Upload trainer_state.json
Browse files- trainer_state.json +1136 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,1136 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.05459373816847801,
|
| 3 |
+
"best_model_checkpoint": "/scratch/lingjzhu_root/lingjzhu1/lingjzhu/g2p/byt5_16_layers_baseline/checkpoint-140000",
|
| 4 |
+
"epoch": 9.917791237204689,
|
| 5 |
+
"global_step": 140000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.07,
|
| 12 |
+
"learning_rate": 0.0003,
|
| 13 |
+
"loss": 7.0422,
|
| 14 |
+
"step": 1000
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.14,
|
| 18 |
+
"learning_rate": 0.0002999623214094654,
|
| 19 |
+
"loss": 1.3957,
|
| 20 |
+
"step": 2000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.21,
|
| 24 |
+
"learning_rate": 0.0002998493045668776,
|
| 25 |
+
"loss": 0.4792,
|
| 26 |
+
"step": 3000
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.28,
|
| 30 |
+
"learning_rate": 0.00029966100624977436,
|
| 31 |
+
"loss": 0.304,
|
| 32 |
+
"step": 4000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.35,
|
| 36 |
+
"learning_rate": 0.00029939752105569156,
|
| 37 |
+
"loss": 0.2449,
|
| 38 |
+
"step": 5000
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.35,
|
| 42 |
+
"eval_cer": 0.21979775329351647,
|
| 43 |
+
"eval_loss": 0.2839755415916443,
|
| 44 |
+
"eval_runtime": 80.4341,
|
| 45 |
+
"eval_samples_per_second": 62.163,
|
| 46 |
+
"eval_steps_per_second": 0.124,
|
| 47 |
+
"eval_wer": 0.5296,
|
| 48 |
+
"step": 5000
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"epoch": 0.43,
|
| 52 |
+
"learning_rate": 0.00029905898135463893,
|
| 53 |
+
"loss": 0.2141,
|
| 54 |
+
"step": 6000
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.5,
|
| 58 |
+
"learning_rate": 0.0002986455572226002,
|
| 59 |
+
"loss": 0.1944,
|
| 60 |
+
"step": 7000
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.57,
|
| 64 |
+
"learning_rate": 0.00029815745635609,
|
| 65 |
+
"loss": 0.1791,
|
| 66 |
+
"step": 8000
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.64,
|
| 70 |
+
"learning_rate": 0.00029759492396781067,
|
| 71 |
+
"loss": 0.1669,
|
| 72 |
+
"step": 9000
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.71,
|
| 76 |
+
"learning_rate": 0.00029695824266346256,
|
| 77 |
+
"loss": 0.1574,
|
| 78 |
+
"step": 10000
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 0.71,
|
| 82 |
+
"eval_cer": 0.16565687056642014,
|
| 83 |
+
"eval_loss": 0.1845117211341858,
|
| 84 |
+
"eval_runtime": 79.5791,
|
| 85 |
+
"eval_samples_per_second": 62.831,
|
| 86 |
+
"eval_steps_per_second": 0.126,
|
| 87 |
+
"eval_wer": 0.4346,
|
| 88 |
+
"step": 10000
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 0.78,
|
| 92 |
+
"learning_rate": 0.000296247732299768,
|
| 93 |
+
"loss": 0.1486,
|
| 94 |
+
"step": 11000
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.85,
|
| 98 |
+
"learning_rate": 0.000295463749823781,
|
| 99 |
+
"loss": 0.1411,
|
| 100 |
+
"step": 12000
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.92,
|
| 104 |
+
"learning_rate": 0.0002946066890935644,
|
| 105 |
+
"loss": 0.1339,
|
| 106 |
+
"step": 13000
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 0.99,
|
| 110 |
+
"learning_rate": 0.0002936769806803222,
|
| 111 |
+
"loss": 0.1287,
|
| 112 |
+
"step": 14000
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 1.06,
|
| 116 |
+
"learning_rate": 0.00029267509165208945,
|
| 117 |
+
"loss": 0.1215,
|
| 118 |
+
"step": 15000
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 1.06,
|
| 122 |
+
"eval_cer": 0.14532439250522866,
|
| 123 |
+
"eval_loss": 0.1390569657087326,
|
| 124 |
+
"eval_runtime": 79.7847,
|
| 125 |
+
"eval_samples_per_second": 62.669,
|
| 126 |
+
"eval_steps_per_second": 0.125,
|
| 127 |
+
"eval_wer": 0.3858,
|
| 128 |
+
"step": 15000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 1.13,
|
| 132 |
+
"learning_rate": 0.00029160152533908566,
|
| 133 |
+
"loss": 0.1181,
|
| 134 |
+
"step": 16000
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 1.2,
|
| 138 |
+
"learning_rate": 0.0002904568210808509,
|
| 139 |
+
"loss": 0.1137,
|
| 140 |
+
"step": 17000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 1.28,
|
| 144 |
+
"learning_rate": 0.00028924155395529237,
|
| 145 |
+
"loss": 0.11,
|
| 146 |
+
"step": 18000
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 1.35,
|
| 150 |
+
"learning_rate": 0.00028795633448977547,
|
| 151 |
+
"loss": 0.1067,
|
| 152 |
+
"step": 19000
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 1.42,
|
| 156 |
+
"learning_rate": 0.0002866018083544069,
|
| 157 |
+
"loss": 0.1042,
|
| 158 |
+
"step": 20000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.42,
|
| 162 |
+
"eval_cer": 0.1343495978783501,
|
| 163 |
+
"eval_loss": 0.11732806265354156,
|
| 164 |
+
"eval_runtime": 79.7707,
|
| 165 |
+
"eval_samples_per_second": 62.68,
|
| 166 |
+
"eval_steps_per_second": 0.125,
|
| 167 |
+
"eval_wer": 0.3624,
|
| 168 |
+
"step": 20000
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 1.49,
|
| 172 |
+
"learning_rate": 0.0002851786560376614,
|
| 173 |
+
"loss": 0.101,
|
| 174 |
+
"step": 21000
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 1.56,
|
| 178 |
+
"learning_rate": 0.00028368759250451786,
|
| 179 |
+
"loss": 0.0983,
|
| 180 |
+
"step": 22000
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 1.63,
|
| 184 |
+
"learning_rate": 0.00028212936683727404,
|
| 185 |
+
"loss": 0.0954,
|
| 186 |
+
"step": 23000
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.7,
|
| 190 |
+
"learning_rate": 0.0002805047618592216,
|
| 191 |
+
"loss": 0.0933,
|
| 192 |
+
"step": 24000
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 1.77,
|
| 196 |
+
"learning_rate": 0.0002788145937413704,
|
| 197 |
+
"loss": 0.0917,
|
| 198 |
+
"step": 25000
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 1.77,
|
| 202 |
+
"eval_cer": 0.1251644063045775,
|
| 203 |
+
"eval_loss": 0.10654760897159576,
|
| 204 |
+
"eval_runtime": 79.786,
|
| 205 |
+
"eval_samples_per_second": 62.668,
|
| 206 |
+
"eval_steps_per_second": 0.125,
|
| 207 |
+
"eval_wer": 0.3488,
|
| 208 |
+
"step": 25000
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 1.84,
|
| 212 |
+
"learning_rate": 0.00027705971159241994,
|
| 213 |
+
"loss": 0.0898,
|
| 214 |
+
"step": 26000
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 1.91,
|
| 218 |
+
"learning_rate": 0.0002752409970321824,
|
| 219 |
+
"loss": 0.0881,
|
| 220 |
+
"step": 27000
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 1.98,
|
| 224 |
+
"learning_rate": 0.000273359363748674,
|
| 225 |
+
"loss": 0.0865,
|
| 226 |
+
"step": 28000
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 2.05,
|
| 230 |
+
"learning_rate": 0.0002714157570390952,
|
| 231 |
+
"loss": 0.0834,
|
| 232 |
+
"step": 29000
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 2.13,
|
| 236 |
+
"learning_rate": 0.0002694111533349308,
|
| 237 |
+
"loss": 0.0821,
|
| 238 |
+
"step": 30000
|
| 239 |
+
},
|
| 240 |
+
{
|
| 241 |
+
"epoch": 2.13,
|
| 242 |
+
"eval_cer": 0.11837253929580198,
|
| 243 |
+
"eval_loss": 0.09495185315608978,
|
| 244 |
+
"eval_runtime": 79.9216,
|
| 245 |
+
"eval_samples_per_second": 62.561,
|
| 246 |
+
"eval_steps_per_second": 0.125,
|
| 247 |
+
"eval_wer": 0.3354,
|
| 248 |
+
"step": 30000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 2.2,
|
| 252 |
+
"learning_rate": 0.00026734655971140956,
|
| 253 |
+
"loss": 0.0806,
|
| 254 |
+
"step": 31000
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 2.27,
|
| 258 |
+
"learning_rate": 0.00026522301338156835,
|
| 259 |
+
"loss": 0.0791,
|
| 260 |
+
"step": 32000
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 2.34,
|
| 264 |
+
"learning_rate": 0.00026304158117517563,
|
| 265 |
+
"loss": 0.0785,
|
| 266 |
+
"step": 33000
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 2.41,
|
| 270 |
+
"learning_rate": 0.0002608033590027766,
|
| 271 |
+
"loss": 0.0773,
|
| 272 |
+
"step": 34000
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 2.48,
|
| 276 |
+
"learning_rate": 0.000258509471305128,
|
| 277 |
+
"loss": 0.0762,
|
| 278 |
+
"step": 35000
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 2.48,
|
| 282 |
+
"eval_cer": 0.1154617391491839,
|
| 283 |
+
"eval_loss": 0.08647703379392624,
|
| 284 |
+
"eval_runtime": 79.5904,
|
| 285 |
+
"eval_samples_per_second": 62.822,
|
| 286 |
+
"eval_steps_per_second": 0.126,
|
| 287 |
+
"eval_wer": 0.3272,
|
| 288 |
+
"step": 35000
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 2.55,
|
| 292 |
+
"learning_rate": 0.00025616107048830045,
|
| 293 |
+
"loss": 0.0747,
|
| 294 |
+
"step": 36000
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 2.62,
|
| 298 |
+
"learning_rate": 0.0002537593363447311,
|
| 299 |
+
"loss": 0.0741,
|
| 300 |
+
"step": 37000
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 2.69,
|
| 304 |
+
"learning_rate": 0.0002513054754605181,
|
| 305 |
+
"loss": 0.0734,
|
| 306 |
+
"step": 38000
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 2.76,
|
| 310 |
+
"learning_rate": 0.0002488007206092547,
|
| 311 |
+
"loss": 0.0724,
|
| 312 |
+
"step": 39000
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 2.83,
|
| 316 |
+
"learning_rate": 0.00024624633013270664,
|
| 317 |
+
"loss": 0.0715,
|
| 318 |
+
"step": 40000
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 2.83,
|
| 322 |
+
"eval_cer": 0.11341339830526748,
|
| 323 |
+
"eval_loss": 0.08288487792015076,
|
| 324 |
+
"eval_runtime": 79.5687,
|
| 325 |
+
"eval_samples_per_second": 62.839,
|
| 326 |
+
"eval_steps_per_second": 0.126,
|
| 327 |
+
"eval_wer": 0.3176,
|
| 328 |
+
"step": 40000
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 2.9,
|
| 332 |
+
"learning_rate": 0.0002436435873086449,
|
| 333 |
+
"loss": 0.0708,
|
| 334 |
+
"step": 41000
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 2.98,
|
| 338 |
+
"learning_rate": 0.00024099379970615137,
|
| 339 |
+
"loss": 0.0695,
|
| 340 |
+
"step": 42000
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 3.05,
|
| 344 |
+
"learning_rate": 0.00023829829852872037,
|
| 345 |
+
"loss": 0.0677,
|
| 346 |
+
"step": 43000
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 3.12,
|
| 350 |
+
"learning_rate": 0.00023555843794548724,
|
| 351 |
+
"loss": 0.0666,
|
| 352 |
+
"step": 44000
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 3.19,
|
| 356 |
+
"learning_rate": 0.00023277559441091917,
|
| 357 |
+
"loss": 0.0661,
|
| 358 |
+
"step": 45000
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 3.19,
|
| 362 |
+
"eval_cer": 0.11099851225770284,
|
| 363 |
+
"eval_loss": 0.07811947166919708,
|
| 364 |
+
"eval_runtime": 79.5629,
|
| 365 |
+
"eval_samples_per_second": 62.843,
|
| 366 |
+
"eval_steps_per_second": 0.126,
|
| 367 |
+
"eval_wer": 0.316,
|
| 368 |
+
"step": 45000
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 3.26,
|
| 372 |
+
"learning_rate": 0.0002299511659733103,
|
| 373 |
+
"loss": 0.0655,
|
| 374 |
+
"step": 46000
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 3.33,
|
| 378 |
+
"learning_rate": 0.0002270865715724285,
|
| 379 |
+
"loss": 0.0651,
|
| 380 |
+
"step": 47000
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 3.4,
|
| 384 |
+
"learning_rate": 0.0002241832503266669,
|
| 385 |
+
"loss": 0.0645,
|
| 386 |
+
"step": 48000
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 3.47,
|
| 390 |
+
"learning_rate": 0.00022124266081005755,
|
| 391 |
+
"loss": 0.0641,
|
| 392 |
+
"step": 49000
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 3.54,
|
| 396 |
+
"learning_rate": 0.0002182662803195115,
|
| 397 |
+
"loss": 0.0636,
|
| 398 |
+
"step": 50000
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 3.54,
|
| 402 |
+
"eval_cer": 0.10931671661743461,
|
| 403 |
+
"eval_loss": 0.07468874752521515,
|
| 404 |
+
"eval_runtime": 80.209,
|
| 405 |
+
"eval_samples_per_second": 62.337,
|
| 406 |
+
"eval_steps_per_second": 0.125,
|
| 407 |
+
"eval_wer": 0.3074,
|
| 408 |
+
"step": 50000
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 3.61,
|
| 412 |
+
"learning_rate": 0.00021525560413265246,
|
| 413 |
+
"loss": 0.0629,
|
| 414 |
+
"step": 51000
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 3.68,
|
| 418 |
+
"learning_rate": 0.00021221214475661745,
|
| 419 |
+
"loss": 0.0619,
|
| 420 |
+
"step": 52000
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 3.75,
|
| 424 |
+
"learning_rate": 0.00020913743116820165,
|
| 425 |
+
"loss": 0.0619,
|
| 426 |
+
"step": 53000
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 3.83,
|
| 430 |
+
"learning_rate": 0.00020603300804572917,
|
| 431 |
+
"loss": 0.0608,
|
| 432 |
+
"step": 54000
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 3.9,
|
| 436 |
+
"learning_rate": 0.00020290043499303573,
|
| 437 |
+
"loss": 0.061,
|
| 438 |
+
"step": 55000
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 3.9,
|
| 442 |
+
"eval_cer": 0.11015761443756873,
|
| 443 |
+
"eval_loss": 0.07217952609062195,
|
| 444 |
+
"eval_runtime": 79.9934,
|
| 445 |
+
"eval_samples_per_second": 62.505,
|
| 446 |
+
"eval_steps_per_second": 0.125,
|
| 447 |
+
"eval_wer": 0.3086,
|
| 448 |
+
"step": 55000
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 3.97,
|
| 452 |
+
"learning_rate": 0.00019974128575595291,
|
| 453 |
+
"loss": 0.06,
|
| 454 |
+
"step": 56000
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 4.04,
|
| 458 |
+
"learning_rate": 0.000196557147431688,
|
| 459 |
+
"loss": 0.0585,
|
| 460 |
+
"step": 57000
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 4.11,
|
| 464 |
+
"learning_rate": 0.00019334961967149597,
|
| 465 |
+
"loss": 0.0577,
|
| 466 |
+
"step": 58000
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 4.18,
|
| 470 |
+
"learning_rate": 0.00019012031387704483,
|
| 471 |
+
"loss": 0.0574,
|
| 472 |
+
"step": 59000
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 4.25,
|
| 476 |
+
"learning_rate": 0.00018687085239087786,
|
| 477 |
+
"loss": 0.0569,
|
| 478 |
+
"step": 60000
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 4.25,
|
| 482 |
+
"eval_cer": 0.10580219495892537,
|
| 483 |
+
"eval_loss": 0.06902794539928436,
|
| 484 |
+
"eval_runtime": 79.9992,
|
| 485 |
+
"eval_samples_per_second": 62.501,
|
| 486 |
+
"eval_steps_per_second": 0.125,
|
| 487 |
+
"eval_wer": 0.2988,
|
| 488 |
+
"step": 60000
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 4.32,
|
| 492 |
+
"learning_rate": 0.00018360286768137894,
|
| 493 |
+
"loss": 0.0567,
|
| 494 |
+
"step": 61000
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 4.39,
|
| 498 |
+
"learning_rate": 0.00018031800152265137,
|
| 499 |
+
"loss": 0.0561,
|
| 500 |
+
"step": 62000
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 4.46,
|
| 504 |
+
"learning_rate": 0.00017701790416972107,
|
| 505 |
+
"loss": 0.0558,
|
| 506 |
+
"step": 63000
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 4.53,
|
| 510 |
+
"learning_rate": 0.00017370423352948002,
|
| 511 |
+
"loss": 0.0557,
|
| 512 |
+
"step": 64000
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 4.6,
|
| 516 |
+
"learning_rate": 0.00017037865432778442,
|
| 517 |
+
"loss": 0.0552,
|
| 518 |
+
"step": 65000
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 4.6,
|
| 522 |
+
"eval_cer": 0.10578063347635784,
|
| 523 |
+
"eval_loss": 0.06745224446058273,
|
| 524 |
+
"eval_runtime": 79.5425,
|
| 525 |
+
"eval_samples_per_second": 62.859,
|
| 526 |
+
"eval_steps_per_second": 0.126,
|
| 527 |
+
"eval_wer": 0.2978,
|
| 528 |
+
"step": 65000
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 4.68,
|
| 532 |
+
"learning_rate": 0.00016704283727312792,
|
| 533 |
+
"loss": 0.0546,
|
| 534 |
+
"step": 66000
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 4.75,
|
| 538 |
+
"learning_rate": 0.00016369845821730933,
|
| 539 |
+
"loss": 0.0544,
|
| 540 |
+
"step": 67000
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 4.82,
|
| 544 |
+
"learning_rate": 0.00016034719731351576,
|
| 545 |
+
"loss": 0.0541,
|
| 546 |
+
"step": 68000
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 4.89,
|
| 550 |
+
"learning_rate": 0.00015699073817224542,
|
| 551 |
+
"loss": 0.0537,
|
| 552 |
+
"step": 69000
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 4.96,
|
| 556 |
+
"learning_rate": 0.00015363076701549341,
|
| 557 |
+
"loss": 0.0531,
|
| 558 |
+
"step": 70000
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 4.96,
|
| 562 |
+
"eval_cer": 0.10444382155717027,
|
| 563 |
+
"eval_loss": 0.06512928009033203,
|
| 564 |
+
"eval_runtime": 79.8169,
|
| 565 |
+
"eval_samples_per_second": 62.643,
|
| 566 |
+
"eval_steps_per_second": 0.125,
|
| 567 |
+
"eval_wer": 0.2962,
|
| 568 |
+
"step": 70000
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 5.03,
|
| 572 |
+
"learning_rate": 0.00015026897182962535,
|
| 573 |
+
"loss": 0.0525,
|
| 574 |
+
"step": 71000
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 5.1,
|
| 578 |
+
"learning_rate": 0.00014690704151736484,
|
| 579 |
+
"loss": 0.0509,
|
| 580 |
+
"step": 72000
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 5.17,
|
| 584 |
+
"learning_rate": 0.0001435466650493204,
|
| 585 |
+
"loss": 0.0509,
|
| 586 |
+
"step": 73000
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 5.24,
|
| 590 |
+
"learning_rate": 0.0001401895306154785,
|
| 591 |
+
"loss": 0.0504,
|
| 592 |
+
"step": 74000
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 5.31,
|
| 596 |
+
"learning_rate": 0.0001368373247770885,
|
| 597 |
+
"loss": 0.0504,
|
| 598 |
+
"step": 75000
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 5.31,
|
| 602 |
+
"eval_cer": 0.10291295629487483,
|
| 603 |
+
"eval_loss": 0.06379809230566025,
|
| 604 |
+
"eval_runtime": 79.7333,
|
| 605 |
+
"eval_samples_per_second": 62.709,
|
| 606 |
+
"eval_steps_per_second": 0.125,
|
| 607 |
+
"eval_wer": 0.2864,
|
| 608 |
+
"step": 75000
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 5.38,
|
| 612 |
+
"learning_rate": 0.00013349173161936601,
|
| 613 |
+
"loss": 0.0502,
|
| 614 |
+
"step": 76000
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 5.45,
|
| 618 |
+
"learning_rate": 0.00013015443190544018,
|
| 619 |
+
"loss": 0.0495,
|
| 620 |
+
"step": 77000
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 5.53,
|
| 624 |
+
"learning_rate": 0.00012682710223196978,
|
| 625 |
+
"loss": 0.0494,
|
| 626 |
+
"step": 78000
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 5.6,
|
| 630 |
+
"learning_rate": 0.00012351141418685273,
|
| 631 |
+
"loss": 0.0496,
|
| 632 |
+
"step": 79000
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 5.67,
|
| 636 |
+
"learning_rate": 0.00012020903350945151,
|
| 637 |
+
"loss": 0.0492,
|
| 638 |
+
"step": 80000
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 5.67,
|
| 642 |
+
"eval_cer": 0.10310700963798271,
|
| 643 |
+
"eval_loss": 0.06157052516937256,
|
| 644 |
+
"eval_runtime": 79.8007,
|
| 645 |
+
"eval_samples_per_second": 62.656,
|
| 646 |
+
"eval_steps_per_second": 0.125,
|
| 647 |
+
"eval_wer": 0.2888,
|
| 648 |
+
"step": 80000
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 5.74,
|
| 652 |
+
"learning_rate": 0.00011692161925375738,
|
| 653 |
+
"loss": 0.0485,
|
| 654 |
+
"step": 81000
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 5.81,
|
| 658 |
+
"learning_rate": 0.00011365082295491233,
|
| 659 |
+
"loss": 0.0484,
|
| 660 |
+
"step": 82000
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 5.88,
|
| 664 |
+
"learning_rate": 0.00011039828779950928,
|
| 665 |
+
"loss": 0.0481,
|
| 666 |
+
"step": 83000
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 5.95,
|
| 670 |
+
"learning_rate": 0.0001071656478000858,
|
| 671 |
+
"loss": 0.048,
|
| 672 |
+
"step": 84000
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 6.02,
|
| 676 |
+
"learning_rate": 0.00010395452697422703,
|
| 677 |
+
"loss": 0.0474,
|
| 678 |
+
"step": 85000
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 6.02,
|
| 682 |
+
"eval_cer": 0.10228767330041613,
|
| 683 |
+
"eval_loss": 0.06011258438229561,
|
| 684 |
+
"eval_runtime": 80.0834,
|
| 685 |
+
"eval_samples_per_second": 62.435,
|
| 686 |
+
"eval_steps_per_second": 0.125,
|
| 687 |
+
"eval_wer": 0.2868,
|
| 688 |
+
"step": 85000
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 6.09,
|
| 692 |
+
"learning_rate": 0.00010076653852868962,
|
| 693 |
+
"loss": 0.0459,
|
| 694 |
+
"step": 86000
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 6.16,
|
| 698 |
+
"learning_rate": 9.760328404895728e-05,
|
| 699 |
+
"loss": 0.0457,
|
| 700 |
+
"step": 87000
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 6.23,
|
| 704 |
+
"learning_rate": 9.446635269463389e-05,
|
| 705 |
+
"loss": 0.0453,
|
| 706 |
+
"step": 88000
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 6.3,
|
| 710 |
+
"learning_rate": 9.135732040107998e-05,
|
| 711 |
+
"loss": 0.0456,
|
| 712 |
+
"step": 89000
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 6.38,
|
| 716 |
+
"learning_rate": 8.82777490876921e-05,
|
| 717 |
+
"loss": 0.0453,
|
| 718 |
+
"step": 90000
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 6.38,
|
| 722 |
+
"eval_cer": 0.1022661118178486,
|
| 723 |
+
"eval_loss": 0.05900218337774277,
|
| 724 |
+
"eval_runtime": 79.8035,
|
| 725 |
+
"eval_samples_per_second": 62.654,
|
| 726 |
+
"eval_steps_per_second": 0.125,
|
| 727 |
+
"eval_wer": 0.284,
|
| 728 |
+
"step": 90000
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 6.45,
|
| 732 |
+
"learning_rate": 8.522918587322418e-05,
|
| 733 |
+
"loss": 0.0452,
|
| 734 |
+
"step": 91000
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 6.52,
|
| 738 |
+
"learning_rate": 8.22131622985438e-05,
|
| 739 |
+
"loss": 0.0451,
|
| 740 |
+
"step": 92000
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 6.59,
|
| 744 |
+
"learning_rate": 7.923119355721526e-05,
|
| 745 |
+
"loss": 0.045,
|
| 746 |
+
"step": 93000
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 6.66,
|
| 750 |
+
"learning_rate": 7.628477773429438e-05,
|
| 751 |
+
"loss": 0.0448,
|
| 752 |
+
"step": 94000
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 6.73,
|
| 756 |
+
"learning_rate": 7.3375395053719e-05,
|
| 757 |
+
"loss": 0.0446,
|
| 758 |
+
"step": 95000
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 6.73,
|
| 762 |
+
"eval_cer": 0.10045494728217512,
|
| 763 |
+
"eval_loss": 0.05785335227847099,
|
| 764 |
+
"eval_runtime": 79.8939,
|
| 765 |
+
"eval_samples_per_second": 62.583,
|
| 766 |
+
"eval_steps_per_second": 0.125,
|
| 767 |
+
"eval_wer": 0.2766,
|
| 768 |
+
"step": 95000
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 6.8,
|
| 772 |
+
"learning_rate": 7.050450713467216e-05,
|
| 773 |
+
"loss": 0.0439,
|
| 774 |
+
"step": 96000
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 6.87,
|
| 778 |
+
"learning_rate": 6.767355625729215e-05,
|
| 779 |
+
"loss": 0.0439,
|
| 780 |
+
"step": 97000
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 6.94,
|
| 784 |
+
"learning_rate": 6.488396463809805e-05,
|
| 785 |
+
"loss": 0.0438,
|
| 786 |
+
"step": 98000
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 7.01,
|
| 790 |
+
"learning_rate": 6.213713371549497e-05,
|
| 791 |
+
"loss": 0.0434,
|
| 792 |
+
"step": 99000
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 7.08,
|
| 796 |
+
"learning_rate": 5.943444344571755e-05,
|
| 797 |
+
"loss": 0.042,
|
| 798 |
+
"step": 100000
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 7.08,
|
| 802 |
+
"eval_cer": 0.09978654132258134,
|
| 803 |
+
"eval_loss": 0.05739114433526993,
|
| 804 |
+
"eval_runtime": 79.8794,
|
| 805 |
+
"eval_samples_per_second": 62.594,
|
| 806 |
+
"eval_steps_per_second": 0.125,
|
| 807 |
+
"eval_wer": 0.277,
|
| 808 |
+
"step": 100000
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 7.15,
|
| 812 |
+
"learning_rate": 5.6777251609566e-05,
|
| 813 |
+
"loss": 0.042,
|
| 814 |
+
"step": 101000
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 7.23,
|
| 818 |
+
"learning_rate": 5.416689313028253e-05,
|
| 819 |
+
"loss": 0.0421,
|
| 820 |
+
"step": 102000
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 7.3,
|
| 824 |
+
"learning_rate": 5.1604679402911015e-05,
|
| 825 |
+
"loss": 0.0419,
|
| 826 |
+
"step": 103000
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 7.37,
|
| 830 |
+
"learning_rate": 4.9091897635476726e-05,
|
| 831 |
+
"loss": 0.0418,
|
| 832 |
+
"step": 104000
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 7.44,
|
| 836 |
+
"learning_rate": 4.6629810202317166e-05,
|
| 837 |
+
"loss": 0.0417,
|
| 838 |
+
"step": 105000
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 7.44,
|
| 842 |
+
"eval_cer": 0.09941999611893314,
|
| 843 |
+
"eval_loss": 0.056080542504787445,
|
| 844 |
+
"eval_runtime": 79.81,
|
| 845 |
+
"eval_samples_per_second": 62.649,
|
| 846 |
+
"eval_steps_per_second": 0.125,
|
| 847 |
+
"eval_wer": 0.2742,
|
| 848 |
+
"step": 105000
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 7.51,
|
| 852 |
+
"learning_rate": 4.421965400988918e-05,
|
| 853 |
+
"loss": 0.0416,
|
| 854 |
+
"step": 106000
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 7.58,
|
| 858 |
+
"learning_rate": 4.1862639875370016e-05,
|
| 859 |
+
"loss": 0.0417,
|
| 860 |
+
"step": 107000
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 7.65,
|
| 864 |
+
"learning_rate": 3.955995191836581e-05,
|
| 865 |
+
"loss": 0.0414,
|
| 866 |
+
"step": 108000
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 7.72,
|
| 870 |
+
"learning_rate": 3.731274696603201e-05,
|
| 871 |
+
"loss": 0.0414,
|
| 872 |
+
"step": 109000
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 7.79,
|
| 876 |
+
"learning_rate": 3.512215397190527e-05,
|
| 877 |
+
"loss": 0.0411,
|
| 878 |
+
"step": 110000
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 7.79,
|
| 882 |
+
"eval_cer": 0.1000237176308243,
|
| 883 |
+
"eval_loss": 0.0557967908680439,
|
| 884 |
+
"eval_runtime": 79.8362,
|
| 885 |
+
"eval_samples_per_second": 62.628,
|
| 886 |
+
"eval_steps_per_second": 0.125,
|
| 887 |
+
"eval_wer": 0.2756,
|
| 888 |
+
"step": 110000
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 7.86,
|
| 892 |
+
"learning_rate": 3.298927344873821e-05,
|
| 893 |
+
"loss": 0.0407,
|
| 894 |
+
"step": 111000
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 7.93,
|
| 898 |
+
"learning_rate": 3.091517691562284e-05,
|
| 899 |
+
"loss": 0.0407,
|
| 900 |
+
"step": 112000
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 8.01,
|
| 904 |
+
"learning_rate": 2.890090635967912e-05,
|
| 905 |
+
"loss": 0.0412,
|
| 906 |
+
"step": 113000
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 8.08,
|
| 910 |
+
"learning_rate": 2.694747371258042e-05,
|
| 911 |
+
"loss": 0.0398,
|
| 912 |
+
"step": 114000
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 8.15,
|
| 916 |
+
"learning_rate": 2.505586034217802e-05,
|
| 917 |
+
"loss": 0.0394,
|
| 918 |
+
"step": 115000
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 8.15,
|
| 922 |
+
"eval_cer": 0.09991591021798658,
|
| 923 |
+
"eval_loss": 0.05538313463330269,
|
| 924 |
+
"eval_runtime": 79.9108,
|
| 925 |
+
"eval_samples_per_second": 62.57,
|
| 926 |
+
"eval_steps_per_second": 0.125,
|
| 927 |
+
"eval_wer": 0.2758,
|
| 928 |
+
"step": 115000
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 8.22,
|
| 932 |
+
"learning_rate": 2.3227016559480356e-05,
|
| 933 |
+
"loss": 0.0396,
|
| 934 |
+
"step": 116000
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 8.29,
|
| 938 |
+
"learning_rate": 2.1461861141234628e-05,
|
| 939 |
+
"loss": 0.0396,
|
| 940 |
+
"step": 117000
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 8.36,
|
| 944 |
+
"learning_rate": 1.9761280868350654e-05,
|
| 945 |
+
"loss": 0.0397,
|
| 946 |
+
"step": 118000
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 8.43,
|
| 950 |
+
"learning_rate": 1.8126130080398725e-05,
|
| 951 |
+
"loss": 0.0395,
|
| 952 |
+
"step": 119000
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 8.5,
|
| 956 |
+
"learning_rate": 1.655723024640553e-05,
|
| 957 |
+
"loss": 0.0394,
|
| 958 |
+
"step": 120000
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 8.5,
|
| 962 |
+
"eval_cer": 0.09993747170055413,
|
| 963 |
+
"eval_loss": 0.05500805750489235,
|
| 964 |
+
"eval_runtime": 79.8857,
|
| 965 |
+
"eval_samples_per_second": 62.589,
|
| 966 |
+
"eval_steps_per_second": 0.125,
|
| 967 |
+
"eval_wer": 0.275,
|
| 968 |
+
"step": 120000
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 8.57,
|
| 972 |
+
"learning_rate": 1.5055369552163509e-05,
|
| 973 |
+
"loss": 0.0394,
|
| 974 |
+
"step": 121000
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 8.64,
|
| 978 |
+
"learning_rate": 1.3621302504261194e-05,
|
| 979 |
+
"loss": 0.0395,
|
| 980 |
+
"step": 122000
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 8.71,
|
| 984 |
+
"learning_rate": 1.2255749551033205e-05,
|
| 985 |
+
"loss": 0.0393,
|
| 986 |
+
"step": 123000
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 8.78,
|
| 990 |
+
"learning_rate": 1.0959396720620555e-05,
|
| 991 |
+
"loss": 0.0393,
|
| 992 |
+
"step": 124000
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 8.86,
|
| 996 |
+
"learning_rate": 9.732895276323099e-06,
|
| 997 |
+
"loss": 0.0392,
|
| 998 |
+
"step": 125000
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 8.86,
|
| 1002 |
+
"eval_cer": 0.09957092649690592,
|
| 1003 |
+
"eval_loss": 0.054800018668174744,
|
| 1004 |
+
"eval_runtime": 80.1483,
|
| 1005 |
+
"eval_samples_per_second": 62.384,
|
| 1006 |
+
"eval_steps_per_second": 0.125,
|
| 1007 |
+
"eval_wer": 0.2756,
|
| 1008 |
+
"step": 125000
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 8.93,
|
| 1012 |
+
"learning_rate": 8.576861389416939e-06,
|
| 1013 |
+
"loss": 0.0391,
|
| 1014 |
+
"step": 126000
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 9.0,
|
| 1018 |
+
"learning_rate": 7.4918758296016734e-06,
|
| 1019 |
+
"loss": 0.039,
|
| 1020 |
+
"step": 127000
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 9.07,
|
| 1024 |
+
"learning_rate": 6.478483673232537e-06,
|
| 1025 |
+
"loss": 0.0388,
|
| 1026 |
+
"step": 128000
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 9.14,
|
| 1030 |
+
"learning_rate": 5.537194029484349e-06,
|
| 1031 |
+
"loss": 0.0383,
|
| 1032 |
+
"step": 129000
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 9.21,
|
| 1036 |
+
"learning_rate": 4.668479784584583e-06,
|
| 1037 |
+
"loss": 0.0383,
|
| 1038 |
+
"step": 130000
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 9.21,
|
| 1042 |
+
"eval_cer": 0.09931218870609543,
|
| 1043 |
+
"eval_loss": 0.054751720279455185,
|
| 1044 |
+
"eval_runtime": 80.4739,
|
| 1045 |
+
"eval_samples_per_second": 62.132,
|
| 1046 |
+
"eval_steps_per_second": 0.124,
|
| 1047 |
+
"eval_wer": 0.2748,
|
| 1048 |
+
"step": 130000
|
| 1049 |
+
},
|
| 1050 |
+
{
|
| 1051 |
+
"epoch": 9.28,
|
| 1052 |
+
"learning_rate": 3.872777364244245e-06,
|
| 1053 |
+
"loss": 0.0387,
|
| 1054 |
+
"step": 131000
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 9.35,
|
| 1058 |
+
"learning_rate": 3.150486514405798e-06,
|
| 1059 |
+
"loss": 0.0385,
|
| 1060 |
+
"step": 132000
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 9.42,
|
| 1064 |
+
"learning_rate": 2.50197010041826e-06,
|
| 1065 |
+
"loss": 0.0385,
|
| 1066 |
+
"step": 133000
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 9.49,
|
| 1070 |
+
"learning_rate": 1.9275539247405437e-06,
|
| 1071 |
+
"loss": 0.0384,
|
| 1072 |
+
"step": 134000
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 9.56,
|
| 1076 |
+
"learning_rate": 1.4275265632643806e-06,
|
| 1077 |
+
"loss": 0.0386,
|
| 1078 |
+
"step": 135000
|
| 1079 |
+
},
|
| 1080 |
+
{
|
| 1081 |
+
"epoch": 9.56,
|
| 1082 |
+
"eval_cer": 0.09924750425839281,
|
| 1083 |
+
"eval_loss": 0.05460319668054581,
|
| 1084 |
+
"eval_runtime": 80.404,
|
| 1085 |
+
"eval_samples_per_second": 62.186,
|
| 1086 |
+
"eval_steps_per_second": 0.124,
|
| 1087 |
+
"eval_wer": 0.2752,
|
| 1088 |
+
"step": 135000
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 9.63,
|
| 1092 |
+
"learning_rate": 1.0021392203391932e-06,
|
| 1093 |
+
"loss": 0.0383,
|
| 1094 |
+
"step": 136000
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 9.71,
|
| 1098 |
+
"learning_rate": 6.516056025718585e-07,
|
| 1099 |
+
"loss": 0.0383,
|
| 1100 |
+
"step": 137000
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 9.78,
|
| 1104 |
+
"learning_rate": 3.76101811464391e-07,
|
| 1105 |
+
"loss": 0.0383,
|
| 1106 |
+
"step": 138000
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 9.85,
|
| 1110 |
+
"learning_rate": 1.7576625494392648e-07,
|
| 1111 |
+
"loss": 0.0384,
|
| 1112 |
+
"step": 139000
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 9.92,
|
| 1116 |
+
"learning_rate": 5.0699577829210744e-08,
|
| 1117 |
+
"loss": 0.0387,
|
| 1118 |
+
"step": 140000
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 9.92,
|
| 1122 |
+
"eval_cer": 0.09937687315379806,
|
| 1123 |
+
"eval_loss": 0.05459373816847801,
|
| 1124 |
+
"eval_runtime": 80.3513,
|
| 1125 |
+
"eval_samples_per_second": 62.227,
|
| 1126 |
+
"eval_steps_per_second": 0.124,
|
| 1127 |
+
"eval_wer": 0.2752,
|
| 1128 |
+
"step": 140000
|
| 1129 |
+
}
|
| 1130 |
+
],
|
| 1131 |
+
"max_steps": 141160,
|
| 1132 |
+
"num_train_epochs": 10,
|
| 1133 |
+
"total_flos": 3.547844813555835e+17,
|
| 1134 |
+
"trial_name": null,
|
| 1135 |
+
"trial_params": null
|
| 1136 |
+
}
|