Commit
•
c790bfc
1
Parent(s):
c2dc058
End of training
Browse files- all_results.json +14 -0
- eval_results.json +9 -0
- pytorch_model.bin +1 -1
- train_results.json +8 -0
- trainer_state.json +670 -0
all_results.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 120.0,
|
| 3 |
+
"eval_loss": 0.21628186106681824,
|
| 4 |
+
"eval_runtime": 72.9369,
|
| 5 |
+
"eval_samples": 1240,
|
| 6 |
+
"eval_samples_per_second": 17.001,
|
| 7 |
+
"eval_steps_per_second": 17.001,
|
| 8 |
+
"eval_wer": 0.198074787115883,
|
| 9 |
+
"train_loss": 1.2021882954097929,
|
| 10 |
+
"train_runtime": 20445.8304,
|
| 11 |
+
"train_samples": 2675,
|
| 12 |
+
"train_samples_per_second": 15.7,
|
| 13 |
+
"train_steps_per_second": 0.493
|
| 14 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 120.0,
|
| 3 |
+
"eval_loss": 0.21628186106681824,
|
| 4 |
+
"eval_runtime": 72.9369,
|
| 5 |
+
"eval_samples": 1240,
|
| 6 |
+
"eval_samples_per_second": 17.001,
|
| 7 |
+
"eval_steps_per_second": 17.001,
|
| 8 |
+
"eval_wer": 0.198074787115883
|
| 9 |
+
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1262087665
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6e7d539bf194d40ff26c948b37add6052429f8267aaf526d65ec3e741825a328
|
| 3 |
size 1262087665
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 120.0,
|
| 3 |
+
"train_loss": 1.2021882954097929,
|
| 4 |
+
"train_runtime": 20445.8304,
|
| 5 |
+
"train_samples": 2675,
|
| 6 |
+
"train_samples_per_second": 15.7,
|
| 7 |
+
"train_steps_per_second": 0.493
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,670 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 120.0,
|
| 5 |
+
"global_step": 10080,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 1.19,
|
| 12 |
+
"learning_rate": 3.3949999999999997e-06,
|
| 13 |
+
"loss": 16.6111,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 2.38,
|
| 18 |
+
"learning_rate": 6.895e-06,
|
| 19 |
+
"loss": 8.8994,
|
| 20 |
+
"step": 200
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 3.57,
|
| 24 |
+
"learning_rate": 1.0394999999999998e-05,
|
| 25 |
+
"loss": 5.2091,
|
| 26 |
+
"step": 300
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 4.76,
|
| 30 |
+
"learning_rate": 1.3895e-05,
|
| 31 |
+
"loss": 4.3004,
|
| 32 |
+
"step": 400
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 5.95,
|
| 36 |
+
"learning_rate": 1.7395e-05,
|
| 37 |
+
"loss": 3.6606,
|
| 38 |
+
"step": 500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 7.14,
|
| 42 |
+
"learning_rate": 2.0894999999999996e-05,
|
| 43 |
+
"loss": 3.2383,
|
| 44 |
+
"step": 600
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 8.33,
|
| 48 |
+
"learning_rate": 2.4394999999999996e-05,
|
| 49 |
+
"loss": 3.0354,
|
| 50 |
+
"step": 700
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 9.52,
|
| 54 |
+
"learning_rate": 2.7895e-05,
|
| 55 |
+
"loss": 2.9742,
|
| 56 |
+
"step": 800
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 10.71,
|
| 60 |
+
"learning_rate": 3.1395e-05,
|
| 61 |
+
"loss": 2.9316,
|
| 62 |
+
"step": 900
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 11.9,
|
| 66 |
+
"learning_rate": 3.4895e-05,
|
| 67 |
+
"loss": 2.8914,
|
| 68 |
+
"step": 1000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 13.1,
|
| 72 |
+
"learning_rate": 3.8394999999999994e-05,
|
| 73 |
+
"loss": 2.8328,
|
| 74 |
+
"step": 1100
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 14.29,
|
| 78 |
+
"learning_rate": 4.1895e-05,
|
| 79 |
+
"loss": 2.8058,
|
| 80 |
+
"step": 1200
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 15.48,
|
| 84 |
+
"learning_rate": 4.5394999999999995e-05,
|
| 85 |
+
"loss": 2.7434,
|
| 86 |
+
"step": 1300
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 16.67,
|
| 90 |
+
"learning_rate": 4.8895e-05,
|
| 91 |
+
"loss": 2.2153,
|
| 92 |
+
"step": 1400
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 17.86,
|
| 96 |
+
"learning_rate": 5.2395e-05,
|
| 97 |
+
"loss": 1.606,
|
| 98 |
+
"step": 1500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 19.05,
|
| 102 |
+
"learning_rate": 5.589499999999999e-05,
|
| 103 |
+
"loss": 1.3918,
|
| 104 |
+
"step": 1600
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 20.24,
|
| 108 |
+
"learning_rate": 5.9394999999999996e-05,
|
| 109 |
+
"loss": 1.2684,
|
| 110 |
+
"step": 1700
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 21.43,
|
| 114 |
+
"learning_rate": 6.289499999999999e-05,
|
| 115 |
+
"loss": 1.193,
|
| 116 |
+
"step": 1800
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 22.62,
|
| 120 |
+
"learning_rate": 6.639499999999999e-05,
|
| 121 |
+
"loss": 1.152,
|
| 122 |
+
"step": 1900
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 23.81,
|
| 126 |
+
"learning_rate": 6.9895e-05,
|
| 127 |
+
"loss": 1.1004,
|
| 128 |
+
"step": 2000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 23.81,
|
| 132 |
+
"eval_loss": 0.371005654335022,
|
| 133 |
+
"eval_runtime": 72.2015,
|
| 134 |
+
"eval_samples_per_second": 17.174,
|
| 135 |
+
"eval_steps_per_second": 17.174,
|
| 136 |
+
"eval_wer": 0.41910403554239173,
|
| 137 |
+
"step": 2000
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 25.0,
|
| 141 |
+
"learning_rate": 6.915965346534654e-05,
|
| 142 |
+
"loss": 1.0522,
|
| 143 |
+
"step": 2100
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 26.19,
|
| 147 |
+
"learning_rate": 6.829331683168317e-05,
|
| 148 |
+
"loss": 1.0103,
|
| 149 |
+
"step": 2200
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 27.38,
|
| 153 |
+
"learning_rate": 6.74269801980198e-05,
|
| 154 |
+
"loss": 0.9936,
|
| 155 |
+
"step": 2300
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 28.57,
|
| 159 |
+
"learning_rate": 6.656064356435643e-05,
|
| 160 |
+
"loss": 0.9501,
|
| 161 |
+
"step": 2400
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 29.76,
|
| 165 |
+
"learning_rate": 6.569430693069306e-05,
|
| 166 |
+
"loss": 0.9198,
|
| 167 |
+
"step": 2500
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 30.95,
|
| 171 |
+
"learning_rate": 6.48279702970297e-05,
|
| 172 |
+
"loss": 0.9018,
|
| 173 |
+
"step": 2600
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 32.14,
|
| 177 |
+
"learning_rate": 6.396163366336633e-05,
|
| 178 |
+
"loss": 0.8733,
|
| 179 |
+
"step": 2700
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 33.33,
|
| 183 |
+
"learning_rate": 6.309529702970296e-05,
|
| 184 |
+
"loss": 0.8516,
|
| 185 |
+
"step": 2800
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 34.52,
|
| 189 |
+
"learning_rate": 6.222896039603959e-05,
|
| 190 |
+
"loss": 0.8474,
|
| 191 |
+
"step": 2900
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 35.71,
|
| 195 |
+
"learning_rate": 6.136262376237622e-05,
|
| 196 |
+
"loss": 0.8208,
|
| 197 |
+
"step": 3000
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 36.9,
|
| 201 |
+
"learning_rate": 6.049628712871287e-05,
|
| 202 |
+
"loss": 0.8079,
|
| 203 |
+
"step": 3100
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 38.1,
|
| 207 |
+
"learning_rate": 5.9629950495049504e-05,
|
| 208 |
+
"loss": 0.7882,
|
| 209 |
+
"step": 3200
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 39.29,
|
| 213 |
+
"learning_rate": 5.8763613861386134e-05,
|
| 214 |
+
"loss": 0.7741,
|
| 215 |
+
"step": 3300
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 40.48,
|
| 219 |
+
"learning_rate": 5.789727722772277e-05,
|
| 220 |
+
"loss": 0.7725,
|
| 221 |
+
"step": 3400
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 41.67,
|
| 225 |
+
"learning_rate": 5.70309405940594e-05,
|
| 226 |
+
"loss": 0.7445,
|
| 227 |
+
"step": 3500
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 42.86,
|
| 231 |
+
"learning_rate": 5.616460396039603e-05,
|
| 232 |
+
"loss": 0.7391,
|
| 233 |
+
"step": 3600
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 44.05,
|
| 237 |
+
"learning_rate": 5.5306930693069306e-05,
|
| 238 |
+
"loss": 0.7449,
|
| 239 |
+
"step": 3700
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 45.24,
|
| 243 |
+
"learning_rate": 5.4440594059405936e-05,
|
| 244 |
+
"loss": 0.7294,
|
| 245 |
+
"step": 3800
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 46.43,
|
| 249 |
+
"learning_rate": 5.357425742574257e-05,
|
| 250 |
+
"loss": 0.7163,
|
| 251 |
+
"step": 3900
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 47.62,
|
| 255 |
+
"learning_rate": 5.27079207920792e-05,
|
| 256 |
+
"loss": 0.7002,
|
| 257 |
+
"step": 4000
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 47.62,
|
| 261 |
+
"eval_loss": 0.23418071866035461,
|
| 262 |
+
"eval_runtime": 71.5351,
|
| 263 |
+
"eval_samples_per_second": 17.334,
|
| 264 |
+
"eval_steps_per_second": 17.334,
|
| 265 |
+
"eval_wer": 0.2562014068863384,
|
| 266 |
+
"step": 4000
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 48.81,
|
| 270 |
+
"learning_rate": 5.1841584158415834e-05,
|
| 271 |
+
"loss": 0.6952,
|
| 272 |
+
"step": 4100
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 50.0,
|
| 276 |
+
"learning_rate": 5.097524752475247e-05,
|
| 277 |
+
"loss": 0.685,
|
| 278 |
+
"step": 4200
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 51.19,
|
| 282 |
+
"learning_rate": 5.01089108910891e-05,
|
| 283 |
+
"loss": 0.6711,
|
| 284 |
+
"step": 4300
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 52.38,
|
| 288 |
+
"learning_rate": 4.924257425742574e-05,
|
| 289 |
+
"loss": 0.6592,
|
| 290 |
+
"step": 4400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 53.57,
|
| 294 |
+
"learning_rate": 4.837623762376237e-05,
|
| 295 |
+
"loss": 0.659,
|
| 296 |
+
"step": 4500
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 54.76,
|
| 300 |
+
"learning_rate": 4.750990099009901e-05,
|
| 301 |
+
"loss": 0.6529,
|
| 302 |
+
"step": 4600
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 55.95,
|
| 306 |
+
"learning_rate": 4.664356435643564e-05,
|
| 307 |
+
"loss": 0.6481,
|
| 308 |
+
"step": 4700
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 57.14,
|
| 312 |
+
"learning_rate": 4.577722772277227e-05,
|
| 313 |
+
"loss": 0.6411,
|
| 314 |
+
"step": 4800
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 58.33,
|
| 318 |
+
"learning_rate": 4.491089108910891e-05,
|
| 319 |
+
"loss": 0.6292,
|
| 320 |
+
"step": 4900
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 59.52,
|
| 324 |
+
"learning_rate": 4.404455445544554e-05,
|
| 325 |
+
"loss": 0.6158,
|
| 326 |
+
"step": 5000
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 60.71,
|
| 330 |
+
"learning_rate": 4.3178217821782176e-05,
|
| 331 |
+
"loss": 0.6152,
|
| 332 |
+
"step": 5100
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 61.9,
|
| 336 |
+
"learning_rate": 4.2311881188118806e-05,
|
| 337 |
+
"loss": 0.6095,
|
| 338 |
+
"step": 5200
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 63.1,
|
| 342 |
+
"learning_rate": 4.144554455445544e-05,
|
| 343 |
+
"loss": 0.5974,
|
| 344 |
+
"step": 5300
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 64.29,
|
| 348 |
+
"learning_rate": 4.057920792079207e-05,
|
| 349 |
+
"loss": 0.5963,
|
| 350 |
+
"step": 5400
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 65.48,
|
| 354 |
+
"learning_rate": 3.9712871287128704e-05,
|
| 355 |
+
"loss": 0.602,
|
| 356 |
+
"step": 5500
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 66.67,
|
| 360 |
+
"learning_rate": 3.884653465346535e-05,
|
| 361 |
+
"loss": 0.5889,
|
| 362 |
+
"step": 5600
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 67.86,
|
| 366 |
+
"learning_rate": 3.798886138613861e-05,
|
| 367 |
+
"loss": 0.5811,
|
| 368 |
+
"step": 5700
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 69.05,
|
| 372 |
+
"learning_rate": 3.7122524752475246e-05,
|
| 373 |
+
"loss": 0.5693,
|
| 374 |
+
"step": 5800
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 70.24,
|
| 378 |
+
"learning_rate": 3.6256188118811876e-05,
|
| 379 |
+
"loss": 0.5674,
|
| 380 |
+
"step": 5900
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 71.43,
|
| 384 |
+
"learning_rate": 3.5389851485148506e-05,
|
| 385 |
+
"loss": 0.5573,
|
| 386 |
+
"step": 6000
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 71.43,
|
| 390 |
+
"eval_loss": 0.21746528148651123,
|
| 391 |
+
"eval_runtime": 71.6386,
|
| 392 |
+
"eval_samples_per_second": 17.309,
|
| 393 |
+
"eval_steps_per_second": 17.309,
|
| 394 |
+
"eval_wer": 0.21769714920399852,
|
| 395 |
+
"step": 6000
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 72.62,
|
| 399 |
+
"learning_rate": 3.452351485148514e-05,
|
| 400 |
+
"loss": 0.5548,
|
| 401 |
+
"step": 6100
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 73.81,
|
| 405 |
+
"learning_rate": 3.365717821782178e-05,
|
| 406 |
+
"loss": 0.5465,
|
| 407 |
+
"step": 6200
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 75.0,
|
| 411 |
+
"learning_rate": 3.279084158415842e-05,
|
| 412 |
+
"loss": 0.5558,
|
| 413 |
+
"step": 6300
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 76.19,
|
| 417 |
+
"learning_rate": 3.192450495049505e-05,
|
| 418 |
+
"loss": 0.5373,
|
| 419 |
+
"step": 6400
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 77.38,
|
| 423 |
+
"learning_rate": 3.1058168316831684e-05,
|
| 424 |
+
"loss": 0.5469,
|
| 425 |
+
"step": 6500
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 78.57,
|
| 429 |
+
"learning_rate": 3.0191831683168314e-05,
|
| 430 |
+
"loss": 0.5386,
|
| 431 |
+
"step": 6600
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 79.76,
|
| 435 |
+
"learning_rate": 2.9325495049504948e-05,
|
| 436 |
+
"loss": 0.5402,
|
| 437 |
+
"step": 6700
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 80.95,
|
| 441 |
+
"learning_rate": 2.845915841584158e-05,
|
| 442 |
+
"loss": 0.5313,
|
| 443 |
+
"step": 6800
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 82.14,
|
| 447 |
+
"learning_rate": 2.7592821782178218e-05,
|
| 448 |
+
"loss": 0.523,
|
| 449 |
+
"step": 6900
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 83.33,
|
| 453 |
+
"learning_rate": 2.672648514851485e-05,
|
| 454 |
+
"loss": 0.5107,
|
| 455 |
+
"step": 7000
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 84.52,
|
| 459 |
+
"learning_rate": 2.5860148514851482e-05,
|
| 460 |
+
"loss": 0.5147,
|
| 461 |
+
"step": 7100
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 85.71,
|
| 465 |
+
"learning_rate": 2.4993811881188116e-05,
|
| 466 |
+
"loss": 0.5185,
|
| 467 |
+
"step": 7200
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 86.9,
|
| 471 |
+
"learning_rate": 2.412747524752475e-05,
|
| 472 |
+
"loss": 0.5066,
|
| 473 |
+
"step": 7300
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 88.1,
|
| 477 |
+
"learning_rate": 2.3261138613861386e-05,
|
| 478 |
+
"loss": 0.4988,
|
| 479 |
+
"step": 7400
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 89.29,
|
| 483 |
+
"learning_rate": 2.239480198019802e-05,
|
| 484 |
+
"loss": 0.4997,
|
| 485 |
+
"step": 7500
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 90.48,
|
| 489 |
+
"learning_rate": 2.152846534653465e-05,
|
| 490 |
+
"loss": 0.5079,
|
| 491 |
+
"step": 7600
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 91.67,
|
| 495 |
+
"learning_rate": 2.0662128712871283e-05,
|
| 496 |
+
"loss": 0.5039,
|
| 497 |
+
"step": 7700
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 92.86,
|
| 501 |
+
"learning_rate": 1.9795792079207917e-05,
|
| 502 |
+
"loss": 0.4924,
|
| 503 |
+
"step": 7800
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 94.05,
|
| 507 |
+
"learning_rate": 1.8929455445544554e-05,
|
| 508 |
+
"loss": 0.4821,
|
| 509 |
+
"step": 7900
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 95.24,
|
| 513 |
+
"learning_rate": 1.8063118811881187e-05,
|
| 514 |
+
"loss": 0.4799,
|
| 515 |
+
"step": 8000
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 95.24,
|
| 519 |
+
"eval_loss": 0.21092435717582703,
|
| 520 |
+
"eval_runtime": 72.1193,
|
| 521 |
+
"eval_samples_per_second": 17.194,
|
| 522 |
+
"eval_steps_per_second": 17.194,
|
| 523 |
+
"eval_wer": 0.19872269529803777,
|
| 524 |
+
"step": 8000
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 96.43,
|
| 528 |
+
"learning_rate": 1.719678217821782e-05,
|
| 529 |
+
"loss": 0.485,
|
| 530 |
+
"step": 8100
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 97.62,
|
| 534 |
+
"learning_rate": 1.6330445544554455e-05,
|
| 535 |
+
"loss": 0.4806,
|
| 536 |
+
"step": 8200
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 98.81,
|
| 540 |
+
"learning_rate": 1.5464108910891088e-05,
|
| 541 |
+
"loss": 0.4726,
|
| 542 |
+
"step": 8300
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 100.0,
|
| 546 |
+
"learning_rate": 1.459777227722772e-05,
|
| 547 |
+
"loss": 0.4649,
|
| 548 |
+
"step": 8400
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 101.19,
|
| 552 |
+
"learning_rate": 1.3731435643564355e-05,
|
| 553 |
+
"loss": 0.4752,
|
| 554 |
+
"step": 8500
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 102.38,
|
| 558 |
+
"learning_rate": 1.2865099009900989e-05,
|
| 559 |
+
"loss": 0.467,
|
| 560 |
+
"step": 8600
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 103.57,
|
| 564 |
+
"learning_rate": 1.1998762376237624e-05,
|
| 565 |
+
"loss": 0.4623,
|
| 566 |
+
"step": 8700
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 104.76,
|
| 570 |
+
"learning_rate": 1.1132425742574256e-05,
|
| 571 |
+
"loss": 0.4651,
|
| 572 |
+
"step": 8800
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 105.95,
|
| 576 |
+
"learning_rate": 1.026608910891089e-05,
|
| 577 |
+
"loss": 0.469,
|
| 578 |
+
"step": 8900
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 107.14,
|
| 582 |
+
"learning_rate": 9.399752475247525e-06,
|
| 583 |
+
"loss": 0.464,
|
| 584 |
+
"step": 9000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 108.33,
|
| 588 |
+
"learning_rate": 8.533415841584157e-06,
|
| 589 |
+
"loss": 0.4614,
|
| 590 |
+
"step": 9100
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 109.52,
|
| 594 |
+
"learning_rate": 7.667079207920792e-06,
|
| 595 |
+
"loss": 0.4526,
|
| 596 |
+
"step": 9200
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 110.71,
|
| 600 |
+
"learning_rate": 6.800742574257426e-06,
|
| 601 |
+
"loss": 0.4485,
|
| 602 |
+
"step": 9300
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 111.9,
|
| 606 |
+
"learning_rate": 5.934405940594059e-06,
|
| 607 |
+
"loss": 0.4532,
|
| 608 |
+
"step": 9400
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 113.1,
|
| 612 |
+
"learning_rate": 5.068069306930692e-06,
|
| 613 |
+
"loss": 0.4417,
|
| 614 |
+
"step": 9500
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 114.29,
|
| 618 |
+
"learning_rate": 4.21039603960396e-06,
|
| 619 |
+
"loss": 0.4453,
|
| 620 |
+
"step": 9600
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 115.48,
|
| 624 |
+
"learning_rate": 3.344059405940594e-06,
|
| 625 |
+
"loss": 0.4517,
|
| 626 |
+
"step": 9700
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 116.67,
|
| 630 |
+
"learning_rate": 2.477722772277227e-06,
|
| 631 |
+
"loss": 0.4399,
|
| 632 |
+
"step": 9800
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 117.86,
|
| 636 |
+
"learning_rate": 1.6113861386138611e-06,
|
| 637 |
+
"loss": 0.4467,
|
| 638 |
+
"step": 9900
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 119.05,
|
| 642 |
+
"learning_rate": 7.45049504950495e-07,
|
| 643 |
+
"loss": 0.4511,
|
| 644 |
+
"step": 10000
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 119.05,
|
| 648 |
+
"eval_loss": 0.2164391726255417,
|
| 649 |
+
"eval_runtime": 71.6824,
|
| 650 |
+
"eval_samples_per_second": 17.299,
|
| 651 |
+
"eval_steps_per_second": 17.299,
|
| 652 |
+
"eval_wer": 0.19751943724546464,
|
| 653 |
+
"step": 10000
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 120.0,
|
| 657 |
+
"step": 10080,
|
| 658 |
+
"total_flos": 5.597929800172578e+19,
|
| 659 |
+
"train_loss": 1.2021882954097929,
|
| 660 |
+
"train_runtime": 20445.8304,
|
| 661 |
+
"train_samples_per_second": 15.7,
|
| 662 |
+
"train_steps_per_second": 0.493
|
| 663 |
+
}
|
| 664 |
+
],
|
| 665 |
+
"max_steps": 10080,
|
| 666 |
+
"num_train_epochs": 120,
|
| 667 |
+
"total_flos": 5.597929800172578e+19,
|
| 668 |
+
"trial_name": null,
|
| 669 |
+
"trial_params": null
|
| 670 |
+
}
|