End of training
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 117.65,
|
| 3 |
+
"eval_loss": 0.50830078125,
|
| 4 |
+
"eval_runtime": 168.297,
|
| 5 |
+
"eval_samples_per_second": 1.616,
|
| 6 |
+
"eval_steps_per_second": 0.101,
|
| 7 |
+
"eval_wer": 10.27117384843982,
|
| 8 |
+
"train_loss": 0.0013260623812675475,
|
| 9 |
+
"train_runtime": 8229.715,
|
| 10 |
+
"train_samples_per_second": 7.777,
|
| 11 |
+
"train_steps_per_second": 0.243
|
| 12 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 117.65,
|
| 3 |
+
"eval_loss": 0.50830078125,
|
| 4 |
+
"eval_runtime": 168.297,
|
| 5 |
+
"eval_samples_per_second": 1.616,
|
| 6 |
+
"eval_steps_per_second": 0.101,
|
| 7 |
+
"eval_wer": 10.27117384843982
|
| 8 |
+
}
|
runs/Dec22_17-47-55_129-146-176-120/events.out.tfevents.1671740046.129-146-176-120.842798.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:988da22e2e9fb901e96bf69b9244244cf0ba0e90ee88cc8fd6ae85d1b5644902
|
| 3 |
+
size 358
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 117.65,
|
| 3 |
+
"train_loss": 0.0013260623812675475,
|
| 4 |
+
"train_runtime": 8229.715,
|
| 5 |
+
"train_samples_per_second": 7.777,
|
| 6 |
+
"train_steps_per_second": 0.243
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,523 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 10.27117384843982,
|
| 3 |
+
"best_model_checkpoint": "./checkpoint-2000",
|
| 4 |
+
"epoch": 117.6470588235294,
|
| 5 |
+
"global_step": 2000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 1.47,
|
| 12 |
+
"learning_rate": 1.5136083400296205e-06,
|
| 13 |
+
"loss": 0.0024,
|
| 14 |
+
"step": 25
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 2.94,
|
| 18 |
+
"learning_rate": 1.8687587131475301e-06,
|
| 19 |
+
"loss": 0.0024,
|
| 20 |
+
"step": 50
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 4.41,
|
| 24 |
+
"learning_rate": 2.0711488350670174e-06,
|
| 25 |
+
"loss": 0.0023,
|
| 26 |
+
"step": 75
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 5.88,
|
| 30 |
+
"learning_rate": 2.213317753617305e-06,
|
| 31 |
+
"loss": 0.0023,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 7.35,
|
| 36 |
+
"learning_rate": 2.3230029693718747e-06,
|
| 37 |
+
"loss": 0.002,
|
| 38 |
+
"step": 125
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 8.82,
|
| 42 |
+
"learning_rate": 2.412322158351148e-06,
|
| 43 |
+
"loss": 0.002,
|
| 44 |
+
"step": 150
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 10.29,
|
| 48 |
+
"learning_rate": 2.4876668872198717e-06,
|
| 49 |
+
"loss": 0.0022,
|
| 50 |
+
"step": 175
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 11.76,
|
| 54 |
+
"learning_rate": 2.552824062407326e-06,
|
| 55 |
+
"loss": 0.0021,
|
| 56 |
+
"step": 200
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 13.24,
|
| 60 |
+
"learning_rate": 2.610223373296667e-06,
|
| 61 |
+
"loss": 0.0034,
|
| 62 |
+
"step": 225
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 14.71,
|
| 66 |
+
"learning_rate": 2.661517182828361e-06,
|
| 67 |
+
"loss": 0.0019,
|
| 68 |
+
"step": 250
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 16.18,
|
| 72 |
+
"learning_rate": 2.7078803874740543e-06,
|
| 73 |
+
"loss": 0.0018,
|
| 74 |
+
"step": 275
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 17.65,
|
| 78 |
+
"learning_rate": 2.750178319990197e-06,
|
| 79 |
+
"loss": 0.0023,
|
| 80 |
+
"step": 300
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 19.12,
|
| 84 |
+
"learning_rate": 2.7890667754365044e-06,
|
| 85 |
+
"loss": 0.0019,
|
| 86 |
+
"step": 325
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 20.59,
|
| 90 |
+
"learning_rate": 2.8250546392106077e-06,
|
| 91 |
+
"loss": 0.0021,
|
| 92 |
+
"step": 350
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 22.06,
|
| 96 |
+
"learning_rate": 2.8585447348549113e-06,
|
| 97 |
+
"loss": 0.0023,
|
| 98 |
+
"step": 375
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 23.53,
|
| 102 |
+
"learning_rate": 2.889861392935294e-06,
|
| 103 |
+
"loss": 0.0021,
|
| 104 |
+
"step": 400
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 25.0,
|
| 108 |
+
"learning_rate": 2.9192696063561725e-06,
|
| 109 |
+
"loss": 0.0016,
|
| 110 |
+
"step": 425
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 26.47,
|
| 114 |
+
"learning_rate": 2.946988676871634e-06,
|
| 115 |
+
"loss": 0.0018,
|
| 116 |
+
"step": 450
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 27.94,
|
| 120 |
+
"learning_rate": 2.973202150939645e-06,
|
| 121 |
+
"loss": 0.0022,
|
| 122 |
+
"step": 475
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 29.41,
|
| 126 |
+
"learning_rate": 2.998065193492142e-06,
|
| 127 |
+
"loss": 0.0018,
|
| 128 |
+
"step": 500
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 30.88,
|
| 132 |
+
"learning_rate": 2.9559999999999997e-06,
|
| 133 |
+
"loss": 0.0018,
|
| 134 |
+
"step": 525
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 32.35,
|
| 138 |
+
"learning_rate": 2.9060000000000002e-06,
|
| 139 |
+
"loss": 0.0021,
|
| 140 |
+
"step": 550
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 33.82,
|
| 144 |
+
"learning_rate": 2.856e-06,
|
| 145 |
+
"loss": 0.0017,
|
| 146 |
+
"step": 575
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 35.29,
|
| 150 |
+
"learning_rate": 2.8060000000000003e-06,
|
| 151 |
+
"loss": 0.0018,
|
| 152 |
+
"step": 600
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 36.76,
|
| 156 |
+
"learning_rate": 2.756e-06,
|
| 157 |
+
"loss": 0.0016,
|
| 158 |
+
"step": 625
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 38.24,
|
| 162 |
+
"learning_rate": 2.706e-06,
|
| 163 |
+
"loss": 0.0017,
|
| 164 |
+
"step": 650
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 39.71,
|
| 168 |
+
"learning_rate": 2.656e-06,
|
| 169 |
+
"loss": 0.0016,
|
| 170 |
+
"step": 675
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 41.18,
|
| 174 |
+
"learning_rate": 2.606e-06,
|
| 175 |
+
"loss": 0.0013,
|
| 176 |
+
"step": 700
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 42.65,
|
| 180 |
+
"learning_rate": 2.556e-06,
|
| 181 |
+
"loss": 0.0013,
|
| 182 |
+
"step": 725
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 44.12,
|
| 186 |
+
"learning_rate": 2.5060000000000002e-06,
|
| 187 |
+
"loss": 0.0012,
|
| 188 |
+
"step": 750
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 45.59,
|
| 192 |
+
"learning_rate": 2.456e-06,
|
| 193 |
+
"loss": 0.0013,
|
| 194 |
+
"step": 775
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 47.06,
|
| 198 |
+
"learning_rate": 2.4060000000000003e-06,
|
| 199 |
+
"loss": 0.0013,
|
| 200 |
+
"step": 800
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 48.53,
|
| 204 |
+
"learning_rate": 2.356e-06,
|
| 205 |
+
"loss": 0.0013,
|
| 206 |
+
"step": 825
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 50.0,
|
| 210 |
+
"learning_rate": 2.306e-06,
|
| 211 |
+
"loss": 0.0011,
|
| 212 |
+
"step": 850
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 51.47,
|
| 216 |
+
"learning_rate": 2.256e-06,
|
| 217 |
+
"loss": 0.0013,
|
| 218 |
+
"step": 875
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 52.94,
|
| 222 |
+
"learning_rate": 2.2059999999999997e-06,
|
| 223 |
+
"loss": 0.0014,
|
| 224 |
+
"step": 900
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 54.41,
|
| 228 |
+
"learning_rate": 2.156e-06,
|
| 229 |
+
"loss": 0.0011,
|
| 230 |
+
"step": 925
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 55.88,
|
| 234 |
+
"learning_rate": 2.106e-06,
|
| 235 |
+
"loss": 0.0024,
|
| 236 |
+
"step": 950
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 57.35,
|
| 240 |
+
"learning_rate": 2.0560000000000003e-06,
|
| 241 |
+
"loss": 0.0014,
|
| 242 |
+
"step": 975
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 58.82,
|
| 246 |
+
"learning_rate": 2.006e-06,
|
| 247 |
+
"loss": 0.0014,
|
| 248 |
+
"step": 1000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 58.82,
|
| 252 |
+
"eval_loss": 0.494384765625,
|
| 253 |
+
"eval_runtime": 154.3699,
|
| 254 |
+
"eval_samples_per_second": 1.762,
|
| 255 |
+
"eval_steps_per_second": 0.11,
|
| 256 |
+
"eval_wer": 10.36404160475483,
|
| 257 |
+
"step": 1000
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 60.29,
|
| 261 |
+
"learning_rate": 1.96e-06,
|
| 262 |
+
"loss": 0.0012,
|
| 263 |
+
"step": 1025
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 61.76,
|
| 267 |
+
"learning_rate": 1.9100000000000003e-06,
|
| 268 |
+
"loss": 0.0009,
|
| 269 |
+
"step": 1050
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 63.24,
|
| 273 |
+
"learning_rate": 1.86e-06,
|
| 274 |
+
"loss": 0.0012,
|
| 275 |
+
"step": 1075
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 64.71,
|
| 279 |
+
"learning_rate": 1.8100000000000002e-06,
|
| 280 |
+
"loss": 0.0009,
|
| 281 |
+
"step": 1100
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 66.18,
|
| 285 |
+
"learning_rate": 1.76e-06,
|
| 286 |
+
"loss": 0.0012,
|
| 287 |
+
"step": 1125
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 67.65,
|
| 291 |
+
"learning_rate": 1.71e-06,
|
| 292 |
+
"loss": 0.0012,
|
| 293 |
+
"step": 1150
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 69.12,
|
| 297 |
+
"learning_rate": 1.66e-06,
|
| 298 |
+
"loss": 0.0009,
|
| 299 |
+
"step": 1175
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 70.59,
|
| 303 |
+
"learning_rate": 1.6099999999999998e-06,
|
| 304 |
+
"loss": 0.0007,
|
| 305 |
+
"step": 1200
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 72.06,
|
| 309 |
+
"learning_rate": 1.56e-06,
|
| 310 |
+
"loss": 0.0011,
|
| 311 |
+
"step": 1225
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 73.53,
|
| 315 |
+
"learning_rate": 1.51e-06,
|
| 316 |
+
"loss": 0.0014,
|
| 317 |
+
"step": 1250
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 75.0,
|
| 321 |
+
"learning_rate": 1.4600000000000002e-06,
|
| 322 |
+
"loss": 0.0008,
|
| 323 |
+
"step": 1275
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 76.47,
|
| 327 |
+
"learning_rate": 1.41e-06,
|
| 328 |
+
"loss": 0.0009,
|
| 329 |
+
"step": 1300
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 77.94,
|
| 333 |
+
"learning_rate": 1.36e-06,
|
| 334 |
+
"loss": 0.0008,
|
| 335 |
+
"step": 1325
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 79.41,
|
| 339 |
+
"learning_rate": 1.31e-06,
|
| 340 |
+
"loss": 0.0009,
|
| 341 |
+
"step": 1350
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 80.88,
|
| 345 |
+
"learning_rate": 1.26e-06,
|
| 346 |
+
"loss": 0.0011,
|
| 347 |
+
"step": 1375
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 82.35,
|
| 351 |
+
"learning_rate": 1.21e-06,
|
| 352 |
+
"loss": 0.0007,
|
| 353 |
+
"step": 1400
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 83.82,
|
| 357 |
+
"learning_rate": 1.16e-06,
|
| 358 |
+
"loss": 0.0011,
|
| 359 |
+
"step": 1425
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 85.29,
|
| 363 |
+
"learning_rate": 1.11e-06,
|
| 364 |
+
"loss": 0.001,
|
| 365 |
+
"step": 1450
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 86.76,
|
| 369 |
+
"learning_rate": 1.06e-06,
|
| 370 |
+
"loss": 0.0008,
|
| 371 |
+
"step": 1475
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 88.24,
|
| 375 |
+
"learning_rate": 1.01e-06,
|
| 376 |
+
"loss": 0.0007,
|
| 377 |
+
"step": 1500
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 89.71,
|
| 381 |
+
"learning_rate": 9.600000000000001e-07,
|
| 382 |
+
"loss": 0.0006,
|
| 383 |
+
"step": 1525
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 91.18,
|
| 387 |
+
"learning_rate": 9.100000000000001e-07,
|
| 388 |
+
"loss": 0.0007,
|
| 389 |
+
"step": 1550
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 92.65,
|
| 393 |
+
"learning_rate": 8.6e-07,
|
| 394 |
+
"loss": 0.0006,
|
| 395 |
+
"step": 1575
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 94.12,
|
| 399 |
+
"learning_rate": 8.100000000000001e-07,
|
| 400 |
+
"loss": 0.0008,
|
| 401 |
+
"step": 1600
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 95.59,
|
| 405 |
+
"learning_rate": 7.600000000000001e-07,
|
| 406 |
+
"loss": 0.0009,
|
| 407 |
+
"step": 1625
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 97.06,
|
| 411 |
+
"learning_rate": 7.1e-07,
|
| 412 |
+
"loss": 0.0006,
|
| 413 |
+
"step": 1650
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 98.53,
|
| 417 |
+
"learning_rate": 6.6e-07,
|
| 418 |
+
"loss": 0.0006,
|
| 419 |
+
"step": 1675
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 100.0,
|
| 423 |
+
"learning_rate": 6.1e-07,
|
| 424 |
+
"loss": 0.0008,
|
| 425 |
+
"step": 1700
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 101.47,
|
| 429 |
+
"learning_rate": 5.6e-07,
|
| 430 |
+
"loss": 0.0006,
|
| 431 |
+
"step": 1725
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 102.94,
|
| 435 |
+
"learning_rate": 5.100000000000001e-07,
|
| 436 |
+
"loss": 0.0008,
|
| 437 |
+
"step": 1750
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 104.41,
|
| 441 |
+
"learning_rate": 4.6e-07,
|
| 442 |
+
"loss": 0.0006,
|
| 443 |
+
"step": 1775
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 105.88,
|
| 447 |
+
"learning_rate": 4.1e-07,
|
| 448 |
+
"loss": 0.0006,
|
| 449 |
+
"step": 1800
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 107.35,
|
| 453 |
+
"learning_rate": 3.6e-07,
|
| 454 |
+
"loss": 0.0007,
|
| 455 |
+
"step": 1825
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 108.82,
|
| 459 |
+
"learning_rate": 3.1e-07,
|
| 460 |
+
"loss": 0.0007,
|
| 461 |
+
"step": 1850
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 110.29,
|
| 465 |
+
"learning_rate": 2.6e-07,
|
| 466 |
+
"loss": 0.0007,
|
| 467 |
+
"step": 1875
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 111.76,
|
| 471 |
+
"learning_rate": 2.1000000000000003e-07,
|
| 472 |
+
"loss": 0.0005,
|
| 473 |
+
"step": 1900
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 113.24,
|
| 477 |
+
"learning_rate": 1.6e-07,
|
| 478 |
+
"loss": 0.0008,
|
| 479 |
+
"step": 1925
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 114.71,
|
| 483 |
+
"learning_rate": 1.1e-07,
|
| 484 |
+
"loss": 0.0006,
|
| 485 |
+
"step": 1950
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 116.18,
|
| 489 |
+
"learning_rate": 6.000000000000001e-08,
|
| 490 |
+
"loss": 0.0006,
|
| 491 |
+
"step": 1975
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 117.65,
|
| 495 |
+
"learning_rate": 1e-08,
|
| 496 |
+
"loss": 0.0007,
|
| 497 |
+
"step": 2000
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 117.65,
|
| 501 |
+
"eval_loss": 0.50830078125,
|
| 502 |
+
"eval_runtime": 170.147,
|
| 503 |
+
"eval_samples_per_second": 1.599,
|
| 504 |
+
"eval_steps_per_second": 0.1,
|
| 505 |
+
"eval_wer": 10.27117384843982,
|
| 506 |
+
"step": 2000
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 117.65,
|
| 510 |
+
"step": 2000,
|
| 511 |
+
"total_flos": 6.161697879136351e+19,
|
| 512 |
+
"train_loss": 0.0013260623812675475,
|
| 513 |
+
"train_runtime": 8229.715,
|
| 514 |
+
"train_samples_per_second": 7.777,
|
| 515 |
+
"train_steps_per_second": 0.243
|
| 516 |
+
}
|
| 517 |
+
],
|
| 518 |
+
"max_steps": 2000,
|
| 519 |
+
"num_train_epochs": 118,
|
| 520 |
+
"total_flos": 6.161697879136351e+19,
|
| 521 |
+
"trial_name": null,
|
| 522 |
+
"trial_params": null
|
| 523 |
+
}
|