Uploaded Model
Browse files- all_results.json +16 -0
- eval_results.json +11 -0
- runs/Jun29_11-40-36_e2bc67d5ceb6/events.out.tfevents.1688039694.e2bc67d5ceb6.572.5 +3 -0
- train_results.json +8 -0
- trainer_state.json +487 -0
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4.0,
|
| 3 |
+
"eval_accuracy": 0.8735042735042735,
|
| 4 |
+
"eval_f1": 0.8735042735042735,
|
| 5 |
+
"eval_loss": 0.5306985974311829,
|
| 6 |
+
"eval_precision": 0.8735042735042735,
|
| 7 |
+
"eval_recall": 0.8735042735042735,
|
| 8 |
+
"eval_runtime": 11.9413,
|
| 9 |
+
"eval_samples_per_second": 48.989,
|
| 10 |
+
"eval_steps_per_second": 6.197,
|
| 11 |
+
"total_flos": 2.410620782808269e+18,
|
| 12 |
+
"train_loss": 0.11269745317058287,
|
| 13 |
+
"train_runtime": 784.6004,
|
| 14 |
+
"train_samples_per_second": 11.216,
|
| 15 |
+
"train_steps_per_second": 0.704
|
| 16 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4.0,
|
| 3 |
+
"eval_accuracy": 0.8735042735042735,
|
| 4 |
+
"eval_f1": 0.8735042735042735,
|
| 5 |
+
"eval_loss": 0.5306985974311829,
|
| 6 |
+
"eval_precision": 0.8735042735042735,
|
| 7 |
+
"eval_recall": 0.8735042735042735,
|
| 8 |
+
"eval_runtime": 11.9413,
|
| 9 |
+
"eval_samples_per_second": 48.989,
|
| 10 |
+
"eval_steps_per_second": 6.197
|
| 11 |
+
}
|
runs/Jun29_11-40-36_e2bc67d5ceb6/events.out.tfevents.1688039694.e2bc67d5ceb6.572.5
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:44bddb0215846f2e4876146dc6fcafe85ef4d64496df019c93698ce586db3d51
|
| 3 |
+
size 560
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4.0,
|
| 3 |
+
"total_flos": 2.410620782808269e+18,
|
| 4 |
+
"train_loss": 0.11269745317058287,
|
| 5 |
+
"train_runtime": 784.6004,
|
| 6 |
+
"train_samples_per_second": 11.216,
|
| 7 |
+
"train_steps_per_second": 0.704
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,487 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.5306985974311829,
|
| 3 |
+
"best_model_checkpoint": "./vit-large-augmented-ph2-patch-16/checkpoint-50",
|
| 4 |
+
"epoch": 4.0,
|
| 5 |
+
"global_step": 552,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.07,
|
| 12 |
+
"learning_rate": 0.00019637681159420293,
|
| 13 |
+
"loss": 0.7939,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.14,
|
| 18 |
+
"learning_rate": 0.0001927536231884058,
|
| 19 |
+
"loss": 0.381,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.22,
|
| 24 |
+
"learning_rate": 0.0001891304347826087,
|
| 25 |
+
"loss": 0.4226,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.29,
|
| 30 |
+
"learning_rate": 0.00018586956521739133,
|
| 31 |
+
"loss": 0.4128,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.36,
|
| 36 |
+
"learning_rate": 0.00018224637681159422,
|
| 37 |
+
"loss": 0.2064,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.36,
|
| 42 |
+
"eval_accuracy": 0.8735042735042735,
|
| 43 |
+
"eval_f1": 0.8735042735042735,
|
| 44 |
+
"eval_loss": 0.5306985974311829,
|
| 45 |
+
"eval_precision": 0.8735042735042735,
|
| 46 |
+
"eval_recall": 0.8735042735042735,
|
| 47 |
+
"eval_runtime": 11.8033,
|
| 48 |
+
"eval_samples_per_second": 49.562,
|
| 49 |
+
"eval_steps_per_second": 6.269,
|
| 50 |
+
"step": 50
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.43,
|
| 54 |
+
"learning_rate": 0.0001786231884057971,
|
| 55 |
+
"loss": 0.132,
|
| 56 |
+
"step": 60
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.51,
|
| 60 |
+
"learning_rate": 0.000175,
|
| 61 |
+
"loss": 0.1388,
|
| 62 |
+
"step": 70
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.58,
|
| 66 |
+
"learning_rate": 0.00017137681159420292,
|
| 67 |
+
"loss": 0.1381,
|
| 68 |
+
"step": 80
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.65,
|
| 72 |
+
"learning_rate": 0.0001677536231884058,
|
| 73 |
+
"loss": 0.1244,
|
| 74 |
+
"step": 90
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.72,
|
| 78 |
+
"learning_rate": 0.0001641304347826087,
|
| 79 |
+
"loss": 0.1145,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.72,
|
| 84 |
+
"eval_accuracy": 0.747008547008547,
|
| 85 |
+
"eval_f1": 0.747008547008547,
|
| 86 |
+
"eval_loss": 0.8837085962295532,
|
| 87 |
+
"eval_precision": 0.747008547008547,
|
| 88 |
+
"eval_recall": 0.747008547008547,
|
| 89 |
+
"eval_runtime": 11.7447,
|
| 90 |
+
"eval_samples_per_second": 49.81,
|
| 91 |
+
"eval_steps_per_second": 6.301,
|
| 92 |
+
"step": 100
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.8,
|
| 96 |
+
"learning_rate": 0.0001605072463768116,
|
| 97 |
+
"loss": 0.2013,
|
| 98 |
+
"step": 110
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.87,
|
| 102 |
+
"learning_rate": 0.0001568840579710145,
|
| 103 |
+
"loss": 0.1908,
|
| 104 |
+
"step": 120
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.94,
|
| 108 |
+
"learning_rate": 0.0001532608695652174,
|
| 109 |
+
"loss": 0.1901,
|
| 110 |
+
"step": 130
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 1.01,
|
| 114 |
+
"learning_rate": 0.0001496376811594203,
|
| 115 |
+
"loss": 0.364,
|
| 116 |
+
"step": 140
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 1.09,
|
| 120 |
+
"learning_rate": 0.0001460144927536232,
|
| 121 |
+
"loss": 0.4187,
|
| 122 |
+
"step": 150
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 1.09,
|
| 126 |
+
"eval_accuracy": 0.6256410256410256,
|
| 127 |
+
"eval_f1": 0.6256410256410256,
|
| 128 |
+
"eval_loss": 0.9484676718711853,
|
| 129 |
+
"eval_precision": 0.6256410256410256,
|
| 130 |
+
"eval_recall": 0.6256410256410256,
|
| 131 |
+
"eval_runtime": 11.6896,
|
| 132 |
+
"eval_samples_per_second": 50.044,
|
| 133 |
+
"eval_steps_per_second": 6.33,
|
| 134 |
+
"step": 150
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 1.16,
|
| 138 |
+
"learning_rate": 0.00014239130434782612,
|
| 139 |
+
"loss": 0.2026,
|
| 140 |
+
"step": 160
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 1.23,
|
| 144 |
+
"learning_rate": 0.00013876811594202898,
|
| 145 |
+
"loss": 0.1367,
|
| 146 |
+
"step": 170
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 1.3,
|
| 150 |
+
"learning_rate": 0.0001351449275362319,
|
| 151 |
+
"loss": 0.0528,
|
| 152 |
+
"step": 180
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 1.38,
|
| 156 |
+
"learning_rate": 0.00013152173913043478,
|
| 157 |
+
"loss": 0.2144,
|
| 158 |
+
"step": 190
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.45,
|
| 162 |
+
"learning_rate": 0.00012789855072463767,
|
| 163 |
+
"loss": 0.0756,
|
| 164 |
+
"step": 200
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 1.45,
|
| 168 |
+
"eval_accuracy": 0.8324786324786325,
|
| 169 |
+
"eval_f1": 0.8324786324786324,
|
| 170 |
+
"eval_loss": 0.6959302425384521,
|
| 171 |
+
"eval_precision": 0.8324786324786325,
|
| 172 |
+
"eval_recall": 0.8324786324786325,
|
| 173 |
+
"eval_runtime": 12.2171,
|
| 174 |
+
"eval_samples_per_second": 47.884,
|
| 175 |
+
"eval_steps_per_second": 6.057,
|
| 176 |
+
"step": 200
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 1.52,
|
| 180 |
+
"learning_rate": 0.0001242753623188406,
|
| 181 |
+
"loss": 0.106,
|
| 182 |
+
"step": 210
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 1.59,
|
| 186 |
+
"learning_rate": 0.00012065217391304348,
|
| 187 |
+
"loss": 0.1342,
|
| 188 |
+
"step": 220
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 1.67,
|
| 192 |
+
"learning_rate": 0.0001170289855072464,
|
| 193 |
+
"loss": 0.1399,
|
| 194 |
+
"step": 230
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 1.74,
|
| 198 |
+
"learning_rate": 0.00011340579710144927,
|
| 199 |
+
"loss": 0.1476,
|
| 200 |
+
"step": 240
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.81,
|
| 204 |
+
"learning_rate": 0.00010978260869565218,
|
| 205 |
+
"loss": 0.0696,
|
| 206 |
+
"step": 250
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 1.81,
|
| 210 |
+
"eval_accuracy": 0.8170940170940171,
|
| 211 |
+
"eval_f1": 0.8170940170940171,
|
| 212 |
+
"eval_loss": 0.7697324156761169,
|
| 213 |
+
"eval_precision": 0.8170940170940171,
|
| 214 |
+
"eval_recall": 0.8170940170940171,
|
| 215 |
+
"eval_runtime": 11.717,
|
| 216 |
+
"eval_samples_per_second": 49.927,
|
| 217 |
+
"eval_steps_per_second": 6.316,
|
| 218 |
+
"step": 250
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 1.88,
|
| 222 |
+
"learning_rate": 0.00010615942028985508,
|
| 223 |
+
"loss": 0.2838,
|
| 224 |
+
"step": 260
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 1.96,
|
| 228 |
+
"learning_rate": 0.00010253623188405797,
|
| 229 |
+
"loss": 0.0627,
|
| 230 |
+
"step": 270
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 2.03,
|
| 234 |
+
"learning_rate": 9.891304347826087e-05,
|
| 235 |
+
"loss": 0.1012,
|
| 236 |
+
"step": 280
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 2.1,
|
| 240 |
+
"learning_rate": 9.528985507246378e-05,
|
| 241 |
+
"loss": 0.0562,
|
| 242 |
+
"step": 290
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 2.17,
|
| 246 |
+
"learning_rate": 9.166666666666667e-05,
|
| 247 |
+
"loss": 0.0251,
|
| 248 |
+
"step": 300
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 2.17,
|
| 252 |
+
"eval_accuracy": 0.8324786324786325,
|
| 253 |
+
"eval_f1": 0.8324786324786324,
|
| 254 |
+
"eval_loss": 0.7361409664154053,
|
| 255 |
+
"eval_precision": 0.8324786324786325,
|
| 256 |
+
"eval_recall": 0.8324786324786325,
|
| 257 |
+
"eval_runtime": 11.7369,
|
| 258 |
+
"eval_samples_per_second": 49.843,
|
| 259 |
+
"eval_steps_per_second": 6.305,
|
| 260 |
+
"step": 300
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 2.25,
|
| 264 |
+
"learning_rate": 8.804347826086957e-05,
|
| 265 |
+
"loss": 0.0332,
|
| 266 |
+
"step": 310
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 2.32,
|
| 270 |
+
"learning_rate": 8.442028985507246e-05,
|
| 271 |
+
"loss": 0.0024,
|
| 272 |
+
"step": 320
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 2.39,
|
| 276 |
+
"learning_rate": 8.079710144927536e-05,
|
| 277 |
+
"loss": 0.002,
|
| 278 |
+
"step": 330
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 2.46,
|
| 282 |
+
"learning_rate": 7.717391304347827e-05,
|
| 283 |
+
"loss": 0.0019,
|
| 284 |
+
"step": 340
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 2.54,
|
| 288 |
+
"learning_rate": 7.355072463768117e-05,
|
| 289 |
+
"loss": 0.0604,
|
| 290 |
+
"step": 350
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 2.54,
|
| 294 |
+
"eval_accuracy": 0.8427350427350427,
|
| 295 |
+
"eval_f1": 0.8427350427350426,
|
| 296 |
+
"eval_loss": 0.9344781637191772,
|
| 297 |
+
"eval_precision": 0.8427350427350427,
|
| 298 |
+
"eval_recall": 0.8427350427350427,
|
| 299 |
+
"eval_runtime": 11.7927,
|
| 300 |
+
"eval_samples_per_second": 49.607,
|
| 301 |
+
"eval_steps_per_second": 6.275,
|
| 302 |
+
"step": 350
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 2.61,
|
| 306 |
+
"learning_rate": 6.992753623188406e-05,
|
| 307 |
+
"loss": 0.0097,
|
| 308 |
+
"step": 360
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 2.68,
|
| 312 |
+
"learning_rate": 6.630434782608695e-05,
|
| 313 |
+
"loss": 0.0034,
|
| 314 |
+
"step": 370
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.75,
|
| 318 |
+
"learning_rate": 6.268115942028985e-05,
|
| 319 |
+
"loss": 0.0225,
|
| 320 |
+
"step": 380
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 2.83,
|
| 324 |
+
"learning_rate": 5.905797101449275e-05,
|
| 325 |
+
"loss": 0.0351,
|
| 326 |
+
"step": 390
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 2.9,
|
| 330 |
+
"learning_rate": 5.5434782608695654e-05,
|
| 331 |
+
"loss": 0.0005,
|
| 332 |
+
"step": 400
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 2.9,
|
| 336 |
+
"eval_accuracy": 0.8512820512820513,
|
| 337 |
+
"eval_f1": 0.8512820512820513,
|
| 338 |
+
"eval_loss": 0.9581409692764282,
|
| 339 |
+
"eval_precision": 0.8512820512820513,
|
| 340 |
+
"eval_recall": 0.8512820512820513,
|
| 341 |
+
"eval_runtime": 11.7794,
|
| 342 |
+
"eval_samples_per_second": 49.663,
|
| 343 |
+
"eval_steps_per_second": 6.282,
|
| 344 |
+
"step": 400
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 2.97,
|
| 348 |
+
"learning_rate": 5.181159420289855e-05,
|
| 349 |
+
"loss": 0.0011,
|
| 350 |
+
"step": 410
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 3.04,
|
| 354 |
+
"learning_rate": 4.818840579710145e-05,
|
| 355 |
+
"loss": 0.0008,
|
| 356 |
+
"step": 420
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 3.12,
|
| 360 |
+
"learning_rate": 4.456521739130435e-05,
|
| 361 |
+
"loss": 0.0004,
|
| 362 |
+
"step": 430
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 3.19,
|
| 366 |
+
"learning_rate": 4.094202898550725e-05,
|
| 367 |
+
"loss": 0.0011,
|
| 368 |
+
"step": 440
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 3.26,
|
| 372 |
+
"learning_rate": 3.7318840579710144e-05,
|
| 373 |
+
"loss": 0.0005,
|
| 374 |
+
"step": 450
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 3.26,
|
| 378 |
+
"eval_accuracy": 0.8444444444444444,
|
| 379 |
+
"eval_f1": 0.8444444444444444,
|
| 380 |
+
"eval_loss": 1.0674240589141846,
|
| 381 |
+
"eval_precision": 0.8444444444444444,
|
| 382 |
+
"eval_recall": 0.8444444444444444,
|
| 383 |
+
"eval_runtime": 11.718,
|
| 384 |
+
"eval_samples_per_second": 49.923,
|
| 385 |
+
"eval_steps_per_second": 6.315,
|
| 386 |
+
"step": 450
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 3.33,
|
| 390 |
+
"learning_rate": 3.369565217391305e-05,
|
| 391 |
+
"loss": 0.0004,
|
| 392 |
+
"step": 460
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 3.41,
|
| 396 |
+
"learning_rate": 3.0072463768115944e-05,
|
| 397 |
+
"loss": 0.001,
|
| 398 |
+
"step": 470
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 3.48,
|
| 402 |
+
"learning_rate": 2.6449275362318844e-05,
|
| 403 |
+
"loss": 0.0003,
|
| 404 |
+
"step": 480
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 3.55,
|
| 408 |
+
"learning_rate": 2.282608695652174e-05,
|
| 409 |
+
"loss": 0.0002,
|
| 410 |
+
"step": 490
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 3.62,
|
| 414 |
+
"learning_rate": 1.9202898550724637e-05,
|
| 415 |
+
"loss": 0.005,
|
| 416 |
+
"step": 500
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 3.62,
|
| 420 |
+
"eval_accuracy": 0.8564102564102564,
|
| 421 |
+
"eval_f1": 0.8564102564102564,
|
| 422 |
+
"eval_loss": 0.9463948607444763,
|
| 423 |
+
"eval_precision": 0.8564102564102564,
|
| 424 |
+
"eval_recall": 0.8564102564102564,
|
| 425 |
+
"eval_runtime": 11.7058,
|
| 426 |
+
"eval_samples_per_second": 49.975,
|
| 427 |
+
"eval_steps_per_second": 6.322,
|
| 428 |
+
"step": 500
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 3.7,
|
| 432 |
+
"learning_rate": 1.5579710144927537e-05,
|
| 433 |
+
"loss": 0.0002,
|
| 434 |
+
"step": 510
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 3.77,
|
| 438 |
+
"learning_rate": 1.1956521739130435e-05,
|
| 439 |
+
"loss": 0.0002,
|
| 440 |
+
"step": 520
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 3.84,
|
| 444 |
+
"learning_rate": 8.333333333333334e-06,
|
| 445 |
+
"loss": 0.0002,
|
| 446 |
+
"step": 530
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 3.91,
|
| 450 |
+
"learning_rate": 4.710144927536232e-06,
|
| 451 |
+
"loss": 0.0002,
|
| 452 |
+
"step": 540
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 3.99,
|
| 456 |
+
"learning_rate": 1.0869565217391306e-06,
|
| 457 |
+
"loss": 0.0002,
|
| 458 |
+
"step": 550
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 3.99,
|
| 462 |
+
"eval_accuracy": 0.8564102564102564,
|
| 463 |
+
"eval_f1": 0.8564102564102564,
|
| 464 |
+
"eval_loss": 0.9574871063232422,
|
| 465 |
+
"eval_precision": 0.8564102564102564,
|
| 466 |
+
"eval_recall": 0.8564102564102564,
|
| 467 |
+
"eval_runtime": 11.6893,
|
| 468 |
+
"eval_samples_per_second": 50.046,
|
| 469 |
+
"eval_steps_per_second": 6.331,
|
| 470 |
+
"step": 550
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 4.0,
|
| 474 |
+
"step": 552,
|
| 475 |
+
"total_flos": 2.410620782808269e+18,
|
| 476 |
+
"train_loss": 0.11269745317058287,
|
| 477 |
+
"train_runtime": 784.6004,
|
| 478 |
+
"train_samples_per_second": 11.216,
|
| 479 |
+
"train_steps_per_second": 0.704
|
| 480 |
+
}
|
| 481 |
+
],
|
| 482 |
+
"max_steps": 552,
|
| 483 |
+
"num_train_epochs": 4,
|
| 484 |
+
"total_flos": 2.410620782808269e+18,
|
| 485 |
+
"trial_name": null,
|
| 486 |
+
"trial_params": null
|
| 487 |
+
}
|