End of training
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.99,
|
| 3 |
+
"eval_accuracy": 0.30788520479242126,
|
| 4 |
+
"eval_loss": 1.6947710514068604,
|
| 5 |
+
"eval_runtime": 17.149,
|
| 6 |
+
"eval_samples_per_second": 209.283,
|
| 7 |
+
"eval_steps_per_second": 6.589,
|
| 8 |
+
"total_flos": 2.054868144086016e+18,
|
| 9 |
+
"train_loss": 1.782613229499292,
|
| 10 |
+
"train_runtime": 1261.819,
|
| 11 |
+
"train_samples_per_second": 76.789,
|
| 12 |
+
"train_steps_per_second": 0.599
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.99,
|
| 3 |
+
"eval_accuracy": 0.30788520479242126,
|
| 4 |
+
"eval_loss": 1.6947710514068604,
|
| 5 |
+
"eval_runtime": 17.149,
|
| 6 |
+
"eval_samples_per_second": 209.283,
|
| 7 |
+
"eval_steps_per_second": 6.589
|
| 8 |
+
}
|
runs/Nov06_02-47-05_9c925b04246e/events.out.tfevents.1699240472.9c925b04246e.8770.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:39b626231fe9cdab98c8a2e0d450c9645ed7c3858cfe33f60a8628e97a27b255
|
| 3 |
+
size 411
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.99,
|
| 3 |
+
"total_flos": 2.054868144086016e+18,
|
| 4 |
+
"train_loss": 1.782613229499292,
|
| 5 |
+
"train_runtime": 1261.819,
|
| 6 |
+
"train_samples_per_second": 76.789,
|
| 7 |
+
"train_steps_per_second": 0.599
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,502 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.30788520479242126,
|
| 3 |
+
"best_model_checkpoint": "microsoft/resnet-50/checkpoint-756",
|
| 4 |
+
"epoch": 2.994059405940594,
|
| 5 |
+
"global_step": 756,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.04,
|
| 12 |
+
"learning_rate": 6.578947368421053e-06,
|
| 13 |
+
"loss": 1.9535,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.08,
|
| 18 |
+
"learning_rate": 1.3157894736842106e-05,
|
| 19 |
+
"loss": 1.9539,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.12,
|
| 24 |
+
"learning_rate": 1.9736842105263158e-05,
|
| 25 |
+
"loss": 1.9476,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.16,
|
| 30 |
+
"learning_rate": 2.6315789473684212e-05,
|
| 31 |
+
"loss": 1.9401,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.2,
|
| 36 |
+
"learning_rate": 3.289473684210527e-05,
|
| 37 |
+
"loss": 1.9308,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.24,
|
| 42 |
+
"learning_rate": 3.9473684210526316e-05,
|
| 43 |
+
"loss": 1.9207,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.28,
|
| 48 |
+
"learning_rate": 4.605263157894737e-05,
|
| 49 |
+
"loss": 1.9125,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.32,
|
| 54 |
+
"learning_rate": 4.970588235294118e-05,
|
| 55 |
+
"loss": 1.8956,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.36,
|
| 60 |
+
"learning_rate": 4.897058823529412e-05,
|
| 61 |
+
"loss": 1.8897,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.4,
|
| 66 |
+
"learning_rate": 4.823529411764706e-05,
|
| 67 |
+
"loss": 1.8761,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.44,
|
| 72 |
+
"learning_rate": 4.75e-05,
|
| 73 |
+
"loss": 1.8658,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.48,
|
| 78 |
+
"learning_rate": 4.6764705882352944e-05,
|
| 79 |
+
"loss": 1.8532,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.51,
|
| 84 |
+
"learning_rate": 4.6029411764705885e-05,
|
| 85 |
+
"loss": 1.8561,
|
| 86 |
+
"step": 130
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.55,
|
| 90 |
+
"learning_rate": 4.5294117647058826e-05,
|
| 91 |
+
"loss": 1.8521,
|
| 92 |
+
"step": 140
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.59,
|
| 96 |
+
"learning_rate": 4.455882352941177e-05,
|
| 97 |
+
"loss": 1.8386,
|
| 98 |
+
"step": 150
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.63,
|
| 102 |
+
"learning_rate": 4.382352941176471e-05,
|
| 103 |
+
"loss": 1.8219,
|
| 104 |
+
"step": 160
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.67,
|
| 108 |
+
"learning_rate": 4.308823529411765e-05,
|
| 109 |
+
"loss": 1.8281,
|
| 110 |
+
"step": 170
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.71,
|
| 114 |
+
"learning_rate": 4.235294117647059e-05,
|
| 115 |
+
"loss": 1.8081,
|
| 116 |
+
"step": 180
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.75,
|
| 120 |
+
"learning_rate": 4.161764705882353e-05,
|
| 121 |
+
"loss": 1.8114,
|
| 122 |
+
"step": 190
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.79,
|
| 126 |
+
"learning_rate": 4.0882352941176474e-05,
|
| 127 |
+
"loss": 1.813,
|
| 128 |
+
"step": 200
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.83,
|
| 132 |
+
"learning_rate": 4.0147058823529415e-05,
|
| 133 |
+
"loss": 1.7975,
|
| 134 |
+
"step": 210
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.87,
|
| 138 |
+
"learning_rate": 3.9411764705882356e-05,
|
| 139 |
+
"loss": 1.8039,
|
| 140 |
+
"step": 220
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.91,
|
| 144 |
+
"learning_rate": 3.86764705882353e-05,
|
| 145 |
+
"loss": 1.7997,
|
| 146 |
+
"step": 230
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.95,
|
| 150 |
+
"learning_rate": 3.794117647058824e-05,
|
| 151 |
+
"loss": 1.7928,
|
| 152 |
+
"step": 240
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.99,
|
| 156 |
+
"learning_rate": 3.720588235294118e-05,
|
| 157 |
+
"loss": 1.7973,
|
| 158 |
+
"step": 250
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.0,
|
| 162 |
+
"eval_accuracy": 0.24770130955697967,
|
| 163 |
+
"eval_loss": 1.7939863204956055,
|
| 164 |
+
"eval_runtime": 18.0712,
|
| 165 |
+
"eval_samples_per_second": 198.604,
|
| 166 |
+
"eval_steps_per_second": 6.253,
|
| 167 |
+
"step": 252
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 1.03,
|
| 171 |
+
"learning_rate": 3.6470588235294114e-05,
|
| 172 |
+
"loss": 1.8085,
|
| 173 |
+
"step": 260
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 1.07,
|
| 177 |
+
"learning_rate": 3.573529411764706e-05,
|
| 178 |
+
"loss": 1.7814,
|
| 179 |
+
"step": 270
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.11,
|
| 183 |
+
"learning_rate": 3.5e-05,
|
| 184 |
+
"loss": 1.8047,
|
| 185 |
+
"step": 280
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 1.15,
|
| 189 |
+
"learning_rate": 3.4264705882352945e-05,
|
| 190 |
+
"loss": 1.801,
|
| 191 |
+
"step": 290
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 1.19,
|
| 195 |
+
"learning_rate": 3.352941176470588e-05,
|
| 196 |
+
"loss": 1.7883,
|
| 197 |
+
"step": 300
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 1.23,
|
| 201 |
+
"learning_rate": 3.279411764705883e-05,
|
| 202 |
+
"loss": 1.7824,
|
| 203 |
+
"step": 310
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 1.27,
|
| 207 |
+
"learning_rate": 3.205882352941177e-05,
|
| 208 |
+
"loss": 1.774,
|
| 209 |
+
"step": 320
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 1.31,
|
| 213 |
+
"learning_rate": 3.132352941176471e-05,
|
| 214 |
+
"loss": 1.7772,
|
| 215 |
+
"step": 330
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 1.35,
|
| 219 |
+
"learning_rate": 3.058823529411765e-05,
|
| 220 |
+
"loss": 1.7584,
|
| 221 |
+
"step": 340
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.39,
|
| 225 |
+
"learning_rate": 2.985294117647059e-05,
|
| 226 |
+
"loss": 1.7609,
|
| 227 |
+
"step": 350
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 1.43,
|
| 231 |
+
"learning_rate": 2.9117647058823534e-05,
|
| 232 |
+
"loss": 1.7864,
|
| 233 |
+
"step": 360
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 1.47,
|
| 237 |
+
"learning_rate": 2.838235294117647e-05,
|
| 238 |
+
"loss": 1.7783,
|
| 239 |
+
"step": 370
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 1.5,
|
| 243 |
+
"learning_rate": 2.7647058823529416e-05,
|
| 244 |
+
"loss": 1.757,
|
| 245 |
+
"step": 380
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 1.54,
|
| 249 |
+
"learning_rate": 2.6911764705882354e-05,
|
| 250 |
+
"loss": 1.7539,
|
| 251 |
+
"step": 390
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 1.58,
|
| 255 |
+
"learning_rate": 2.6176470588235295e-05,
|
| 256 |
+
"loss": 1.7694,
|
| 257 |
+
"step": 400
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 1.62,
|
| 261 |
+
"learning_rate": 2.5441176470588236e-05,
|
| 262 |
+
"loss": 1.7601,
|
| 263 |
+
"step": 410
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.66,
|
| 267 |
+
"learning_rate": 2.4705882352941178e-05,
|
| 268 |
+
"loss": 1.7417,
|
| 269 |
+
"step": 420
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 1.7,
|
| 273 |
+
"learning_rate": 2.397058823529412e-05,
|
| 274 |
+
"loss": 1.744,
|
| 275 |
+
"step": 430
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 1.74,
|
| 279 |
+
"learning_rate": 2.323529411764706e-05,
|
| 280 |
+
"loss": 1.7491,
|
| 281 |
+
"step": 440
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 1.78,
|
| 285 |
+
"learning_rate": 2.25e-05,
|
| 286 |
+
"loss": 1.7387,
|
| 287 |
+
"step": 450
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 1.82,
|
| 291 |
+
"learning_rate": 2.1764705882352943e-05,
|
| 292 |
+
"loss": 1.7526,
|
| 293 |
+
"step": 460
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 1.86,
|
| 297 |
+
"learning_rate": 2.1029411764705884e-05,
|
| 298 |
+
"loss": 1.7361,
|
| 299 |
+
"step": 470
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 1.9,
|
| 303 |
+
"learning_rate": 2.0294117647058825e-05,
|
| 304 |
+
"loss": 1.7344,
|
| 305 |
+
"step": 480
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 1.94,
|
| 309 |
+
"learning_rate": 1.9558823529411766e-05,
|
| 310 |
+
"loss": 1.7211,
|
| 311 |
+
"step": 490
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 1.98,
|
| 315 |
+
"learning_rate": 1.8823529411764708e-05,
|
| 316 |
+
"loss": 1.7299,
|
| 317 |
+
"step": 500
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 2.0,
|
| 321 |
+
"eval_accuracy": 0.2803009194761772,
|
| 322 |
+
"eval_loss": 1.727730393409729,
|
| 323 |
+
"eval_runtime": 17.3747,
|
| 324 |
+
"eval_samples_per_second": 206.564,
|
| 325 |
+
"eval_steps_per_second": 6.504,
|
| 326 |
+
"step": 505
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 2.02,
|
| 330 |
+
"learning_rate": 1.808823529411765e-05,
|
| 331 |
+
"loss": 1.7469,
|
| 332 |
+
"step": 510
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 2.06,
|
| 336 |
+
"learning_rate": 1.735294117647059e-05,
|
| 337 |
+
"loss": 1.744,
|
| 338 |
+
"step": 520
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 2.1,
|
| 342 |
+
"learning_rate": 1.6617647058823528e-05,
|
| 343 |
+
"loss": 1.7489,
|
| 344 |
+
"step": 530
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 2.14,
|
| 348 |
+
"learning_rate": 1.588235294117647e-05,
|
| 349 |
+
"loss": 1.7356,
|
| 350 |
+
"step": 540
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 2.18,
|
| 354 |
+
"learning_rate": 1.5147058823529412e-05,
|
| 355 |
+
"loss": 1.7327,
|
| 356 |
+
"step": 550
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.22,
|
| 360 |
+
"learning_rate": 1.4411764705882352e-05,
|
| 361 |
+
"loss": 1.7045,
|
| 362 |
+
"step": 560
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 2.26,
|
| 366 |
+
"learning_rate": 1.3676470588235296e-05,
|
| 367 |
+
"loss": 1.717,
|
| 368 |
+
"step": 570
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 2.3,
|
| 372 |
+
"learning_rate": 1.2941176470588238e-05,
|
| 373 |
+
"loss": 1.7335,
|
| 374 |
+
"step": 580
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 2.34,
|
| 378 |
+
"learning_rate": 1.2205882352941177e-05,
|
| 379 |
+
"loss": 1.7338,
|
| 380 |
+
"step": 590
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 2.38,
|
| 384 |
+
"learning_rate": 1.1470588235294118e-05,
|
| 385 |
+
"loss": 1.7145,
|
| 386 |
+
"step": 600
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 2.42,
|
| 390 |
+
"learning_rate": 1.0735294117647058e-05,
|
| 391 |
+
"loss": 1.7349,
|
| 392 |
+
"step": 610
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 2.46,
|
| 396 |
+
"learning_rate": 1e-05,
|
| 397 |
+
"loss": 1.7293,
|
| 398 |
+
"step": 620
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 2.5,
|
| 402 |
+
"learning_rate": 9.264705882352942e-06,
|
| 403 |
+
"loss": 1.7291,
|
| 404 |
+
"step": 630
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 2.53,
|
| 408 |
+
"learning_rate": 8.529411764705883e-06,
|
| 409 |
+
"loss": 1.7192,
|
| 410 |
+
"step": 640
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 2.57,
|
| 414 |
+
"learning_rate": 7.794117647058825e-06,
|
| 415 |
+
"loss": 1.7203,
|
| 416 |
+
"step": 650
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 2.61,
|
| 420 |
+
"learning_rate": 7.058823529411765e-06,
|
| 421 |
+
"loss": 1.7403,
|
| 422 |
+
"step": 660
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 2.65,
|
| 426 |
+
"learning_rate": 6.323529411764706e-06,
|
| 427 |
+
"loss": 1.7077,
|
| 428 |
+
"step": 670
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 2.69,
|
| 432 |
+
"learning_rate": 5.588235294117647e-06,
|
| 433 |
+
"loss": 1.712,
|
| 434 |
+
"step": 680
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 2.73,
|
| 438 |
+
"learning_rate": 4.852941176470589e-06,
|
| 439 |
+
"loss": 1.7327,
|
| 440 |
+
"step": 690
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 2.77,
|
| 444 |
+
"learning_rate": 4.11764705882353e-06,
|
| 445 |
+
"loss": 1.7021,
|
| 446 |
+
"step": 700
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 2.81,
|
| 450 |
+
"learning_rate": 3.3823529411764707e-06,
|
| 451 |
+
"loss": 1.7295,
|
| 452 |
+
"step": 710
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 2.85,
|
| 456 |
+
"learning_rate": 2.647058823529412e-06,
|
| 457 |
+
"loss": 1.7288,
|
| 458 |
+
"step": 720
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 2.89,
|
| 462 |
+
"learning_rate": 1.911764705882353e-06,
|
| 463 |
+
"loss": 1.6857,
|
| 464 |
+
"step": 730
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 2.93,
|
| 468 |
+
"learning_rate": 1.1764705882352942e-06,
|
| 469 |
+
"loss": 1.7052,
|
| 470 |
+
"step": 740
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 2.97,
|
| 474 |
+
"learning_rate": 4.411764705882353e-07,
|
| 475 |
+
"loss": 1.7096,
|
| 476 |
+
"step": 750
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 2.99,
|
| 480 |
+
"eval_accuracy": 0.30788520479242126,
|
| 481 |
+
"eval_loss": 1.6947710514068604,
|
| 482 |
+
"eval_runtime": 16.9402,
|
| 483 |
+
"eval_samples_per_second": 211.863,
|
| 484 |
+
"eval_steps_per_second": 6.671,
|
| 485 |
+
"step": 756
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 2.99,
|
| 489 |
+
"step": 756,
|
| 490 |
+
"total_flos": 2.054868144086016e+18,
|
| 491 |
+
"train_loss": 1.782613229499292,
|
| 492 |
+
"train_runtime": 1261.819,
|
| 493 |
+
"train_samples_per_second": 76.789,
|
| 494 |
+
"train_steps_per_second": 0.599
|
| 495 |
+
}
|
| 496 |
+
],
|
| 497 |
+
"max_steps": 756,
|
| 498 |
+
"num_train_epochs": 3,
|
| 499 |
+
"total_flos": 2.054868144086016e+18,
|
| 500 |
+
"trial_name": null,
|
| 501 |
+
"trial_params": null
|
| 502 |
+
}
|