paul
commited on
Commit
•
5a2e2ee
1
Parent(s):
f9f80dc
End of training
Browse files- all_results.json +8 -0
- pytorch_model.bin +1 -1
- runs/Oct21_22-24-56_teesta/events.out.tfevents.1666371324.teesta.31301.0 +2 -2
- train_results.json +8 -0
- trainer_state.json +505 -0
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 19.99,
|
| 3 |
+
"total_flos": 5.432231359979311e+18,
|
| 4 |
+
"train_loss": 0.6680345350503921,
|
| 5 |
+
"train_runtime": 1446.6101,
|
| 6 |
+
"train_samples_per_second": 71.533,
|
| 7 |
+
"train_steps_per_second": 0.277
|
| 8 |
+
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 233549753
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ba9137a731f10b5b65349f6bd59255dff66fca1ef6ac1dedd12da9b212cf0703
|
| 3 |
size 233549753
|
runs/Oct21_22-24-56_teesta/events.out.tfevents.1666371324.teesta.31301.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2b5dc00af85f6e4b9200adefc729928cb70b4740d4e9e745ee7c23d9810ba4f9
|
| 3 |
+
size 19828
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 19.99,
|
| 3 |
+
"total_flos": 5.432231359979311e+18,
|
| 4 |
+
"train_loss": 0.6680345350503921,
|
| 5 |
+
"train_runtime": 1446.6101,
|
| 6 |
+
"train_samples_per_second": 71.533,
|
| 7 |
+
"train_steps_per_second": 0.277
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,505 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.758887171561051,
|
| 3 |
+
"best_model_checkpoint": "resnet152-FV-finetuned-memes/checkpoint-300",
|
| 4 |
+
"epoch": 19.987654320987655,
|
| 5 |
+
"global_step": 400,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.49,
|
| 12 |
+
"learning_rate": 3e-05,
|
| 13 |
+
"loss": 1.605,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.99,
|
| 18 |
+
"learning_rate": 6e-05,
|
| 19 |
+
"loss": 1.5739,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.99,
|
| 24 |
+
"eval_accuracy": 0.4520865533230294,
|
| 25 |
+
"eval_f1": 0.2879543301590467,
|
| 26 |
+
"eval_loss": 1.542733907699585,
|
| 27 |
+
"eval_precision": 0.31310616306027816,
|
| 28 |
+
"eval_recall": 0.4520865533230294,
|
| 29 |
+
"eval_runtime": 13.5797,
|
| 30 |
+
"eval_samples_per_second": 95.289,
|
| 31 |
+
"eval_steps_per_second": 1.546,
|
| 32 |
+
"step": 20
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 1.49,
|
| 36 |
+
"learning_rate": 9e-05,
|
| 37 |
+
"loss": 1.557,
|
| 38 |
+
"step": 30
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 1.99,
|
| 42 |
+
"learning_rate": 0.00012,
|
| 43 |
+
"loss": 1.4353,
|
| 44 |
+
"step": 40
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 1.99,
|
| 48 |
+
"eval_accuracy": 0.44899536321483774,
|
| 49 |
+
"eval_f1": 0.27913559166246477,
|
| 50 |
+
"eval_loss": 1.378554105758667,
|
| 51 |
+
"eval_precision": 0.3849846576082604,
|
| 52 |
+
"eval_recall": 0.44899536321483774,
|
| 53 |
+
"eval_runtime": 11.7486,
|
| 54 |
+
"eval_samples_per_second": 110.141,
|
| 55 |
+
"eval_steps_per_second": 1.787,
|
| 56 |
+
"step": 40
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 2.49,
|
| 60 |
+
"learning_rate": 0.00011666666666666667,
|
| 61 |
+
"loss": 1.3861,
|
| 62 |
+
"step": 50
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 2.99,
|
| 66 |
+
"learning_rate": 0.00011333333333333333,
|
| 67 |
+
"loss": 1.3026,
|
| 68 |
+
"step": 60
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 2.99,
|
| 72 |
+
"eval_accuracy": 0.47990726429675423,
|
| 73 |
+
"eval_f1": 0.33925239665034806,
|
| 74 |
+
"eval_loss": 1.2733628749847412,
|
| 75 |
+
"eval_precision": 0.3072903900501765,
|
| 76 |
+
"eval_recall": 0.47990726429675423,
|
| 77 |
+
"eval_runtime": 11.82,
|
| 78 |
+
"eval_samples_per_second": 109.475,
|
| 79 |
+
"eval_steps_per_second": 1.777,
|
| 80 |
+
"step": 60
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 3.49,
|
| 84 |
+
"learning_rate": 0.00011,
|
| 85 |
+
"loss": 1.2728,
|
| 86 |
+
"step": 70
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 3.99,
|
| 90 |
+
"learning_rate": 0.00010666666666666667,
|
| 91 |
+
"loss": 1.1579,
|
| 92 |
+
"step": 80
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 3.99,
|
| 96 |
+
"eval_accuracy": 0.5278207109737248,
|
| 97 |
+
"eval_f1": 0.41425802390052474,
|
| 98 |
+
"eval_loss": 1.137792706489563,
|
| 99 |
+
"eval_precision": 0.4299797234929115,
|
| 100 |
+
"eval_recall": 0.5278207109737248,
|
| 101 |
+
"eval_runtime": 11.7797,
|
| 102 |
+
"eval_samples_per_second": 109.85,
|
| 103 |
+
"eval_steps_per_second": 1.783,
|
| 104 |
+
"step": 80
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 4.49,
|
| 108 |
+
"learning_rate": 0.00010333333333333334,
|
| 109 |
+
"loss": 1.1096,
|
| 110 |
+
"step": 90
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 4.99,
|
| 114 |
+
"learning_rate": 0.0001,
|
| 115 |
+
"loss": 1.0276,
|
| 116 |
+
"step": 100
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 4.99,
|
| 120 |
+
"eval_accuracy": 0.5734157650695518,
|
| 121 |
+
"eval_f1": 0.4864664230611172,
|
| 122 |
+
"eval_loss": 1.0231207609176636,
|
| 123 |
+
"eval_precision": 0.449744448551538,
|
| 124 |
+
"eval_recall": 0.5734157650695518,
|
| 125 |
+
"eval_runtime": 11.7657,
|
| 126 |
+
"eval_samples_per_second": 109.98,
|
| 127 |
+
"eval_steps_per_second": 1.785,
|
| 128 |
+
"step": 100
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 5.49,
|
| 132 |
+
"learning_rate": 9.666666666666667e-05,
|
| 133 |
+
"loss": 0.9751,
|
| 134 |
+
"step": 110
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 5.99,
|
| 138 |
+
"learning_rate": 9.333333333333334e-05,
|
| 139 |
+
"loss": 0.8826,
|
| 140 |
+
"step": 120
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 5.99,
|
| 144 |
+
"eval_accuracy": 0.625193199381762,
|
| 145 |
+
"eval_f1": 0.5636892396732486,
|
| 146 |
+
"eval_loss": 0.9227971434593201,
|
| 147 |
+
"eval_precision": 0.5982698615367767,
|
| 148 |
+
"eval_recall": 0.625193199381762,
|
| 149 |
+
"eval_runtime": 12.7817,
|
| 150 |
+
"eval_samples_per_second": 101.239,
|
| 151 |
+
"eval_steps_per_second": 1.643,
|
| 152 |
+
"step": 120
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 6.49,
|
| 156 |
+
"learning_rate": 9e-05,
|
| 157 |
+
"loss": 0.8232,
|
| 158 |
+
"step": 130
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 6.99,
|
| 162 |
+
"learning_rate": 8.666666666666667e-05,
|
| 163 |
+
"loss": 0.766,
|
| 164 |
+
"step": 140
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 6.99,
|
| 168 |
+
"eval_accuracy": 0.6661514683153014,
|
| 169 |
+
"eval_f1": 0.6320349212780727,
|
| 170 |
+
"eval_loss": 0.8440618515014648,
|
| 171 |
+
"eval_precision": 0.647375804405681,
|
| 172 |
+
"eval_recall": 0.6661514683153014,
|
| 173 |
+
"eval_runtime": 11.8583,
|
| 174 |
+
"eval_samples_per_second": 109.122,
|
| 175 |
+
"eval_steps_per_second": 1.771,
|
| 176 |
+
"step": 140
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 7.49,
|
| 180 |
+
"learning_rate": 8.333333333333333e-05,
|
| 181 |
+
"loss": 0.7145,
|
| 182 |
+
"step": 150
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 7.99,
|
| 186 |
+
"learning_rate": 7.999999999999999e-05,
|
| 187 |
+
"loss": 0.6732,
|
| 188 |
+
"step": 160
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 7.99,
|
| 192 |
+
"eval_accuracy": 0.6901081916537867,
|
| 193 |
+
"eval_f1": 0.6704100670249821,
|
| 194 |
+
"eval_loss": 0.8009499311447144,
|
| 195 |
+
"eval_precision": 0.6758956033590393,
|
| 196 |
+
"eval_recall": 0.6901081916537867,
|
| 197 |
+
"eval_runtime": 11.8229,
|
| 198 |
+
"eval_samples_per_second": 109.449,
|
| 199 |
+
"eval_steps_per_second": 1.776,
|
| 200 |
+
"step": 160
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 8.49,
|
| 204 |
+
"learning_rate": 7.666666666666667e-05,
|
| 205 |
+
"loss": 0.635,
|
| 206 |
+
"step": 170
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 8.99,
|
| 210 |
+
"learning_rate": 7.333333333333334e-05,
|
| 211 |
+
"loss": 0.5653,
|
| 212 |
+
"step": 180
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 8.99,
|
| 216 |
+
"eval_accuracy": 0.7217928902627512,
|
| 217 |
+
"eval_f1": 0.7129313347452736,
|
| 218 |
+
"eval_loss": 0.7535430192947388,
|
| 219 |
+
"eval_precision": 0.7141409959676454,
|
| 220 |
+
"eval_recall": 0.7217928902627512,
|
| 221 |
+
"eval_runtime": 11.8519,
|
| 222 |
+
"eval_samples_per_second": 109.181,
|
| 223 |
+
"eval_steps_per_second": 1.772,
|
| 224 |
+
"step": 180
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 9.49,
|
| 228 |
+
"learning_rate": 7.000000000000001e-05,
|
| 229 |
+
"loss": 0.5358,
|
| 230 |
+
"step": 190
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 9.99,
|
| 234 |
+
"learning_rate": 6.666666666666667e-05,
|
| 235 |
+
"loss": 0.4957,
|
| 236 |
+
"step": 200
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 9.99,
|
| 240 |
+
"eval_accuracy": 0.7256568778979907,
|
| 241 |
+
"eval_f1": 0.7199928490312443,
|
| 242 |
+
"eval_loss": 0.731657087802887,
|
| 243 |
+
"eval_precision": 0.7247700086876532,
|
| 244 |
+
"eval_recall": 0.7256568778979907,
|
| 245 |
+
"eval_runtime": 12.0809,
|
| 246 |
+
"eval_samples_per_second": 107.111,
|
| 247 |
+
"eval_steps_per_second": 1.738,
|
| 248 |
+
"step": 200
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 10.49,
|
| 252 |
+
"learning_rate": 6.333333333333333e-05,
|
| 253 |
+
"loss": 0.471,
|
| 254 |
+
"step": 210
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 10.99,
|
| 258 |
+
"learning_rate": 6e-05,
|
| 259 |
+
"loss": 0.4534,
|
| 260 |
+
"step": 220
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 10.99,
|
| 264 |
+
"eval_accuracy": 0.7434312210200927,
|
| 265 |
+
"eval_f1": 0.7389905178601613,
|
| 266 |
+
"eval_loss": 0.6807736158370972,
|
| 267 |
+
"eval_precision": 0.7405180153510328,
|
| 268 |
+
"eval_recall": 0.7434312210200927,
|
| 269 |
+
"eval_runtime": 12.457,
|
| 270 |
+
"eval_samples_per_second": 103.877,
|
| 271 |
+
"eval_steps_per_second": 1.686,
|
| 272 |
+
"step": 220
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 11.49,
|
| 276 |
+
"learning_rate": 5.6666666666666664e-05,
|
| 277 |
+
"loss": 0.4102,
|
| 278 |
+
"step": 230
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 11.99,
|
| 282 |
+
"learning_rate": 5.333333333333333e-05,
|
| 283 |
+
"loss": 0.3792,
|
| 284 |
+
"step": 240
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 11.99,
|
| 288 |
+
"eval_accuracy": 0.7449768160741885,
|
| 289 |
+
"eval_f1": 0.7398936491076405,
|
| 290 |
+
"eval_loss": 0.694913387298584,
|
| 291 |
+
"eval_precision": 0.7453650213081748,
|
| 292 |
+
"eval_recall": 0.7449768160741885,
|
| 293 |
+
"eval_runtime": 11.855,
|
| 294 |
+
"eval_samples_per_second": 109.153,
|
| 295 |
+
"eval_steps_per_second": 1.771,
|
| 296 |
+
"step": 240
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 12.49,
|
| 300 |
+
"learning_rate": 5e-05,
|
| 301 |
+
"loss": 0.3766,
|
| 302 |
+
"step": 250
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 12.99,
|
| 306 |
+
"learning_rate": 4.666666666666667e-05,
|
| 307 |
+
"loss": 0.3489,
|
| 308 |
+
"step": 260
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 12.99,
|
| 312 |
+
"eval_accuracy": 0.749613601236476,
|
| 313 |
+
"eval_f1": 0.7474367490892962,
|
| 314 |
+
"eval_loss": 0.6745789647102356,
|
| 315 |
+
"eval_precision": 0.751086737820255,
|
| 316 |
+
"eval_recall": 0.749613601236476,
|
| 317 |
+
"eval_runtime": 11.7383,
|
| 318 |
+
"eval_samples_per_second": 110.237,
|
| 319 |
+
"eval_steps_per_second": 1.789,
|
| 320 |
+
"step": 260
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 13.49,
|
| 324 |
+
"learning_rate": 4.3333333333333334e-05,
|
| 325 |
+
"loss": 0.3487,
|
| 326 |
+
"step": 270
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 13.99,
|
| 330 |
+
"learning_rate": 3.9999999999999996e-05,
|
| 331 |
+
"loss": 0.3113,
|
| 332 |
+
"step": 280
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 13.99,
|
| 336 |
+
"eval_accuracy": 0.7573415765069552,
|
| 337 |
+
"eval_f1": 0.7578669772980416,
|
| 338 |
+
"eval_loss": 0.6637156009674072,
|
| 339 |
+
"eval_precision": 0.7638354392114911,
|
| 340 |
+
"eval_recall": 0.7573415765069552,
|
| 341 |
+
"eval_runtime": 11.864,
|
| 342 |
+
"eval_samples_per_second": 109.069,
|
| 343 |
+
"eval_steps_per_second": 1.77,
|
| 344 |
+
"step": 280
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 14.49,
|
| 348 |
+
"learning_rate": 3.666666666666667e-05,
|
| 349 |
+
"loss": 0.3036,
|
| 350 |
+
"step": 290
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 14.99,
|
| 354 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 355 |
+
"loss": 0.2947,
|
| 356 |
+
"step": 300
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 14.99,
|
| 360 |
+
"eval_accuracy": 0.758887171561051,
|
| 361 |
+
"eval_f1": 0.7609777447379339,
|
| 362 |
+
"eval_loss": 0.6451451182365417,
|
| 363 |
+
"eval_precision": 0.7666890251959252,
|
| 364 |
+
"eval_recall": 0.758887171561051,
|
| 365 |
+
"eval_runtime": 11.5648,
|
| 366 |
+
"eval_samples_per_second": 111.891,
|
| 367 |
+
"eval_steps_per_second": 1.816,
|
| 368 |
+
"step": 300
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 15.49,
|
| 372 |
+
"learning_rate": 3e-05,
|
| 373 |
+
"loss": 0.2748,
|
| 374 |
+
"step": 310
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 15.99,
|
| 378 |
+
"learning_rate": 2.6666666666666667e-05,
|
| 379 |
+
"loss": 0.2776,
|
| 380 |
+
"step": 320
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 15.99,
|
| 384 |
+
"eval_accuracy": 0.7542503863987635,
|
| 385 |
+
"eval_f1": 0.7525049520726458,
|
| 386 |
+
"eval_loss": 0.6754108667373657,
|
| 387 |
+
"eval_precision": 0.7565187938926003,
|
| 388 |
+
"eval_recall": 0.7542503863987635,
|
| 389 |
+
"eval_runtime": 12.7697,
|
| 390 |
+
"eval_samples_per_second": 101.334,
|
| 391 |
+
"eval_steps_per_second": 1.645,
|
| 392 |
+
"step": 320
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 16.49,
|
| 396 |
+
"learning_rate": 2.3333333333333336e-05,
|
| 397 |
+
"loss": 0.272,
|
| 398 |
+
"step": 330
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 16.99,
|
| 402 |
+
"learning_rate": 1.9999999999999998e-05,
|
| 403 |
+
"loss": 0.2611,
|
| 404 |
+
"step": 340
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 16.99,
|
| 408 |
+
"eval_accuracy": 0.7550231839258115,
|
| 409 |
+
"eval_f1": 0.7529156321916382,
|
| 410 |
+
"eval_loss": 0.6808292269706726,
|
| 411 |
+
"eval_precision": 0.7606561956666212,
|
| 412 |
+
"eval_recall": 0.7550231839258115,
|
| 413 |
+
"eval_runtime": 11.6273,
|
| 414 |
+
"eval_samples_per_second": 111.29,
|
| 415 |
+
"eval_steps_per_second": 1.806,
|
| 416 |
+
"step": 340
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 17.49,
|
| 420 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 421 |
+
"loss": 0.2508,
|
| 422 |
+
"step": 350
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 17.99,
|
| 426 |
+
"learning_rate": 1.3333333333333333e-05,
|
| 427 |
+
"loss": 0.2428,
|
| 428 |
+
"step": 360
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 17.99,
|
| 432 |
+
"eval_accuracy": 0.7457496136012365,
|
| 433 |
+
"eval_f1": 0.7403755418120831,
|
| 434 |
+
"eval_loss": 0.7005118727684021,
|
| 435 |
+
"eval_precision": 0.7497269917356472,
|
| 436 |
+
"eval_recall": 0.7457496136012365,
|
| 437 |
+
"eval_runtime": 11.7515,
|
| 438 |
+
"eval_samples_per_second": 110.114,
|
| 439 |
+
"eval_steps_per_second": 1.787,
|
| 440 |
+
"step": 360
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 18.49,
|
| 444 |
+
"learning_rate": 9.999999999999999e-06,
|
| 445 |
+
"loss": 0.2479,
|
| 446 |
+
"step": 370
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 18.99,
|
| 450 |
+
"learning_rate": 6.666666666666667e-06,
|
| 451 |
+
"loss": 0.2346,
|
| 452 |
+
"step": 380
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 18.99,
|
| 456 |
+
"eval_accuracy": 0.7573415765069552,
|
| 457 |
+
"eval_f1": 0.7590201528020784,
|
| 458 |
+
"eval_loss": 0.6597253680229187,
|
| 459 |
+
"eval_precision": 0.7642128778135023,
|
| 460 |
+
"eval_recall": 0.7573415765069552,
|
| 461 |
+
"eval_runtime": 11.869,
|
| 462 |
+
"eval_samples_per_second": 109.024,
|
| 463 |
+
"eval_steps_per_second": 1.769,
|
| 464 |
+
"step": 380
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 19.49,
|
| 468 |
+
"learning_rate": 3.3333333333333333e-06,
|
| 469 |
+
"loss": 0.2312,
|
| 470 |
+
"step": 390
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 19.99,
|
| 474 |
+
"learning_rate": 0.0,
|
| 475 |
+
"loss": 0.2367,
|
| 476 |
+
"step": 400
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 19.99,
|
| 480 |
+
"eval_accuracy": 0.7557959814528593,
|
| 481 |
+
"eval_f1": 0.7545674798253312,
|
| 482 |
+
"eval_loss": 0.6771848797798157,
|
| 483 |
+
"eval_precision": 0.7556690736625777,
|
| 484 |
+
"eval_recall": 0.7557959814528593,
|
| 485 |
+
"eval_runtime": 12.018,
|
| 486 |
+
"eval_samples_per_second": 107.672,
|
| 487 |
+
"eval_steps_per_second": 1.747,
|
| 488 |
+
"step": 400
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 19.99,
|
| 492 |
+
"step": 400,
|
| 493 |
+
"total_flos": 5.432231359979311e+18,
|
| 494 |
+
"train_loss": 0.6680345350503921,
|
| 495 |
+
"train_runtime": 1446.6101,
|
| 496 |
+
"train_samples_per_second": 71.533,
|
| 497 |
+
"train_steps_per_second": 0.277
|
| 498 |
+
}
|
| 499 |
+
],
|
| 500 |
+
"max_steps": 400,
|
| 501 |
+
"num_train_epochs": 20,
|
| 502 |
+
"total_flos": 5.432231359979311e+18,
|
| 503 |
+
"trial_name": null,
|
| 504 |
+
"trial_params": null
|
| 505 |
+
}
|