End of training
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.99,
|
| 3 |
+
"eval_accuracy": 0.9638691322901849,
|
| 4 |
+
"eval_loss": 0.15102514624595642,
|
| 5 |
+
"eval_runtime": 44.5453,
|
| 6 |
+
"eval_samples_per_second": 157.817,
|
| 7 |
+
"eval_steps_per_second": 1.594,
|
| 8 |
+
"total_flos": 1.01609365487616e+18,
|
| 9 |
+
"train_loss": 0.711250480468766,
|
| 10 |
+
"train_runtime": 3427.9798,
|
| 11 |
+
"train_samples_per_second": 110.733,
|
| 12 |
+
"train_steps_per_second": 0.277
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.99,
|
| 3 |
+
"eval_accuracy": 0.9638691322901849,
|
| 4 |
+
"eval_loss": 0.15102514624595642,
|
| 5 |
+
"eval_runtime": 44.5453,
|
| 6 |
+
"eval_samples_per_second": 157.817,
|
| 7 |
+
"eval_steps_per_second": 1.594
|
| 8 |
+
}
|
runs/Dec22_09-06-12_17aa6734d17f/events.out.tfevents.1703239606.17aa6734d17f.604.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b7d16a9c46a6db572f85f1f3c75a707187ac2519aea905bc0d506d3f02ca2e5d
|
| 3 |
+
size 411
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.99,
|
| 3 |
+
"total_flos": 1.01609365487616e+18,
|
| 4 |
+
"train_loss": 0.711250480468766,
|
| 5 |
+
"train_runtime": 3427.9798,
|
| 6 |
+
"train_samples_per_second": 110.733,
|
| 7 |
+
"train_steps_per_second": 0.277
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,646 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9638691322901849,
|
| 3 |
+
"best_model_checkpoint": "mobilenet_v2_1.0_224-plant-disease2/checkpoint-948",
|
| 4 |
+
"epoch": 5.990521327014218,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 948,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.06,
|
| 13 |
+
"learning_rate": 5.263157894736842e-06,
|
| 14 |
+
"loss": 3.7263,
|
| 15 |
+
"step": 10
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.13,
|
| 19 |
+
"learning_rate": 1.0526315789473684e-05,
|
| 20 |
+
"loss": 3.7035,
|
| 21 |
+
"step": 20
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.19,
|
| 25 |
+
"learning_rate": 1.5789473684210526e-05,
|
| 26 |
+
"loss": 3.6641,
|
| 27 |
+
"step": 30
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.25,
|
| 31 |
+
"learning_rate": 2.105263157894737e-05,
|
| 32 |
+
"loss": 3.5615,
|
| 33 |
+
"step": 40
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.32,
|
| 37 |
+
"learning_rate": 2.6315789473684212e-05,
|
| 38 |
+
"loss": 3.4717,
|
| 39 |
+
"step": 50
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.38,
|
| 43 |
+
"learning_rate": 3.157894736842105e-05,
|
| 44 |
+
"loss": 3.3354,
|
| 45 |
+
"step": 60
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.44,
|
| 49 |
+
"learning_rate": 3.6842105263157895e-05,
|
| 50 |
+
"loss": 3.1731,
|
| 51 |
+
"step": 70
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.51,
|
| 55 |
+
"learning_rate": 4.210526315789474e-05,
|
| 56 |
+
"loss": 2.9675,
|
| 57 |
+
"step": 80
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.57,
|
| 61 |
+
"learning_rate": 4.736842105263158e-05,
|
| 62 |
+
"loss": 2.6944,
|
| 63 |
+
"step": 90
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.63,
|
| 67 |
+
"learning_rate": 4.970691676436108e-05,
|
| 68 |
+
"loss": 2.427,
|
| 69 |
+
"step": 100
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.7,
|
| 73 |
+
"learning_rate": 4.912075029308324e-05,
|
| 74 |
+
"loss": 2.1254,
|
| 75 |
+
"step": 110
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.76,
|
| 79 |
+
"learning_rate": 4.8534583821805394e-05,
|
| 80 |
+
"loss": 1.8635,
|
| 81 |
+
"step": 120
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.82,
|
| 85 |
+
"learning_rate": 4.794841735052755e-05,
|
| 86 |
+
"loss": 1.6268,
|
| 87 |
+
"step": 130
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.88,
|
| 91 |
+
"learning_rate": 4.736225087924971e-05,
|
| 92 |
+
"loss": 1.4245,
|
| 93 |
+
"step": 140
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.95,
|
| 97 |
+
"learning_rate": 4.6776084407971864e-05,
|
| 98 |
+
"loss": 1.2171,
|
| 99 |
+
"step": 150
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 1.0,
|
| 103 |
+
"eval_accuracy": 0.8187766714082504,
|
| 104 |
+
"eval_loss": 1.059464693069458,
|
| 105 |
+
"eval_runtime": 29.4475,
|
| 106 |
+
"eval_samples_per_second": 238.73,
|
| 107 |
+
"eval_steps_per_second": 2.411,
|
| 108 |
+
"step": 158
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 1.01,
|
| 112 |
+
"learning_rate": 4.6189917936694024e-05,
|
| 113 |
+
"loss": 1.1133,
|
| 114 |
+
"step": 160
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 1.07,
|
| 118 |
+
"learning_rate": 4.5603751465416176e-05,
|
| 119 |
+
"loss": 1.012,
|
| 120 |
+
"step": 170
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 1.14,
|
| 124 |
+
"learning_rate": 4.501758499413834e-05,
|
| 125 |
+
"loss": 0.8743,
|
| 126 |
+
"step": 180
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 1.2,
|
| 130 |
+
"learning_rate": 4.4431418522860494e-05,
|
| 131 |
+
"loss": 0.8055,
|
| 132 |
+
"step": 190
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 1.26,
|
| 136 |
+
"learning_rate": 4.3845252051582654e-05,
|
| 137 |
+
"loss": 0.7388,
|
| 138 |
+
"step": 200
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 1.33,
|
| 142 |
+
"learning_rate": 4.3259085580304806e-05,
|
| 143 |
+
"loss": 0.6954,
|
| 144 |
+
"step": 210
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.39,
|
| 148 |
+
"learning_rate": 4.2672919109026965e-05,
|
| 149 |
+
"loss": 0.6384,
|
| 150 |
+
"step": 220
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 1.45,
|
| 154 |
+
"learning_rate": 4.2086752637749124e-05,
|
| 155 |
+
"loss": 0.5831,
|
| 156 |
+
"step": 230
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 1.52,
|
| 160 |
+
"learning_rate": 4.150058616647128e-05,
|
| 161 |
+
"loss": 0.5561,
|
| 162 |
+
"step": 240
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 1.58,
|
| 166 |
+
"learning_rate": 4.0914419695193436e-05,
|
| 167 |
+
"loss": 0.5289,
|
| 168 |
+
"step": 250
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 1.64,
|
| 172 |
+
"learning_rate": 4.0328253223915595e-05,
|
| 173 |
+
"loss": 0.5063,
|
| 174 |
+
"step": 260
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 1.71,
|
| 178 |
+
"learning_rate": 3.9742086752637754e-05,
|
| 179 |
+
"loss": 0.4788,
|
| 180 |
+
"step": 270
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 1.77,
|
| 184 |
+
"learning_rate": 3.915592028135991e-05,
|
| 185 |
+
"loss": 0.4552,
|
| 186 |
+
"step": 280
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.83,
|
| 190 |
+
"learning_rate": 3.8569753810082066e-05,
|
| 191 |
+
"loss": 0.4645,
|
| 192 |
+
"step": 290
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 1.9,
|
| 196 |
+
"learning_rate": 3.7983587338804225e-05,
|
| 197 |
+
"loss": 0.4267,
|
| 198 |
+
"step": 300
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 1.96,
|
| 202 |
+
"learning_rate": 3.739742086752638e-05,
|
| 203 |
+
"loss": 0.4082,
|
| 204 |
+
"step": 310
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 2.0,
|
| 208 |
+
"eval_accuracy": 0.9386913229018492,
|
| 209 |
+
"eval_loss": 0.3154248893260956,
|
| 210 |
+
"eval_runtime": 29.2493,
|
| 211 |
+
"eval_samples_per_second": 240.347,
|
| 212 |
+
"eval_steps_per_second": 2.427,
|
| 213 |
+
"step": 316
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 2.02,
|
| 217 |
+
"learning_rate": 3.681125439624854e-05,
|
| 218 |
+
"loss": 0.4,
|
| 219 |
+
"step": 320
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 2.09,
|
| 223 |
+
"learning_rate": 3.622508792497069e-05,
|
| 224 |
+
"loss": 0.3832,
|
| 225 |
+
"step": 330
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 2.15,
|
| 229 |
+
"learning_rate": 3.5638921453692855e-05,
|
| 230 |
+
"loss": 0.3888,
|
| 231 |
+
"step": 340
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 2.21,
|
| 235 |
+
"learning_rate": 3.505275498241501e-05,
|
| 236 |
+
"loss": 0.3595,
|
| 237 |
+
"step": 350
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 2.27,
|
| 241 |
+
"learning_rate": 3.446658851113717e-05,
|
| 242 |
+
"loss": 0.3454,
|
| 243 |
+
"step": 360
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 2.34,
|
| 247 |
+
"learning_rate": 3.388042203985932e-05,
|
| 248 |
+
"loss": 0.3392,
|
| 249 |
+
"step": 370
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 2.4,
|
| 253 |
+
"learning_rate": 3.329425556858148e-05,
|
| 254 |
+
"loss": 0.3466,
|
| 255 |
+
"step": 380
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 2.46,
|
| 259 |
+
"learning_rate": 3.270808909730364e-05,
|
| 260 |
+
"loss": 0.345,
|
| 261 |
+
"step": 390
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 2.53,
|
| 265 |
+
"learning_rate": 3.212192262602579e-05,
|
| 266 |
+
"loss": 0.3102,
|
| 267 |
+
"step": 400
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 2.59,
|
| 271 |
+
"learning_rate": 3.153575615474795e-05,
|
| 272 |
+
"loss": 0.3045,
|
| 273 |
+
"step": 410
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 2.65,
|
| 277 |
+
"learning_rate": 3.094958968347011e-05,
|
| 278 |
+
"loss": 0.3242,
|
| 279 |
+
"step": 420
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 2.72,
|
| 283 |
+
"learning_rate": 3.0363423212192264e-05,
|
| 284 |
+
"loss": 0.3022,
|
| 285 |
+
"step": 430
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 2.78,
|
| 289 |
+
"learning_rate": 2.9777256740914423e-05,
|
| 290 |
+
"loss": 0.3098,
|
| 291 |
+
"step": 440
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 2.84,
|
| 295 |
+
"learning_rate": 2.919109026963658e-05,
|
| 296 |
+
"loss": 0.2845,
|
| 297 |
+
"step": 450
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 2.91,
|
| 301 |
+
"learning_rate": 2.8604923798358735e-05,
|
| 302 |
+
"loss": 0.2807,
|
| 303 |
+
"step": 460
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 2.97,
|
| 307 |
+
"learning_rate": 2.801875732708089e-05,
|
| 308 |
+
"loss": 0.295,
|
| 309 |
+
"step": 470
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 3.0,
|
| 313 |
+
"eval_accuracy": 0.9554765291607397,
|
| 314 |
+
"eval_loss": 0.21907542645931244,
|
| 315 |
+
"eval_runtime": 28.5527,
|
| 316 |
+
"eval_samples_per_second": 246.211,
|
| 317 |
+
"eval_steps_per_second": 2.487,
|
| 318 |
+
"step": 474
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 3.03,
|
| 322 |
+
"learning_rate": 2.7432590855803047e-05,
|
| 323 |
+
"loss": 0.2667,
|
| 324 |
+
"step": 480
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 3.1,
|
| 328 |
+
"learning_rate": 2.6846424384525202e-05,
|
| 329 |
+
"loss": 0.281,
|
| 330 |
+
"step": 490
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 3.16,
|
| 334 |
+
"learning_rate": 2.6260257913247365e-05,
|
| 335 |
+
"loss": 0.268,
|
| 336 |
+
"step": 500
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 3.22,
|
| 340 |
+
"learning_rate": 2.567409144196952e-05,
|
| 341 |
+
"loss": 0.2458,
|
| 342 |
+
"step": 510
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 3.29,
|
| 346 |
+
"learning_rate": 2.508792497069168e-05,
|
| 347 |
+
"loss": 0.2493,
|
| 348 |
+
"step": 520
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 3.35,
|
| 352 |
+
"learning_rate": 2.4501758499413836e-05,
|
| 353 |
+
"loss": 0.2607,
|
| 354 |
+
"step": 530
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 3.41,
|
| 358 |
+
"learning_rate": 2.391559202813599e-05,
|
| 359 |
+
"loss": 0.2553,
|
| 360 |
+
"step": 540
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 3.48,
|
| 364 |
+
"learning_rate": 2.3329425556858147e-05,
|
| 365 |
+
"loss": 0.2481,
|
| 366 |
+
"step": 550
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 3.54,
|
| 370 |
+
"learning_rate": 2.2743259085580307e-05,
|
| 371 |
+
"loss": 0.2537,
|
| 372 |
+
"step": 560
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 3.6,
|
| 376 |
+
"learning_rate": 2.2157092614302462e-05,
|
| 377 |
+
"loss": 0.2407,
|
| 378 |
+
"step": 570
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 3.67,
|
| 382 |
+
"learning_rate": 2.1570926143024618e-05,
|
| 383 |
+
"loss": 0.2465,
|
| 384 |
+
"step": 580
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 3.73,
|
| 388 |
+
"learning_rate": 2.0984759671746777e-05,
|
| 389 |
+
"loss": 0.2562,
|
| 390 |
+
"step": 590
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 3.79,
|
| 394 |
+
"learning_rate": 2.0398593200468937e-05,
|
| 395 |
+
"loss": 0.2471,
|
| 396 |
+
"step": 600
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 3.85,
|
| 400 |
+
"learning_rate": 1.9812426729191092e-05,
|
| 401 |
+
"loss": 0.2453,
|
| 402 |
+
"step": 610
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 3.92,
|
| 406 |
+
"learning_rate": 1.9226260257913248e-05,
|
| 407 |
+
"loss": 0.2292,
|
| 408 |
+
"step": 620
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 3.98,
|
| 412 |
+
"learning_rate": 1.8640093786635404e-05,
|
| 413 |
+
"loss": 0.2266,
|
| 414 |
+
"step": 630
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 4.0,
|
| 418 |
+
"eval_accuracy": 0.9594594594594594,
|
| 419 |
+
"eval_loss": 0.1747303009033203,
|
| 420 |
+
"eval_runtime": 28.6742,
|
| 421 |
+
"eval_samples_per_second": 245.168,
|
| 422 |
+
"eval_steps_per_second": 2.476,
|
| 423 |
+
"step": 633
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 4.04,
|
| 427 |
+
"learning_rate": 1.8053927315357563e-05,
|
| 428 |
+
"loss": 0.2208,
|
| 429 |
+
"step": 640
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 4.11,
|
| 433 |
+
"learning_rate": 1.746776084407972e-05,
|
| 434 |
+
"loss": 0.2339,
|
| 435 |
+
"step": 650
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 4.17,
|
| 439 |
+
"learning_rate": 1.6881594372801875e-05,
|
| 440 |
+
"loss": 0.2395,
|
| 441 |
+
"step": 660
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 4.23,
|
| 445 |
+
"learning_rate": 1.6295427901524034e-05,
|
| 446 |
+
"loss": 0.2314,
|
| 447 |
+
"step": 670
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 4.3,
|
| 451 |
+
"learning_rate": 1.5709261430246193e-05,
|
| 452 |
+
"loss": 0.2208,
|
| 453 |
+
"step": 680
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 4.36,
|
| 457 |
+
"learning_rate": 1.5123094958968347e-05,
|
| 458 |
+
"loss": 0.229,
|
| 459 |
+
"step": 690
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 4.42,
|
| 463 |
+
"learning_rate": 1.4536928487690505e-05,
|
| 464 |
+
"loss": 0.2251,
|
| 465 |
+
"step": 700
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 4.49,
|
| 469 |
+
"learning_rate": 1.395076201641266e-05,
|
| 470 |
+
"loss": 0.232,
|
| 471 |
+
"step": 710
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 4.55,
|
| 475 |
+
"learning_rate": 1.336459554513482e-05,
|
| 476 |
+
"loss": 0.2342,
|
| 477 |
+
"step": 720
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 4.61,
|
| 481 |
+
"learning_rate": 1.2778429073856976e-05,
|
| 482 |
+
"loss": 0.1995,
|
| 483 |
+
"step": 730
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 4.68,
|
| 487 |
+
"learning_rate": 1.2192262602579133e-05,
|
| 488 |
+
"loss": 0.2119,
|
| 489 |
+
"step": 740
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 4.74,
|
| 493 |
+
"learning_rate": 1.160609613130129e-05,
|
| 494 |
+
"loss": 0.2185,
|
| 495 |
+
"step": 750
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 4.8,
|
| 499 |
+
"learning_rate": 1.1019929660023446e-05,
|
| 500 |
+
"loss": 0.2079,
|
| 501 |
+
"step": 760
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 4.87,
|
| 505 |
+
"learning_rate": 1.0433763188745604e-05,
|
| 506 |
+
"loss": 0.2113,
|
| 507 |
+
"step": 770
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 4.93,
|
| 511 |
+
"learning_rate": 9.847596717467761e-06,
|
| 512 |
+
"loss": 0.2177,
|
| 513 |
+
"step": 780
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 4.99,
|
| 517 |
+
"learning_rate": 9.261430246189919e-06,
|
| 518 |
+
"loss": 0.2168,
|
| 519 |
+
"step": 790
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 5.0,
|
| 523 |
+
"eval_accuracy": 0.9499288762446657,
|
| 524 |
+
"eval_loss": 0.21349148452281952,
|
| 525 |
+
"eval_runtime": 28.3255,
|
| 526 |
+
"eval_samples_per_second": 248.186,
|
| 527 |
+
"eval_steps_per_second": 2.507,
|
| 528 |
+
"step": 791
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 5.06,
|
| 532 |
+
"learning_rate": 8.675263774912075e-06,
|
| 533 |
+
"loss": 0.2072,
|
| 534 |
+
"step": 800
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 5.12,
|
| 538 |
+
"learning_rate": 8.089097303634232e-06,
|
| 539 |
+
"loss": 0.2083,
|
| 540 |
+
"step": 810
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 5.18,
|
| 544 |
+
"learning_rate": 7.502930832356389e-06,
|
| 545 |
+
"loss": 0.1997,
|
| 546 |
+
"step": 820
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 5.24,
|
| 550 |
+
"learning_rate": 6.916764361078546e-06,
|
| 551 |
+
"loss": 0.1917,
|
| 552 |
+
"step": 830
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 5.31,
|
| 556 |
+
"learning_rate": 6.330597889800703e-06,
|
| 557 |
+
"loss": 0.2098,
|
| 558 |
+
"step": 840
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 5.37,
|
| 562 |
+
"learning_rate": 5.7444314185228606e-06,
|
| 563 |
+
"loss": 0.2001,
|
| 564 |
+
"step": 850
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 5.43,
|
| 568 |
+
"learning_rate": 5.158264947245018e-06,
|
| 569 |
+
"loss": 0.2105,
|
| 570 |
+
"step": 860
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 5.5,
|
| 574 |
+
"learning_rate": 4.572098475967175e-06,
|
| 575 |
+
"loss": 0.2168,
|
| 576 |
+
"step": 870
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 5.56,
|
| 580 |
+
"learning_rate": 3.985932004689332e-06,
|
| 581 |
+
"loss": 0.2111,
|
| 582 |
+
"step": 880
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 5.62,
|
| 586 |
+
"learning_rate": 3.3997655334114893e-06,
|
| 587 |
+
"loss": 0.1999,
|
| 588 |
+
"step": 890
|
| 589 |
+
},
|
| 590 |
+
{
|
| 591 |
+
"epoch": 5.69,
|
| 592 |
+
"learning_rate": 2.8135990621336464e-06,
|
| 593 |
+
"loss": 0.2069,
|
| 594 |
+
"step": 900
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 5.75,
|
| 598 |
+
"learning_rate": 2.2274325908558035e-06,
|
| 599 |
+
"loss": 0.2302,
|
| 600 |
+
"step": 910
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 5.81,
|
| 604 |
+
"learning_rate": 1.6412661195779603e-06,
|
| 605 |
+
"loss": 0.2072,
|
| 606 |
+
"step": 920
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 5.88,
|
| 610 |
+
"learning_rate": 1.0550996483001172e-06,
|
| 611 |
+
"loss": 0.2042,
|
| 612 |
+
"step": 930
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 5.94,
|
| 616 |
+
"learning_rate": 4.689331770222743e-07,
|
| 617 |
+
"loss": 0.2091,
|
| 618 |
+
"step": 940
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 5.99,
|
| 622 |
+
"eval_accuracy": 0.9638691322901849,
|
| 623 |
+
"eval_loss": 0.15102514624595642,
|
| 624 |
+
"eval_runtime": 27.9653,
|
| 625 |
+
"eval_samples_per_second": 251.383,
|
| 626 |
+
"eval_steps_per_second": 2.539,
|
| 627 |
+
"step": 948
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 5.99,
|
| 631 |
+
"step": 948,
|
| 632 |
+
"total_flos": 1.01609365487616e+18,
|
| 633 |
+
"train_loss": 0.711250480468766,
|
| 634 |
+
"train_runtime": 3427.9798,
|
| 635 |
+
"train_samples_per_second": 110.733,
|
| 636 |
+
"train_steps_per_second": 0.277
|
| 637 |
+
}
|
| 638 |
+
],
|
| 639 |
+
"logging_steps": 10,
|
| 640 |
+
"max_steps": 948,
|
| 641 |
+
"num_train_epochs": 6,
|
| 642 |
+
"save_steps": 500,
|
| 643 |
+
"total_flos": 1.01609365487616e+18,
|
| 644 |
+
"trial_name": null,
|
| 645 |
+
"trial_params": null
|
| 646 |
+
}
|