Commit
·
8847357
1
Parent(s):
dfa018f
End of training
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.13,
|
| 3 |
+
"eval_loss": 0.05638869106769562,
|
| 4 |
+
"eval_runtime": 7650.395,
|
| 5 |
+
"eval_samples_per_second": 1.256,
|
| 6 |
+
"eval_steps_per_second": 0.157,
|
| 7 |
+
"eval_wer": 8.493092929242826,
|
| 8 |
+
"train_loss": 0.11752363123212542,
|
| 9 |
+
"train_runtime": 521775.8612,
|
| 10 |
+
"train_samples_per_second": 1.717,
|
| 11 |
+
"train_steps_per_second": 0.054
|
| 12 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.13,
|
| 3 |
+
"eval_loss": 0.05638869106769562,
|
| 4 |
+
"eval_runtime": 7650.395,
|
| 5 |
+
"eval_samples_per_second": 1.256,
|
| 6 |
+
"eval_steps_per_second": 0.157,
|
| 7 |
+
"eval_wer": 8.493092929242826
|
| 8 |
+
}
|
runs/Jul17_11-19-44_DESKTOP-D9D4DSN/events.out.tfevents.1690111714.DESKTOP-D9D4DSN.15996.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:495342ff7e2b5346a39200a3f4d19364e6b24577783215e3f2771e8d6e69eef3
|
| 3 |
+
size 412
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.13,
|
| 3 |
+
"train_loss": 0.11752363123212542,
|
| 4 |
+
"train_runtime": 521775.8612,
|
| 5 |
+
"train_samples_per_second": 1.717,
|
| 6 |
+
"train_steps_per_second": 0.054
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,487 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 8.493092929242826,
|
| 3 |
+
"best_model_checkpoint": "./whisper-small-bn-3ds\\checkpoint-26000",
|
| 4 |
+
"epoch": 3.1294531839391992,
|
| 5 |
+
"global_step": 28000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.06,
|
| 12 |
+
"learning_rate": 5.58659217877095e-07,
|
| 13 |
+
"loss": 1.8271,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.11,
|
| 18 |
+
"learning_rate": 1.11731843575419e-06,
|
| 19 |
+
"loss": 0.6492,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.17,
|
| 24 |
+
"learning_rate": 1.675977653631285e-06,
|
| 25 |
+
"loss": 0.358,
|
| 26 |
+
"step": 1500
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.22,
|
| 30 |
+
"learning_rate": 2.23463687150838e-06,
|
| 31 |
+
"loss": 0.2337,
|
| 32 |
+
"step": 2000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.22,
|
| 36 |
+
"eval_loss": 0.2297607660293579,
|
| 37 |
+
"eval_runtime": 8187.9402,
|
| 38 |
+
"eval_samples_per_second": 1.174,
|
| 39 |
+
"eval_steps_per_second": 0.147,
|
| 40 |
+
"eval_wer": 31.773352182072205,
|
| 41 |
+
"step": 2000
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 0.28,
|
| 45 |
+
"learning_rate": 2.793296089385475e-06,
|
| 46 |
+
"loss": 0.2002,
|
| 47 |
+
"step": 2500
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 0.34,
|
| 51 |
+
"learning_rate": 3.35195530726257e-06,
|
| 52 |
+
"loss": 0.1755,
|
| 53 |
+
"step": 3000
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.39,
|
| 57 |
+
"learning_rate": 3.910614525139665e-06,
|
| 58 |
+
"loss": 0.1562,
|
| 59 |
+
"step": 3500
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.45,
|
| 63 |
+
"learning_rate": 4.46927374301676e-06,
|
| 64 |
+
"loss": 0.1444,
|
| 65 |
+
"step": 4000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.45,
|
| 69 |
+
"eval_loss": 0.1516258269548416,
|
| 70 |
+
"eval_runtime": 8249.6732,
|
| 71 |
+
"eval_samples_per_second": 1.165,
|
| 72 |
+
"eval_steps_per_second": 0.146,
|
| 73 |
+
"eval_wer": 22.238205070972114,
|
| 74 |
+
"step": 4000
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.5,
|
| 78 |
+
"learning_rate": 5.027932960893855e-06,
|
| 79 |
+
"loss": 0.134,
|
| 80 |
+
"step": 4500
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.56,
|
| 84 |
+
"learning_rate": 5.58659217877095e-06,
|
| 85 |
+
"loss": 0.1231,
|
| 86 |
+
"step": 5000
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.61,
|
| 90 |
+
"learning_rate": 6.145251396648045e-06,
|
| 91 |
+
"loss": 0.1162,
|
| 92 |
+
"step": 5500
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.67,
|
| 96 |
+
"learning_rate": 6.70391061452514e-06,
|
| 97 |
+
"loss": 0.1106,
|
| 98 |
+
"step": 6000
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.67,
|
| 102 |
+
"eval_loss": 0.1180974692106247,
|
| 103 |
+
"eval_runtime": 8273.9229,
|
| 104 |
+
"eval_samples_per_second": 1.162,
|
| 105 |
+
"eval_steps_per_second": 0.145,
|
| 106 |
+
"eval_wer": 17.223035142977604,
|
| 107 |
+
"step": 6000
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.73,
|
| 111 |
+
"learning_rate": 7.262569832402235e-06,
|
| 112 |
+
"loss": 0.1069,
|
| 113 |
+
"step": 6500
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 0.78,
|
| 117 |
+
"learning_rate": 7.82122905027933e-06,
|
| 118 |
+
"loss": 0.1014,
|
| 119 |
+
"step": 7000
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 0.84,
|
| 123 |
+
"learning_rate": 8.379888268156426e-06,
|
| 124 |
+
"loss": 0.0964,
|
| 125 |
+
"step": 7500
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 0.89,
|
| 129 |
+
"learning_rate": 8.93854748603352e-06,
|
| 130 |
+
"loss": 0.0933,
|
| 131 |
+
"step": 8000
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 0.89,
|
| 135 |
+
"eval_loss": 0.10048126429319382,
|
| 136 |
+
"eval_runtime": 9253.512,
|
| 137 |
+
"eval_samples_per_second": 1.039,
|
| 138 |
+
"eval_steps_per_second": 0.13,
|
| 139 |
+
"eval_wer": 14.800806858295513,
|
| 140 |
+
"step": 8000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.95,
|
| 144 |
+
"learning_rate": 9.497206703910616e-06,
|
| 145 |
+
"loss": 0.0908,
|
| 146 |
+
"step": 8500
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 1.01,
|
| 150 |
+
"learning_rate": 9.973753280839897e-06,
|
| 151 |
+
"loss": 0.0859,
|
| 152 |
+
"step": 9000
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 1.06,
|
| 156 |
+
"learning_rate": 9.711286089238846e-06,
|
| 157 |
+
"loss": 0.0743,
|
| 158 |
+
"step": 9500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.12,
|
| 162 |
+
"learning_rate": 9.448818897637797e-06,
|
| 163 |
+
"loss": 0.0747,
|
| 164 |
+
"step": 10000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 1.12,
|
| 168 |
+
"eval_loss": 0.08645793795585632,
|
| 169 |
+
"eval_runtime": 8690.2436,
|
| 170 |
+
"eval_samples_per_second": 1.106,
|
| 171 |
+
"eval_steps_per_second": 0.138,
|
| 172 |
+
"eval_wer": 12.718351892727528,
|
| 173 |
+
"step": 10000
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 1.17,
|
| 177 |
+
"learning_rate": 9.186351706036746e-06,
|
| 178 |
+
"loss": 0.0706,
|
| 179 |
+
"step": 10500
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.23,
|
| 183 |
+
"learning_rate": 8.923884514435697e-06,
|
| 184 |
+
"loss": 0.0687,
|
| 185 |
+
"step": 11000
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 1.29,
|
| 189 |
+
"learning_rate": 8.661417322834647e-06,
|
| 190 |
+
"loss": 0.068,
|
| 191 |
+
"step": 11500
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 1.34,
|
| 195 |
+
"learning_rate": 8.398950131233596e-06,
|
| 196 |
+
"loss": 0.0642,
|
| 197 |
+
"step": 12000
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 1.34,
|
| 201 |
+
"eval_loss": 0.07882149517536163,
|
| 202 |
+
"eval_runtime": 9343.6844,
|
| 203 |
+
"eval_samples_per_second": 1.029,
|
| 204 |
+
"eval_steps_per_second": 0.129,
|
| 205 |
+
"eval_wer": 11.644056447012723,
|
| 206 |
+
"step": 12000
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 1.4,
|
| 210 |
+
"learning_rate": 8.136482939632546e-06,
|
| 211 |
+
"loss": 0.0644,
|
| 212 |
+
"step": 12500
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 1.45,
|
| 216 |
+
"learning_rate": 7.874015748031496e-06,
|
| 217 |
+
"loss": 0.0631,
|
| 218 |
+
"step": 13000
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 1.51,
|
| 222 |
+
"learning_rate": 7.611548556430447e-06,
|
| 223 |
+
"loss": 0.0615,
|
| 224 |
+
"step": 13500
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 1.56,
|
| 228 |
+
"learning_rate": 7.349081364829397e-06,
|
| 229 |
+
"loss": 0.0613,
|
| 230 |
+
"step": 14000
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 1.56,
|
| 234 |
+
"eval_loss": 0.07153687626123428,
|
| 235 |
+
"eval_runtime": 9348.8718,
|
| 236 |
+
"eval_samples_per_second": 1.028,
|
| 237 |
+
"eval_steps_per_second": 0.129,
|
| 238 |
+
"eval_wer": 10.652430908624952,
|
| 239 |
+
"step": 14000
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 1.62,
|
| 243 |
+
"learning_rate": 7.086614173228347e-06,
|
| 244 |
+
"loss": 0.0597,
|
| 245 |
+
"step": 14500
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 1.68,
|
| 249 |
+
"learning_rate": 6.824146981627298e-06,
|
| 250 |
+
"loss": 0.0586,
|
| 251 |
+
"step": 15000
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 1.73,
|
| 255 |
+
"learning_rate": 6.561679790026248e-06,
|
| 256 |
+
"loss": 0.0569,
|
| 257 |
+
"step": 15500
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 1.79,
|
| 261 |
+
"learning_rate": 6.299212598425197e-06,
|
| 262 |
+
"loss": 0.0565,
|
| 263 |
+
"step": 16000
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.79,
|
| 267 |
+
"eval_loss": 0.0665777176618576,
|
| 268 |
+
"eval_runtime": 8488.0108,
|
| 269 |
+
"eval_samples_per_second": 1.132,
|
| 270 |
+
"eval_steps_per_second": 0.142,
|
| 271 |
+
"eval_wer": 9.88070732372709,
|
| 272 |
+
"step": 16000
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 1.84,
|
| 276 |
+
"learning_rate": 6.036745406824147e-06,
|
| 277 |
+
"loss": 0.0559,
|
| 278 |
+
"step": 16500
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 1.9,
|
| 282 |
+
"learning_rate": 5.774278215223098e-06,
|
| 283 |
+
"loss": 0.056,
|
| 284 |
+
"step": 17000
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 1.96,
|
| 288 |
+
"learning_rate": 5.511811023622048e-06,
|
| 289 |
+
"loss": 0.0537,
|
| 290 |
+
"step": 17500
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 2.01,
|
| 294 |
+
"learning_rate": 5.2493438320209976e-06,
|
| 295 |
+
"loss": 0.0494,
|
| 296 |
+
"step": 18000
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 2.01,
|
| 300 |
+
"eval_loss": 0.06404710561037064,
|
| 301 |
+
"eval_runtime": 9359.4591,
|
| 302 |
+
"eval_samples_per_second": 1.027,
|
| 303 |
+
"eval_steps_per_second": 0.128,
|
| 304 |
+
"eval_wer": 9.52646677083075,
|
| 305 |
+
"step": 18000
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 2.07,
|
| 309 |
+
"learning_rate": 4.986876640419948e-06,
|
| 310 |
+
"loss": 0.0369,
|
| 311 |
+
"step": 18500
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 2.12,
|
| 315 |
+
"learning_rate": 4.724409448818898e-06,
|
| 316 |
+
"loss": 0.0368,
|
| 317 |
+
"step": 19000
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 2.18,
|
| 321 |
+
"learning_rate": 4.461942257217848e-06,
|
| 322 |
+
"loss": 0.0367,
|
| 323 |
+
"step": 19500
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 2.24,
|
| 327 |
+
"learning_rate": 4.199475065616798e-06,
|
| 328 |
+
"loss": 0.037,
|
| 329 |
+
"step": 20000
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 2.24,
|
| 333 |
+
"eval_loss": 0.06194682419300079,
|
| 334 |
+
"eval_runtime": 9390.5519,
|
| 335 |
+
"eval_samples_per_second": 1.023,
|
| 336 |
+
"eval_steps_per_second": 0.128,
|
| 337 |
+
"eval_wer": 9.031687375478452,
|
| 338 |
+
"step": 20000
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 2.29,
|
| 342 |
+
"learning_rate": 3.937007874015748e-06,
|
| 343 |
+
"loss": 0.0363,
|
| 344 |
+
"step": 20500
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 2.35,
|
| 348 |
+
"learning_rate": 3.6745406824146986e-06,
|
| 349 |
+
"loss": 0.0358,
|
| 350 |
+
"step": 21000
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 2.4,
|
| 354 |
+
"learning_rate": 3.412073490813649e-06,
|
| 355 |
+
"loss": 0.0364,
|
| 356 |
+
"step": 21500
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.46,
|
| 360 |
+
"learning_rate": 3.1496062992125985e-06,
|
| 361 |
+
"loss": 0.0348,
|
| 362 |
+
"step": 22000
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 2.46,
|
| 366 |
+
"eval_loss": 0.05995701625943184,
|
| 367 |
+
"eval_runtime": 8568.802,
|
| 368 |
+
"eval_samples_per_second": 1.122,
|
| 369 |
+
"eval_steps_per_second": 0.14,
|
| 370 |
+
"eval_wer": 9.032927424088358,
|
| 371 |
+
"step": 22000
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 2.51,
|
| 375 |
+
"learning_rate": 2.887139107611549e-06,
|
| 376 |
+
"loss": 0.0348,
|
| 377 |
+
"step": 22500
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 2.57,
|
| 381 |
+
"learning_rate": 2.6246719160104988e-06,
|
| 382 |
+
"loss": 0.0334,
|
| 383 |
+
"step": 23000
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 2.63,
|
| 387 |
+
"learning_rate": 2.362204724409449e-06,
|
| 388 |
+
"loss": 0.0333,
|
| 389 |
+
"step": 23500
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 2.68,
|
| 393 |
+
"learning_rate": 2.099737532808399e-06,
|
| 394 |
+
"loss": 0.0329,
|
| 395 |
+
"step": 24000
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 2.68,
|
| 399 |
+
"eval_loss": 0.058194514364004135,
|
| 400 |
+
"eval_runtime": 7732.5254,
|
| 401 |
+
"eval_samples_per_second": 1.243,
|
| 402 |
+
"eval_steps_per_second": 0.155,
|
| 403 |
+
"eval_wer": 8.840719889553004,
|
| 404 |
+
"step": 24000
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 2.74,
|
| 408 |
+
"learning_rate": 1.8372703412073493e-06,
|
| 409 |
+
"loss": 0.0332,
|
| 410 |
+
"step": 24500
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 2.79,
|
| 414 |
+
"learning_rate": 1.5748031496062992e-06,
|
| 415 |
+
"loss": 0.0326,
|
| 416 |
+
"step": 25000
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 2.85,
|
| 420 |
+
"learning_rate": 1.3123359580052494e-06,
|
| 421 |
+
"loss": 0.0321,
|
| 422 |
+
"step": 25500
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 2.91,
|
| 426 |
+
"learning_rate": 1.0498687664041996e-06,
|
| 427 |
+
"loss": 0.0321,
|
| 428 |
+
"step": 26000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 2.91,
|
| 432 |
+
"eval_loss": 0.05638869106769562,
|
| 433 |
+
"eval_runtime": 7807.0833,
|
| 434 |
+
"eval_samples_per_second": 1.231,
|
| 435 |
+
"eval_steps_per_second": 0.154,
|
| 436 |
+
"eval_wer": 8.493092929242826,
|
| 437 |
+
"step": 26000
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 2.96,
|
| 441 |
+
"learning_rate": 7.874015748031496e-07,
|
| 442 |
+
"loss": 0.0318,
|
| 443 |
+
"step": 26500
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 3.02,
|
| 447 |
+
"learning_rate": 5.249343832020998e-07,
|
| 448 |
+
"loss": 0.0285,
|
| 449 |
+
"step": 27000
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 3.07,
|
| 453 |
+
"learning_rate": 2.624671916010499e-07,
|
| 454 |
+
"loss": 0.0228,
|
| 455 |
+
"step": 27500
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 3.13,
|
| 459 |
+
"learning_rate": 0.0,
|
| 460 |
+
"loss": 0.0227,
|
| 461 |
+
"step": 28000
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 3.13,
|
| 465 |
+
"eval_loss": 0.05735579878091812,
|
| 466 |
+
"eval_runtime": 7923.7534,
|
| 467 |
+
"eval_samples_per_second": 1.213,
|
| 468 |
+
"eval_steps_per_second": 0.152,
|
| 469 |
+
"eval_wer": 8.50384001719534,
|
| 470 |
+
"step": 28000
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 3.13,
|
| 474 |
+
"step": 28000,
|
| 475 |
+
"total_flos": 2.5857251868672e+20,
|
| 476 |
+
"train_loss": 0.11752363123212542,
|
| 477 |
+
"train_runtime": 521775.8612,
|
| 478 |
+
"train_samples_per_second": 1.717,
|
| 479 |
+
"train_steps_per_second": 0.054
|
| 480 |
+
}
|
| 481 |
+
],
|
| 482 |
+
"max_steps": 28000,
|
| 483 |
+
"num_train_epochs": 4,
|
| 484 |
+
"total_flos": 2.5857251868672e+20,
|
| 485 |
+
"trial_name": null,
|
| 486 |
+
"trial_params": null
|
| 487 |
+
}
|