Commit
•
20ca64c
1
Parent(s):
d269680
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 +712 -0
all_results.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 100.0,
|
| 3 |
+
"eval_loss": 0.5816978812217712,
|
| 4 |
+
"eval_runtime": 80.9703,
|
| 5 |
+
"eval_samples": 1613,
|
| 6 |
+
"eval_samples_per_second": 19.921,
|
| 7 |
+
"eval_steps_per_second": 19.921,
|
| 8 |
+
"eval_wer": 0.40963193554294086,
|
| 9 |
+
"train_loss": 1.4013342235244324,
|
| 10 |
+
"train_runtime": 14378.2108,
|
| 11 |
+
"train_samples": 3400,
|
| 12 |
+
"train_samples_per_second": 23.647,
|
| 13 |
+
"train_steps_per_second": 0.744
|
| 14 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 100.0,
|
| 3 |
+
"eval_loss": 0.5816978812217712,
|
| 4 |
+
"eval_runtime": 80.9703,
|
| 5 |
+
"eval_samples": 1613,
|
| 6 |
+
"eval_samples_per_second": 19.921,
|
| 7 |
+
"eval_steps_per_second": 19.921,
|
| 8 |
+
"eval_wer": 0.40963193554294086
|
| 9 |
+
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1262099953
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e49ac3b9a4f3441f823e4bdab20bf08f988f7ddf6211b4063d669740d1b0118e
|
| 3 |
size 1262099953
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 100.0,
|
| 3 |
+
"train_loss": 1.4013342235244324,
|
| 4 |
+
"train_runtime": 14378.2108,
|
| 5 |
+
"train_samples": 3400,
|
| 6 |
+
"train_samples_per_second": 23.647,
|
| 7 |
+
"train_steps_per_second": 0.744
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,712 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 100.0,
|
| 5 |
+
"global_step": 10700,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.93,
|
| 12 |
+
"learning_rate": 1.47e-05,
|
| 13 |
+
"loss": 9.2459,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 1.87,
|
| 18 |
+
"learning_rate": 2.97e-05,
|
| 19 |
+
"loss": 4.0521,
|
| 20 |
+
"step": 200
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 2.8,
|
| 24 |
+
"learning_rate": 4.4699999999999996e-05,
|
| 25 |
+
"loss": 3.4098,
|
| 26 |
+
"step": 300
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 3.74,
|
| 30 |
+
"learning_rate": 5.97e-05,
|
| 31 |
+
"loss": 3.2533,
|
| 32 |
+
"step": 400
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 4.67,
|
| 36 |
+
"learning_rate": 7.47e-05,
|
| 37 |
+
"loss": 3.1982,
|
| 38 |
+
"step": 500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 5.61,
|
| 42 |
+
"learning_rate": 8.969999999999998e-05,
|
| 43 |
+
"loss": 3.1301,
|
| 44 |
+
"step": 600
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 6.54,
|
| 48 |
+
"learning_rate": 0.00010469999999999998,
|
| 49 |
+
"loss": 3.0863,
|
| 50 |
+
"step": 700
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 7.48,
|
| 54 |
+
"learning_rate": 0.0001197,
|
| 55 |
+
"loss": 3.0034,
|
| 56 |
+
"step": 800
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 8.41,
|
| 60 |
+
"learning_rate": 0.0001347,
|
| 61 |
+
"loss": 2.5449,
|
| 62 |
+
"step": 900
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 9.35,
|
| 66 |
+
"learning_rate": 0.00014969999999999998,
|
| 67 |
+
"loss": 2.0192,
|
| 68 |
+
"step": 1000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 10.28,
|
| 72 |
+
"learning_rate": 0.0001647,
|
| 73 |
+
"loss": 1.846,
|
| 74 |
+
"step": 1100
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 11.21,
|
| 78 |
+
"learning_rate": 0.00017969999999999998,
|
| 79 |
+
"loss": 1.7371,
|
| 80 |
+
"step": 1200
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 12.15,
|
| 84 |
+
"learning_rate": 0.0001947,
|
| 85 |
+
"loss": 1.6727,
|
| 86 |
+
"step": 1300
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 13.08,
|
| 90 |
+
"learning_rate": 0.00020969999999999997,
|
| 91 |
+
"loss": 1.6512,
|
| 92 |
+
"step": 1400
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 14.02,
|
| 96 |
+
"learning_rate": 0.0002247,
|
| 97 |
+
"loss": 1.6324,
|
| 98 |
+
"step": 1500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 14.95,
|
| 102 |
+
"learning_rate": 0.0002397,
|
| 103 |
+
"loss": 1.5982,
|
| 104 |
+
"step": 1600
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 15.89,
|
| 108 |
+
"learning_rate": 0.00025469999999999996,
|
| 109 |
+
"loss": 1.564,
|
| 110 |
+
"step": 1700
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 16.82,
|
| 114 |
+
"learning_rate": 0.0002697,
|
| 115 |
+
"loss": 1.5777,
|
| 116 |
+
"step": 1800
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 17.76,
|
| 120 |
+
"learning_rate": 0.0002847,
|
| 121 |
+
"loss": 1.5636,
|
| 122 |
+
"step": 1900
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 18.69,
|
| 126 |
+
"learning_rate": 0.00029969999999999997,
|
| 127 |
+
"loss": 1.5412,
|
| 128 |
+
"step": 2000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 18.69,
|
| 132 |
+
"eval_loss": 0.6161174178123474,
|
| 133 |
+
"eval_runtime": 82.7828,
|
| 134 |
+
"eval_samples_per_second": 19.485,
|
| 135 |
+
"eval_steps_per_second": 19.485,
|
| 136 |
+
"eval_wer": 0.5747115912836477,
|
| 137 |
+
"step": 2000
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 19.63,
|
| 141 |
+
"learning_rate": 0.00029662068965517237,
|
| 142 |
+
"loss": 1.5532,
|
| 143 |
+
"step": 2100
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 20.56,
|
| 147 |
+
"learning_rate": 0.0002931724137931034,
|
| 148 |
+
"loss": 1.547,
|
| 149 |
+
"step": 2200
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 21.5,
|
| 153 |
+
"learning_rate": 0.00028975862068965515,
|
| 154 |
+
"loss": 1.5271,
|
| 155 |
+
"step": 2300
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 22.43,
|
| 159 |
+
"learning_rate": 0.00028631034482758615,
|
| 160 |
+
"loss": 1.5141,
|
| 161 |
+
"step": 2400
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 23.36,
|
| 165 |
+
"learning_rate": 0.0002828620689655172,
|
| 166 |
+
"loss": 1.492,
|
| 167 |
+
"step": 2500
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 24.3,
|
| 171 |
+
"learning_rate": 0.00027941379310344826,
|
| 172 |
+
"loss": 1.4687,
|
| 173 |
+
"step": 2600
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 25.23,
|
| 177 |
+
"learning_rate": 0.00027596551724137926,
|
| 178 |
+
"loss": 1.4519,
|
| 179 |
+
"step": 2700
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 26.17,
|
| 183 |
+
"learning_rate": 0.0002725172413793103,
|
| 184 |
+
"loss": 1.4215,
|
| 185 |
+
"step": 2800
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 27.1,
|
| 189 |
+
"learning_rate": 0.0002690689655172414,
|
| 190 |
+
"loss": 1.4192,
|
| 191 |
+
"step": 2900
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 28.04,
|
| 195 |
+
"learning_rate": 0.00026562068965517243,
|
| 196 |
+
"loss": 1.418,
|
| 197 |
+
"step": 3000
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 28.97,
|
| 201 |
+
"learning_rate": 0.00026217241379310343,
|
| 202 |
+
"loss": 1.3969,
|
| 203 |
+
"step": 3100
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 29.91,
|
| 207 |
+
"learning_rate": 0.00025875862068965515,
|
| 208 |
+
"loss": 1.3693,
|
| 209 |
+
"step": 3200
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 30.84,
|
| 213 |
+
"learning_rate": 0.0002553448275862069,
|
| 214 |
+
"loss": 1.3633,
|
| 215 |
+
"step": 3300
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 31.78,
|
| 219 |
+
"learning_rate": 0.0002518965517241379,
|
| 220 |
+
"loss": 1.3744,
|
| 221 |
+
"step": 3400
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 32.71,
|
| 225 |
+
"learning_rate": 0.00024844827586206894,
|
| 226 |
+
"loss": 1.3554,
|
| 227 |
+
"step": 3500
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 33.64,
|
| 231 |
+
"learning_rate": 0.000245,
|
| 232 |
+
"loss": 1.3262,
|
| 233 |
+
"step": 3600
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 34.58,
|
| 237 |
+
"learning_rate": 0.00024155172413793102,
|
| 238 |
+
"loss": 1.321,
|
| 239 |
+
"step": 3700
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 35.51,
|
| 243 |
+
"learning_rate": 0.00023810344827586205,
|
| 244 |
+
"loss": 1.3375,
|
| 245 |
+
"step": 3800
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 36.45,
|
| 249 |
+
"learning_rate": 0.0002346551724137931,
|
| 250 |
+
"loss": 1.3009,
|
| 251 |
+
"step": 3900
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 37.38,
|
| 255 |
+
"learning_rate": 0.00023120689655172413,
|
| 256 |
+
"loss": 1.311,
|
| 257 |
+
"step": 4000
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 37.38,
|
| 261 |
+
"eval_loss": 0.5707210302352905,
|
| 262 |
+
"eval_runtime": 86.6018,
|
| 263 |
+
"eval_samples_per_second": 18.625,
|
| 264 |
+
"eval_steps_per_second": 18.625,
|
| 265 |
+
"eval_wer": 0.5070499908441677,
|
| 266 |
+
"step": 4000
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 38.32,
|
| 270 |
+
"learning_rate": 0.00022775862068965516,
|
| 271 |
+
"loss": 1.2889,
|
| 272 |
+
"step": 4100
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 39.25,
|
| 276 |
+
"learning_rate": 0.00022431034482758616,
|
| 277 |
+
"loss": 1.3253,
|
| 278 |
+
"step": 4200
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 40.19,
|
| 282 |
+
"learning_rate": 0.00022086206896551722,
|
| 283 |
+
"loss": 1.2888,
|
| 284 |
+
"step": 4300
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 41.12,
|
| 288 |
+
"learning_rate": 0.00021741379310344825,
|
| 289 |
+
"loss": 1.2567,
|
| 290 |
+
"step": 4400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 42.06,
|
| 294 |
+
"learning_rate": 0.0002139655172413793,
|
| 295 |
+
"loss": 1.2445,
|
| 296 |
+
"step": 4500
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 42.99,
|
| 300 |
+
"learning_rate": 0.00021051724137931033,
|
| 301 |
+
"loss": 1.2608,
|
| 302 |
+
"step": 4600
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 43.93,
|
| 306 |
+
"learning_rate": 0.00020706896551724136,
|
| 307 |
+
"loss": 1.2272,
|
| 308 |
+
"step": 4700
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 44.86,
|
| 312 |
+
"learning_rate": 0.0002036206896551724,
|
| 313 |
+
"loss": 1.2186,
|
| 314 |
+
"step": 4800
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 45.79,
|
| 318 |
+
"learning_rate": 0.0002001724137931034,
|
| 319 |
+
"loss": 1.2086,
|
| 320 |
+
"step": 4900
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 46.73,
|
| 324 |
+
"learning_rate": 0.00019672413793103444,
|
| 325 |
+
"loss": 1.2042,
|
| 326 |
+
"step": 5000
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 47.66,
|
| 330 |
+
"learning_rate": 0.0001932758620689655,
|
| 331 |
+
"loss": 1.1856,
|
| 332 |
+
"step": 5100
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 48.6,
|
| 336 |
+
"learning_rate": 0.00018982758620689653,
|
| 337 |
+
"loss": 1.1965,
|
| 338 |
+
"step": 5200
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 49.53,
|
| 342 |
+
"learning_rate": 0.00018637931034482758,
|
| 343 |
+
"loss": 1.2104,
|
| 344 |
+
"step": 5300
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 50.47,
|
| 348 |
+
"learning_rate": 0.0001829310344827586,
|
| 349 |
+
"loss": 1.1631,
|
| 350 |
+
"step": 5400
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 51.4,
|
| 354 |
+
"learning_rate": 0.00017948275862068964,
|
| 355 |
+
"loss": 1.1535,
|
| 356 |
+
"step": 5500
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 52.34,
|
| 360 |
+
"learning_rate": 0.0001760344827586207,
|
| 361 |
+
"loss": 1.1615,
|
| 362 |
+
"step": 5600
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 53.27,
|
| 366 |
+
"learning_rate": 0.0001725862068965517,
|
| 367 |
+
"loss": 1.1632,
|
| 368 |
+
"step": 5700
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 54.21,
|
| 372 |
+
"learning_rate": 0.00016913793103448272,
|
| 373 |
+
"loss": 1.1431,
|
| 374 |
+
"step": 5800
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 55.14,
|
| 378 |
+
"learning_rate": 0.00016568965517241378,
|
| 379 |
+
"loss": 1.1245,
|
| 380 |
+
"step": 5900
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 56.07,
|
| 384 |
+
"learning_rate": 0.0001622413793103448,
|
| 385 |
+
"loss": 1.1367,
|
| 386 |
+
"step": 6000
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 56.07,
|
| 390 |
+
"eval_loss": 0.5372028946876526,
|
| 391 |
+
"eval_runtime": 83.9889,
|
| 392 |
+
"eval_samples_per_second": 19.205,
|
| 393 |
+
"eval_steps_per_second": 19.205,
|
| 394 |
+
"eval_wer": 0.46639809558688883,
|
| 395 |
+
"step": 6000
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 57.01,
|
| 399 |
+
"learning_rate": 0.00015879310344827586,
|
| 400 |
+
"loss": 1.1242,
|
| 401 |
+
"step": 6100
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 57.94,
|
| 405 |
+
"learning_rate": 0.0001553448275862069,
|
| 406 |
+
"loss": 1.1016,
|
| 407 |
+
"step": 6200
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 58.88,
|
| 411 |
+
"learning_rate": 0.00015189655172413792,
|
| 412 |
+
"loss": 1.0946,
|
| 413 |
+
"step": 6300
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 59.81,
|
| 417 |
+
"learning_rate": 0.00014844827586206895,
|
| 418 |
+
"loss": 1.0953,
|
| 419 |
+
"step": 6400
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 60.75,
|
| 423 |
+
"learning_rate": 0.000145,
|
| 424 |
+
"loss": 1.1016,
|
| 425 |
+
"step": 6500
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 61.68,
|
| 429 |
+
"learning_rate": 0.00014155172413793103,
|
| 430 |
+
"loss": 1.0745,
|
| 431 |
+
"step": 6600
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 62.62,
|
| 435 |
+
"learning_rate": 0.00013810344827586206,
|
| 436 |
+
"loss": 1.072,
|
| 437 |
+
"step": 6700
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 63.55,
|
| 441 |
+
"learning_rate": 0.00013465517241379309,
|
| 442 |
+
"loss": 1.0621,
|
| 443 |
+
"step": 6800
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 64.49,
|
| 447 |
+
"learning_rate": 0.00013120689655172414,
|
| 448 |
+
"loss": 1.0659,
|
| 449 |
+
"step": 6900
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 65.42,
|
| 453 |
+
"learning_rate": 0.00012775862068965517,
|
| 454 |
+
"loss": 1.0271,
|
| 455 |
+
"step": 7000
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 66.36,
|
| 459 |
+
"learning_rate": 0.0001243103448275862,
|
| 460 |
+
"loss": 1.0381,
|
| 461 |
+
"step": 7100
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 67.29,
|
| 465 |
+
"learning_rate": 0.00012086206896551723,
|
| 466 |
+
"loss": 1.0332,
|
| 467 |
+
"step": 7200
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 68.22,
|
| 471 |
+
"learning_rate": 0.00011741379310344827,
|
| 472 |
+
"loss": 1.0279,
|
| 473 |
+
"step": 7300
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 69.16,
|
| 477 |
+
"learning_rate": 0.00011396551724137931,
|
| 478 |
+
"loss": 1.0174,
|
| 479 |
+
"step": 7400
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 70.09,
|
| 483 |
+
"learning_rate": 0.00011051724137931032,
|
| 484 |
+
"loss": 1.0124,
|
| 485 |
+
"step": 7500
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 71.03,
|
| 489 |
+
"learning_rate": 0.00010706896551724137,
|
| 490 |
+
"loss": 1.0142,
|
| 491 |
+
"step": 7600
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 71.96,
|
| 495 |
+
"learning_rate": 0.00010362068965517241,
|
| 496 |
+
"loss": 0.9809,
|
| 497 |
+
"step": 7700
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 72.9,
|
| 501 |
+
"learning_rate": 0.00010017241379310345,
|
| 502 |
+
"loss": 0.9609,
|
| 503 |
+
"step": 7800
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 73.83,
|
| 507 |
+
"learning_rate": 9.672413793103446e-05,
|
| 508 |
+
"loss": 0.9808,
|
| 509 |
+
"step": 7900
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 74.77,
|
| 513 |
+
"learning_rate": 9.32758620689655e-05,
|
| 514 |
+
"loss": 0.9696,
|
| 515 |
+
"step": 8000
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 74.77,
|
| 519 |
+
"eval_loss": 0.5442632436752319,
|
| 520 |
+
"eval_runtime": 84.0493,
|
| 521 |
+
"eval_samples_per_second": 19.191,
|
| 522 |
+
"eval_steps_per_second": 19.191,
|
| 523 |
+
"eval_wer": 0.4327961911737777,
|
| 524 |
+
"step": 8000
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 75.7,
|
| 528 |
+
"learning_rate": 8.982758620689655e-05,
|
| 529 |
+
"loss": 0.9559,
|
| 530 |
+
"step": 8100
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 76.64,
|
| 534 |
+
"learning_rate": 8.637931034482759e-05,
|
| 535 |
+
"loss": 0.9507,
|
| 536 |
+
"step": 8200
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 77.57,
|
| 540 |
+
"learning_rate": 8.29310344827586e-05,
|
| 541 |
+
"loss": 0.9358,
|
| 542 |
+
"step": 8300
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 78.5,
|
| 546 |
+
"learning_rate": 7.948275862068965e-05,
|
| 547 |
+
"loss": 0.9367,
|
| 548 |
+
"step": 8400
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 79.44,
|
| 552 |
+
"learning_rate": 7.603448275862069e-05,
|
| 553 |
+
"loss": 0.9223,
|
| 554 |
+
"step": 8500
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 80.37,
|
| 558 |
+
"learning_rate": 7.258620689655172e-05,
|
| 559 |
+
"loss": 0.9258,
|
| 560 |
+
"step": 8600
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 81.31,
|
| 564 |
+
"learning_rate": 6.913793103448276e-05,
|
| 565 |
+
"loss": 0.8865,
|
| 566 |
+
"step": 8700
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 82.24,
|
| 570 |
+
"learning_rate": 6.568965517241379e-05,
|
| 571 |
+
"loss": 0.9062,
|
| 572 |
+
"step": 8800
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 83.18,
|
| 576 |
+
"learning_rate": 6.224137931034483e-05,
|
| 577 |
+
"loss": 0.88,
|
| 578 |
+
"step": 8900
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 84.11,
|
| 582 |
+
"learning_rate": 5.8793103448275856e-05,
|
| 583 |
+
"loss": 0.8794,
|
| 584 |
+
"step": 9000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 85.05,
|
| 588 |
+
"learning_rate": 5.534482758620689e-05,
|
| 589 |
+
"loss": 0.8892,
|
| 590 |
+
"step": 9100
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 85.98,
|
| 594 |
+
"learning_rate": 5.1896551724137926e-05,
|
| 595 |
+
"loss": 0.8746,
|
| 596 |
+
"step": 9200
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 86.92,
|
| 600 |
+
"learning_rate": 4.844827586206896e-05,
|
| 601 |
+
"loss": 0.8465,
|
| 602 |
+
"step": 9300
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 87.85,
|
| 606 |
+
"learning_rate": 4.5034482758620686e-05,
|
| 607 |
+
"loss": 0.8453,
|
| 608 |
+
"step": 9400
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 88.79,
|
| 612 |
+
"learning_rate": 4.162068965517241e-05,
|
| 613 |
+
"loss": 0.8472,
|
| 614 |
+
"step": 9500
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 89.72,
|
| 618 |
+
"learning_rate": 3.817241379310344e-05,
|
| 619 |
+
"loss": 0.8638,
|
| 620 |
+
"step": 9600
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 90.65,
|
| 624 |
+
"learning_rate": 3.4724137931034475e-05,
|
| 625 |
+
"loss": 0.8354,
|
| 626 |
+
"step": 9700
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 91.59,
|
| 630 |
+
"learning_rate": 3.127586206896551e-05,
|
| 631 |
+
"loss": 0.825,
|
| 632 |
+
"step": 9800
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 92.52,
|
| 636 |
+
"learning_rate": 2.7827586206896548e-05,
|
| 637 |
+
"loss": 0.8395,
|
| 638 |
+
"step": 9900
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 93.46,
|
| 642 |
+
"learning_rate": 2.4379310344827583e-05,
|
| 643 |
+
"loss": 0.8163,
|
| 644 |
+
"step": 10000
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 93.46,
|
| 648 |
+
"eval_loss": 0.5916153788566589,
|
| 649 |
+
"eval_runtime": 83.5748,
|
| 650 |
+
"eval_samples_per_second": 19.3,
|
| 651 |
+
"eval_steps_per_second": 19.3,
|
| 652 |
+
"eval_wer": 0.4123786852224867,
|
| 653 |
+
"step": 10000
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 94.39,
|
| 657 |
+
"learning_rate": 2.093103448275862e-05,
|
| 658 |
+
"loss": 0.8116,
|
| 659 |
+
"step": 10100
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 95.33,
|
| 663 |
+
"learning_rate": 1.7482758620689653e-05,
|
| 664 |
+
"loss": 0.8103,
|
| 665 |
+
"step": 10200
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 96.26,
|
| 669 |
+
"learning_rate": 1.4034482758620688e-05,
|
| 670 |
+
"loss": 0.8252,
|
| 671 |
+
"step": 10300
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 97.2,
|
| 675 |
+
"learning_rate": 1.0586206896551723e-05,
|
| 676 |
+
"loss": 0.805,
|
| 677 |
+
"step": 10400
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 98.13,
|
| 681 |
+
"learning_rate": 7.1379310344827575e-06,
|
| 682 |
+
"loss": 0.7983,
|
| 683 |
+
"step": 10500
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 99.07,
|
| 687 |
+
"learning_rate": 3.689655172413793e-06,
|
| 688 |
+
"loss": 0.7991,
|
| 689 |
+
"step": 10600
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 100.0,
|
| 693 |
+
"learning_rate": 2.413793103448276e-07,
|
| 694 |
+
"loss": 0.8191,
|
| 695 |
+
"step": 10700
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 100.0,
|
| 699 |
+
"step": 10700,
|
| 700 |
+
"total_flos": 4.070849231885728e+19,
|
| 701 |
+
"train_loss": 1.4013342235244324,
|
| 702 |
+
"train_runtime": 14378.2108,
|
| 703 |
+
"train_samples_per_second": 23.647,
|
| 704 |
+
"train_steps_per_second": 0.744
|
| 705 |
+
}
|
| 706 |
+
],
|
| 707 |
+
"max_steps": 10700,
|
| 708 |
+
"num_train_epochs": 100,
|
| 709 |
+
"total_flos": 4.070849231885728e+19,
|
| 710 |
+
"trial_name": null,
|
| 711 |
+
"trial_params": null
|
| 712 |
+
}
|