Moreno La Quatra
commited on
Commit
·
b50b456
1
Parent(s):
3b2cf11
End of training
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 75.76,
|
| 3 |
+
"eval_loss": 0.7925559282302856,
|
| 4 |
+
"eval_runtime": 432.5691,
|
| 5 |
+
"eval_samples_per_second": 4.931,
|
| 6 |
+
"eval_steps_per_second": 0.155,
|
| 7 |
+
"eval_wer": 32.3250920568122,
|
| 8 |
+
"train_loss": 0.07575882525146008,
|
| 9 |
+
"train_runtime": 14911.1216,
|
| 10 |
+
"train_samples_per_second": 21.46,
|
| 11 |
+
"train_steps_per_second": 0.335
|
| 12 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 75.76,
|
| 3 |
+
"eval_loss": 0.7925559282302856,
|
| 4 |
+
"eval_runtime": 432.5691,
|
| 5 |
+
"eval_samples_per_second": 4.931,
|
| 6 |
+
"eval_steps_per_second": 0.155,
|
| 7 |
+
"eval_wer": 32.3250920568122
|
| 8 |
+
}
|
runs/Dec20_09-58-55_150-136-42-72/events.out.tfevents.1671545891.150-136-42-72.1115156.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0453ecbf9b07bcad02beb08ce0bf28efbc78b2761d9cdac7783a24df932752ba
|
| 3 |
+
size 358
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 75.76,
|
| 3 |
+
"train_loss": 0.07575882525146008,
|
| 4 |
+
"train_runtime": 14911.1216,
|
| 5 |
+
"train_samples_per_second": 21.46,
|
| 6 |
+
"train_steps_per_second": 0.335
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,670 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 31.6096791162546,
|
| 3 |
+
"best_model_checkpoint": "whisper-da-small-augmented/checkpoint-4000",
|
| 4 |
+
"epoch": 75.75757575757575,
|
| 5 |
+
"global_step": 5000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.76,
|
| 12 |
+
"learning_rate": 4.6e-06,
|
| 13 |
+
"loss": 2.9766,
|
| 14 |
+
"step": 50
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 1.52,
|
| 18 |
+
"learning_rate": 9.600000000000001e-06,
|
| 19 |
+
"loss": 1.3858,
|
| 20 |
+
"step": 100
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 2.27,
|
| 24 |
+
"learning_rate": 1.4599999999999999e-05,
|
| 25 |
+
"loss": 0.8213,
|
| 26 |
+
"step": 150
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 3.03,
|
| 30 |
+
"learning_rate": 1.9600000000000002e-05,
|
| 31 |
+
"loss": 0.5669,
|
| 32 |
+
"step": 200
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 3.79,
|
| 36 |
+
"learning_rate": 2.46e-05,
|
| 37 |
+
"loss": 0.3486,
|
| 38 |
+
"step": 250
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 4.55,
|
| 42 |
+
"learning_rate": 2.96e-05,
|
| 43 |
+
"loss": 0.2351,
|
| 44 |
+
"step": 300
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 5.3,
|
| 48 |
+
"learning_rate": 3.46e-05,
|
| 49 |
+
"loss": 0.182,
|
| 50 |
+
"step": 350
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 6.06,
|
| 54 |
+
"learning_rate": 3.960000000000001e-05,
|
| 55 |
+
"loss": 0.1346,
|
| 56 |
+
"step": 400
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 6.82,
|
| 60 |
+
"learning_rate": 4.46e-05,
|
| 61 |
+
"loss": 0.102,
|
| 62 |
+
"step": 450
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 7.58,
|
| 66 |
+
"learning_rate": 4.96e-05,
|
| 67 |
+
"loss": 0.0888,
|
| 68 |
+
"step": 500
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 8.33,
|
| 72 |
+
"learning_rate": 4.948888888888889e-05,
|
| 73 |
+
"loss": 0.0801,
|
| 74 |
+
"step": 550
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 9.09,
|
| 78 |
+
"learning_rate": 4.8933333333333335e-05,
|
| 79 |
+
"loss": 0.0665,
|
| 80 |
+
"step": 600
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 9.85,
|
| 84 |
+
"learning_rate": 4.837777777777778e-05,
|
| 85 |
+
"loss": 0.0491,
|
| 86 |
+
"step": 650
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 10.61,
|
| 90 |
+
"learning_rate": 4.782222222222222e-05,
|
| 91 |
+
"loss": 0.0454,
|
| 92 |
+
"step": 700
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 11.36,
|
| 96 |
+
"learning_rate": 4.726666666666667e-05,
|
| 97 |
+
"loss": 0.0423,
|
| 98 |
+
"step": 750
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 12.12,
|
| 102 |
+
"learning_rate": 4.671111111111111e-05,
|
| 103 |
+
"loss": 0.0363,
|
| 104 |
+
"step": 800
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 12.88,
|
| 108 |
+
"learning_rate": 4.615555555555556e-05,
|
| 109 |
+
"loss": 0.0309,
|
| 110 |
+
"step": 850
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 13.64,
|
| 114 |
+
"learning_rate": 4.5600000000000004e-05,
|
| 115 |
+
"loss": 0.0291,
|
| 116 |
+
"step": 900
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 14.39,
|
| 120 |
+
"learning_rate": 4.504444444444445e-05,
|
| 121 |
+
"loss": 0.0246,
|
| 122 |
+
"step": 950
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 15.15,
|
| 126 |
+
"learning_rate": 4.448888888888889e-05,
|
| 127 |
+
"loss": 0.0232,
|
| 128 |
+
"step": 1000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 15.15,
|
| 132 |
+
"eval_loss": 0.7538474798202515,
|
| 133 |
+
"eval_runtime": 460.7386,
|
| 134 |
+
"eval_samples_per_second": 4.63,
|
| 135 |
+
"eval_steps_per_second": 0.145,
|
| 136 |
+
"eval_wer": 35.58127301420305,
|
| 137 |
+
"step": 1000
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 15.91,
|
| 141 |
+
"learning_rate": 4.3933333333333335e-05,
|
| 142 |
+
"loss": 0.0213,
|
| 143 |
+
"step": 1050
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 16.67,
|
| 147 |
+
"learning_rate": 4.337777777777778e-05,
|
| 148 |
+
"loss": 0.0209,
|
| 149 |
+
"step": 1100
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 17.42,
|
| 153 |
+
"learning_rate": 4.282222222222222e-05,
|
| 154 |
+
"loss": 0.0171,
|
| 155 |
+
"step": 1150
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 18.18,
|
| 159 |
+
"learning_rate": 4.226666666666667e-05,
|
| 160 |
+
"loss": 0.0182,
|
| 161 |
+
"step": 1200
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 18.94,
|
| 165 |
+
"learning_rate": 4.171111111111111e-05,
|
| 166 |
+
"loss": 0.0157,
|
| 167 |
+
"step": 1250
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 19.7,
|
| 171 |
+
"learning_rate": 4.115555555555556e-05,
|
| 172 |
+
"loss": 0.0162,
|
| 173 |
+
"step": 1300
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 20.45,
|
| 177 |
+
"learning_rate": 4.0600000000000004e-05,
|
| 178 |
+
"loss": 0.0141,
|
| 179 |
+
"step": 1350
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 21.21,
|
| 183 |
+
"learning_rate": 4.004444444444445e-05,
|
| 184 |
+
"loss": 0.0127,
|
| 185 |
+
"step": 1400
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 21.97,
|
| 189 |
+
"learning_rate": 3.948888888888889e-05,
|
| 190 |
+
"loss": 0.0103,
|
| 191 |
+
"step": 1450
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 22.73,
|
| 195 |
+
"learning_rate": 3.8933333333333336e-05,
|
| 196 |
+
"loss": 0.0098,
|
| 197 |
+
"step": 1500
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 23.48,
|
| 201 |
+
"learning_rate": 3.837777777777778e-05,
|
| 202 |
+
"loss": 0.0099,
|
| 203 |
+
"step": 1550
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 24.24,
|
| 207 |
+
"learning_rate": 3.782222222222222e-05,
|
| 208 |
+
"loss": 0.0109,
|
| 209 |
+
"step": 1600
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 25.0,
|
| 213 |
+
"learning_rate": 3.726666666666667e-05,
|
| 214 |
+
"loss": 0.0078,
|
| 215 |
+
"step": 1650
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 25.76,
|
| 219 |
+
"learning_rate": 3.671111111111111e-05,
|
| 220 |
+
"loss": 0.0073,
|
| 221 |
+
"step": 1700
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 26.52,
|
| 225 |
+
"learning_rate": 3.615555555555556e-05,
|
| 226 |
+
"loss": 0.0057,
|
| 227 |
+
"step": 1750
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 27.27,
|
| 231 |
+
"learning_rate": 3.56e-05,
|
| 232 |
+
"loss": 0.0068,
|
| 233 |
+
"step": 1800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 28.03,
|
| 237 |
+
"learning_rate": 3.504444444444445e-05,
|
| 238 |
+
"loss": 0.0052,
|
| 239 |
+
"step": 1850
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 28.79,
|
| 243 |
+
"learning_rate": 3.448888888888889e-05,
|
| 244 |
+
"loss": 0.0084,
|
| 245 |
+
"step": 1900
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 29.55,
|
| 249 |
+
"learning_rate": 3.3933333333333336e-05,
|
| 250 |
+
"loss": 0.006,
|
| 251 |
+
"step": 1950
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 30.3,
|
| 255 |
+
"learning_rate": 3.337777777777778e-05,
|
| 256 |
+
"loss": 0.0061,
|
| 257 |
+
"step": 2000
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 30.3,
|
| 261 |
+
"eval_loss": 0.7932922840118408,
|
| 262 |
+
"eval_runtime": 424.2221,
|
| 263 |
+
"eval_samples_per_second": 5.028,
|
| 264 |
+
"eval_steps_per_second": 0.158,
|
| 265 |
+
"eval_wer": 34.3766438716465,
|
| 266 |
+
"step": 2000
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 31.06,
|
| 270 |
+
"learning_rate": 3.2822222222222223e-05,
|
| 271 |
+
"loss": 0.0067,
|
| 272 |
+
"step": 2050
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 31.82,
|
| 276 |
+
"learning_rate": 3.226666666666667e-05,
|
| 277 |
+
"loss": 0.0056,
|
| 278 |
+
"step": 2100
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 32.58,
|
| 282 |
+
"learning_rate": 3.171111111111111e-05,
|
| 283 |
+
"loss": 0.0044,
|
| 284 |
+
"step": 2150
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 33.33,
|
| 288 |
+
"learning_rate": 3.1155555555555555e-05,
|
| 289 |
+
"loss": 0.0041,
|
| 290 |
+
"step": 2200
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 34.09,
|
| 294 |
+
"learning_rate": 3.06e-05,
|
| 295 |
+
"loss": 0.0037,
|
| 296 |
+
"step": 2250
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 34.85,
|
| 300 |
+
"learning_rate": 3.004444444444445e-05,
|
| 301 |
+
"loss": 0.0025,
|
| 302 |
+
"step": 2300
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 35.61,
|
| 306 |
+
"learning_rate": 2.948888888888889e-05,
|
| 307 |
+
"loss": 0.0029,
|
| 308 |
+
"step": 2350
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 36.36,
|
| 312 |
+
"learning_rate": 2.8933333333333333e-05,
|
| 313 |
+
"loss": 0.0025,
|
| 314 |
+
"step": 2400
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 37.12,
|
| 318 |
+
"learning_rate": 2.837777777777778e-05,
|
| 319 |
+
"loss": 0.0031,
|
| 320 |
+
"step": 2450
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 37.88,
|
| 324 |
+
"learning_rate": 2.782222222222222e-05,
|
| 325 |
+
"loss": 0.0025,
|
| 326 |
+
"step": 2500
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 38.64,
|
| 330 |
+
"learning_rate": 2.7266666666666668e-05,
|
| 331 |
+
"loss": 0.0016,
|
| 332 |
+
"step": 2550
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 39.39,
|
| 336 |
+
"learning_rate": 2.6711111111111115e-05,
|
| 337 |
+
"loss": 0.003,
|
| 338 |
+
"step": 2600
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 40.15,
|
| 342 |
+
"learning_rate": 2.6155555555555555e-05,
|
| 343 |
+
"loss": 0.0015,
|
| 344 |
+
"step": 2650
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 40.91,
|
| 348 |
+
"learning_rate": 2.5600000000000002e-05,
|
| 349 |
+
"loss": 0.0016,
|
| 350 |
+
"step": 2700
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 41.67,
|
| 354 |
+
"learning_rate": 2.504444444444445e-05,
|
| 355 |
+
"loss": 0.0014,
|
| 356 |
+
"step": 2750
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 42.42,
|
| 360 |
+
"learning_rate": 2.448888888888889e-05,
|
| 361 |
+
"loss": 0.0013,
|
| 362 |
+
"step": 2800
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 43.18,
|
| 366 |
+
"learning_rate": 2.3933333333333337e-05,
|
| 367 |
+
"loss": 0.0027,
|
| 368 |
+
"step": 2850
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 43.94,
|
| 372 |
+
"learning_rate": 2.337777777777778e-05,
|
| 373 |
+
"loss": 0.0016,
|
| 374 |
+
"step": 2900
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 44.7,
|
| 378 |
+
"learning_rate": 2.282222222222222e-05,
|
| 379 |
+
"loss": 0.0025,
|
| 380 |
+
"step": 2950
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 45.45,
|
| 384 |
+
"learning_rate": 2.2266666666666668e-05,
|
| 385 |
+
"loss": 0.0016,
|
| 386 |
+
"step": 3000
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 45.45,
|
| 390 |
+
"eval_loss": 0.7993177771568298,
|
| 391 |
+
"eval_runtime": 430.3696,
|
| 392 |
+
"eval_samples_per_second": 4.956,
|
| 393 |
+
"eval_steps_per_second": 0.156,
|
| 394 |
+
"eval_wer": 33.58232509205681,
|
| 395 |
+
"step": 3000
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 46.21,
|
| 399 |
+
"learning_rate": 2.1711111111111112e-05,
|
| 400 |
+
"loss": 0.0018,
|
| 401 |
+
"step": 3050
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 46.97,
|
| 405 |
+
"learning_rate": 2.1155555555555556e-05,
|
| 406 |
+
"loss": 0.001,
|
| 407 |
+
"step": 3100
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 47.73,
|
| 411 |
+
"learning_rate": 2.06e-05,
|
| 412 |
+
"loss": 0.0011,
|
| 413 |
+
"step": 3150
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 48.48,
|
| 417 |
+
"learning_rate": 2.0044444444444446e-05,
|
| 418 |
+
"loss": 0.001,
|
| 419 |
+
"step": 3200
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 49.24,
|
| 423 |
+
"learning_rate": 1.948888888888889e-05,
|
| 424 |
+
"loss": 0.0008,
|
| 425 |
+
"step": 3250
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 50.0,
|
| 429 |
+
"learning_rate": 1.8933333333333334e-05,
|
| 430 |
+
"loss": 0.0011,
|
| 431 |
+
"step": 3300
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 50.76,
|
| 435 |
+
"learning_rate": 1.837777777777778e-05,
|
| 436 |
+
"loss": 0.0007,
|
| 437 |
+
"step": 3350
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 51.52,
|
| 441 |
+
"learning_rate": 1.7822222222222225e-05,
|
| 442 |
+
"loss": 0.001,
|
| 443 |
+
"step": 3400
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 52.27,
|
| 447 |
+
"learning_rate": 1.726666666666667e-05,
|
| 448 |
+
"loss": 0.0014,
|
| 449 |
+
"step": 3450
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 53.03,
|
| 453 |
+
"learning_rate": 1.6711111111111112e-05,
|
| 454 |
+
"loss": 0.0009,
|
| 455 |
+
"step": 3500
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 53.79,
|
| 459 |
+
"learning_rate": 1.6155555555555556e-05,
|
| 460 |
+
"loss": 0.0008,
|
| 461 |
+
"step": 3550
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 54.55,
|
| 465 |
+
"learning_rate": 1.56e-05,
|
| 466 |
+
"loss": 0.0003,
|
| 467 |
+
"step": 3600
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 55.3,
|
| 471 |
+
"learning_rate": 1.5044444444444445e-05,
|
| 472 |
+
"loss": 0.0004,
|
| 473 |
+
"step": 3650
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 56.06,
|
| 477 |
+
"learning_rate": 1.448888888888889e-05,
|
| 478 |
+
"loss": 0.0004,
|
| 479 |
+
"step": 3700
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 56.82,
|
| 483 |
+
"learning_rate": 1.3933333333333334e-05,
|
| 484 |
+
"loss": 0.0005,
|
| 485 |
+
"step": 3750
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 57.58,
|
| 489 |
+
"learning_rate": 1.3377777777777778e-05,
|
| 490 |
+
"loss": 0.0005,
|
| 491 |
+
"step": 3800
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 58.33,
|
| 495 |
+
"learning_rate": 1.2822222222222222e-05,
|
| 496 |
+
"loss": 0.0006,
|
| 497 |
+
"step": 3850
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 59.09,
|
| 501 |
+
"learning_rate": 1.2266666666666667e-05,
|
| 502 |
+
"loss": 0.0005,
|
| 503 |
+
"step": 3900
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 59.85,
|
| 507 |
+
"learning_rate": 1.1711111111111111e-05,
|
| 508 |
+
"loss": 0.0004,
|
| 509 |
+
"step": 3950
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 60.61,
|
| 513 |
+
"learning_rate": 1.1155555555555556e-05,
|
| 514 |
+
"loss": 0.0003,
|
| 515 |
+
"step": 4000
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 60.61,
|
| 519 |
+
"eval_loss": 0.7985624074935913,
|
| 520 |
+
"eval_runtime": 420.8978,
|
| 521 |
+
"eval_samples_per_second": 5.068,
|
| 522 |
+
"eval_steps_per_second": 0.159,
|
| 523 |
+
"eval_wer": 31.6096791162546,
|
| 524 |
+
"step": 4000
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 61.36,
|
| 528 |
+
"learning_rate": 1.06e-05,
|
| 529 |
+
"loss": 0.0002,
|
| 530 |
+
"step": 4050
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 62.12,
|
| 534 |
+
"learning_rate": 1.0044444444444446e-05,
|
| 535 |
+
"loss": 0.0003,
|
| 536 |
+
"step": 4100
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 62.88,
|
| 540 |
+
"learning_rate": 9.48888888888889e-06,
|
| 541 |
+
"loss": 0.0003,
|
| 542 |
+
"step": 4150
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 63.64,
|
| 546 |
+
"learning_rate": 8.933333333333333e-06,
|
| 547 |
+
"loss": 0.0002,
|
| 548 |
+
"step": 4200
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 64.39,
|
| 552 |
+
"learning_rate": 8.377777777777779e-06,
|
| 553 |
+
"loss": 0.0002,
|
| 554 |
+
"step": 4250
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 65.15,
|
| 558 |
+
"learning_rate": 7.822222222222222e-06,
|
| 559 |
+
"loss": 0.0002,
|
| 560 |
+
"step": 4300
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 65.91,
|
| 564 |
+
"learning_rate": 7.266666666666668e-06,
|
| 565 |
+
"loss": 0.0002,
|
| 566 |
+
"step": 4350
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 66.67,
|
| 570 |
+
"learning_rate": 6.711111111111111e-06,
|
| 571 |
+
"loss": 0.0002,
|
| 572 |
+
"step": 4400
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 67.42,
|
| 576 |
+
"learning_rate": 6.155555555555556e-06,
|
| 577 |
+
"loss": 0.0002,
|
| 578 |
+
"step": 4450
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 68.18,
|
| 582 |
+
"learning_rate": 5.600000000000001e-06,
|
| 583 |
+
"loss": 0.0002,
|
| 584 |
+
"step": 4500
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 68.94,
|
| 588 |
+
"learning_rate": 5.044444444444444e-06,
|
| 589 |
+
"loss": 0.0003,
|
| 590 |
+
"step": 4550
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 69.7,
|
| 594 |
+
"learning_rate": 4.488888888888889e-06,
|
| 595 |
+
"loss": 0.0003,
|
| 596 |
+
"step": 4600
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 70.45,
|
| 600 |
+
"learning_rate": 3.9333333333333335e-06,
|
| 601 |
+
"loss": 0.0002,
|
| 602 |
+
"step": 4650
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 71.21,
|
| 606 |
+
"learning_rate": 3.3777777777777777e-06,
|
| 607 |
+
"loss": 0.0002,
|
| 608 |
+
"step": 4700
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 71.97,
|
| 612 |
+
"learning_rate": 2.8222222222222223e-06,
|
| 613 |
+
"loss": 0.0001,
|
| 614 |
+
"step": 4750
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 72.73,
|
| 618 |
+
"learning_rate": 2.266666666666667e-06,
|
| 619 |
+
"loss": 0.0001,
|
| 620 |
+
"step": 4800
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 73.48,
|
| 624 |
+
"learning_rate": 1.7111111111111112e-06,
|
| 625 |
+
"loss": 0.0001,
|
| 626 |
+
"step": 4850
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 74.24,
|
| 630 |
+
"learning_rate": 1.1555555555555556e-06,
|
| 631 |
+
"loss": 0.0002,
|
| 632 |
+
"step": 4900
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 75.0,
|
| 636 |
+
"learning_rate": 6.000000000000001e-07,
|
| 637 |
+
"loss": 0.0003,
|
| 638 |
+
"step": 4950
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 75.76,
|
| 642 |
+
"learning_rate": 4.444444444444445e-08,
|
| 643 |
+
"loss": 0.0002,
|
| 644 |
+
"step": 5000
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 75.76,
|
| 648 |
+
"eval_loss": 0.7901055812835693,
|
| 649 |
+
"eval_runtime": 421.2754,
|
| 650 |
+
"eval_samples_per_second": 5.063,
|
| 651 |
+
"eval_steps_per_second": 0.159,
|
| 652 |
+
"eval_wer": 32.13571804313519,
|
| 653 |
+
"step": 5000
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 75.76,
|
| 657 |
+
"step": 5000,
|
| 658 |
+
"total_flos": 9.2044313432064e+19,
|
| 659 |
+
"train_loss": 0.07575882525146008,
|
| 660 |
+
"train_runtime": 14911.1216,
|
| 661 |
+
"train_samples_per_second": 21.46,
|
| 662 |
+
"train_steps_per_second": 0.335
|
| 663 |
+
}
|
| 664 |
+
],
|
| 665 |
+
"max_steps": 5000,
|
| 666 |
+
"num_train_epochs": 76,
|
| 667 |
+
"total_flos": 9.2044313432064e+19,
|
| 668 |
+
"trial_name": null,
|
| 669 |
+
"trial_params": null
|
| 670 |
+
}
|