End of training
Browse files- all_results.json +17 -0
- eval_results.json +12 -0
- train_results.json +8 -0
- trainer_state.json +451 -0
all_results.json
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"eval_accuracy": 0.983217164440637,
|
| 4 |
+
"eval_f1": 0.8968406938741306,
|
| 5 |
+
"eval_loss": 0.0666249617934227,
|
| 6 |
+
"eval_precision": 0.8931730929727926,
|
| 7 |
+
"eval_recall": 0.9005385392123864,
|
| 8 |
+
"eval_runtime": 21.6325,
|
| 9 |
+
"eval_samples": 3250,
|
| 10 |
+
"eval_samples_per_second": 150.237,
|
| 11 |
+
"eval_steps_per_second": 4.715,
|
| 12 |
+
"train_loss": 0.09878404265384196,
|
| 13 |
+
"train_runtime": 1144.6421,
|
| 14 |
+
"train_samples": 14041,
|
| 15 |
+
"train_samples_per_second": 24.533,
|
| 16 |
+
"train_steps_per_second": 0.767
|
| 17 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"eval_accuracy": 0.983217164440637,
|
| 4 |
+
"eval_f1": 0.8968406938741306,
|
| 5 |
+
"eval_loss": 0.0666249617934227,
|
| 6 |
+
"eval_precision": 0.8931730929727926,
|
| 7 |
+
"eval_recall": 0.9005385392123864,
|
| 8 |
+
"eval_runtime": 21.6325,
|
| 9 |
+
"eval_samples": 3250,
|
| 10 |
+
"eval_samples_per_second": 150.237,
|
| 11 |
+
"eval_steps_per_second": 4.715
|
| 12 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"train_loss": 0.09878404265384196,
|
| 4 |
+
"train_runtime": 1144.6421,
|
| 5 |
+
"train_samples": 14041,
|
| 6 |
+
"train_samples_per_second": 24.533,
|
| 7 |
+
"train_steps_per_second": 0.767
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,451 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.8968406938741306,
|
| 3 |
+
"best_model_checkpoint": "hmBERT-CoNLL-cp2/checkpoint-500",
|
| 4 |
+
"epoch": 2.0,
|
| 5 |
+
"global_step": 878,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.06,
|
| 12 |
+
"eval_accuracy": 0.9005295743935205,
|
| 13 |
+
"eval_f1": 0.36743180306054557,
|
| 14 |
+
"eval_loss": 0.4116312265396118,
|
| 15 |
+
"eval_precision": 0.36320289378493914,
|
| 16 |
+
"eval_recall": 0.3717603500504881,
|
| 17 |
+
"eval_runtime": 21.4683,
|
| 18 |
+
"eval_samples_per_second": 151.386,
|
| 19 |
+
"eval_steps_per_second": 4.751,
|
| 20 |
+
"step": 25
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.11,
|
| 24 |
+
"eval_accuracy": 0.9458549121918929,
|
| 25 |
+
"eval_f1": 0.6632702571567201,
|
| 26 |
+
"eval_loss": 0.22468078136444092,
|
| 27 |
+
"eval_precision": 0.638387297633873,
|
| 28 |
+
"eval_recall": 0.6901716593739482,
|
| 29 |
+
"eval_runtime": 20.8937,
|
| 30 |
+
"eval_samples_per_second": 155.549,
|
| 31 |
+
"eval_steps_per_second": 4.882,
|
| 32 |
+
"step": 50
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.17,
|
| 36 |
+
"eval_accuracy": 0.9579845021611308,
|
| 37 |
+
"eval_f1": 0.746131050378663,
|
| 38 |
+
"eval_loss": 0.16239947080612183,
|
| 39 |
+
"eval_precision": 0.7302610377054464,
|
| 40 |
+
"eval_recall": 0.7627061595422416,
|
| 41 |
+
"eval_runtime": 21.6024,
|
| 42 |
+
"eval_samples_per_second": 150.446,
|
| 43 |
+
"eval_steps_per_second": 4.722,
|
| 44 |
+
"step": 75
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.23,
|
| 48 |
+
"eval_accuracy": 0.9588022273275962,
|
| 49 |
+
"eval_f1": 0.7508835374373305,
|
| 50 |
+
"eval_loss": 0.15414278209209442,
|
| 51 |
+
"eval_precision": 0.7338152610441767,
|
| 52 |
+
"eval_recall": 0.7687647256815887,
|
| 53 |
+
"eval_runtime": 21.4049,
|
| 54 |
+
"eval_samples_per_second": 151.835,
|
| 55 |
+
"eval_steps_per_second": 4.765,
|
| 56 |
+
"step": 100
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.28,
|
| 60 |
+
"eval_accuracy": 0.9642537284373661,
|
| 61 |
+
"eval_f1": 0.7844736198320149,
|
| 62 |
+
"eval_loss": 0.13490867614746094,
|
| 63 |
+
"eval_precision": 0.7609555450087011,
|
| 64 |
+
"eval_recall": 0.8094917536183104,
|
| 65 |
+
"eval_runtime": 21.4103,
|
| 66 |
+
"eval_samples_per_second": 151.796,
|
| 67 |
+
"eval_steps_per_second": 4.764,
|
| 68 |
+
"step": 125
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.34,
|
| 72 |
+
"eval_accuracy": 0.9693547759043651,
|
| 73 |
+
"eval_f1": 0.8115174582161179,
|
| 74 |
+
"eval_loss": 0.12301833927631378,
|
| 75 |
+
"eval_precision": 0.7981770833333334,
|
| 76 |
+
"eval_recall": 0.8253113429821609,
|
| 77 |
+
"eval_runtime": 21.3799,
|
| 78 |
+
"eval_samples_per_second": 152.012,
|
| 79 |
+
"eval_steps_per_second": 4.771,
|
| 80 |
+
"step": 150
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.4,
|
| 84 |
+
"eval_accuracy": 0.97272302480433,
|
| 85 |
+
"eval_f1": 0.8234421364985163,
|
| 86 |
+
"eval_loss": 0.0997077226638794,
|
| 87 |
+
"eval_precision": 0.8069466882067852,
|
| 88 |
+
"eval_recall": 0.8406260518343992,
|
| 89 |
+
"eval_runtime": 21.0809,
|
| 90 |
+
"eval_samples_per_second": 154.168,
|
| 91 |
+
"eval_steps_per_second": 4.839,
|
| 92 |
+
"step": 175
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.46,
|
| 96 |
+
"eval_accuracy": 0.9731902963280246,
|
| 97 |
+
"eval_f1": 0.8308945793149318,
|
| 98 |
+
"eval_loss": 0.10443305224180222,
|
| 99 |
+
"eval_precision": 0.8210647387446599,
|
| 100 |
+
"eval_recall": 0.8409626388421407,
|
| 101 |
+
"eval_runtime": 21.7924,
|
| 102 |
+
"eval_samples_per_second": 149.135,
|
| 103 |
+
"eval_steps_per_second": 4.681,
|
| 104 |
+
"step": 200
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.51,
|
| 108 |
+
"eval_accuracy": 0.9759939254701919,
|
| 109 |
+
"eval_f1": 0.8507239141288069,
|
| 110 |
+
"eval_loss": 0.08712451159954071,
|
| 111 |
+
"eval_precision": 0.8413429888084266,
|
| 112 |
+
"eval_recall": 0.860316391787277,
|
| 113 |
+
"eval_runtime": 21.6266,
|
| 114 |
+
"eval_samples_per_second": 150.278,
|
| 115 |
+
"eval_steps_per_second": 4.716,
|
| 116 |
+
"step": 225
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.57,
|
| 120 |
+
"eval_accuracy": 0.9732876445621277,
|
| 121 |
+
"eval_f1": 0.8375655648988427,
|
| 122 |
+
"eval_loss": 0.10657607018947601,
|
| 123 |
+
"eval_precision": 0.8288021090789257,
|
| 124 |
+
"eval_recall": 0.8465163244698755,
|
| 125 |
+
"eval_runtime": 21.4498,
|
| 126 |
+
"eval_samples_per_second": 151.517,
|
| 127 |
+
"eval_steps_per_second": 4.755,
|
| 128 |
+
"step": 250
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.63,
|
| 132 |
+
"eval_accuracy": 0.9765780148748102,
|
| 133 |
+
"eval_f1": 0.8623576691225721,
|
| 134 |
+
"eval_loss": 0.08719414472579956,
|
| 135 |
+
"eval_precision": 0.8580473175608131,
|
| 136 |
+
"eval_recall": 0.8667115449343655,
|
| 137 |
+
"eval_runtime": 21.04,
|
| 138 |
+
"eval_samples_per_second": 154.468,
|
| 139 |
+
"eval_steps_per_second": 4.848,
|
| 140 |
+
"step": 275
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.68,
|
| 144 |
+
"eval_accuracy": 0.9773178614539932,
|
| 145 |
+
"eval_f1": 0.8613053613053614,
|
| 146 |
+
"eval_loss": 0.08342944085597992,
|
| 147 |
+
"eval_precision": 0.8522240527182866,
|
| 148 |
+
"eval_recall": 0.8705822955233928,
|
| 149 |
+
"eval_runtime": 20.9406,
|
| 150 |
+
"eval_samples_per_second": 155.201,
|
| 151 |
+
"eval_steps_per_second": 4.871,
|
| 152 |
+
"step": 300
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.74,
|
| 156 |
+
"eval_accuracy": 0.9782718741482029,
|
| 157 |
+
"eval_f1": 0.868680182043856,
|
| 158 |
+
"eval_loss": 0.08315661549568176,
|
| 159 |
+
"eval_precision": 0.8544685007325411,
|
| 160 |
+
"eval_recall": 0.8833726018175698,
|
| 161 |
+
"eval_runtime": 21.3907,
|
| 162 |
+
"eval_samples_per_second": 151.935,
|
| 163 |
+
"eval_steps_per_second": 4.768,
|
| 164 |
+
"step": 325
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.8,
|
| 168 |
+
"eval_accuracy": 0.9787002063782563,
|
| 169 |
+
"eval_f1": 0.8685364441135103,
|
| 170 |
+
"eval_loss": 0.07756651937961578,
|
| 171 |
+
"eval_precision": 0.8541903986981285,
|
| 172 |
+
"eval_recall": 0.8833726018175698,
|
| 173 |
+
"eval_runtime": 21.9914,
|
| 174 |
+
"eval_samples_per_second": 147.785,
|
| 175 |
+
"eval_steps_per_second": 4.638,
|
| 176 |
+
"step": 350
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.85,
|
| 180 |
+
"eval_accuracy": 0.98010202094934,
|
| 181 |
+
"eval_f1": 0.8760357971494862,
|
| 182 |
+
"eval_loss": 0.07601791620254517,
|
| 183 |
+
"eval_precision": 0.8628795298726738,
|
| 184 |
+
"eval_recall": 0.8895994614607876,
|
| 185 |
+
"eval_runtime": 21.6439,
|
| 186 |
+
"eval_samples_per_second": 150.158,
|
| 187 |
+
"eval_steps_per_second": 4.713,
|
| 188 |
+
"step": 375
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.91,
|
| 192 |
+
"eval_accuracy": 0.9823994392741716,
|
| 193 |
+
"eval_f1": 0.8887781377190795,
|
| 194 |
+
"eval_loss": 0.06729913502931595,
|
| 195 |
+
"eval_precision": 0.8774807282269969,
|
| 196 |
+
"eval_recall": 0.9003702457085156,
|
| 197 |
+
"eval_runtime": 20.8932,
|
| 198 |
+
"eval_samples_per_second": 155.553,
|
| 199 |
+
"eval_steps_per_second": 4.882,
|
| 200 |
+
"step": 400
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.97,
|
| 204 |
+
"eval_accuracy": 0.9816595926949885,
|
| 205 |
+
"eval_f1": 0.8882013546283134,
|
| 206 |
+
"eval_loss": 0.06811415404081345,
|
| 207 |
+
"eval_precision": 0.8826657802891806,
|
| 208 |
+
"eval_recall": 0.8938067990575563,
|
| 209 |
+
"eval_runtime": 21.516,
|
| 210 |
+
"eval_samples_per_second": 151.051,
|
| 211 |
+
"eval_steps_per_second": 4.741,
|
| 212 |
+
"step": 425
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 1.03,
|
| 216 |
+
"eval_accuracy": 0.9824189089209922,
|
| 217 |
+
"eval_f1": 0.8896695943120033,
|
| 218 |
+
"eval_loss": 0.06585536897182465,
|
| 219 |
+
"eval_precision": 0.8844170962913687,
|
| 220 |
+
"eval_recall": 0.8949848535846516,
|
| 221 |
+
"eval_runtime": 21.1661,
|
| 222 |
+
"eval_samples_per_second": 153.547,
|
| 223 |
+
"eval_steps_per_second": 4.819,
|
| 224 |
+
"step": 450
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 1.08,
|
| 228 |
+
"eval_accuracy": 0.9831587555001752,
|
| 229 |
+
"eval_f1": 0.8923128175231116,
|
| 230 |
+
"eval_loss": 0.06903570890426636,
|
| 231 |
+
"eval_precision": 0.8832646331409728,
|
| 232 |
+
"eval_recall": 0.9015483002356109,
|
| 233 |
+
"eval_runtime": 21.4553,
|
| 234 |
+
"eval_samples_per_second": 151.478,
|
| 235 |
+
"eval_steps_per_second": 4.754,
|
| 236 |
+
"step": 475
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 1.14,
|
| 240 |
+
"learning_rate": 2.152619589977221e-05,
|
| 241 |
+
"loss": 0.1399,
|
| 242 |
+
"step": 500
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.14,
|
| 246 |
+
"eval_accuracy": 0.983217164440637,
|
| 247 |
+
"eval_f1": 0.8968406938741306,
|
| 248 |
+
"eval_loss": 0.0666249617934227,
|
| 249 |
+
"eval_precision": 0.8931730929727926,
|
| 250 |
+
"eval_recall": 0.9005385392123864,
|
| 251 |
+
"eval_runtime": 21.9318,
|
| 252 |
+
"eval_samples_per_second": 148.187,
|
| 253 |
+
"eval_steps_per_second": 4.651,
|
| 254 |
+
"step": 500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 1.2,
|
| 258 |
+
"eval_accuracy": 0.982477317861454,
|
| 259 |
+
"eval_f1": 0.8943538268506901,
|
| 260 |
+
"eval_loss": 0.06665532290935516,
|
| 261 |
+
"eval_precision": 0.8890736737069682,
|
| 262 |
+
"eval_recall": 0.8996970716930327,
|
| 263 |
+
"eval_runtime": 21.7087,
|
| 264 |
+
"eval_samples_per_second": 149.71,
|
| 265 |
+
"eval_steps_per_second": 4.699,
|
| 266 |
+
"step": 525
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 1.25,
|
| 270 |
+
"eval_accuracy": 0.9820100463377595,
|
| 271 |
+
"eval_f1": 0.8851177106729889,
|
| 272 |
+
"eval_loss": 0.0698634684085846,
|
| 273 |
+
"eval_precision": 0.8751439381477216,
|
| 274 |
+
"eval_recall": 0.8953214405923932,
|
| 275 |
+
"eval_runtime": 21.8836,
|
| 276 |
+
"eval_samples_per_second": 148.513,
|
| 277 |
+
"eval_steps_per_second": 4.661,
|
| 278 |
+
"step": 550
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 1.31,
|
| 282 |
+
"eval_accuracy": 0.984015419960282,
|
| 283 |
+
"eval_f1": 0.900702106318957,
|
| 284 |
+
"eval_loss": 0.06173388287425041,
|
| 285 |
+
"eval_precision": 0.8947193623380937,
|
| 286 |
+
"eval_recall": 0.9067653988556041,
|
| 287 |
+
"eval_runtime": 21.5086,
|
| 288 |
+
"eval_samples_per_second": 151.102,
|
| 289 |
+
"eval_steps_per_second": 4.742,
|
| 290 |
+
"step": 575
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 1.37,
|
| 294 |
+
"eval_accuracy": 0.9840543592539231,
|
| 295 |
+
"eval_f1": 0.9028686462003019,
|
| 296 |
+
"eval_loss": 0.06334252655506134,
|
| 297 |
+
"eval_precision": 0.9,
|
| 298 |
+
"eval_recall": 0.9057556378323797,
|
| 299 |
+
"eval_runtime": 21.1745,
|
| 300 |
+
"eval_samples_per_second": 153.487,
|
| 301 |
+
"eval_steps_per_second": 4.817,
|
| 302 |
+
"step": 600
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 1.42,
|
| 306 |
+
"eval_accuracy": 0.9843269343094116,
|
| 307 |
+
"eval_f1": 0.9040387182910546,
|
| 308 |
+
"eval_loss": 0.06390959024429321,
|
| 309 |
+
"eval_precision": 0.8965574313141343,
|
| 310 |
+
"eval_recall": 0.9116459104678559,
|
| 311 |
+
"eval_runtime": 21.7996,
|
| 312 |
+
"eval_samples_per_second": 149.086,
|
| 313 |
+
"eval_steps_per_second": 4.679,
|
| 314 |
+
"step": 625
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 1.48,
|
| 318 |
+
"eval_accuracy": 0.9845021611307971,
|
| 319 |
+
"eval_f1": 0.9040501043841336,
|
| 320 |
+
"eval_loss": 0.06244850531220436,
|
| 321 |
+
"eval_precision": 0.8972318912647108,
|
| 322 |
+
"eval_recall": 0.910972736452373,
|
| 323 |
+
"eval_runtime": 21.6575,
|
| 324 |
+
"eval_samples_per_second": 150.064,
|
| 325 |
+
"eval_steps_per_second": 4.71,
|
| 326 |
+
"step": 650
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 1.54,
|
| 330 |
+
"eval_accuracy": 0.9841517074880262,
|
| 331 |
+
"eval_f1": 0.90302066772655,
|
| 332 |
+
"eval_loss": 0.06191794201731682,
|
| 333 |
+
"eval_precision": 0.8979863538026294,
|
| 334 |
+
"eval_recall": 0.9081117468865701,
|
| 335 |
+
"eval_runtime": 21.8896,
|
| 336 |
+
"eval_samples_per_second": 148.473,
|
| 337 |
+
"eval_steps_per_second": 4.66,
|
| 338 |
+
"step": 675
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 1.59,
|
| 342 |
+
"eval_accuracy": 0.9842685253689498,
|
| 343 |
+
"eval_f1": 0.904538603249037,
|
| 344 |
+
"eval_loss": 0.06151856482028961,
|
| 345 |
+
"eval_precision": 0.9001666666666667,
|
| 346 |
+
"eval_recall": 0.9089532144059239,
|
| 347 |
+
"eval_runtime": 21.9582,
|
| 348 |
+
"eval_samples_per_second": 148.008,
|
| 349 |
+
"eval_steps_per_second": 4.645,
|
| 350 |
+
"step": 700
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 1.65,
|
| 354 |
+
"eval_accuracy": 0.9849889023013123,
|
| 355 |
+
"eval_f1": 0.9082384460817148,
|
| 356 |
+
"eval_loss": 0.060051579028367996,
|
| 357 |
+
"eval_precision": 0.9036987670776407,
|
| 358 |
+
"eval_recall": 0.9128239649949512,
|
| 359 |
+
"eval_runtime": 21.5715,
|
| 360 |
+
"eval_samples_per_second": 150.661,
|
| 361 |
+
"eval_steps_per_second": 4.728,
|
| 362 |
+
"step": 725
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 1.71,
|
| 366 |
+
"eval_accuracy": 0.984949963007671,
|
| 367 |
+
"eval_f1": 0.9085891109810152,
|
| 368 |
+
"eval_loss": 0.058469709008932114,
|
| 369 |
+
"eval_precision": 0.9030756442227764,
|
| 370 |
+
"eval_recall": 0.9141703130259172,
|
| 371 |
+
"eval_runtime": 21.259,
|
| 372 |
+
"eval_samples_per_second": 152.877,
|
| 373 |
+
"eval_steps_per_second": 4.798,
|
| 374 |
+
"step": 750
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 1.77,
|
| 378 |
+
"eval_accuracy": 0.9850862505354153,
|
| 379 |
+
"eval_f1": 0.9089084065244667,
|
| 380 |
+
"eval_loss": 0.05824408307671547,
|
| 381 |
+
"eval_precision": 0.903542324962581,
|
| 382 |
+
"eval_recall": 0.9143386065297879,
|
| 383 |
+
"eval_runtime": 21.7846,
|
| 384 |
+
"eval_samples_per_second": 149.188,
|
| 385 |
+
"eval_steps_per_second": 4.682,
|
| 386 |
+
"step": 775
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 1.82,
|
| 390 |
+
"eval_accuracy": 0.9852809470036213,
|
| 391 |
+
"eval_f1": 0.9100183977253721,
|
| 392 |
+
"eval_loss": 0.0580282025039196,
|
| 393 |
+
"eval_precision": 0.9044215425531915,
|
| 394 |
+
"eval_recall": 0.9156849545607539,
|
| 395 |
+
"eval_runtime": 21.4108,
|
| 396 |
+
"eval_samples_per_second": 151.792,
|
| 397 |
+
"eval_steps_per_second": 4.764,
|
| 398 |
+
"step": 800
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 1.88,
|
| 402 |
+
"eval_accuracy": 0.9851446594758771,
|
| 403 |
+
"eval_f1": 0.9096682543661736,
|
| 404 |
+
"eval_loss": 0.058311279863119125,
|
| 405 |
+
"eval_precision": 0.903402489626556,
|
| 406 |
+
"eval_recall": 0.9160215415684955,
|
| 407 |
+
"eval_runtime": 21.1305,
|
| 408 |
+
"eval_samples_per_second": 153.806,
|
| 409 |
+
"eval_steps_per_second": 4.827,
|
| 410 |
+
"step": 825
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 1.94,
|
| 414 |
+
"eval_accuracy": 0.9853782952377244,
|
| 415 |
+
"eval_f1": 0.9113564141160729,
|
| 416 |
+
"eval_loss": 0.05776725709438324,
|
| 417 |
+
"eval_precision": 0.905751329787234,
|
| 418 |
+
"eval_recall": 0.9170313025917199,
|
| 419 |
+
"eval_runtime": 21.1284,
|
| 420 |
+
"eval_samples_per_second": 153.821,
|
| 421 |
+
"eval_steps_per_second": 4.828,
|
| 422 |
+
"step": 850
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 1.99,
|
| 426 |
+
"eval_accuracy": 0.9852420077099802,
|
| 427 |
+
"eval_f1": 0.9112356730527901,
|
| 428 |
+
"eval_loss": 0.05755457654595375,
|
| 429 |
+
"eval_precision": 0.9060056562967892,
|
| 430 |
+
"eval_recall": 0.9165264220801077,
|
| 431 |
+
"eval_runtime": 21.4936,
|
| 432 |
+
"eval_samples_per_second": 151.208,
|
| 433 |
+
"eval_steps_per_second": 4.746,
|
| 434 |
+
"step": 875
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 2.0,
|
| 438 |
+
"step": 878,
|
| 439 |
+
"total_flos": 913386421477854.0,
|
| 440 |
+
"train_loss": 0.09878404265384196,
|
| 441 |
+
"train_runtime": 1144.6421,
|
| 442 |
+
"train_samples_per_second": 24.533,
|
| 443 |
+
"train_steps_per_second": 0.767
|
| 444 |
+
}
|
| 445 |
+
],
|
| 446 |
+
"max_steps": 878,
|
| 447 |
+
"num_train_epochs": 2,
|
| 448 |
+
"total_flos": 913386421477854.0,
|
| 449 |
+
"trial_name": null,
|
| 450 |
+
"trial_params": null
|
| 451 |
+
}
|