auc": 0.8715800817488106,
593
+ "eval_runtime": 525.0872,
594
+ "eval_samples_per_second": 5.496,
595
+ "eval_steps_per_second": 0.173,
596
+ "learning_rate": 0.0001,
597
+ "step": 9555
598
+ },
599
+ {
600
+ "epoch": 36.0,
601
+ "eval_accuracy": 0.28274428274428276,
602
+ "eval_f1_macro": 0.6876394944988364,
603
+ "eval_f1_micro": 0.8070400273399119,
604
+ "eval_loss": 0.12687553465366364,
605
+ "eval_roc_auc": 0.8657418371913091,
606
+ "eval_runtime": 513.0757,
607
+ "eval_samples_per_second": 5.625,
608
+ "eval_steps_per_second": 0.177,
609
+ "learning_rate": 0.0001,
610
+ "step": 9828
611
+ },
612
+ {
613
+ "epoch": 36.63003663003663,
614
+ "grad_norm": 0.20557202398777008,
615
+ "learning_rate": 0.0001,
616
+ "loss": 0.1339,
617
+ "step": 10000
618
+ },
619
+ {
620
+ "epoch": 37.0,
621
+ "eval_accuracy": 0.28655578655578656,
622
+ "eval_f1_macro": 0.6833930255395054,
623
+ "eval_f1_micro": 0.8081597960050999,
624
+ "eval_loss": 0.12656189501285553,
625
+ "eval_roc_auc": 0.8678163688633396,
626
+ "eval_runtime": 515.3436,
627
+ "eval_samples_per_second": 5.6,
628
+ "eval_steps_per_second": 0.177,
629
+ "learning_rate": 0.0001,
630
+ "step": 10101
631
+ },
632
+ {
633
+ "epoch": 38.0,
634
+ "eval_accuracy": 0.2955647955647956,
635
+ "eval_f1_macro": 0.6936175483283518,
636
+ "eval_f1_micro": 0.8106371284826448,
637
+ "eval_loss": 0.12547720968723297,
638
+ "eval_roc_auc": 0.8706625538294134,
639
+ "eval_runtime": 512.1358,
640
+ "eval_samples_per_second": 5.635,
641
+ "eval_steps_per_second": 0.178,
642
+ "learning_rate": 0.0001,
643
+ "step": 10374
644
+ },
645
+ {
646
+ "epoch": 38.46153846153846,
647
+ "grad_norm": 0.2112371176481247,
648
+ "learning_rate": 0.0001,
649
+ "loss": 0.1307,
650
+ "step": 10500
651
+ },
652
+ {
653
+ "epoch": 39.0,
654
+ "eval_accuracy": 0.2927927927927928,
655
+ "eval_f1_macro": 0.6985657340894045,
656
+ "eval_f1_micro": 0.8141880626875626,
657
+ "eval_loss": 0.12485096603631973,
658
+ "eval_roc_auc": 0.8767653445350737,
659
+ "eval_runtime": 512.6109,
660
+ "eval_samples_per_second": 5.63,
661
+ "eval_steps_per_second": 0.178,
662
+ "learning_rate": 0.0001,
663
+ "step": 10647
664
+ },
665
+ {
666
+ "epoch": 40.0,
667
+ "eval_accuracy": 0.2934857934857935,
668
+ "eval_f1_macro": 0.6989554260935754,
669
+ "eval_f1_micro": 0.8138017044273539,
670
+ "eval_loss": 0.1257668137550354,
671
+ "eval_roc_auc": 0.8773247787534647,
672
+ "eval_runtime": 513.8833,
673
+ "eval_samples_per_second": 5.616,
674
+ "eval_steps_per_second": 0.177,
675
+ "learning_rate": 0.0001,
676
+ "step": 10920
677
+ },
678
+ {
679
+ "epoch": 40.29304029304029,
680
+ "grad_norm": 0.23032954335212708,
681
+ "learning_rate": 0.0001,
682
+ "loss": 0.1317,
683
+ "step": 11000
684
+ },
685
+ {
686
+ "epoch": 41.0,
687
+ "eval_accuracy": 0.29244629244629244,
688
+ "eval_f1_macro": 0.6923923602014324,
689
+ "eval_f1_micro": 0.8101351925856646,
690
+ "eval_loss": 0.12528541684150696,
691
+ "eval_roc_auc": 0.8687915491174283,
692
+ "eval_runtime": 513.0005,
693
+ "eval_samples_per_second": 5.626,
694
+ "eval_steps_per_second": 0.177,
695
+ "learning_rate": 0.0001,
696
+ "step": 11193
697
+ },
698
+ {
699
+ "epoch": 42.0,
700
+ "eval_accuracy": 0.3004158004158004,
701
+ "eval_f1_macro": 0.6970236383039276,
702
+ "eval_f1_micro": 0.8138018093835474,
703
+ "eval_loss": 0.12443084269762039,
704
+ "eval_roc_auc": 0.8737649281720051,
705
+ "eval_runtime": 525.5315,
706
+ "eval_samples_per_second": 5.492,
707
+ "eval_steps_per_second": 0.173,
708
+ "learning_rate": 0.0001,
709
+ "step": 11466
710
+ },
711
+ {
712
+ "epoch": 42.124542124542124,
713
+ "grad_norm": 0.23487386107444763,
714
+ "learning_rate": 0.0001,
715
+ "loss": 0.1308,
716
+ "step": 11500
717
+ },
718
+ {
719
+ "epoch": 43.0,
720
+ "eval_accuracy": 0.2948717948717949,
721
+ "eval_f1_macro": 0.6956334056896907,
722
+ "eval_f1_micro": 0.8131470414948238,
723
+ "eval_loss": 0.12451612949371338,
724
+ "eval_roc_auc": 0.8733690344991142,
725
+ "eval_runtime": 514.3778,
726
+ "eval_samples_per_second": 5.611,
727
+ "eval_steps_per_second": 0.177,
728
+ "learning_rate": 0.0001,
729
+ "step": 11739
730
+ },
731
+ {
732
+ "epoch": 43.956043956043956,
733
+ "grad_norm": 0.25621357560157776,
734
+ "learning_rate": 0.0001,
735
+ "loss": 0.1307,
736
+ "step": 12000
737
+ },
738
+ {
739
+ "epoch": 44.0,
740
+ "eval_accuracy": 0.2966042966042966,
741
+ "eval_f1_macro": 0.6915470420512126,
742
+ "eval_f1_micro": 0.812950847173293,
743
+ "eval_loss": 0.12501148879528046,
744
+ "eval_roc_auc": 0.8742664283667729,
745
+ "eval_runtime": 519.3764,
746
+ "eval_samples_per_second": 5.557,
747
+ "eval_steps_per_second": 0.175,
748
+ "learning_rate": 0.0001,
749
+ "step": 12012
750
+ },
751
+ {
752
+ "epoch": 45.0,
753
+ "eval_accuracy": 0.29625779625779625,
754
+ "eval_f1_macro": 0.7050548840380568,
755
+ "eval_f1_micro": 0.8136846971798428,
756
+ "eval_loss": 0.12397606670856476,
757
+ "eval_roc_auc": 0.8740443367647517,
758
+ "eval_runtime": 515.8997,
759
+ "eval_samples_per_second": 5.594,
760
+ "eval_steps_per_second": 0.176,
761
+ "learning_rate": 0.0001,
762
+ "step": 12285
763
+ },
764
+ {
765
+ "epoch": 45.78754578754579,
766
+ "grad_norm": 0.22914335131645203,
767
+ "learning_rate": 0.0001,
768
+ "loss": 0.1295,
769
+ "step": 12500
770
+ },
771
+ {
772
+ "epoch": 46.0,
773
+ "eval_accuracy": 0.29764379764379767,
774
+ "eval_f1_macro": 0.6987723620069867,
775
+ "eval_f1_micro": 0.8130628734954971,
776
+ "eval_loss": 0.12409698963165283,
777
+ "eval_roc_auc": 0.8733228777555885,
778
+ "eval_runtime": 516.6269,
779
+ "eval_samples_per_second": 5.586,
780
+ "eval_steps_per_second": 0.176,
781
+ "learning_rate": 0.0001,
782
+ "step": 12558
783
+ },
784
+ {
785
+ "epoch": 47.0,
786
+ "eval_accuracy": 0.2955647955647956,
787
+ "eval_f1_macro": 0.6957628076563835,
788
+ "eval_f1_micro": 0.811911298838437,
789
+ "eval_loss": 0.12429661303758621,
790
+ "eval_roc_auc": 0.8716271908692008,
791
+ "eval_runtime": 518.0917,
792
+ "eval_samples_per_second": 5.57,
793
+ "eval_steps_per_second": 0.176,
794
+ "learning_rate": 0.0001,
795
+ "step": 12831
796
+ },
797
+ {
798
+ "epoch": 47.61904761904762,
799
+ "grad_norm": 0.25639113783836365,
800
+ "learning_rate": 0.0001,
801
+ "loss": 0.1293,
802
+ "step": 13000
803
+ },
804
+ {
805
+ "epoch": 48.0,
806
+ "eval_accuracy": 0.2955647955647956,
807
+ "eval_f1_macro": 0.6990296569974817,
808
+ "eval_f1_micro": 0.8135280295401142,
809
+ "eval_loss": 0.12393072247505188,
810
+ "eval_roc_auc": 0.874436809929186,
811
+ "eval_runtime": 517.4634,
812
+ "eval_samples_per_second": 5.577,
813
+ "eval_steps_per_second": 0.176,
814
+ "learning_rate": 0.0001,
815
+ "step": 13104
816
+ },
817
+ {
818
+ "epoch": 49.0,
819
+ "eval_accuracy": 0.29972279972279975,
820
+ "eval_f1_macro": 0.7007060102949784,
821
+ "eval_f1_micro": 0.8152993625265614,
822
+ "eval_loss": 0.1242954283952713,
823
+ "eval_roc_auc": 0.8774914581184896,
824
+ "eval_runtime": 511.3749,
825
+ "eval_samples_per_second": 5.644,
826
+ "eval_steps_per_second": 0.178,
827
+ "learning_rate": 0.0001,
828
+ "step": 13377
829
+ },
830
+ {
831
+ "epoch": 49.45054945054945,
832
+ "grad_norm": 0.27197974920272827,
833
+ "learning_rate": 0.0001,
834
+ "loss": 0.1274,
835
+ "step": 13500
836
+ },
837
+ {
838
+ "epoch": 50.0,
839
+ "eval_accuracy": 0.29799029799029797,
840
+ "eval_f1_macro": 0.6999734070385492,
841
+ "eval_f1_micro": 0.8151919866444074,
842
+ "eval_loss": 0.12405084818601608,
843
+ "eval_roc_auc": 0.8769273693258459,
844
+ "eval_runtime": 509.3276,
845
+ "eval_samples_per_second": 5.666,
846
+ "eval_steps_per_second": 0.179,
847
+ "learning_rate": 0.0001,
848
+ "step": 13650
849
+ },
850
+ {
851
+ "epoch": 51.0,
852
+ "eval_accuracy": 0.3011088011088011,
853
+ "eval_f1_macro": 0.7055935576453343,
854
+ "eval_f1_micro": 0.8153039745759215,
855
+ "eval_loss": 0.12483017891645432,
856
+ "eval_roc_auc": 0.8803007418345086,
857
+ "eval_runtime": 511.0056,
858
+ "eval_samples_per_second": 5.648,
859
+ "eval_steps_per_second": 0.178,
860
+ "learning_rate": 0.0001,
861
+ "step": 13923
862
+ },
863
+ {
864
+ "epoch": 51.282051282051285,
865
+ "grad_norm": 0.23091430962085724,
866
+ "learning_rate": 0.0001,
867
+ "loss": 0.1271,
868
+ "step": 14000
869
+ },
870
+ {
871
+ "epoch": 52.0,
872
+ "eval_accuracy": 0.3049203049203049,
873
+ "eval_f1_macro": 0.7035566403965832,
874
+ "eval_f1_micro": 0.8157241959217996,
875
+ "eval_loss": 0.12426182627677917,
876
+ "eval_roc_auc": 0.8750656737623661,
877
+ "eval_runtime": 511.1647,
878
+ "eval_samples_per_second": 5.646,
879
+ "eval_steps_per_second": 0.178,
880
+ "learning_rate": 0.0001,
881
+ "step": 14196
882
+ },
883
+ {
884
+ "epoch": 53.0,
885
+ "eval_accuracy": 0.30214830214830213,
886
+ "eval_f1_macro": 0.7031528349086803,
887
+ "eval_f1_micro": 0.8152648882600192,
888
+ "eval_loss": 0.12408608943223953,
889
+ "eval_roc_auc": 0.8778170234547618,
890
+ "eval_runtime": 520.4,
891
+ "eval_samples_per_second": 5.546,
892
+ "eval_steps_per_second": 0.175,
893
+ "learning_rate": 0.0001,
894
+ "step": 14469
895
+ },
896
+ {
897
+ "epoch": 53.11355311355312,
898
+ "grad_norm": 0.23177389800548553,
899
+ "learning_rate": 0.0001,
900
+ "loss": 0.1275,
901
+ "step": 14500
902
+ },
903
+ {
904
+ "epoch": 54.0,
905
+ "eval_accuracy": 0.30214830214830213,
906
+ "eval_f1_macro": 0.7067666695453366,
907
+ "eval_f1_micro": 0.8152251458307105,
908
+ "eval_loss": 0.12344320118427277,
909
+ "eval_roc_auc": 0.8753333050750151,
910
+ "eval_runtime": 522.8329,
911
+ "eval_samples_per_second": 5.52,
912
+ "eval_steps_per_second": 0.174,
913
+ "learning_rate": 0.0001,
914
+ "step": 14742
915
+ },
916
+ {
917
+ "epoch": 54.94505494505494,
918
+ "grad_norm": 0.3403611481189728,
919
+ "learning_rate": 0.0001,
920
+ "loss": 0.1256,
921
+ "step": 15000
922
+ },
923
+ {
924
+ "epoch": 55.0,
925
+ "eval_accuracy": 0.30180180180180183,
926
+ "eval_f1_macro": 0.7075536762185066,
927
+ "eval_f1_micro": 0.8166332665330662,
928
+ "eval_loss": 0.12307523190975189,
929
+ "eval_roc_auc": 0.8776256091187804,
930
+ "eval_runtime": 513.5394,
931
+ "eval_samples_per_second": 5.62,
932
+ "eval_steps_per_second": 0.177,
933
+ "learning_rate": 0.0001,
934
+ "step": 15015
935
+ },
936
+ {
937
+ "epoch": 56.0,
938
+ "eval_accuracy": 0.30665280665280664,
939
+ "eval_f1_macro": 0.7087921855865761,
940
+ "eval_f1_micro": 0.8189626693095475,
941
+ "eval_loss": 0.12282071262598038,
942
+ "eval_roc_auc": 0.8821854285803199,
943
+ "eval_runtime": 519.2592,
944
+ "eval_samples_per_second": 5.558,
945
+ "eval_steps_per_second": 0.175,
946
+ "learning_rate": 0.0001,
947
+ "step": 15288
948
+ },
949
+ {
950
+ "epoch": 56.776556776556774,
951
+ "grad_norm": 0.28649473190307617,
952
+ "learning_rate": 0.0001,
953
+ "loss": 0.1258,
954
+ "step": 15500
955
+ },
956
+ {
957
+ "epoch": 57.0,
958
+ "eval_accuracy": 0.306999306999307,
959
+ "eval_f1_macro": 0.7079839879234633,
960
+ "eval_f1_micro": 0.8160328019748128,
961
+ "eval_loss": 0.12259934842586517,
962
+ "eval_roc_auc": 0.8766650096203477,
963
+ "eval_runtime": 523.8952,
964
+ "eval_samples_per_second": 5.509,
965
+ "eval_steps_per_second": 0.174,
966
+ "learning_rate": 0.0001,
967
+ "step": 15561
968
+ },
969
+ {
970
+ "epoch": 58.0,
971
+ "eval_accuracy": 0.30214830214830213,
972
+ "eval_f1_macro": 0.7072503847729165,
973
+ "eval_f1_micro": 0.8170145133631687,
974
+ "eval_loss": 0.12334763258695602,
975
+ "eval_roc_auc": 0.8773053153896588,
976
+ "eval_runtime": 522.7463,
977
+ "eval_samples_per_second": 5.521,
978
+ "eval_steps_per_second": 0.174,
979
+ "learning_rate": 0.0001,
980
+ "step": 15834
981
+ },
982
+ {
983
+ "epoch": 58.608058608058606,
984
+ "grad_norm": 0.2677023112773895,
985
+ "learning_rate": 0.0001,
986
+ "loss": 0.1258,
987
+ "step": 16000
988
+ },
989
+ {
990
+ "epoch": 59.0,
991
+ "eval_accuracy": 0.30214830214830213,
992
+ "eval_f1_macro": 0.713532815646164,
993
+ "eval_f1_micro": 0.8172105834237543,
994
+ "eval_loss": 0.12272054702043533,
995
+ "eval_roc_auc": 0.8780682765680952,
996
+ "eval_runtime": 524.8476,
997
+ "eval_samples_per_second": 5.499,
998
+ "eval_steps_per_second": 0.173,
999
+ "learning_rate": 0.0001,
1000
+ "step": 16107
1001
+ },
1002
+ {
1003
+ "epoch": 60.0,
1004
+ "eval_accuracy": 0.30214830214830213,
1005
+ "eval_f1_macro": 0.7039801220819605,
1006
+ "eval_f1_micro": 0.8142579609764339,
1007
+ "eval_loss": 0.12334387749433517,
1008
+ "eval_roc_auc": 0.8729462194126062,
1009
+ "eval_runtime": 526.97,
1010
+ "eval_samples_per_second": 5.477,
1011
+ "eval_steps_per_second": 0.173,
1012
+ "learning_rate": 0.0001,
1013
+ "step": 16380
1014
+ },
1015
+ {
1016
+ "epoch": 60.43956043956044,
1017
+ "grad_norm": 0.273879736661911,
1018
+ "learning_rate": 0.0001,
1019
+ "loss": 0.1252,
1020
+ "step": 16500
1021
+ },
1022
+ {
1023
+ "epoch": 61.0,
1024
+ "eval_accuracy": 0.3042273042273042,
1025
+ "eval_f1_macro": 0.7120578542808926,
1026
+ "eval_f1_micro": 0.816814564846061,
1027
+ "eval_loss": 0.12339764833450317,
1028
+ "eval_roc_auc": 0.8783554248995846,
1029
+ "eval_runtime": 524.4656,
1030
+ "eval_samples_per_second": 5.503,
1031
+ "eval_steps_per_second": 0.174,
1032
+ "learning_rate": 0.0001,
1033
+ "step": 16653
1034
+ },
1035
+ {
1036
+ "epoch": 62.0,
1037
+ "eval_accuracy": 0.3049203049203049,
1038
+ "eval_f1_macro": 0.7124854785684515,
1039
+ "eval_f1_micro": 0.8169309505831026,
1040
+ "eval_loss": 0.12234435975551605,
1041
+ "eval_roc_auc": 0.876382515863111,
1042
+ "eval_runtime": 518.389,
1043
+ "eval_samples_per_second": 5.567,
1044
+ "eval_steps_per_second": 0.176,
1045
+ "learning_rate": 0.0001,
1046
+ "step": 16926
1047
+ },
1048
+ {
1049
+ "epoch": 62.27106227106227,
1050
+ "grad_norm": 0.2836596667766571,
1051
+ "learning_rate": 0.0001,
1052
+ "loss": 0.1238,
1053
+ "step": 17000
1054
+ },
1055
+ {
1056
+ "epoch": 63.0,
1057
+ "eval_accuracy": 0.30353430353430355,
1058
+ "eval_f1_macro": 0.709030237195192,
1059
+ "eval_f1_micro": 0.8151443922095366,
1060
+ "eval_loss": 0.12311259657144547,
1061
+ "eval_roc_auc": 0.875227363209172,
1062
+ "eval_runtime": 523.1406,
1063
+ "eval_samples_per_second": 5.517,
1064
+ "eval_steps_per_second": 0.174,
1065
+ "learning_rate": 0.0001,
1066
+ "step": 17199
1067
+ },
1068
+ {
1069
+ "epoch": 64.0,
1070
+ "eval_accuracy": 0.30665280665280664,
1071
+ "eval_f1_macro": 0.7114197657112039,
1072
+ "eval_f1_micro": 0.8183222681531587,
1073
+ "eval_loss": 0.12282687425613403,
1074
+ "eval_roc_auc": 0.8785221042646094,
1075
+ "eval_runtime": 525.9879,
1076
+ "eval_samples_per_second": 5.487,
1077
+ "eval_steps_per_second": 0.173,
1078
+ "learning_rate": 0.0001,
1079
+ "step": 17472
1080
+ },
1081
+ {
1082
+ "epoch": 64.1025641025641,
1083
+ "grad_norm": 0.327009916305542,
1084
+ "learning_rate": 0.0001,
1085
+ "loss": 0.1247,
1086
+ "step": 17500
1087
+ },
1088
+ {
1089
+ "epoch": 65.0,
1090
+ "eval_accuracy": 0.30353430353430355,
1091
+ "eval_f1_macro": 0.715610525327271,
1092
+ "eval_f1_micro": 0.8185065204751224,
1093
+ "eval_loss": 0.12305620312690735,
1094
+ "eval_roc_auc": 0.8802214933483853,
1095
+ "eval_runtime": 527.6963,
1096
+ "eval_samples_per_second": 5.469,
1097
+ "eval_steps_per_second": 0.172,
1098
+ "learning_rate": 0.0001,
1099
+ "step": 17745
1100
+ },
1101
+ {
1102
+ "epoch": 65.93406593406593,
1103
+ "grad_norm": 0.3439556360244751,
1104
+ "learning_rate": 0.0001,
1105
+ "loss": 0.123,
1106
+ "step": 18000
1107
+ },
1108
+ {
1109
+ "epoch": 66.0,
1110
+ "eval_accuracy": 0.30214830214830213,
1111
+ "eval_f1_macro": 0.7083957677770276,
1112
+ "eval_f1_micro": 0.8193021036471515,
1113
+ "eval_loss": 0.12252139300107956,
1114
+ "eval_roc_auc": 0.8809488409975973,
1115
+ "eval_runtime": 523.6027,
1116
+ "eval_samples_per_second": 5.512,
1117
+ "eval_steps_per_second": 0.174,
1118
+ "learning_rate": 0.0001,
1119
+ "step": 18018
1120
+ },
1121
+ {
1122
+ "epoch": 67.0,
1123
+ "eval_accuracy": 0.3031878031878032,
1124
+ "eval_f1_macro": 0.713563304331985,
1125
+ "eval_f1_micro": 0.8185542268382505,
1126
+ "eval_loss": 0.12215397506952286,
1127
+ "eval_roc_auc": 0.8813502879665707,
1128
+ "eval_runtime": 528.5406,
1129
+ "eval_samples_per_second": 5.46,
1130
+ "eval_steps_per_second": 0.172,
1131
+ "learning_rate": 0.0001,
1132
+ "step": 18291
1133
+ },
1134
+ {
1135
+ "epoch": 67.76556776556777,
1136
+ "grad_norm": 0.3434881269931793,
1137
+ "learning_rate": 0.0001,
1138
+ "loss": 0.1224,
1139
+ "step": 18500
1140
+ },
1141
+ {
1142
+ "epoch": 68.0,
1143
+ "eval_accuracy": 0.3090783090783091,
1144
+ "eval_f1_macro": 0.7169216330412181,
1145
+ "eval_f1_micro": 0.8201218248870841,
1146
+ "eval_loss": 0.12200037389993668,
1147
+ "eval_roc_auc": 0.8818022645643908,
1148
+ "eval_runtime": 525.6971,
1149
+ "eval_samples_per_second": 5.49,
1150
+ "eval_steps_per_second": 0.173,
1151
+ "learning_rate": 0.0001,
1152
+ "step": 18564
1153
+ },
1154
+ {
1155
+ "epoch": 69.0,
1156
+ "eval_accuracy": 0.30180180180180183,
1157
+ "eval_f1_macro": 0.7165157275423649,
1158
+ "eval_f1_micro": 0.8171493231633209,
1159
+ "eval_loss": 0.12282921373844147,
1160
+ "eval_roc_auc": 0.8767867663076429,
1161
+ "eval_runtime": 539.1574,
1162
+ "eval_samples_per_second": 5.353,
1163
+ "eval_steps_per_second": 0.169,
1164
+ "learning_rate": 0.0001,
1165
+ "step": 18837
1166
+ },
1167
+ {
1168
+ "epoch": 69.59706959706959,
1169
+ "grad_norm": 0.2773456275463104,
1170
+ "learning_rate": 0.0001,
1171
+ "loss": 0.1228,
1172
+ "step": 19000
1173
+ },
1174
+ {
1175
+ "epoch": 70.0,
1176
+ "eval_accuracy": 0.3042273042273042,
1177
+ "eval_f1_macro": 0.7130922408537738,
1178
+ "eval_f1_micro": 0.8176893032631977,
1179
+ "eval_loss": 0.12265007942914963,
1180
+ "eval_roc_auc": 0.8764658555456234,
1181
+ "eval_runtime": 532.0042,
1182
+ "eval_samples_per_second": 5.425,
1183
+ "eval_steps_per_second": 0.171,
1184
+ "learning_rate": 0.0001,
1185
+ "step": 19110
1186
+ },
1187
+ {
1188
+ "epoch": 71.0,
1189
+ "eval_accuracy": 0.29799029799029797,
1190
+ "eval_f1_macro": 0.7123118599173115,
1191
+ "eval_f1_micro": 0.8155257705805251,
1192
+ "eval_loss": 0.12318737804889679,
1193
+ "eval_roc_auc": 0.8733064995562728,
1194
+ "eval_runtime": 512.5227,
1195
+ "eval_samples_per_second": 5.631,
1196
+ "eval_steps_per_second": 0.178,
1197
+ "learning_rate": 0.0001,
1198
+ "step": 19383
1199
+ },
1200
+ {
1201
+ "epoch": 71.42857142857143,
1202
+ "grad_norm": 0.32921841740608215,
1203
+ "learning_rate": 0.0001,
1204
+ "loss": 0.1224,
1205
+ "step": 19500
1206
+ },
1207
+ {
1208
+ "epoch": 72.0,
1209
+ "eval_accuracy": 0.30561330561330563,
1210
+ "eval_f1_macro": 0.7181217472368024,
1211
+ "eval_f1_micro": 0.8177146438270315,
1212
+ "eval_loss": 0.12224896252155304,
1213
+ "eval_roc_auc": 0.8780131460200304,
1214
+ "eval_runtime": 526.8353,
1215
+ "eval_samples_per_second": 5.478,
1216
+ "eval_steps_per_second": 0.173,
1217
+ "learning_rate": 0.0001,
1218
+ "step": 19656
1219
+ },
1220
+ {
1221
+ "epoch": 73.0,
1222
+ "eval_accuracy": 0.3076923076923077,
1223
+ "eval_f1_macro": 0.7046690012290543,
1224
+ "eval_f1_micro": 0.8161570403926011,
1225
+ "eval_loss": 0.12214501202106476,
1226
+ "eval_roc_auc": 0.8759937448960649,
1227
+ "eval_runtime": 523.4325,
1228
+ "eval_samples_per_second": 5.514,
1229
+ "eval_steps_per_second": 0.174,
1230
+ "learning_rate": 0.0001,
1231
+ "step": 19929
1232
+ },
1233
+ {
1234
+ "epoch": 73.26007326007326,
1235
+ "grad_norm": 0.27500712871551514,
1236
+ "learning_rate": 0.0001,
1237
+ "loss": 0.122,
1238
+ "step": 20000
1239
+ },
1240
+ {
1241
+ "epoch": 74.0,
1242
+ "eval_accuracy": 0.2972972972972973,
1243
+ "eval_f1_macro": 0.7070482653980339,
1244
+ "eval_f1_micro": 0.8147835269271382,
1245
+ "eval_loss": 0.12297073751688004,
1246
+ "eval_roc_auc": 0.8731965201490751,
1247
+ "eval_runtime": 521.6748,
1248
+ "eval_samples_per_second": 5.532,
1249
+ "eval_steps_per_second": 0.174,
1250
+ "learning_rate": 0.0001,
1251
+ "step": 20202
1252
+ },
1253
+ {
1254
+ "epoch": 75.0,
1255
+ "eval_accuracy": 0.3049203049203049,
1256
+ "eval_f1_macro": 0.7123584497861349,
1257
+ "eval_f1_micro": 0.8175831550689987,
1258
+ "eval_loss": 0.12141965329647064,
1259
+ "eval_roc_auc": 0.876778409536002,
1260
+ "eval_runtime": 521.9582,
1261
+ "eval_samples_per_second": 5.529,
1262
+ "eval_steps_per_second": 0.174,
1263
+ "learning_rate": 1e-05,
1264
+ "step": 20475
1265
+ },
1266
+ {
1267
+ "epoch": 75.0915750915751,
1268
+ "grad_norm": 0.34586936235427856,
1269
+ "learning_rate": 1e-05,
1270
+ "loss": 0.1201,
1271
+ "step": 20500
1272
+ },
1273
+ {
1274
+ "epoch": 76.0,
1275
+ "eval_accuracy": 0.30665280665280664,
1276
+ "eval_f1_macro": 0.7265282519195887,
1277
+ "eval_f1_micro": 0.8212704324436167,
1278
+ "eval_loss": 0.12091591954231262,
1279
+ "eval_roc_auc": 0.8828403151052873,
1280
+ "eval_runtime": 515.687,
1281
+ "eval_samples_per_second": 5.596,
1282
+ "eval_steps_per_second": 0.176,
1283
+ "learning_rate": 1e-05,
1284
+ "step": 20748
1285
+ },
1286
+ {
1287
+ "epoch": 76.92307692307692,
1288
+ "grad_norm": 0.3650946617126465,
1289
+ "learning_rate": 1e-05,
1290
+ "loss": 0.1192,
1291
+ "step": 21000
1292
+ },
1293
+ {
1294
+ "epoch": 77.0,
1295
+ "eval_accuracy": 0.30734580734580735,
1296
+ "eval_f1_macro": 0.7249141687532618,
1297
+ "eval_f1_micro": 0.8221009885557243,
1298
+ "eval_loss": 0.12162773311138153,
1299
+ "eval_roc_auc": 0.88597146196019,
1300
+ "eval_runtime": 505.8066,
1301
+ "eval_samples_per_second": 5.706,
1302
+ "eval_steps_per_second": 0.18,
1303
+ "learning_rate": 1e-05,
1304
+ "step": 21021
1305
+ },
1306
+ {
1307
+ "epoch": 78.0,
1308
+ "eval_accuracy": 0.30561330561330563,
1309
+ "eval_f1_macro": 0.7232913822219021,
1310
+ "eval_f1_micro": 0.821013443640124,
1311
+ "eval_loss": 0.12114103883504868,
1312
+ "eval_roc_auc": 0.8828214151193448,
1313
+ "eval_runtime": 515.423,
1314
+ "eval_samples_per_second": 5.599,
1315
+ "eval_steps_per_second": 0.177,
1316
+ "learning_rate": 1e-05,
1317
+ "step": 21294
1318
+ },
1319
+ {
1320
+ "epoch": 78.75457875457876,
1321
+ "grad_norm": 0.3805921673774719,
1322
+ "learning_rate": 1e-05,
1323
+ "loss": 0.1178,
1324
+ "step": 21500
1325
+ },
1326
+ {
1327
+ "epoch": 79.0,
1328
+ "eval_accuracy": 0.30561330561330563,
1329
+ "eval_f1_macro": 0.7157592534107864,
1330
+ "eval_f1_micro": 0.8181284095677717,
1331
+ "eval_loss": 0.1210767850279808,
1332
+ "eval_roc_auc": 0.8769422854254683,
1333
+ "eval_runtime": 524.7026,
1334
+ "eval_samples_per_second": 5.5,
1335
+ "eval_steps_per_second": 0.173,
1336
+ "learning_rate": 1e-05,
1337
+ "step": 21567
1338
+ },
1339
+ {
1340
+ "epoch": 80.0,
1341
+ "eval_accuracy": 0.3090783090783091,
1342
+ "eval_f1_macro": 0.7196736600383237,
1343
+ "eval_f1_micro": 0.8200463116109824,
1344
+ "eval_loss": 0.12099559605121613,
1345
+ "eval_roc_auc": 0.8823936101146178,
1346
+ "eval_runtime": 518.5996,
1347
+ "eval_samples_per_second": 5.565,
1348
+ "eval_steps_per_second": 0.175,
1349
+ "learning_rate": 1e-05,
1350
+ "step": 21840
1351
+ },
1352
+ {
1353
+ "epoch": 80.58608058608058,
1354
+ "grad_norm": 0.38496658205986023,
1355
+ "learning_rate": 1e-05,
1356
+ "loss": 0.1178,
1357
+ "step": 22000
1358
+ },
1359
+ {
1360
+ "epoch": 81.0,
1361
+ "eval_accuracy": 0.31046431046431044,
1362
+ "eval_f1_macro": 0.7194056763702963,
1363
+ "eval_f1_micro": 0.8189727287937092,
1364
+ "eval_loss": 0.12053155153989792,
1365
+ "eval_roc_auc": 0.8783734261636972,
1366
+ "eval_runtime": 517.5249,
1367
+ "eval_samples_per_second": 5.577,
1368
+ "eval_steps_per_second": 0.176,
1369
+ "learning_rate": 1e-05,
1370
+ "step": 22113
1371
+ },
1372
+ {
1373
+ "epoch": 82.0,
1374
+ "eval_accuracy": 0.306999306999307,
1375
+ "eval_f1_macro": 0.7212694332008583,
1376
+ "eval_f1_micro": 0.8186875235267054,
1377
+ "eval_loss": 0.12050338089466095,
1378
+ "eval_roc_auc": 0.8782284502601733,
1379
+ "eval_runtime": 511.5713,
1380
+ "eval_samples_per_second": 5.641,
1381
+ "eval_steps_per_second": 0.178,
1382
+ "learning_rate": 1e-05,
1383
+ "step": 22386
1384
+ },
1385
+ {
1386
+ "epoch": 82.41758241758242,
1387
+ "grad_norm": 0.29807013273239136,
1388
+ "learning_rate": 1e-05,
1389
+ "loss": 0.1162,
1390
+ "step": 22500
1391
+ },
1392
+ {
1393
+ "epoch": 83.0,
1394
+ "eval_accuracy": 0.3049203049203049,
1395
+ "eval_f1_macro": 0.7136069207682542,
1396
+ "eval_f1_micro": 0.817129142279675,
1397
+ "eval_loss": 0.12153622508049011,
1398
+ "eval_roc_auc": 0.8753921914755026,
1399
+ "eval_runtime": 514.5554,
1400
+ "eval_samples_per_second": 5.609,
1401
+ "eval_steps_per_second": 0.177,
1402
+ "learning_rate": 1e-05,
1403
+ "step": 22659
1404
+ },
1405
+ {
1406
+ "epoch": 84.0,
1407
+ "eval_accuracy": 0.3115038115038115,
1408
+ "eval_f1_macro": 0.72263281374496,
1409
+ "eval_f1_micro": 0.8212135055442501,
1410
+ "eval_loss": 0.12091034650802612,
1411
+ "eval_roc_auc": 0.8817381602117871,
1412
+ "eval_runtime": 514.2801,
1413
+ "eval_samples_per_second": 5.612,
1414
+ "eval_steps_per_second": 0.177,
1415
+ "learning_rate": 1e-05,
1416
+ "step": 22932
1417
+ },
1418
+ {
1419
+ "epoch": 84.24908424908425,
1420
+ "grad_norm": 0.4926730692386627,
1421
+ "learning_rate": 1e-05,
1422
+ "loss": 0.1174,
1423
+ "step": 23000
1424
+ },
1425
+ {
1426
+ "epoch": 85.0,
1427
+ "eval_accuracy": 0.30942480942480943,
1428
+ "eval_f1_macro": 0.7219026145386024,
1429
+ "eval_f1_micro": 0.8212908842183808,
1430
+ "eval_loss": 0.12058679759502411,
1431
+ "eval_roc_auc": 0.8823288887291161,
1432
+ "eval_runtime": 513.2258,
1433
+ "eval_samples_per_second": 5.623,
1434
+ "eval_steps_per_second": 0.177,
1435
+ "learning_rate": 1e-05,
1436
+ "step": 23205
1437
+ },
1438
+ {
1439
+ "epoch": 86.0,
1440
+ "eval_accuracy": 0.30838530838530837,
1441
+ "eval_f1_macro": 0.7255503995321377,
1442
+ "eval_f1_micro": 0.8206727371003285,
1443
+ "eval_loss": 0.1210218220949173,
1444
+ "eval_roc_auc": 0.8810894976708349,
1445
+ "eval_runtime": 516.6336,
1446
+ "eval_samples_per_second": 5.586,
1447
+ "eval_steps_per_second": 0.176,
1448
+ "learning_rate": 1e-05,
1449
+ "step": 23478
1450
+ },
1451
+ {
1452
+ "epoch": 86.08058608058609,
1453
+ "grad_norm": 0.3941400647163391,
1454
+ "learning_rate": 1e-05,
1455
+ "loss": 0.1167,
1456
+ "step": 23500
1457
+ },
1458
+ {
1459
+ "epoch": 87.0,
1460
+ "eval_accuracy": 0.30734580734580735,
1461
+ "eval_f1_macro": 0.7163464112504625,
1462
+ "eval_f1_micro": 0.81919187715867,
1463
+ "eval_loss": 0.12097787857055664,
1464
+ "eval_roc_auc": 0.8800195357981024,
1465
+ "eval_runtime": 514.309,
1466
+ "eval_samples_per_second": 5.611,
1467
+ "eval_steps_per_second": 0.177,
1468
+ "learning_rate": 1e-05,
1469
+ "step": 23751
1470
+ },
1471
+ {
1472
+ "epoch": 87.91208791208791,
1473
+ "grad_norm": 0.30446189641952515,
1474
+ "learning_rate": 1e-05,
1475
+ "loss": 0.116,
1476
+ "step": 24000
1477
+ },
1478
+ {
1479
+ "epoch": 88.0,
1480
+ "eval_accuracy": 0.30942480942480943,
1481
+ "eval_f1_macro": 0.7179611359738045,
1482
+ "eval_f1_micro": 0.8219223445649475,
1483
+ "eval_loss": 0.12078534066677094,
1484
+ "eval_roc_auc": 0.8831166385335435,
1485
+ "eval_runtime": 513.9044,
1486
+ "eval_samples_per_second": 5.616,
1487
+ "eval_steps_per_second": 0.177,
1488
+ "learning_rate": 1e-05,
1489
+ "step": 24024
1490
+ },
1491
+ {
1492
+ "epoch": 89.0,
1493
+ "eval_accuracy": 0.3125433125433125,
1494
+ "eval_f1_macro": 0.7293063087262872,
1495
+ "eval_f1_micro": 0.8235824319895118,
1496
+ "eval_loss": 0.1213160827755928,
1497
+ "eval_roc_auc": 0.8871674997505042,
1498
+ "eval_runtime": 514.1086,
1499
+ "eval_samples_per_second": 5.614,
1500
+ "eval_steps_per_second": 0.177,
1501
+ "learning_rate": 1.0000000000000002e-06,
1502
+ "step": 24297
1503
+ },
1504
+ {
1505
+ "epoch": 89.74358974358974,
1506
+ "grad_norm": 0.2996889054775238,
1507
+ "learning_rate": 1.0000000000000002e-06,
1508
+ "loss": 0.1161,
1509
+ "step": 24500
1510
+ },
1511
+ {
1512
+ "epoch": 90.0,
1513
+ "eval_accuracy": 0.3108108108108108,
1514
+ "eval_f1_macro": 0.7249894355418997,
1515
+ "eval_f1_micro": 0.8228019165403988,
1516
+ "eval_loss": 0.12110408395528793,
1517
+ "eval_roc_auc": 0.8868651536304606,
1518
+ "eval_runtime": 511.9928,
1519
+ "eval_samples_per_second": 5.637,
1520
+ "eval_steps_per_second": 0.178,
1521
+ "learning_rate": 1.0000000000000002e-06,
1522
+ "step": 24570
1523
+ },
1524
+ {
1525
+ "epoch": 91.0,
1526
+ "eval_accuracy": 0.31046431046431044,
1527
+ "eval_f1_macro": 0.7187027508297176,
1528
+ "eval_f1_micro": 0.8191074795725959,
1529
+ "eval_loss": 0.1205781027674675,
1530
+ "eval_roc_auc": 0.8779146622039986,
1531
+ "eval_runtime": 513.1929,
1532
+ "eval_samples_per_second": 5.624,
1533
+ "eval_steps_per_second": 0.177,
1534
+ "learning_rate": 1.0000000000000002e-06,
1535
+ "step": 24843
1536
+ },
1537
+ {
1538
+ "epoch": 91.57509157509158,
1539
+ "grad_norm": 0.30445897579193115,
1540
+ "learning_rate": 1.0000000000000002e-06,
1541
+ "loss": 0.1162,
1542
+ "step": 25000
1543
+ },
1544
+ {
1545
+ "epoch": 92.0,
1546
+ "eval_accuracy": 0.31046431046431044,
1547
+ "eval_f1_macro": 0.7150284118631205,
1548
+ "eval_f1_micro": 0.8196009683612989,
1549
+ "eval_loss": 0.12076584249734879,
1550
+ "eval_roc_auc": 0.8793405313350767,
1551
+ "eval_runtime": 520.7671,
1552
+ "eval_samples_per_second": 5.542,
1553
+ "eval_steps_per_second": 0.175,
1554
+ "learning_rate": 1.0000000000000002e-06,
1555
+ "step": 25116
1556
+ },
1557
+ {
1558
+ "epoch": 92.0,
1559
+ "learning_rate": 1.0000000000000002e-06,
1560
+ "step": 25116,
1561
+ "total_flos": 1.1890234809282512e+21,
1562
+ "train_loss": 0.1360613288991788,
1563
+ "train_runtime": 194834.2342,
1564
+ "train_samples_per_second": 6.71,
1565
+ "train_steps_per_second": 0.21
1566
+ }
1567
+ ],
1568
+ "logging_steps": 500,
1569
+ "max_steps": 40950,
1570
+ "num_input_tokens_seen": 0,
1571
+ "num_train_epochs": 150,
1572
+ "save_steps": 500,
1573
+ "stateful_callbacks": {
1574
+ "EarlyStoppingCallback": {
1575
+ "args": {
1576
+ "early_stopping_patience": 10,
1577
+ "early_stopping_threshold": 0.0
1578
+ },
1579
+ "attributes": {
1580
+ "early_stopping_patience_counter": 0
1581
+ }
1582
+ },
1583
+ "TrainerControl": {
1584
+ "args": {
1585
+ "should_epoch_stop": false,
1586
+ "should_evaluate": false,
1587
+ "should_log": false,
1588
+ "should_save": true,
1589
+ "should_training_stop": true
1590
+ },
1591
+ "attributes": {}
1592
+ }
1593
+ },
1594
+ "total_flos": 1.1890234809282512e+21,
1595
+ "train_batch_size": 32,
1596
+ "trial_name": null,
1597
+ "trial_params": null
1598
+ }