End of training
Browse files- all_results.json +12 -0
- eval_results.json +8 -0
- runs/Aug13_20-29-08_xc-pc/events.out.tfevents.1660396011.xc-pc.6149.2 +3 -0
- train_results.json +7 -0
- trainer_state.json +730 -0
all_results.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.9811111111111112,
|
| 4 |
+
"eval_loss": 0.04829270392656326,
|
| 5 |
+
"eval_runtime": 23.7444,
|
| 6 |
+
"eval_samples_per_second": 113.711,
|
| 7 |
+
"eval_steps_per_second": 7.117,
|
| 8 |
+
"train_loss": 0.29242842561231963,
|
| 9 |
+
"train_runtime": 1937.8226,
|
| 10 |
+
"train_samples_per_second": 37.62,
|
| 11 |
+
"train_steps_per_second": 0.587
|
| 12 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.9811111111111112,
|
| 4 |
+
"eval_loss": 0.04829270392656326,
|
| 5 |
+
"eval_runtime": 23.7444,
|
| 6 |
+
"eval_samples_per_second": 113.711,
|
| 7 |
+
"eval_steps_per_second": 7.117
|
| 8 |
+
}
|
runs/Aug13_20-29-08_xc-pc/events.out.tfevents.1660396011.xc-pc.6149.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:476f5ea308046a655388b31d88f973ca7f4a76f5a9bf6f078544b10a957fd838
|
| 3 |
+
size 363
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"train_loss": 0.29242842561231963,
|
| 4 |
+
"train_runtime": 1937.8226,
|
| 5 |
+
"train_samples_per_second": 37.62,
|
| 6 |
+
"train_steps_per_second": 0.587
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,730 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9811111111111112,
|
| 3 |
+
"best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned-eurosat/checkpoint-1137",
|
| 4 |
+
"epoch": 2.9980250164581963,
|
| 5 |
+
"global_step": 1137,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.03,
|
| 12 |
+
"learning_rate": 4.3859649122807014e-06,
|
| 13 |
+
"loss": 2.2879,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.05,
|
| 18 |
+
"learning_rate": 8.771929824561403e-06,
|
| 19 |
+
"loss": 2.1937,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.08,
|
| 24 |
+
"learning_rate": 1.3157894736842106e-05,
|
| 25 |
+
"loss": 1.9682,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.11,
|
| 30 |
+
"learning_rate": 1.7543859649122806e-05,
|
| 31 |
+
"loss": 1.5989,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.13,
|
| 36 |
+
"learning_rate": 2.1929824561403507e-05,
|
| 37 |
+
"loss": 1.1674,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.16,
|
| 42 |
+
"learning_rate": 2.6315789473684212e-05,
|
| 43 |
+
"loss": 0.8328,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.18,
|
| 48 |
+
"learning_rate": 3.0701754385964913e-05,
|
| 49 |
+
"loss": 0.5801,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.21,
|
| 54 |
+
"learning_rate": 3.508771929824561e-05,
|
| 55 |
+
"loss": 0.5215,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.24,
|
| 60 |
+
"learning_rate": 3.9473684210526316e-05,
|
| 61 |
+
"loss": 0.5077,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.26,
|
| 66 |
+
"learning_rate": 4.3859649122807014e-05,
|
| 67 |
+
"loss": 0.3814,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.29,
|
| 72 |
+
"learning_rate": 4.824561403508772e-05,
|
| 73 |
+
"loss": 0.4916,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.32,
|
| 78 |
+
"learning_rate": 4.9706744868035194e-05,
|
| 79 |
+
"loss": 0.471,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.34,
|
| 84 |
+
"learning_rate": 4.921798631476051e-05,
|
| 85 |
+
"loss": 0.3859,
|
| 86 |
+
"step": 130
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.37,
|
| 90 |
+
"learning_rate": 4.8729227761485826e-05,
|
| 91 |
+
"loss": 0.4545,
|
| 92 |
+
"step": 140
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.39,
|
| 96 |
+
"learning_rate": 4.824046920821115e-05,
|
| 97 |
+
"loss": 0.3339,
|
| 98 |
+
"step": 150
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.42,
|
| 102 |
+
"learning_rate": 4.7751710654936466e-05,
|
| 103 |
+
"loss": 0.3497,
|
| 104 |
+
"step": 160
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.45,
|
| 108 |
+
"learning_rate": 4.726295210166178e-05,
|
| 109 |
+
"loss": 0.3524,
|
| 110 |
+
"step": 170
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.47,
|
| 114 |
+
"learning_rate": 4.67741935483871e-05,
|
| 115 |
+
"loss": 0.3657,
|
| 116 |
+
"step": 180
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.5,
|
| 120 |
+
"learning_rate": 4.6285434995112415e-05,
|
| 121 |
+
"loss": 0.3355,
|
| 122 |
+
"step": 190
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.53,
|
| 126 |
+
"learning_rate": 4.579667644183773e-05,
|
| 127 |
+
"loss": 0.3126,
|
| 128 |
+
"step": 200
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.55,
|
| 132 |
+
"learning_rate": 4.530791788856305e-05,
|
| 133 |
+
"loss": 0.2599,
|
| 134 |
+
"step": 210
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.58,
|
| 138 |
+
"learning_rate": 4.481915933528837e-05,
|
| 139 |
+
"loss": 0.3003,
|
| 140 |
+
"step": 220
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.61,
|
| 144 |
+
"learning_rate": 4.433040078201369e-05,
|
| 145 |
+
"loss": 0.2953,
|
| 146 |
+
"step": 230
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.63,
|
| 150 |
+
"learning_rate": 4.3841642228739e-05,
|
| 151 |
+
"loss": 0.299,
|
| 152 |
+
"step": 240
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.66,
|
| 156 |
+
"learning_rate": 4.3352883675464326e-05,
|
| 157 |
+
"loss": 0.3176,
|
| 158 |
+
"step": 250
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.68,
|
| 162 |
+
"learning_rate": 4.286412512218964e-05,
|
| 163 |
+
"loss": 0.3457,
|
| 164 |
+
"step": 260
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.71,
|
| 168 |
+
"learning_rate": 4.237536656891496e-05,
|
| 169 |
+
"loss": 0.2836,
|
| 170 |
+
"step": 270
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.74,
|
| 174 |
+
"learning_rate": 4.1886608015640275e-05,
|
| 175 |
+
"loss": 0.276,
|
| 176 |
+
"step": 280
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.76,
|
| 180 |
+
"learning_rate": 4.13978494623656e-05,
|
| 181 |
+
"loss": 0.286,
|
| 182 |
+
"step": 290
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.79,
|
| 186 |
+
"learning_rate": 4.0909090909090915e-05,
|
| 187 |
+
"loss": 0.2599,
|
| 188 |
+
"step": 300
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.82,
|
| 192 |
+
"learning_rate": 4.0420332355816224e-05,
|
| 193 |
+
"loss": 0.2062,
|
| 194 |
+
"step": 310
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.84,
|
| 198 |
+
"learning_rate": 3.993157380254155e-05,
|
| 199 |
+
"loss": 0.2683,
|
| 200 |
+
"step": 320
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.87,
|
| 204 |
+
"learning_rate": 3.9442815249266864e-05,
|
| 205 |
+
"loss": 0.2461,
|
| 206 |
+
"step": 330
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.9,
|
| 210 |
+
"learning_rate": 3.895405669599218e-05,
|
| 211 |
+
"loss": 0.2316,
|
| 212 |
+
"step": 340
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.92,
|
| 216 |
+
"learning_rate": 3.8465298142717496e-05,
|
| 217 |
+
"loss": 0.2065,
|
| 218 |
+
"step": 350
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.95,
|
| 222 |
+
"learning_rate": 3.797653958944282e-05,
|
| 223 |
+
"loss": 0.2518,
|
| 224 |
+
"step": 360
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.97,
|
| 228 |
+
"learning_rate": 3.7487781036168136e-05,
|
| 229 |
+
"loss": 0.2623,
|
| 230 |
+
"step": 370
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 1.0,
|
| 234 |
+
"eval_accuracy": 0.9674074074074074,
|
| 235 |
+
"eval_loss": 0.10060614347457886,
|
| 236 |
+
"eval_runtime": 39.1856,
|
| 237 |
+
"eval_samples_per_second": 68.903,
|
| 238 |
+
"eval_steps_per_second": 4.313,
|
| 239 |
+
"step": 379
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 1.0,
|
| 243 |
+
"learning_rate": 3.699902248289345e-05,
|
| 244 |
+
"loss": 0.2489,
|
| 245 |
+
"step": 380
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 1.03,
|
| 249 |
+
"learning_rate": 3.651026392961877e-05,
|
| 250 |
+
"loss": 0.1956,
|
| 251 |
+
"step": 390
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 1.06,
|
| 255 |
+
"learning_rate": 3.602150537634409e-05,
|
| 256 |
+
"loss": 0.2072,
|
| 257 |
+
"step": 400
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 1.08,
|
| 261 |
+
"learning_rate": 3.553274682306941e-05,
|
| 262 |
+
"loss": 0.2942,
|
| 263 |
+
"step": 410
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.11,
|
| 267 |
+
"learning_rate": 3.5043988269794724e-05,
|
| 268 |
+
"loss": 0.2417,
|
| 269 |
+
"step": 420
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 1.13,
|
| 273 |
+
"learning_rate": 3.455522971652004e-05,
|
| 274 |
+
"loss": 0.2737,
|
| 275 |
+
"step": 430
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 1.16,
|
| 279 |
+
"learning_rate": 3.406647116324536e-05,
|
| 280 |
+
"loss": 0.188,
|
| 281 |
+
"step": 440
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 1.19,
|
| 285 |
+
"learning_rate": 3.357771260997067e-05,
|
| 286 |
+
"loss": 0.1887,
|
| 287 |
+
"step": 450
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 1.21,
|
| 291 |
+
"learning_rate": 3.308895405669599e-05,
|
| 292 |
+
"loss": 0.162,
|
| 293 |
+
"step": 460
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 1.24,
|
| 297 |
+
"learning_rate": 3.260019550342131e-05,
|
| 298 |
+
"loss": 0.2178,
|
| 299 |
+
"step": 470
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 1.27,
|
| 303 |
+
"learning_rate": 3.211143695014663e-05,
|
| 304 |
+
"loss": 0.21,
|
| 305 |
+
"step": 480
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 1.29,
|
| 309 |
+
"learning_rate": 3.1622678396871946e-05,
|
| 310 |
+
"loss": 0.2613,
|
| 311 |
+
"step": 490
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 1.32,
|
| 315 |
+
"learning_rate": 3.113391984359726e-05,
|
| 316 |
+
"loss": 0.2346,
|
| 317 |
+
"step": 500
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 1.34,
|
| 321 |
+
"learning_rate": 3.0645161290322585e-05,
|
| 322 |
+
"loss": 0.2232,
|
| 323 |
+
"step": 510
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 1.37,
|
| 327 |
+
"learning_rate": 3.01564027370479e-05,
|
| 328 |
+
"loss": 0.2075,
|
| 329 |
+
"step": 520
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 1.4,
|
| 333 |
+
"learning_rate": 2.9667644183773218e-05,
|
| 334 |
+
"loss": 0.183,
|
| 335 |
+
"step": 530
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 1.42,
|
| 339 |
+
"learning_rate": 2.9178885630498537e-05,
|
| 340 |
+
"loss": 0.191,
|
| 341 |
+
"step": 540
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 1.45,
|
| 345 |
+
"learning_rate": 2.8690127077223854e-05,
|
| 346 |
+
"loss": 0.1784,
|
| 347 |
+
"step": 550
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 1.48,
|
| 351 |
+
"learning_rate": 2.8201368523949173e-05,
|
| 352 |
+
"loss": 0.1796,
|
| 353 |
+
"step": 560
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 1.5,
|
| 357 |
+
"learning_rate": 2.7712609970674486e-05,
|
| 358 |
+
"loss": 0.2068,
|
| 359 |
+
"step": 570
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 1.53,
|
| 363 |
+
"learning_rate": 2.7223851417399803e-05,
|
| 364 |
+
"loss": 0.1868,
|
| 365 |
+
"step": 580
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 1.56,
|
| 369 |
+
"learning_rate": 2.6735092864125122e-05,
|
| 370 |
+
"loss": 0.1637,
|
| 371 |
+
"step": 590
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 1.58,
|
| 375 |
+
"learning_rate": 2.624633431085044e-05,
|
| 376 |
+
"loss": 0.1847,
|
| 377 |
+
"step": 600
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 1.61,
|
| 381 |
+
"learning_rate": 2.575757575757576e-05,
|
| 382 |
+
"loss": 0.2313,
|
| 383 |
+
"step": 610
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 1.63,
|
| 387 |
+
"learning_rate": 2.5268817204301075e-05,
|
| 388 |
+
"loss": 0.1774,
|
| 389 |
+
"step": 620
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 1.66,
|
| 393 |
+
"learning_rate": 2.4780058651026395e-05,
|
| 394 |
+
"loss": 0.1533,
|
| 395 |
+
"step": 630
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 1.69,
|
| 399 |
+
"learning_rate": 2.429130009775171e-05,
|
| 400 |
+
"loss": 0.1553,
|
| 401 |
+
"step": 640
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 1.71,
|
| 405 |
+
"learning_rate": 2.380254154447703e-05,
|
| 406 |
+
"loss": 0.1509,
|
| 407 |
+
"step": 650
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 1.74,
|
| 411 |
+
"learning_rate": 2.3313782991202347e-05,
|
| 412 |
+
"loss": 0.1703,
|
| 413 |
+
"step": 660
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 1.77,
|
| 417 |
+
"learning_rate": 2.2825024437927663e-05,
|
| 418 |
+
"loss": 0.1582,
|
| 419 |
+
"step": 670
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 1.79,
|
| 423 |
+
"learning_rate": 2.2336265884652983e-05,
|
| 424 |
+
"loss": 0.1765,
|
| 425 |
+
"step": 680
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 1.82,
|
| 429 |
+
"learning_rate": 2.18475073313783e-05,
|
| 430 |
+
"loss": 0.1067,
|
| 431 |
+
"step": 690
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 1.85,
|
| 435 |
+
"learning_rate": 2.135874877810362e-05,
|
| 436 |
+
"loss": 0.2041,
|
| 437 |
+
"step": 700
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 1.87,
|
| 441 |
+
"learning_rate": 2.0869990224828935e-05,
|
| 442 |
+
"loss": 0.1287,
|
| 443 |
+
"step": 710
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 1.9,
|
| 447 |
+
"learning_rate": 2.0381231671554255e-05,
|
| 448 |
+
"loss": 0.1599,
|
| 449 |
+
"step": 720
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 1.92,
|
| 453 |
+
"learning_rate": 1.989247311827957e-05,
|
| 454 |
+
"loss": 0.1947,
|
| 455 |
+
"step": 730
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 1.95,
|
| 459 |
+
"learning_rate": 1.9403714565004888e-05,
|
| 460 |
+
"loss": 0.128,
|
| 461 |
+
"step": 740
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 1.98,
|
| 465 |
+
"learning_rate": 1.8914956011730208e-05,
|
| 466 |
+
"loss": 0.1712,
|
| 467 |
+
"step": 750
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 2.0,
|
| 471 |
+
"eval_accuracy": 0.9803703703703703,
|
| 472 |
+
"eval_loss": 0.06200645864009857,
|
| 473 |
+
"eval_runtime": 23.9639,
|
| 474 |
+
"eval_samples_per_second": 112.67,
|
| 475 |
+
"eval_steps_per_second": 7.052,
|
| 476 |
+
"step": 758
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 2.01,
|
| 480 |
+
"learning_rate": 1.8426197458455524e-05,
|
| 481 |
+
"loss": 0.1996,
|
| 482 |
+
"step": 760
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 2.03,
|
| 486 |
+
"learning_rate": 1.7937438905180844e-05,
|
| 487 |
+
"loss": 0.1536,
|
| 488 |
+
"step": 770
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 2.06,
|
| 492 |
+
"learning_rate": 1.744868035190616e-05,
|
| 493 |
+
"loss": 0.1541,
|
| 494 |
+
"step": 780
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 2.08,
|
| 498 |
+
"learning_rate": 1.6959921798631476e-05,
|
| 499 |
+
"loss": 0.1816,
|
| 500 |
+
"step": 790
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 2.11,
|
| 504 |
+
"learning_rate": 1.6471163245356793e-05,
|
| 505 |
+
"loss": 0.1873,
|
| 506 |
+
"step": 800
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 2.14,
|
| 510 |
+
"learning_rate": 1.5982404692082112e-05,
|
| 511 |
+
"loss": 0.1438,
|
| 512 |
+
"step": 810
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 2.16,
|
| 516 |
+
"learning_rate": 1.549364613880743e-05,
|
| 517 |
+
"loss": 0.1331,
|
| 518 |
+
"step": 820
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 2.19,
|
| 522 |
+
"learning_rate": 1.5004887585532748e-05,
|
| 523 |
+
"loss": 0.1349,
|
| 524 |
+
"step": 830
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 2.22,
|
| 528 |
+
"learning_rate": 1.4516129032258066e-05,
|
| 529 |
+
"loss": 0.1604,
|
| 530 |
+
"step": 840
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 2.24,
|
| 534 |
+
"learning_rate": 1.4027370478983381e-05,
|
| 535 |
+
"loss": 0.1458,
|
| 536 |
+
"step": 850
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 2.27,
|
| 540 |
+
"learning_rate": 1.3538611925708699e-05,
|
| 541 |
+
"loss": 0.1664,
|
| 542 |
+
"step": 860
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 2.29,
|
| 546 |
+
"learning_rate": 1.3049853372434017e-05,
|
| 547 |
+
"loss": 0.1661,
|
| 548 |
+
"step": 870
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 2.32,
|
| 552 |
+
"learning_rate": 1.2561094819159335e-05,
|
| 553 |
+
"loss": 0.1145,
|
| 554 |
+
"step": 880
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 2.35,
|
| 558 |
+
"learning_rate": 1.2072336265884653e-05,
|
| 559 |
+
"loss": 0.1451,
|
| 560 |
+
"step": 890
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 2.37,
|
| 564 |
+
"learning_rate": 1.1583577712609971e-05,
|
| 565 |
+
"loss": 0.125,
|
| 566 |
+
"step": 900
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 2.4,
|
| 570 |
+
"learning_rate": 1.109481915933529e-05,
|
| 571 |
+
"loss": 0.166,
|
| 572 |
+
"step": 910
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 2.43,
|
| 576 |
+
"learning_rate": 1.0606060606060607e-05,
|
| 577 |
+
"loss": 0.1224,
|
| 578 |
+
"step": 920
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 2.45,
|
| 582 |
+
"learning_rate": 1.0117302052785924e-05,
|
| 583 |
+
"loss": 0.1164,
|
| 584 |
+
"step": 930
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 2.48,
|
| 588 |
+
"learning_rate": 9.628543499511242e-06,
|
| 589 |
+
"loss": 0.1776,
|
| 590 |
+
"step": 940
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 2.51,
|
| 594 |
+
"learning_rate": 9.13978494623656e-06,
|
| 595 |
+
"loss": 0.1205,
|
| 596 |
+
"step": 950
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 2.53,
|
| 600 |
+
"learning_rate": 8.651026392961878e-06,
|
| 601 |
+
"loss": 0.1365,
|
| 602 |
+
"step": 960
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 2.56,
|
| 606 |
+
"learning_rate": 8.162267839687194e-06,
|
| 607 |
+
"loss": 0.1259,
|
| 608 |
+
"step": 970
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 2.58,
|
| 612 |
+
"learning_rate": 7.673509286412512e-06,
|
| 613 |
+
"loss": 0.1319,
|
| 614 |
+
"step": 980
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 2.61,
|
| 618 |
+
"learning_rate": 7.184750733137831e-06,
|
| 619 |
+
"loss": 0.1263,
|
| 620 |
+
"step": 990
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 2.64,
|
| 624 |
+
"learning_rate": 6.695992179863147e-06,
|
| 625 |
+
"loss": 0.1881,
|
| 626 |
+
"step": 1000
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 2.66,
|
| 630 |
+
"learning_rate": 6.207233626588465e-06,
|
| 631 |
+
"loss": 0.1263,
|
| 632 |
+
"step": 1010
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 2.69,
|
| 636 |
+
"learning_rate": 5.7184750733137834e-06,
|
| 637 |
+
"loss": 0.1082,
|
| 638 |
+
"step": 1020
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 2.72,
|
| 642 |
+
"learning_rate": 5.2297165200391015e-06,
|
| 643 |
+
"loss": 0.1434,
|
| 644 |
+
"step": 1030
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 2.74,
|
| 648 |
+
"learning_rate": 4.740957966764419e-06,
|
| 649 |
+
"loss": 0.0859,
|
| 650 |
+
"step": 1040
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 2.77,
|
| 654 |
+
"learning_rate": 4.252199413489736e-06,
|
| 655 |
+
"loss": 0.1362,
|
| 656 |
+
"step": 1050
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 2.8,
|
| 660 |
+
"learning_rate": 3.763440860215054e-06,
|
| 661 |
+
"loss": 0.1036,
|
| 662 |
+
"step": 1060
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 2.82,
|
| 666 |
+
"learning_rate": 3.2746823069403715e-06,
|
| 667 |
+
"loss": 0.1185,
|
| 668 |
+
"step": 1070
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 2.85,
|
| 672 |
+
"learning_rate": 2.785923753665689e-06,
|
| 673 |
+
"loss": 0.1534,
|
| 674 |
+
"step": 1080
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 2.87,
|
| 678 |
+
"learning_rate": 2.297165200391007e-06,
|
| 679 |
+
"loss": 0.1346,
|
| 680 |
+
"step": 1090
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 2.9,
|
| 684 |
+
"learning_rate": 1.8084066471163247e-06,
|
| 685 |
+
"loss": 0.1079,
|
| 686 |
+
"step": 1100
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 2.93,
|
| 690 |
+
"learning_rate": 1.3196480938416423e-06,
|
| 691 |
+
"loss": 0.1347,
|
| 692 |
+
"step": 1110
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 2.95,
|
| 696 |
+
"learning_rate": 8.308895405669599e-07,
|
| 697 |
+
"loss": 0.1662,
|
| 698 |
+
"step": 1120
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 2.98,
|
| 702 |
+
"learning_rate": 3.4213098729227765e-07,
|
| 703 |
+
"loss": 0.1206,
|
| 704 |
+
"step": 1130
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 3.0,
|
| 708 |
+
"eval_accuracy": 0.9811111111111112,
|
| 709 |
+
"eval_loss": 0.04829270392656326,
|
| 710 |
+
"eval_runtime": 23.9411,
|
| 711 |
+
"eval_samples_per_second": 112.777,
|
| 712 |
+
"eval_steps_per_second": 7.059,
|
| 713 |
+
"step": 1137
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 3.0,
|
| 717 |
+
"step": 1137,
|
| 718 |
+
"total_flos": 1.8113127425839596e+18,
|
| 719 |
+
"train_loss": 0.29242842561231963,
|
| 720 |
+
"train_runtime": 1937.8226,
|
| 721 |
+
"train_samples_per_second": 37.62,
|
| 722 |
+
"train_steps_per_second": 0.587
|
| 723 |
+
}
|
| 724 |
+
],
|
| 725 |
+
"max_steps": 1137,
|
| 726 |
+
"num_train_epochs": 3,
|
| 727 |
+
"total_flos": 1.8113127425839596e+18,
|
| 728 |
+
"trial_name": null,
|
| 729 |
+
"trial_params": null
|
| 730 |
+
}
|