Commit
•
4cad449
1
Parent(s):
0b4f67c
End of training
Browse files- all_results.json +14 -0
- eval_results.json +9 -0
- train_results.json +8 -0
- trainer_state.json +835 -0
all_results.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"eval_loss": 0.30670487880706787,
|
| 4 |
+
"eval_runtime": 85.7358,
|
| 5 |
+
"eval_samples": 2217,
|
| 6 |
+
"eval_samples_per_second": 25.858,
|
| 7 |
+
"eval_steps_per_second": 1.621,
|
| 8 |
+
"eval_wer": 0.2678153846153846,
|
| 9 |
+
"train_loss": 1.4761058049029614,
|
| 10 |
+
"train_runtime": 17861.6167,
|
| 11 |
+
"train_samples": 5281,
|
| 12 |
+
"train_samples_per_second": 17.74,
|
| 13 |
+
"train_steps_per_second": 0.558
|
| 14 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"eval_loss": 0.30670487880706787,
|
| 4 |
+
"eval_runtime": 85.7358,
|
| 5 |
+
"eval_samples": 2217,
|
| 6 |
+
"eval_samples_per_second": 25.858,
|
| 7 |
+
"eval_steps_per_second": 1.621,
|
| 8 |
+
"eval_wer": 0.2678153846153846
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"train_loss": 1.4761058049029614,
|
| 4 |
+
"train_runtime": 17861.6167,
|
| 5 |
+
"train_samples": 5281,
|
| 6 |
+
"train_samples_per_second": 17.74,
|
| 7 |
+
"train_steps_per_second": 0.558
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,835 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 60.0,
|
| 5 |
+
"global_step": 9960,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.6,
|
| 12 |
+
"learning_rate": 4.8e-06,
|
| 13 |
+
"loss": 27.2555,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 1.2,
|
| 18 |
+
"learning_rate": 9.799999999999998e-06,
|
| 19 |
+
"loss": 12.0667,
|
| 20 |
+
"step": 200
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 1.81,
|
| 24 |
+
"learning_rate": 1.4799999999999999e-05,
|
| 25 |
+
"loss": 6.8407,
|
| 26 |
+
"step": 300
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 2.41,
|
| 30 |
+
"learning_rate": 1.98e-05,
|
| 31 |
+
"loss": 5.175,
|
| 32 |
+
"step": 400
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 2.41,
|
| 36 |
+
"eval_loss": 4.690936088562012,
|
| 37 |
+
"eval_runtime": 87.6676,
|
| 38 |
+
"eval_samples_per_second": 25.289,
|
| 39 |
+
"eval_steps_per_second": 1.586,
|
| 40 |
+
"eval_wer": 1.0,
|
| 41 |
+
"step": 400
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 3.01,
|
| 45 |
+
"learning_rate": 2.4799999999999996e-05,
|
| 46 |
+
"loss": 4.1143,
|
| 47 |
+
"step": 500
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 3.61,
|
| 51 |
+
"learning_rate": 2.9799999999999996e-05,
|
| 52 |
+
"loss": 3.5613,
|
| 53 |
+
"step": 600
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 4.22,
|
| 57 |
+
"learning_rate": 3.48e-05,
|
| 58 |
+
"loss": 3.4854,
|
| 59 |
+
"step": 700
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 4.82,
|
| 63 |
+
"learning_rate": 3.979999999999999e-05,
|
| 64 |
+
"loss": 3.3785,
|
| 65 |
+
"step": 800
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 4.82,
|
| 69 |
+
"eval_loss": 3.308009147644043,
|
| 70 |
+
"eval_runtime": 87.3792,
|
| 71 |
+
"eval_samples_per_second": 25.372,
|
| 72 |
+
"eval_steps_per_second": 1.591,
|
| 73 |
+
"eval_wer": 1.0,
|
| 74 |
+
"step": 800
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 5.42,
|
| 78 |
+
"learning_rate": 4.48e-05,
|
| 79 |
+
"loss": 3.2155,
|
| 80 |
+
"step": 900
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 6.02,
|
| 84 |
+
"learning_rate": 4.98e-05,
|
| 85 |
+
"loss": 3.1587,
|
| 86 |
+
"step": 1000
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 6.63,
|
| 90 |
+
"learning_rate": 5.48e-05,
|
| 91 |
+
"loss": 3.0407,
|
| 92 |
+
"step": 1100
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 7.23,
|
| 96 |
+
"learning_rate": 5.98e-05,
|
| 97 |
+
"loss": 2.6964,
|
| 98 |
+
"step": 1200
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 7.23,
|
| 102 |
+
"eval_loss": 2.0650923252105713,
|
| 103 |
+
"eval_runtime": 86.9803,
|
| 104 |
+
"eval_samples_per_second": 25.489,
|
| 105 |
+
"eval_steps_per_second": 1.598,
|
| 106 |
+
"eval_wer": 1.105476923076923,
|
| 107 |
+
"step": 1200
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 7.83,
|
| 111 |
+
"learning_rate": 6.479999999999999e-05,
|
| 112 |
+
"loss": 2.0851,
|
| 113 |
+
"step": 1300
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 8.43,
|
| 117 |
+
"learning_rate": 6.979999999999999e-05,
|
| 118 |
+
"loss": 1.6595,
|
| 119 |
+
"step": 1400
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 9.04,
|
| 123 |
+
"learning_rate": 7.479999999999999e-05,
|
| 124 |
+
"loss": 1.4478,
|
| 125 |
+
"step": 1500
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 9.64,
|
| 129 |
+
"learning_rate": 7.414893617021276e-05,
|
| 130 |
+
"loss": 1.3008,
|
| 131 |
+
"step": 1600
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 9.64,
|
| 135 |
+
"eval_loss": 0.5845397114753723,
|
| 136 |
+
"eval_runtime": 86.5846,
|
| 137 |
+
"eval_samples_per_second": 25.605,
|
| 138 |
+
"eval_steps_per_second": 1.605,
|
| 139 |
+
"eval_wer": 0.6206769230769231,
|
| 140 |
+
"step": 1600
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 10.24,
|
| 144 |
+
"learning_rate": 7.326241134751773e-05,
|
| 145 |
+
"loss": 1.2563,
|
| 146 |
+
"step": 1700
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 10.84,
|
| 150 |
+
"learning_rate": 7.237588652482268e-05,
|
| 151 |
+
"loss": 1.1552,
|
| 152 |
+
"step": 1800
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 11.45,
|
| 156 |
+
"learning_rate": 7.148936170212766e-05,
|
| 157 |
+
"loss": 1.1357,
|
| 158 |
+
"step": 1900
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 12.05,
|
| 162 |
+
"learning_rate": 7.060283687943261e-05,
|
| 163 |
+
"loss": 1.1185,
|
| 164 |
+
"step": 2000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 12.05,
|
| 168 |
+
"eval_loss": 0.4195278584957123,
|
| 169 |
+
"eval_runtime": 87.4194,
|
| 170 |
+
"eval_samples_per_second": 25.361,
|
| 171 |
+
"eval_steps_per_second": 1.59,
|
| 172 |
+
"eval_wer": 0.4193230769230769,
|
| 173 |
+
"step": 2000
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 12.65,
|
| 177 |
+
"learning_rate": 6.971631205673758e-05,
|
| 178 |
+
"loss": 1.0362,
|
| 179 |
+
"step": 2100
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 13.25,
|
| 183 |
+
"learning_rate": 6.882978723404255e-05,
|
| 184 |
+
"loss": 1.0465,
|
| 185 |
+
"step": 2200
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 13.86,
|
| 189 |
+
"learning_rate": 6.79432624113475e-05,
|
| 190 |
+
"loss": 0.984,
|
| 191 |
+
"step": 2300
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 14.46,
|
| 195 |
+
"learning_rate": 6.705673758865247e-05,
|
| 196 |
+
"loss": 1.0252,
|
| 197 |
+
"step": 2400
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 14.46,
|
| 201 |
+
"eval_loss": 0.3823728859424591,
|
| 202 |
+
"eval_runtime": 87.0092,
|
| 203 |
+
"eval_samples_per_second": 25.48,
|
| 204 |
+
"eval_steps_per_second": 1.598,
|
| 205 |
+
"eval_wer": 0.35704615384615385,
|
| 206 |
+
"step": 2400
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 15.06,
|
| 210 |
+
"learning_rate": 6.617021276595744e-05,
|
| 211 |
+
"loss": 0.9813,
|
| 212 |
+
"step": 2500
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 15.66,
|
| 216 |
+
"learning_rate": 6.52836879432624e-05,
|
| 217 |
+
"loss": 0.9642,
|
| 218 |
+
"step": 2600
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 16.27,
|
| 222 |
+
"learning_rate": 6.439716312056737e-05,
|
| 223 |
+
"loss": 0.9494,
|
| 224 |
+
"step": 2700
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 16.87,
|
| 228 |
+
"learning_rate": 6.351063829787234e-05,
|
| 229 |
+
"loss": 0.935,
|
| 230 |
+
"step": 2800
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 16.87,
|
| 234 |
+
"eval_loss": 0.36926984786987305,
|
| 235 |
+
"eval_runtime": 88.3011,
|
| 236 |
+
"eval_samples_per_second": 25.107,
|
| 237 |
+
"eval_steps_per_second": 1.574,
|
| 238 |
+
"eval_wer": 0.3462153846153846,
|
| 239 |
+
"step": 2800
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 17.47,
|
| 243 |
+
"learning_rate": 6.26241134751773e-05,
|
| 244 |
+
"loss": 0.9282,
|
| 245 |
+
"step": 2900
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 18.07,
|
| 249 |
+
"learning_rate": 6.173758865248226e-05,
|
| 250 |
+
"loss": 0.8868,
|
| 251 |
+
"step": 3000
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 18.67,
|
| 255 |
+
"learning_rate": 6.0851063829787224e-05,
|
| 256 |
+
"loss": 0.8898,
|
| 257 |
+
"step": 3100
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 19.28,
|
| 261 |
+
"learning_rate": 5.99645390070922e-05,
|
| 262 |
+
"loss": 0.8818,
|
| 263 |
+
"step": 3200
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 19.28,
|
| 267 |
+
"eval_loss": 0.3587293326854706,
|
| 268 |
+
"eval_runtime": 87.4375,
|
| 269 |
+
"eval_samples_per_second": 25.355,
|
| 270 |
+
"eval_steps_per_second": 1.59,
|
| 271 |
+
"eval_wer": 0.3318153846153846,
|
| 272 |
+
"step": 3200
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 19.88,
|
| 276 |
+
"learning_rate": 5.907801418439716e-05,
|
| 277 |
+
"loss": 0.8633,
|
| 278 |
+
"step": 3300
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 20.48,
|
| 282 |
+
"learning_rate": 5.819148936170212e-05,
|
| 283 |
+
"loss": 0.8942,
|
| 284 |
+
"step": 3400
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 21.08,
|
| 288 |
+
"learning_rate": 5.730496453900709e-05,
|
| 289 |
+
"loss": 0.856,
|
| 290 |
+
"step": 3500
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 21.69,
|
| 294 |
+
"learning_rate": 5.641843971631205e-05,
|
| 295 |
+
"loss": 0.8534,
|
| 296 |
+
"step": 3600
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 21.69,
|
| 300 |
+
"eval_loss": 0.3420199751853943,
|
| 301 |
+
"eval_runtime": 88.0755,
|
| 302 |
+
"eval_samples_per_second": 25.172,
|
| 303 |
+
"eval_steps_per_second": 1.578,
|
| 304 |
+
"eval_wer": 0.31803076923076923,
|
| 305 |
+
"step": 3600
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 22.29,
|
| 309 |
+
"learning_rate": 5.5531914893617016e-05,
|
| 310 |
+
"loss": 0.8509,
|
| 311 |
+
"step": 3700
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 22.89,
|
| 315 |
+
"learning_rate": 5.4645390070921984e-05,
|
| 316 |
+
"loss": 0.8185,
|
| 317 |
+
"step": 3800
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 23.49,
|
| 321 |
+
"learning_rate": 5.375886524822695e-05,
|
| 322 |
+
"loss": 0.8319,
|
| 323 |
+
"step": 3900
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 24.1,
|
| 327 |
+
"learning_rate": 5.287234042553191e-05,
|
| 328 |
+
"loss": 0.8137,
|
| 329 |
+
"step": 4000
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 24.1,
|
| 333 |
+
"eval_loss": 0.34257009625434875,
|
| 334 |
+
"eval_runtime": 89.8454,
|
| 335 |
+
"eval_samples_per_second": 24.676,
|
| 336 |
+
"eval_steps_per_second": 1.547,
|
| 337 |
+
"eval_wer": 0.31298461538461536,
|
| 338 |
+
"step": 4000
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 24.7,
|
| 342 |
+
"learning_rate": 5.198581560283687e-05,
|
| 343 |
+
"loss": 0.8089,
|
| 344 |
+
"step": 4100
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 25.3,
|
| 348 |
+
"learning_rate": 5.110815602836879e-05,
|
| 349 |
+
"loss": 0.8291,
|
| 350 |
+
"step": 4200
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 25.9,
|
| 354 |
+
"learning_rate": 5.0221631205673754e-05,
|
| 355 |
+
"loss": 0.7874,
|
| 356 |
+
"step": 4300
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 26.51,
|
| 360 |
+
"learning_rate": 4.9335106382978715e-05,
|
| 361 |
+
"loss": 0.7968,
|
| 362 |
+
"step": 4400
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 26.51,
|
| 366 |
+
"eval_loss": 0.334939181804657,
|
| 367 |
+
"eval_runtime": 85.418,
|
| 368 |
+
"eval_samples_per_second": 25.955,
|
| 369 |
+
"eval_steps_per_second": 1.627,
|
| 370 |
+
"eval_wer": 0.31015384615384617,
|
| 371 |
+
"step": 4400
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 27.11,
|
| 375 |
+
"learning_rate": 4.844858156028368e-05,
|
| 376 |
+
"loss": 0.7791,
|
| 377 |
+
"step": 4500
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 27.71,
|
| 381 |
+
"learning_rate": 4.756205673758865e-05,
|
| 382 |
+
"loss": 0.7802,
|
| 383 |
+
"step": 4600
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 28.31,
|
| 387 |
+
"learning_rate": 4.667553191489362e-05,
|
| 388 |
+
"loss": 0.7639,
|
| 389 |
+
"step": 4700
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 28.92,
|
| 393 |
+
"learning_rate": 4.579787234042553e-05,
|
| 394 |
+
"loss": 0.7558,
|
| 395 |
+
"step": 4800
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 28.92,
|
| 399 |
+
"eval_loss": 0.3215944170951843,
|
| 400 |
+
"eval_runtime": 85.8433,
|
| 401 |
+
"eval_samples_per_second": 25.826,
|
| 402 |
+
"eval_steps_per_second": 1.619,
|
| 403 |
+
"eval_wer": 0.3019076923076923,
|
| 404 |
+
"step": 4800
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 29.52,
|
| 408 |
+
"learning_rate": 4.49113475177305e-05,
|
| 409 |
+
"loss": 0.8011,
|
| 410 |
+
"step": 4900
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 30.12,
|
| 414 |
+
"learning_rate": 4.402482269503546e-05,
|
| 415 |
+
"loss": 0.7381,
|
| 416 |
+
"step": 5000
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 30.72,
|
| 420 |
+
"learning_rate": 4.313829787234042e-05,
|
| 421 |
+
"loss": 0.7412,
|
| 422 |
+
"step": 5100
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 31.33,
|
| 426 |
+
"learning_rate": 4.225177304964539e-05,
|
| 427 |
+
"loss": 0.7313,
|
| 428 |
+
"step": 5200
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 31.33,
|
| 432 |
+
"eval_loss": 0.3450605571269989,
|
| 433 |
+
"eval_runtime": 86.5107,
|
| 434 |
+
"eval_samples_per_second": 25.627,
|
| 435 |
+
"eval_steps_per_second": 1.607,
|
| 436 |
+
"eval_wer": 0.30596923076923077,
|
| 437 |
+
"step": 5200
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 31.93,
|
| 441 |
+
"learning_rate": 4.136524822695035e-05,
|
| 442 |
+
"loss": 0.7421,
|
| 443 |
+
"step": 5300
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 32.53,
|
| 447 |
+
"learning_rate": 4.047872340425531e-05,
|
| 448 |
+
"loss": 0.7639,
|
| 449 |
+
"step": 5400
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 33.13,
|
| 453 |
+
"learning_rate": 3.9592198581560283e-05,
|
| 454 |
+
"loss": 0.7198,
|
| 455 |
+
"step": 5500
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 33.73,
|
| 459 |
+
"learning_rate": 3.8705673758865244e-05,
|
| 460 |
+
"loss": 0.7358,
|
| 461 |
+
"step": 5600
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 33.73,
|
| 465 |
+
"eval_loss": 0.32715940475463867,
|
| 466 |
+
"eval_runtime": 89.1066,
|
| 467 |
+
"eval_samples_per_second": 24.88,
|
| 468 |
+
"eval_steps_per_second": 1.56,
|
| 469 |
+
"eval_wer": 0.29673846153846156,
|
| 470 |
+
"step": 5600
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 34.34,
|
| 474 |
+
"learning_rate": 3.781914893617021e-05,
|
| 475 |
+
"loss": 0.7053,
|
| 476 |
+
"step": 5700
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 34.94,
|
| 480 |
+
"learning_rate": 3.693262411347517e-05,
|
| 481 |
+
"loss": 0.7084,
|
| 482 |
+
"step": 5800
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 35.54,
|
| 486 |
+
"learning_rate": 3.604609929078014e-05,
|
| 487 |
+
"loss": 0.7029,
|
| 488 |
+
"step": 5900
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 36.14,
|
| 492 |
+
"learning_rate": 3.51595744680851e-05,
|
| 493 |
+
"loss": 0.718,
|
| 494 |
+
"step": 6000
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 36.14,
|
| 498 |
+
"eval_loss": 0.3315288722515106,
|
| 499 |
+
"eval_runtime": 85.6491,
|
| 500 |
+
"eval_samples_per_second": 25.885,
|
| 501 |
+
"eval_steps_per_second": 1.623,
|
| 502 |
+
"eval_wer": 0.2882461538461538,
|
| 503 |
+
"step": 6000
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 36.75,
|
| 507 |
+
"learning_rate": 3.427304964539007e-05,
|
| 508 |
+
"loss": 0.7136,
|
| 509 |
+
"step": 6100
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 37.35,
|
| 513 |
+
"learning_rate": 3.338652482269503e-05,
|
| 514 |
+
"loss": 0.6807,
|
| 515 |
+
"step": 6200
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 37.95,
|
| 519 |
+
"learning_rate": 3.25e-05,
|
| 520 |
+
"loss": 0.6968,
|
| 521 |
+
"step": 6300
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 38.55,
|
| 525 |
+
"learning_rate": 3.1613475177304965e-05,
|
| 526 |
+
"loss": 0.6991,
|
| 527 |
+
"step": 6400
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 38.55,
|
| 531 |
+
"eval_loss": 0.3299025297164917,
|
| 532 |
+
"eval_runtime": 91.3334,
|
| 533 |
+
"eval_samples_per_second": 24.274,
|
| 534 |
+
"eval_steps_per_second": 1.522,
|
| 535 |
+
"eval_wer": 0.28295384615384617,
|
| 536 |
+
"step": 6400
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 39.16,
|
| 540 |
+
"learning_rate": 3.0726950354609926e-05,
|
| 541 |
+
"loss": 0.6666,
|
| 542 |
+
"step": 6500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 39.76,
|
| 546 |
+
"learning_rate": 2.9840425531914893e-05,
|
| 547 |
+
"loss": 0.669,
|
| 548 |
+
"step": 6600
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 40.36,
|
| 552 |
+
"learning_rate": 2.8953900709219857e-05,
|
| 553 |
+
"loss": 0.6367,
|
| 554 |
+
"step": 6700
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 40.96,
|
| 558 |
+
"learning_rate": 2.8067375886524818e-05,
|
| 559 |
+
"loss": 0.6529,
|
| 560 |
+
"step": 6800
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 40.96,
|
| 564 |
+
"eval_loss": 0.3140033185482025,
|
| 565 |
+
"eval_runtime": 86.1836,
|
| 566 |
+
"eval_samples_per_second": 25.724,
|
| 567 |
+
"eval_steps_per_second": 1.613,
|
| 568 |
+
"eval_wer": 0.2835692307692308,
|
| 569 |
+
"step": 6800
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 41.57,
|
| 573 |
+
"learning_rate": 2.7180851063829786e-05,
|
| 574 |
+
"loss": 0.6539,
|
| 575 |
+
"step": 6900
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 42.17,
|
| 579 |
+
"learning_rate": 2.629432624113475e-05,
|
| 580 |
+
"loss": 0.6421,
|
| 581 |
+
"step": 7000
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 42.77,
|
| 585 |
+
"learning_rate": 2.5407801418439714e-05,
|
| 586 |
+
"loss": 0.6533,
|
| 587 |
+
"step": 7100
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 43.37,
|
| 591 |
+
"learning_rate": 2.4521276595744675e-05,
|
| 592 |
+
"loss": 0.6225,
|
| 593 |
+
"step": 7200
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 43.37,
|
| 597 |
+
"eval_loss": 0.31283286213874817,
|
| 598 |
+
"eval_runtime": 89.5905,
|
| 599 |
+
"eval_samples_per_second": 24.746,
|
| 600 |
+
"eval_steps_per_second": 1.552,
|
| 601 |
+
"eval_wer": 0.27507692307692305,
|
| 602 |
+
"step": 7200
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 43.98,
|
| 606 |
+
"learning_rate": 2.3634751773049643e-05,
|
| 607 |
+
"loss": 0.6544,
|
| 608 |
+
"step": 7300
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 44.58,
|
| 612 |
+
"learning_rate": 2.2748226950354607e-05,
|
| 613 |
+
"loss": 0.626,
|
| 614 |
+
"step": 7400
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 45.18,
|
| 618 |
+
"learning_rate": 2.186170212765957e-05,
|
| 619 |
+
"loss": 0.6182,
|
| 620 |
+
"step": 7500
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 45.78,
|
| 624 |
+
"learning_rate": 2.097517730496454e-05,
|
| 625 |
+
"loss": 0.633,
|
| 626 |
+
"step": 7600
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 45.78,
|
| 630 |
+
"eval_loss": 0.3211357891559601,
|
| 631 |
+
"eval_runtime": 86.4578,
|
| 632 |
+
"eval_samples_per_second": 25.643,
|
| 633 |
+
"eval_steps_per_second": 1.608,
|
| 634 |
+
"eval_wer": 0.2774153846153846,
|
| 635 |
+
"step": 7600
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 46.39,
|
| 639 |
+
"learning_rate": 2.0097517730496455e-05,
|
| 640 |
+
"loss": 0.6123,
|
| 641 |
+
"step": 7700
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 46.99,
|
| 645 |
+
"learning_rate": 1.9210992907801416e-05,
|
| 646 |
+
"loss": 0.6536,
|
| 647 |
+
"step": 7800
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 47.59,
|
| 651 |
+
"learning_rate": 1.832446808510638e-05,
|
| 652 |
+
"loss": 0.614,
|
| 653 |
+
"step": 7900
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 48.19,
|
| 657 |
+
"learning_rate": 1.7437943262411345e-05,
|
| 658 |
+
"loss": 0.5876,
|
| 659 |
+
"step": 8000
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 48.19,
|
| 663 |
+
"eval_loss": 0.3162022531032562,
|
| 664 |
+
"eval_runtime": 87.3839,
|
| 665 |
+
"eval_samples_per_second": 25.371,
|
| 666 |
+
"eval_steps_per_second": 1.591,
|
| 667 |
+
"eval_wer": 0.27643076923076926,
|
| 668 |
+
"step": 8000
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 48.8,
|
| 672 |
+
"learning_rate": 1.656028368794326e-05,
|
| 673 |
+
"loss": 0.6021,
|
| 674 |
+
"step": 8100
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 49.4,
|
| 678 |
+
"learning_rate": 1.5673758865248225e-05,
|
| 679 |
+
"loss": 0.5854,
|
| 680 |
+
"step": 8200
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 50.0,
|
| 684 |
+
"learning_rate": 1.4787234042553191e-05,
|
| 685 |
+
"loss": 0.6204,
|
| 686 |
+
"step": 8300
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 50.6,
|
| 690 |
+
"learning_rate": 1.3900709219858154e-05,
|
| 691 |
+
"loss": 0.588,
|
| 692 |
+
"step": 8400
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 50.6,
|
| 696 |
+
"eval_loss": 0.30821505188941956,
|
| 697 |
+
"eval_runtime": 91.283,
|
| 698 |
+
"eval_samples_per_second": 24.287,
|
| 699 |
+
"eval_steps_per_second": 1.523,
|
| 700 |
+
"eval_wer": 0.27224615384615386,
|
| 701 |
+
"step": 8400
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 51.2,
|
| 705 |
+
"learning_rate": 1.301418439716312e-05,
|
| 706 |
+
"loss": 0.5872,
|
| 707 |
+
"step": 8500
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 51.81,
|
| 711 |
+
"learning_rate": 1.2127659574468086e-05,
|
| 712 |
+
"loss": 0.5996,
|
| 713 |
+
"step": 8600
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 52.41,
|
| 717 |
+
"learning_rate": 1.1241134751773048e-05,
|
| 718 |
+
"loss": 0.6042,
|
| 719 |
+
"step": 8700
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 53.01,
|
| 723 |
+
"learning_rate": 1.0354609929078014e-05,
|
| 724 |
+
"loss": 0.5915,
|
| 725 |
+
"step": 8800
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 53.01,
|
| 729 |
+
"eval_loss": 0.3120293617248535,
|
| 730 |
+
"eval_runtime": 85.3554,
|
| 731 |
+
"eval_samples_per_second": 25.974,
|
| 732 |
+
"eval_steps_per_second": 1.628,
|
| 733 |
+
"eval_wer": 0.26806153846153846,
|
| 734 |
+
"step": 8800
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 53.61,
|
| 738 |
+
"learning_rate": 9.468085106382977e-06,
|
| 739 |
+
"loss": 0.5774,
|
| 740 |
+
"step": 8900
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 54.22,
|
| 744 |
+
"learning_rate": 8.581560283687943e-06,
|
| 745 |
+
"loss": 0.5936,
|
| 746 |
+
"step": 9000
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 54.82,
|
| 750 |
+
"learning_rate": 7.695035460992907e-06,
|
| 751 |
+
"loss": 0.5876,
|
| 752 |
+
"step": 9100
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 55.42,
|
| 756 |
+
"learning_rate": 6.808510638297872e-06,
|
| 757 |
+
"loss": 0.5798,
|
| 758 |
+
"step": 9200
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 55.42,
|
| 762 |
+
"eval_loss": 0.3132975995540619,
|
| 763 |
+
"eval_runtime": 86.5587,
|
| 764 |
+
"eval_samples_per_second": 25.613,
|
| 765 |
+
"eval_steps_per_second": 1.606,
|
| 766 |
+
"eval_wer": 0.2708923076923077,
|
| 767 |
+
"step": 9200
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 56.02,
|
| 771 |
+
"learning_rate": 5.921985815602837e-06,
|
| 772 |
+
"loss": 0.5921,
|
| 773 |
+
"step": 9300
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 56.63,
|
| 777 |
+
"learning_rate": 5.035460992907801e-06,
|
| 778 |
+
"loss": 0.5603,
|
| 779 |
+
"step": 9400
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 57.23,
|
| 783 |
+
"learning_rate": 4.148936170212765e-06,
|
| 784 |
+
"loss": 0.5659,
|
| 785 |
+
"step": 9500
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 57.83,
|
| 789 |
+
"learning_rate": 3.2624113475177303e-06,
|
| 790 |
+
"loss": 0.5736,
|
| 791 |
+
"step": 9600
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 57.83,
|
| 795 |
+
"eval_loss": 0.3086037039756775,
|
| 796 |
+
"eval_runtime": 85.004,
|
| 797 |
+
"eval_samples_per_second": 26.081,
|
| 798 |
+
"eval_steps_per_second": 1.635,
|
| 799 |
+
"eval_wer": 0.2675692307692308,
|
| 800 |
+
"step": 9600
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 58.43,
|
| 804 |
+
"learning_rate": 2.3758865248226945e-06,
|
| 805 |
+
"loss": 0.5714,
|
| 806 |
+
"step": 9700
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 59.04,
|
| 810 |
+
"learning_rate": 1.4893617021276596e-06,
|
| 811 |
+
"loss": 0.5835,
|
| 812 |
+
"step": 9800
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 59.64,
|
| 816 |
+
"learning_rate": 6.02836879432624e-07,
|
| 817 |
+
"loss": 0.5691,
|
| 818 |
+
"step": 9900
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 60.0,
|
| 822 |
+
"step": 9960,
|
| 823 |
+
"total_flos": 3.851174581370918e+19,
|
| 824 |
+
"train_loss": 1.4761058049029614,
|
| 825 |
+
"train_runtime": 17861.6167,
|
| 826 |
+
"train_samples_per_second": 17.74,
|
| 827 |
+
"train_steps_per_second": 0.558
|
| 828 |
+
}
|
| 829 |
+
],
|
| 830 |
+
"max_steps": 9960,
|
| 831 |
+
"num_train_epochs": 60,
|
| 832 |
+
"total_flos": 3.851174581370918e+19,
|
| 833 |
+
"trial_name": null,
|
| 834 |
+
"trial_params": null
|
| 835 |
+
}
|