End of training
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 14.99,
|
| 3 |
+
"eval_accuracy": 1.0,
|
| 4 |
+
"eval_loss": 0.007474538870155811,
|
| 5 |
+
"eval_runtime": 6.689,
|
| 6 |
+
"eval_samples_per_second": 62.341,
|
| 7 |
+
"eval_steps_per_second": 4.036,
|
| 8 |
+
"total_flos": 5.035680667331113e+18,
|
| 9 |
+
"train_loss": 0.13053628388996147,
|
| 10 |
+
"train_runtime": 2520.4596,
|
| 11 |
+
"train_samples_per_second": 25.799,
|
| 12 |
+
"train_steps_per_second": 0.399
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 14.99,
|
| 3 |
+
"eval_accuracy": 1.0,
|
| 4 |
+
"eval_loss": 0.007474538870155811,
|
| 5 |
+
"eval_runtime": 6.689,
|
| 6 |
+
"eval_samples_per_second": 62.341,
|
| 7 |
+
"eval_steps_per_second": 4.036
|
| 8 |
+
}
|
runs/Jul15_08-20-36_69b1e7626d29/events.out.tfevents.1657876260.69b1e7626d29.72.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3046375eca418249b13fec1f241595ab18dfa6dc24f54a96f5ca5324a1438516
|
| 3 |
+
size 363
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 14.99,
|
| 3 |
+
"total_flos": 5.035680667331113e+18,
|
| 4 |
+
"train_loss": 0.13053628388996147,
|
| 5 |
+
"train_runtime": 2520.4596,
|
| 6 |
+
"train_samples_per_second": 25.799,
|
| 7 |
+
"train_steps_per_second": 0.399
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,760 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 1.0,
|
| 3 |
+
"best_model_checkpoint": "ViTFineTuned/checkpoint-603",
|
| 4 |
+
"epoch": 14.988929889298893,
|
| 5 |
+
"global_step": 1005,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.15,
|
| 12 |
+
"learning_rate": 4.950495049504951e-05,
|
| 13 |
+
"loss": 2.3113,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.3,
|
| 18 |
+
"learning_rate": 9.900990099009902e-05,
|
| 19 |
+
"loss": 1.8782,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.44,
|
| 24 |
+
"learning_rate": 0.0001485148514851485,
|
| 25 |
+
"loss": 1.2023,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.59,
|
| 30 |
+
"learning_rate": 0.00019801980198019803,
|
| 31 |
+
"loss": 0.6597,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.74,
|
| 36 |
+
"learning_rate": 0.00024752475247524753,
|
| 37 |
+
"loss": 0.4134,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.89,
|
| 42 |
+
"learning_rate": 0.000297029702970297,
|
| 43 |
+
"loss": 0.2859,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.99,
|
| 48 |
+
"eval_accuracy": 0.9784172661870504,
|
| 49 |
+
"eval_loss": 0.21802514791488647,
|
| 50 |
+
"eval_runtime": 6.2997,
|
| 51 |
+
"eval_samples_per_second": 66.193,
|
| 52 |
+
"eval_steps_per_second": 4.286,
|
| 53 |
+
"step": 67
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 1.04,
|
| 57 |
+
"learning_rate": 0.0003465346534653465,
|
| 58 |
+
"loss": 0.285,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 1.19,
|
| 63 |
+
"learning_rate": 0.00039603960396039607,
|
| 64 |
+
"loss": 0.261,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 1.34,
|
| 69 |
+
"learning_rate": 0.00044554455445544556,
|
| 70 |
+
"loss": 0.2659,
|
| 71 |
+
"step": 90
|
| 72 |
+
},
|
| 73 |
+
{
|
| 74 |
+
"epoch": 1.49,
|
| 75 |
+
"learning_rate": 0.0004950495049504951,
|
| 76 |
+
"loss": 0.2812,
|
| 77 |
+
"step": 100
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 1.63,
|
| 81 |
+
"learning_rate": 0.0004950221238938053,
|
| 82 |
+
"loss": 0.2512,
|
| 83 |
+
"step": 110
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 1.78,
|
| 87 |
+
"learning_rate": 0.0004894911504424779,
|
| 88 |
+
"loss": 0.2033,
|
| 89 |
+
"step": 120
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 1.93,
|
| 93 |
+
"learning_rate": 0.0004839601769911505,
|
| 94 |
+
"loss": 0.293,
|
| 95 |
+
"step": 130
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 1.99,
|
| 99 |
+
"eval_accuracy": 0.9184652278177458,
|
| 100 |
+
"eval_loss": 0.3308344781398773,
|
| 101 |
+
"eval_runtime": 6.9556,
|
| 102 |
+
"eval_samples_per_second": 59.952,
|
| 103 |
+
"eval_steps_per_second": 3.882,
|
| 104 |
+
"step": 134
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 2.09,
|
| 108 |
+
"learning_rate": 0.000478429203539823,
|
| 109 |
+
"loss": 0.2754,
|
| 110 |
+
"step": 140
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 2.24,
|
| 114 |
+
"learning_rate": 0.0004728982300884956,
|
| 115 |
+
"loss": 0.2746,
|
| 116 |
+
"step": 150
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 2.38,
|
| 120 |
+
"learning_rate": 0.00046736725663716817,
|
| 121 |
+
"loss": 0.1477,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 2.53,
|
| 126 |
+
"learning_rate": 0.0004618362831858407,
|
| 127 |
+
"loss": 0.1984,
|
| 128 |
+
"step": 170
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 2.68,
|
| 132 |
+
"learning_rate": 0.00045630530973451327,
|
| 133 |
+
"loss": 0.279,
|
| 134 |
+
"step": 180
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 2.83,
|
| 138 |
+
"learning_rate": 0.00045077433628318585,
|
| 139 |
+
"loss": 0.1846,
|
| 140 |
+
"step": 190
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 2.97,
|
| 144 |
+
"learning_rate": 0.0004452433628318584,
|
| 145 |
+
"loss": 0.1444,
|
| 146 |
+
"step": 200
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 2.99,
|
| 150 |
+
"eval_accuracy": 0.9568345323741008,
|
| 151 |
+
"eval_loss": 0.15321111679077148,
|
| 152 |
+
"eval_runtime": 6.3227,
|
| 153 |
+
"eval_samples_per_second": 65.953,
|
| 154 |
+
"eval_steps_per_second": 4.27,
|
| 155 |
+
"step": 201
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 3.13,
|
| 159 |
+
"learning_rate": 0.000439712389380531,
|
| 160 |
+
"loss": 0.1136,
|
| 161 |
+
"step": 210
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 3.28,
|
| 165 |
+
"learning_rate": 0.0004341814159292036,
|
| 166 |
+
"loss": 0.1482,
|
| 167 |
+
"step": 220
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 3.43,
|
| 171 |
+
"learning_rate": 0.0004286504424778761,
|
| 172 |
+
"loss": 0.1124,
|
| 173 |
+
"step": 230
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 3.58,
|
| 177 |
+
"learning_rate": 0.0004231194690265487,
|
| 178 |
+
"loss": 0.2128,
|
| 179 |
+
"step": 240
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 3.72,
|
| 183 |
+
"learning_rate": 0.00041758849557522125,
|
| 184 |
+
"loss": 0.0859,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 3.87,
|
| 189 |
+
"learning_rate": 0.0004120575221238938,
|
| 190 |
+
"loss": 0.0833,
|
| 191 |
+
"step": 260
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 3.99,
|
| 195 |
+
"eval_accuracy": 0.9856115107913669,
|
| 196 |
+
"eval_loss": 0.051548197865486145,
|
| 197 |
+
"eval_runtime": 6.2776,
|
| 198 |
+
"eval_samples_per_second": 66.426,
|
| 199 |
+
"eval_steps_per_second": 4.301,
|
| 200 |
+
"step": 268
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 4.03,
|
| 204 |
+
"learning_rate": 0.0004065265486725664,
|
| 205 |
+
"loss": 0.1025,
|
| 206 |
+
"step": 270
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 4.18,
|
| 210 |
+
"learning_rate": 0.000400995575221239,
|
| 211 |
+
"loss": 0.0917,
|
| 212 |
+
"step": 280
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 4.32,
|
| 216 |
+
"learning_rate": 0.0003954646017699115,
|
| 217 |
+
"loss": 0.0812,
|
| 218 |
+
"step": 290
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 4.47,
|
| 222 |
+
"learning_rate": 0.0003899336283185841,
|
| 223 |
+
"loss": 0.0879,
|
| 224 |
+
"step": 300
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 4.62,
|
| 228 |
+
"learning_rate": 0.0003844026548672566,
|
| 229 |
+
"loss": 0.0763,
|
| 230 |
+
"step": 310
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 4.77,
|
| 234 |
+
"learning_rate": 0.0003788716814159292,
|
| 235 |
+
"loss": 0.0582,
|
| 236 |
+
"step": 320
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 4.92,
|
| 240 |
+
"learning_rate": 0.00037334070796460176,
|
| 241 |
+
"loss": 0.1007,
|
| 242 |
+
"step": 330
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 4.99,
|
| 246 |
+
"eval_accuracy": 0.9904076738609112,
|
| 247 |
+
"eval_loss": 0.029486695304512978,
|
| 248 |
+
"eval_runtime": 6.2974,
|
| 249 |
+
"eval_samples_per_second": 66.218,
|
| 250 |
+
"eval_steps_per_second": 4.288,
|
| 251 |
+
"step": 335
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 5.07,
|
| 255 |
+
"learning_rate": 0.00036780973451327434,
|
| 256 |
+
"loss": 0.0564,
|
| 257 |
+
"step": 340
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 5.22,
|
| 261 |
+
"learning_rate": 0.0003622787610619469,
|
| 262 |
+
"loss": 0.0506,
|
| 263 |
+
"step": 350
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 5.37,
|
| 267 |
+
"learning_rate": 0.0003567477876106195,
|
| 268 |
+
"loss": 0.0611,
|
| 269 |
+
"step": 360
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 5.52,
|
| 273 |
+
"learning_rate": 0.000351216814159292,
|
| 274 |
+
"loss": 0.0576,
|
| 275 |
+
"step": 370
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 5.66,
|
| 279 |
+
"learning_rate": 0.0003456858407079646,
|
| 280 |
+
"loss": 0.0832,
|
| 281 |
+
"step": 380
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 5.81,
|
| 285 |
+
"learning_rate": 0.0003401548672566372,
|
| 286 |
+
"loss": 0.0561,
|
| 287 |
+
"step": 390
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 5.96,
|
| 291 |
+
"learning_rate": 0.0003346238938053097,
|
| 292 |
+
"loss": 0.0372,
|
| 293 |
+
"step": 400
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 5.99,
|
| 297 |
+
"eval_accuracy": 0.9808153477218226,
|
| 298 |
+
"eval_loss": 0.05741920694708824,
|
| 299 |
+
"eval_runtime": 6.2799,
|
| 300 |
+
"eval_samples_per_second": 66.402,
|
| 301 |
+
"eval_steps_per_second": 4.299,
|
| 302 |
+
"step": 402
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 6.12,
|
| 306 |
+
"learning_rate": 0.00032909292035398233,
|
| 307 |
+
"loss": 0.0749,
|
| 308 |
+
"step": 410
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 6.27,
|
| 312 |
+
"learning_rate": 0.0003235619469026549,
|
| 313 |
+
"loss": 0.0464,
|
| 314 |
+
"step": 420
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 6.41,
|
| 318 |
+
"learning_rate": 0.00031803097345132743,
|
| 319 |
+
"loss": 0.0538,
|
| 320 |
+
"step": 430
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 6.56,
|
| 324 |
+
"learning_rate": 0.0003125,
|
| 325 |
+
"loss": 0.073,
|
| 326 |
+
"step": 440
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 6.71,
|
| 330 |
+
"learning_rate": 0.0003069690265486726,
|
| 331 |
+
"loss": 0.0533,
|
| 332 |
+
"step": 450
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 6.86,
|
| 336 |
+
"learning_rate": 0.0003014380530973451,
|
| 337 |
+
"loss": 0.0919,
|
| 338 |
+
"step": 460
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 6.99,
|
| 342 |
+
"eval_accuracy": 0.988009592326139,
|
| 343 |
+
"eval_loss": 0.05372486636042595,
|
| 344 |
+
"eval_runtime": 6.3165,
|
| 345 |
+
"eval_samples_per_second": 66.017,
|
| 346 |
+
"eval_steps_per_second": 4.275,
|
| 347 |
+
"step": 469
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 7.01,
|
| 351 |
+
"learning_rate": 0.0002959070796460177,
|
| 352 |
+
"loss": 0.0518,
|
| 353 |
+
"step": 470
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 7.16,
|
| 357 |
+
"learning_rate": 0.0002903761061946903,
|
| 358 |
+
"loss": 0.0233,
|
| 359 |
+
"step": 480
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 7.31,
|
| 363 |
+
"learning_rate": 0.00028484513274336284,
|
| 364 |
+
"loss": 0.0501,
|
| 365 |
+
"step": 490
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 7.46,
|
| 369 |
+
"learning_rate": 0.0002793141592920354,
|
| 370 |
+
"loss": 0.0685,
|
| 371 |
+
"step": 500
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 7.61,
|
| 375 |
+
"learning_rate": 0.000273783185840708,
|
| 376 |
+
"loss": 0.0116,
|
| 377 |
+
"step": 510
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 7.75,
|
| 381 |
+
"learning_rate": 0.0002682522123893805,
|
| 382 |
+
"loss": 0.0228,
|
| 383 |
+
"step": 520
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 7.9,
|
| 387 |
+
"learning_rate": 0.0002627212389380531,
|
| 388 |
+
"loss": 0.0135,
|
| 389 |
+
"step": 530
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 7.99,
|
| 393 |
+
"eval_accuracy": 0.9952038369304557,
|
| 394 |
+
"eval_loss": 0.011652274057269096,
|
| 395 |
+
"eval_runtime": 6.3118,
|
| 396 |
+
"eval_samples_per_second": 66.067,
|
| 397 |
+
"eval_steps_per_second": 4.278,
|
| 398 |
+
"step": 536
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 8.06,
|
| 402 |
+
"learning_rate": 0.00025719026548672567,
|
| 403 |
+
"loss": 0.0091,
|
| 404 |
+
"step": 540
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 8.21,
|
| 408 |
+
"learning_rate": 0.00025165929203539825,
|
| 409 |
+
"loss": 0.0469,
|
| 410 |
+
"step": 550
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 8.35,
|
| 414 |
+
"learning_rate": 0.0002461283185840708,
|
| 415 |
+
"loss": 0.0157,
|
| 416 |
+
"step": 560
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 8.5,
|
| 420 |
+
"learning_rate": 0.00024059734513274338,
|
| 421 |
+
"loss": 0.0172,
|
| 422 |
+
"step": 570
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 8.65,
|
| 426 |
+
"learning_rate": 0.00023506637168141593,
|
| 427 |
+
"loss": 0.0202,
|
| 428 |
+
"step": 580
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 8.8,
|
| 432 |
+
"learning_rate": 0.0002295353982300885,
|
| 433 |
+
"loss": 0.0303,
|
| 434 |
+
"step": 590
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 8.94,
|
| 438 |
+
"learning_rate": 0.00022400442477876108,
|
| 439 |
+
"loss": 0.0472,
|
| 440 |
+
"step": 600
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 8.99,
|
| 444 |
+
"eval_accuracy": 1.0,
|
| 445 |
+
"eval_loss": 0.007474538870155811,
|
| 446 |
+
"eval_runtime": 6.2502,
|
| 447 |
+
"eval_samples_per_second": 66.718,
|
| 448 |
+
"eval_steps_per_second": 4.32,
|
| 449 |
+
"step": 603
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 9.1,
|
| 453 |
+
"learning_rate": 0.00021847345132743363,
|
| 454 |
+
"loss": 0.0147,
|
| 455 |
+
"step": 610
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 9.25,
|
| 459 |
+
"learning_rate": 0.0002129424778761062,
|
| 460 |
+
"loss": 0.0133,
|
| 461 |
+
"step": 620
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 9.4,
|
| 465 |
+
"learning_rate": 0.00020741150442477876,
|
| 466 |
+
"loss": 0.0264,
|
| 467 |
+
"step": 630
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 9.55,
|
| 471 |
+
"learning_rate": 0.00020188053097345134,
|
| 472 |
+
"loss": 0.053,
|
| 473 |
+
"step": 640
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 9.69,
|
| 477 |
+
"learning_rate": 0.0001963495575221239,
|
| 478 |
+
"loss": 0.0056,
|
| 479 |
+
"step": 650
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 9.84,
|
| 483 |
+
"learning_rate": 0.00019081858407079646,
|
| 484 |
+
"loss": 0.0128,
|
| 485 |
+
"step": 660
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 9.99,
|
| 489 |
+
"learning_rate": 0.00018528761061946904,
|
| 490 |
+
"loss": 0.0151,
|
| 491 |
+
"step": 670
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 9.99,
|
| 495 |
+
"eval_accuracy": 1.0,
|
| 496 |
+
"eval_loss": 0.004758651368319988,
|
| 497 |
+
"eval_runtime": 6.2595,
|
| 498 |
+
"eval_samples_per_second": 66.619,
|
| 499 |
+
"eval_steps_per_second": 4.313,
|
| 500 |
+
"step": 670
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 10.15,
|
| 504 |
+
"learning_rate": 0.0001797566371681416,
|
| 505 |
+
"loss": 0.0135,
|
| 506 |
+
"step": 680
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 10.3,
|
| 510 |
+
"learning_rate": 0.00017422566371681417,
|
| 511 |
+
"loss": 0.0394,
|
| 512 |
+
"step": 690
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 10.44,
|
| 516 |
+
"learning_rate": 0.00016869469026548672,
|
| 517 |
+
"loss": 0.0122,
|
| 518 |
+
"step": 700
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 10.59,
|
| 522 |
+
"learning_rate": 0.0001631637168141593,
|
| 523 |
+
"loss": 0.0136,
|
| 524 |
+
"step": 710
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 10.74,
|
| 528 |
+
"learning_rate": 0.00015763274336283187,
|
| 529 |
+
"loss": 0.01,
|
| 530 |
+
"step": 720
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 10.89,
|
| 534 |
+
"learning_rate": 0.00015210176991150442,
|
| 535 |
+
"loss": 0.0052,
|
| 536 |
+
"step": 730
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 10.99,
|
| 540 |
+
"eval_accuracy": 0.9976019184652278,
|
| 541 |
+
"eval_loss": 0.007309095934033394,
|
| 542 |
+
"eval_runtime": 6.2533,
|
| 543 |
+
"eval_samples_per_second": 66.685,
|
| 544 |
+
"eval_steps_per_second": 4.318,
|
| 545 |
+
"step": 737
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 11.04,
|
| 549 |
+
"learning_rate": 0.00014657079646017697,
|
| 550 |
+
"loss": 0.0113,
|
| 551 |
+
"step": 740
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 11.19,
|
| 555 |
+
"learning_rate": 0.00014103982300884958,
|
| 556 |
+
"loss": 0.0325,
|
| 557 |
+
"step": 750
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 11.34,
|
| 561 |
+
"learning_rate": 0.00013550884955752213,
|
| 562 |
+
"loss": 0.012,
|
| 563 |
+
"step": 760
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 11.49,
|
| 567 |
+
"learning_rate": 0.00012997787610619468,
|
| 568 |
+
"loss": 0.0185,
|
| 569 |
+
"step": 770
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 11.63,
|
| 573 |
+
"learning_rate": 0.00012444690265486725,
|
| 574 |
+
"loss": 0.0049,
|
| 575 |
+
"step": 780
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 11.78,
|
| 579 |
+
"learning_rate": 0.00011891592920353983,
|
| 580 |
+
"loss": 0.0038,
|
| 581 |
+
"step": 790
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 11.93,
|
| 585 |
+
"learning_rate": 0.00011338495575221238,
|
| 586 |
+
"loss": 0.0109,
|
| 587 |
+
"step": 800
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 11.99,
|
| 591 |
+
"eval_accuracy": 0.9952038369304557,
|
| 592 |
+
"eval_loss": 0.01980188488960266,
|
| 593 |
+
"eval_runtime": 6.279,
|
| 594 |
+
"eval_samples_per_second": 66.411,
|
| 595 |
+
"eval_steps_per_second": 4.3,
|
| 596 |
+
"step": 804
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 12.09,
|
| 600 |
+
"learning_rate": 0.00010785398230088496,
|
| 601 |
+
"loss": 0.0096,
|
| 602 |
+
"step": 810
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 12.24,
|
| 606 |
+
"learning_rate": 0.00010232300884955754,
|
| 607 |
+
"loss": 0.0115,
|
| 608 |
+
"step": 820
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 12.38,
|
| 612 |
+
"learning_rate": 9.679203539823009e-05,
|
| 613 |
+
"loss": 0.0171,
|
| 614 |
+
"step": 830
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 12.53,
|
| 618 |
+
"learning_rate": 9.126106194690266e-05,
|
| 619 |
+
"loss": 0.004,
|
| 620 |
+
"step": 840
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 12.68,
|
| 624 |
+
"learning_rate": 8.573008849557521e-05,
|
| 625 |
+
"loss": 0.0073,
|
| 626 |
+
"step": 850
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 12.83,
|
| 630 |
+
"learning_rate": 8.019911504424779e-05,
|
| 631 |
+
"loss": 0.0045,
|
| 632 |
+
"step": 860
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 12.97,
|
| 636 |
+
"learning_rate": 7.466814159292036e-05,
|
| 637 |
+
"loss": 0.0033,
|
| 638 |
+
"step": 870
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 12.99,
|
| 642 |
+
"eval_accuracy": 0.9976019184652278,
|
| 643 |
+
"eval_loss": 0.006568592973053455,
|
| 644 |
+
"eval_runtime": 6.2861,
|
| 645 |
+
"eval_samples_per_second": 66.336,
|
| 646 |
+
"eval_steps_per_second": 4.295,
|
| 647 |
+
"step": 871
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 13.13,
|
| 651 |
+
"learning_rate": 6.913716814159292e-05,
|
| 652 |
+
"loss": 0.0077,
|
| 653 |
+
"step": 880
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 13.28,
|
| 657 |
+
"learning_rate": 6.360619469026548e-05,
|
| 658 |
+
"loss": 0.0027,
|
| 659 |
+
"step": 890
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 13.43,
|
| 663 |
+
"learning_rate": 5.8075221238938054e-05,
|
| 664 |
+
"loss": 0.0028,
|
| 665 |
+
"step": 900
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 13.58,
|
| 669 |
+
"learning_rate": 5.254424778761062e-05,
|
| 670 |
+
"loss": 0.0025,
|
| 671 |
+
"step": 910
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 13.72,
|
| 675 |
+
"learning_rate": 4.701327433628318e-05,
|
| 676 |
+
"loss": 0.0024,
|
| 677 |
+
"step": 920
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 13.87,
|
| 681 |
+
"learning_rate": 4.148230088495576e-05,
|
| 682 |
+
"loss": 0.011,
|
| 683 |
+
"step": 930
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 13.99,
|
| 687 |
+
"eval_accuracy": 0.9976019184652278,
|
| 688 |
+
"eval_loss": 0.006684896536171436,
|
| 689 |
+
"eval_runtime": 6.258,
|
| 690 |
+
"eval_samples_per_second": 66.635,
|
| 691 |
+
"eval_steps_per_second": 4.314,
|
| 692 |
+
"step": 938
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 14.03,
|
| 696 |
+
"learning_rate": 3.595132743362832e-05,
|
| 697 |
+
"loss": 0.0027,
|
| 698 |
+
"step": 940
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 14.18,
|
| 702 |
+
"learning_rate": 3.0420353982300886e-05,
|
| 703 |
+
"loss": 0.0092,
|
| 704 |
+
"step": 950
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 14.32,
|
| 708 |
+
"learning_rate": 2.4889380530973453e-05,
|
| 709 |
+
"loss": 0.0025,
|
| 710 |
+
"step": 960
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 14.47,
|
| 714 |
+
"learning_rate": 1.9358407079646017e-05,
|
| 715 |
+
"loss": 0.0024,
|
| 716 |
+
"step": 970
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 14.62,
|
| 720 |
+
"learning_rate": 1.3827433628318586e-05,
|
| 721 |
+
"loss": 0.0061,
|
| 722 |
+
"step": 980
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 14.77,
|
| 726 |
+
"learning_rate": 8.296460176991151e-06,
|
| 727 |
+
"loss": 0.0028,
|
| 728 |
+
"step": 990
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 14.92,
|
| 732 |
+
"learning_rate": 2.765486725663717e-06,
|
| 733 |
+
"loss": 0.0032,
|
| 734 |
+
"step": 1000
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 14.99,
|
| 738 |
+
"eval_accuracy": 0.9976019184652278,
|
| 739 |
+
"eval_loss": 0.00603157514706254,
|
| 740 |
+
"eval_runtime": 6.2764,
|
| 741 |
+
"eval_samples_per_second": 66.44,
|
| 742 |
+
"eval_steps_per_second": 4.302,
|
| 743 |
+
"step": 1005
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 14.99,
|
| 747 |
+
"step": 1005,
|
| 748 |
+
"total_flos": 5.035680667331113e+18,
|
| 749 |
+
"train_loss": 0.13053628388996147,
|
| 750 |
+
"train_runtime": 2520.4596,
|
| 751 |
+
"train_samples_per_second": 25.799,
|
| 752 |
+
"train_steps_per_second": 0.399
|
| 753 |
+
}
|
| 754 |
+
],
|
| 755 |
+
"max_steps": 1005,
|
| 756 |
+
"num_train_epochs": 15,
|
| 757 |
+
"total_flos": 5.035680667331113e+18,
|
| 758 |
+
"trial_name": null,
|
| 759 |
+
"trial_params": null
|
| 760 |
+
}
|