End of training
Browse files- README.md +3 -3
- all_results.json +16 -0
- eval--in domain set_results.json +10 -0
- train_results.json +9 -0
- trainer_state.json +1062 -0
README.md
CHANGED
|
@@ -16,9 +16,9 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 16 |
|
| 17 |
This model was trained from scratch on an unknown dataset.
|
| 18 |
It achieves the following results on the evaluation set:
|
| 19 |
-
- Loss: 156.
|
| 20 |
-
- Wer: 0.
|
| 21 |
-
- Cer: 0.
|
| 22 |
|
| 23 |
## Model description
|
| 24 |
|
|
|
|
| 16 |
|
| 17 |
This model was trained from scratch on an unknown dataset.
|
| 18 |
It achieves the following results on the evaluation set:
|
| 19 |
+
- Loss: 156.9329
|
| 20 |
+
- Wer: 0.1783
|
| 21 |
+
- Cer: 0.0520
|
| 22 |
|
| 23 |
## Model description
|
| 24 |
|
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"eval_cer": 0.05199546446303709,
|
| 4 |
+
"eval_loss": 156.9329376220703,
|
| 5 |
+
"eval_runtime": 51.7119,
|
| 6 |
+
"eval_samples": 1727,
|
| 7 |
+
"eval_samples_per_second": 33.397,
|
| 8 |
+
"eval_steps_per_second": 4.177,
|
| 9 |
+
"eval_wer": 0.1783226683277649,
|
| 10 |
+
"total_flos": 4.979407221481362e+19,
|
| 11 |
+
"train_loss": 513.291294528035,
|
| 12 |
+
"train_runtime": 37892.4227,
|
| 13 |
+
"train_samples": 15540,
|
| 14 |
+
"train_samples_per_second": 24.607,
|
| 15 |
+
"train_steps_per_second": 1.539
|
| 16 |
+
}
|
eval--in domain set_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"eval_cer": 0.05199546446303709,
|
| 4 |
+
"eval_loss": 156.9329376220703,
|
| 5 |
+
"eval_runtime": 51.7119,
|
| 6 |
+
"eval_samples": 1727,
|
| 7 |
+
"eval_samples_per_second": 33.397,
|
| 8 |
+
"eval_steps_per_second": 4.177,
|
| 9 |
+
"eval_wer": 0.1783226683277649
|
| 10 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"total_flos": 4.979407221481362e+19,
|
| 4 |
+
"train_loss": 513.291294528035,
|
| 5 |
+
"train_runtime": 37892.4227,
|
| 6 |
+
"train_samples": 15540,
|
| 7 |
+
"train_samples_per_second": 24.607,
|
| 8 |
+
"train_steps_per_second": 1.539
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1062 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.1786624384166714,
|
| 3 |
+
"best_model_checkpoint": "/scratch/elec/puhe/p/palp3/sami_ASR/base_model_output/wav2vec2-base-sami-22k/widv/widv_2dips_validate/60epoch_default/outputs/checkpoint-58320",
|
| 4 |
+
"epoch": 60.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 58320,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 1.0,
|
| 13 |
+
"grad_norm": 658.7919311523438,
|
| 14 |
+
"learning_rate": 3.275034293552812e-05,
|
| 15 |
+
"loss": 4165.6836,
|
| 16 |
+
"step": 972
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 1.0,
|
| 20 |
+
"eval_cer": 0.12357620217499614,
|
| 21 |
+
"eval_loss": 257.47174072265625,
|
| 22 |
+
"eval_runtime": 52.781,
|
| 23 |
+
"eval_samples_per_second": 32.72,
|
| 24 |
+
"eval_steps_per_second": 4.092,
|
| 25 |
+
"eval_wer": 0.5061441757743926,
|
| 26 |
+
"step": 972
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 2.0,
|
| 30 |
+
"grad_norm": 529.923583984375,
|
| 31 |
+
"learning_rate": 6.608367626886145e-05,
|
| 32 |
+
"loss": 711.8623,
|
| 33 |
+
"step": 1944
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 2.0,
|
| 37 |
+
"eval_cer": 0.07524009139794183,
|
| 38 |
+
"eval_loss": 178.0095672607422,
|
| 39 |
+
"eval_runtime": 52.2311,
|
| 40 |
+
"eval_samples_per_second": 33.065,
|
| 41 |
+
"eval_steps_per_second": 4.135,
|
| 42 |
+
"eval_wer": 0.30018687354889856,
|
| 43 |
+
"step": 1944
|
| 44 |
+
},
|
| 45 |
+
{
|
| 46 |
+
"epoch": 3.0,
|
| 47 |
+
"grad_norm": 597.9530639648438,
|
| 48 |
+
"learning_rate": 9.941700960219479e-05,
|
| 49 |
+
"loss": 536.9188,
|
| 50 |
+
"step": 2916
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 3.0,
|
| 54 |
+
"eval_cer": 0.07335887436219012,
|
| 55 |
+
"eval_loss": 177.63063049316406,
|
| 56 |
+
"eval_runtime": 51.8439,
|
| 57 |
+
"eval_samples_per_second": 33.312,
|
| 58 |
+
"eval_steps_per_second": 4.166,
|
| 59 |
+
"eval_wer": 0.3027351492156974,
|
| 60 |
+
"step": 2916
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 4.0,
|
| 64 |
+
"grad_norm": 74.35365295410156,
|
| 65 |
+
"learning_rate": 0.00013275034293552812,
|
| 66 |
+
"loss": 499.5686,
|
| 67 |
+
"step": 3888
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 4.0,
|
| 71 |
+
"eval_cer": 0.06541309464497397,
|
| 72 |
+
"eval_loss": 160.8972625732422,
|
| 73 |
+
"eval_runtime": 52.9987,
|
| 74 |
+
"eval_samples_per_second": 32.586,
|
| 75 |
+
"eval_steps_per_second": 4.076,
|
| 76 |
+
"eval_wer": 0.2519961492723257,
|
| 77 |
+
"step": 3888
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 5.0,
|
| 81 |
+
"grad_norm": 652.4288940429688,
|
| 82 |
+
"learning_rate": 0.00016608367626886144,
|
| 83 |
+
"loss": 489.7358,
|
| 84 |
+
"step": 4860
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 5.0,
|
| 88 |
+
"eval_cer": 0.0774906798151425,
|
| 89 |
+
"eval_loss": 174.74952697753906,
|
| 90 |
+
"eval_runtime": 51.7437,
|
| 91 |
+
"eval_samples_per_second": 33.376,
|
| 92 |
+
"eval_steps_per_second": 4.174,
|
| 93 |
+
"eval_wer": 0.28999377088170336,
|
| 94 |
+
"step": 4860
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 6.0,
|
| 98 |
+
"grad_norm": 275.6253356933594,
|
| 99 |
+
"learning_rate": 0.00019941700960219482,
|
| 100 |
+
"loss": 508.3668,
|
| 101 |
+
"step": 5832
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 6.0,
|
| 105 |
+
"eval_cer": 0.07830673286718093,
|
| 106 |
+
"eval_loss": 193.48646545410156,
|
| 107 |
+
"eval_runtime": 80.1664,
|
| 108 |
+
"eval_samples_per_second": 21.543,
|
| 109 |
+
"eval_steps_per_second": 2.694,
|
| 110 |
+
"eval_wer": 0.29441078203748794,
|
| 111 |
+
"step": 5832
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 7.0,
|
| 115 |
+
"grad_norm": 890.0252685546875,
|
| 116 |
+
"learning_rate": 0.00023275034293552814,
|
| 117 |
+
"loss": 537.328,
|
| 118 |
+
"step": 6804
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 7.0,
|
| 122 |
+
"eval_cer": 0.08461181644819352,
|
| 123 |
+
"eval_loss": 230.1122589111328,
|
| 124 |
+
"eval_runtime": 52.6861,
|
| 125 |
+
"eval_samples_per_second": 32.779,
|
| 126 |
+
"eval_steps_per_second": 4.1,
|
| 127 |
+
"eval_wer": 0.31037997621609376,
|
| 128 |
+
"step": 6804
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 8.0,
|
| 132 |
+
"grad_norm": 165.17703247070312,
|
| 133 |
+
"learning_rate": 0.00026608367626886146,
|
| 134 |
+
"loss": 553.6925,
|
| 135 |
+
"step": 7776
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 8.0,
|
| 139 |
+
"eval_cer": 0.10000515401927604,
|
| 140 |
+
"eval_loss": 256.9770812988281,
|
| 141 |
+
"eval_runtime": 52.1076,
|
| 142 |
+
"eval_samples_per_second": 33.143,
|
| 143 |
+
"eval_steps_per_second": 4.145,
|
| 144 |
+
"eval_wer": 0.3464522339883346,
|
| 145 |
+
"step": 7776
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 9.0,
|
| 149 |
+
"grad_norm": 932.0779418945312,
|
| 150 |
+
"learning_rate": 0.0002994170096021948,
|
| 151 |
+
"loss": 581.7494,
|
| 152 |
+
"step": 8748
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 9.0,
|
| 156 |
+
"eval_cer": 0.10558867490164413,
|
| 157 |
+
"eval_loss": 254.68865966796875,
|
| 158 |
+
"eval_runtime": 52.1145,
|
| 159 |
+
"eval_samples_per_second": 33.139,
|
| 160 |
+
"eval_steps_per_second": 4.145,
|
| 161 |
+
"eval_wer": 0.3680276346338977,
|
| 162 |
+
"step": 8748
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 10.0,
|
| 166 |
+
"grad_norm": 767.27197265625,
|
| 167 |
+
"learning_rate": 0.00033275034293552816,
|
| 168 |
+
"loss": 623.7695,
|
| 169 |
+
"step": 9720
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 10.0,
|
| 173 |
+
"eval_cer": 0.1115415671654612,
|
| 174 |
+
"eval_loss": 268.8912658691406,
|
| 175 |
+
"eval_runtime": 52.2444,
|
| 176 |
+
"eval_samples_per_second": 33.056,
|
| 177 |
+
"eval_steps_per_second": 4.134,
|
| 178 |
+
"eval_wer": 0.38099552636049605,
|
| 179 |
+
"step": 9720
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 11.0,
|
| 183 |
+
"grad_norm": 251.00375366210938,
|
| 184 |
+
"learning_rate": 0.00036608367626886145,
|
| 185 |
+
"loss": 652.2123,
|
| 186 |
+
"step": 10692
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 11.0,
|
| 190 |
+
"eval_cer": 0.11990825845688663,
|
| 191 |
+
"eval_loss": 315.2755432128906,
|
| 192 |
+
"eval_runtime": 51.9211,
|
| 193 |
+
"eval_samples_per_second": 33.262,
|
| 194 |
+
"eval_steps_per_second": 4.16,
|
| 195 |
+
"eval_wer": 0.40659153972478623,
|
| 196 |
+
"step": 10692
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 12.0,
|
| 200 |
+
"grad_norm": 208.37503051757812,
|
| 201 |
+
"learning_rate": 0.0003993827160493827,
|
| 202 |
+
"loss": 702.8481,
|
| 203 |
+
"step": 11664
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 12.0,
|
| 207 |
+
"eval_cer": 0.12610167162025185,
|
| 208 |
+
"eval_loss": 314.96868896484375,
|
| 209 |
+
"eval_runtime": 53.0922,
|
| 210 |
+
"eval_samples_per_second": 32.528,
|
| 211 |
+
"eval_steps_per_second": 4.068,
|
| 212 |
+
"eval_wer": 0.4256186647035506,
|
| 213 |
+
"step": 11664
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 13.0,
|
| 217 |
+
"grad_norm": 410.18218994140625,
|
| 218 |
+
"learning_rate": 0.000432681755829904,
|
| 219 |
+
"loss": 743.4121,
|
| 220 |
+
"step": 12636
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 13.0,
|
| 224 |
+
"eval_cer": 0.16132939337193122,
|
| 225 |
+
"eval_loss": 311.7789611816406,
|
| 226 |
+
"eval_runtime": 52.4611,
|
| 227 |
+
"eval_samples_per_second": 32.92,
|
| 228 |
+
"eval_steps_per_second": 4.117,
|
| 229 |
+
"eval_wer": 0.5520697661249221,
|
| 230 |
+
"step": 12636
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 14.0,
|
| 234 |
+
"grad_norm": 260.27325439453125,
|
| 235 |
+
"learning_rate": 0.0004660150891632373,
|
| 236 |
+
"loss": 799.5601,
|
| 237 |
+
"step": 13608
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 14.0,
|
| 241 |
+
"eval_cer": 0.13851426804336248,
|
| 242 |
+
"eval_loss": 359.827880859375,
|
| 243 |
+
"eval_runtime": 52.426,
|
| 244 |
+
"eval_samples_per_second": 32.942,
|
| 245 |
+
"eval_steps_per_second": 4.12,
|
| 246 |
+
"eval_wer": 0.45455575060875475,
|
| 247 |
+
"step": 13608
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 15.0,
|
| 251 |
+
"grad_norm": 272.1807556152344,
|
| 252 |
+
"learning_rate": 0.0004993484224965707,
|
| 253 |
+
"loss": 816.097,
|
| 254 |
+
"step": 14580
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 15.0,
|
| 258 |
+
"eval_cer": 0.18363770680502345,
|
| 259 |
+
"eval_loss": 376.90533447265625,
|
| 260 |
+
"eval_runtime": 51.4749,
|
| 261 |
+
"eval_samples_per_second": 33.55,
|
| 262 |
+
"eval_steps_per_second": 4.196,
|
| 263 |
+
"eval_wer": 0.572059573022255,
|
| 264 |
+
"step": 14580
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 16.0,
|
| 268 |
+
"grad_norm": 301.32891845703125,
|
| 269 |
+
"learning_rate": 0.0004891060813900321,
|
| 270 |
+
"loss": 815.892,
|
| 271 |
+
"step": 15552
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 16.0,
|
| 275 |
+
"eval_cer": 0.1669558644149329,
|
| 276 |
+
"eval_loss": 378.6718444824219,
|
| 277 |
+
"eval_runtime": 51.8917,
|
| 278 |
+
"eval_samples_per_second": 33.281,
|
| 279 |
+
"eval_steps_per_second": 4.163,
|
| 280 |
+
"eval_wer": 0.534571606546237,
|
| 281 |
+
"step": 15552
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 17.0,
|
| 285 |
+
"grad_norm": 509.3703918457031,
|
| 286 |
+
"learning_rate": 0.00047799497027892093,
|
| 287 |
+
"loss": 803.2213,
|
| 288 |
+
"step": 16524
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 17.0,
|
| 292 |
+
"eval_cer": 0.14506846255604997,
|
| 293 |
+
"eval_loss": 358.10638427734375,
|
| 294 |
+
"eval_runtime": 51.7272,
|
| 295 |
+
"eval_samples_per_second": 33.387,
|
| 296 |
+
"eval_steps_per_second": 4.176,
|
| 297 |
+
"eval_wer": 0.4638427997055326,
|
| 298 |
+
"step": 16524
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 18.0,
|
| 302 |
+
"grad_norm": 278.8409423828125,
|
| 303 |
+
"learning_rate": 0.0004668952903520805,
|
| 304 |
+
"loss": 783.7681,
|
| 305 |
+
"step": 17496
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 18.0,
|
| 309 |
+
"eval_cer": 0.14475063136736133,
|
| 310 |
+
"eval_loss": 332.414794921875,
|
| 311 |
+
"eval_runtime": 51.8709,
|
| 312 |
+
"eval_samples_per_second": 33.294,
|
| 313 |
+
"eval_steps_per_second": 4.164,
|
| 314 |
+
"eval_wer": 0.4774902316099439,
|
| 315 |
+
"step": 17496
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 19.0,
|
| 319 |
+
"grad_norm": 225.4127960205078,
|
| 320 |
+
"learning_rate": 0.0004557841792409694,
|
| 321 |
+
"loss": 730.573,
|
| 322 |
+
"step": 18468
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 19.0,
|
| 326 |
+
"eval_cer": 0.13914134038861306,
|
| 327 |
+
"eval_loss": 312.7887268066406,
|
| 328 |
+
"eval_runtime": 53.4549,
|
| 329 |
+
"eval_samples_per_second": 32.308,
|
| 330 |
+
"eval_steps_per_second": 4.041,
|
| 331 |
+
"eval_wer": 0.45478226400135907,
|
| 332 |
+
"step": 18468
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 20.0,
|
| 336 |
+
"grad_norm": 144.8686981201172,
|
| 337 |
+
"learning_rate": 0.00044467306812985823,
|
| 338 |
+
"loss": 722.3447,
|
| 339 |
+
"step": 19440
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 20.0,
|
| 343 |
+
"eval_cer": 0.1279485285274967,
|
| 344 |
+
"eval_loss": 316.9328918457031,
|
| 345 |
+
"eval_runtime": 51.7766,
|
| 346 |
+
"eval_samples_per_second": 33.355,
|
| 347 |
+
"eval_steps_per_second": 4.172,
|
| 348 |
+
"eval_wer": 0.42692111671102556,
|
| 349 |
+
"step": 19440
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 21.0,
|
| 353 |
+
"grad_norm": 164.9013214111328,
|
| 354 |
+
"learning_rate": 0.00043357338820301783,
|
| 355 |
+
"loss": 683.925,
|
| 356 |
+
"step": 20412
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 21.0,
|
| 360 |
+
"eval_cer": 0.12733863624649955,
|
| 361 |
+
"eval_loss": 290.0935974121094,
|
| 362 |
+
"eval_runtime": 51.8702,
|
| 363 |
+
"eval_samples_per_second": 33.295,
|
| 364 |
+
"eval_steps_per_second": 4.164,
|
| 365 |
+
"eval_wer": 0.414462880117787,
|
| 366 |
+
"step": 20412
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 22.0,
|
| 370 |
+
"grad_norm": 61.14198684692383,
|
| 371 |
+
"learning_rate": 0.00042247370827617743,
|
| 372 |
+
"loss": 650.4165,
|
| 373 |
+
"step": 21384
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 22.0,
|
| 377 |
+
"eval_cer": 0.11940144656141014,
|
| 378 |
+
"eval_loss": 288.4570007324219,
|
| 379 |
+
"eval_runtime": 51.3929,
|
| 380 |
+
"eval_samples_per_second": 33.604,
|
| 381 |
+
"eval_steps_per_second": 4.203,
|
| 382 |
+
"eval_wer": 0.39889008437623874,
|
| 383 |
+
"step": 21384
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 23.0,
|
| 387 |
+
"grad_norm": 462.71295166015625,
|
| 388 |
+
"learning_rate": 0.0004113625971650663,
|
| 389 |
+
"loss": 631.0532,
|
| 390 |
+
"step": 22356
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 23.0,
|
| 394 |
+
"eval_cer": 0.12083598192657241,
|
| 395 |
+
"eval_loss": 271.81787109375,
|
| 396 |
+
"eval_runtime": 51.8764,
|
| 397 |
+
"eval_samples_per_second": 33.291,
|
| 398 |
+
"eval_steps_per_second": 4.164,
|
| 399 |
+
"eval_wer": 0.4042697774505918,
|
| 400 |
+
"step": 22356
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 24.0,
|
| 404 |
+
"grad_norm": 70.46085357666016,
|
| 405 |
+
"learning_rate": 0.0004002514860539552,
|
| 406 |
+
"loss": 617.0345,
|
| 407 |
+
"step": 23328
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 24.0,
|
| 411 |
+
"eval_cer": 0.11391241603243596,
|
| 412 |
+
"eval_loss": 240.84332275390625,
|
| 413 |
+
"eval_runtime": 52.4166,
|
| 414 |
+
"eval_samples_per_second": 32.948,
|
| 415 |
+
"eval_steps_per_second": 4.121,
|
| 416 |
+
"eval_wer": 0.38796081318307946,
|
| 417 |
+
"step": 23328
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 25.0,
|
| 421 |
+
"grad_norm": 116.7966079711914,
|
| 422 |
+
"learning_rate": 0.0003891403749428441,
|
| 423 |
+
"loss": 579.2521,
|
| 424 |
+
"step": 24300
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 25.0,
|
| 428 |
+
"eval_cer": 0.11367189513288779,
|
| 429 |
+
"eval_loss": 261.6251525878906,
|
| 430 |
+
"eval_runtime": 51.9507,
|
| 431 |
+
"eval_samples_per_second": 33.243,
|
| 432 |
+
"eval_steps_per_second": 4.158,
|
| 433 |
+
"eval_wer": 0.3953791267908715,
|
| 434 |
+
"step": 24300
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 26.0,
|
| 438 |
+
"grad_norm": 43.10677719116211,
|
| 439 |
+
"learning_rate": 0.0003780406950160037,
|
| 440 |
+
"loss": 571.9061,
|
| 441 |
+
"step": 25272
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 26.0,
|
| 445 |
+
"eval_cer": 0.10684281959214527,
|
| 446 |
+
"eval_loss": 248.2078094482422,
|
| 447 |
+
"eval_runtime": 52.057,
|
| 448 |
+
"eval_samples_per_second": 33.175,
|
| 449 |
+
"eval_steps_per_second": 4.149,
|
| 450 |
+
"eval_wer": 0.36508296053004136,
|
| 451 |
+
"step": 25272
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 27.0,
|
| 455 |
+
"grad_norm": 309.9073181152344,
|
| 456 |
+
"learning_rate": 0.0003669295839048926,
|
| 457 |
+
"loss": 535.8284,
|
| 458 |
+
"step": 26244
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 27.0,
|
| 462 |
+
"eval_cer": 0.10169739034823991,
|
| 463 |
+
"eval_loss": 210.3215789794922,
|
| 464 |
+
"eval_runtime": 51.8198,
|
| 465 |
+
"eval_samples_per_second": 33.327,
|
| 466 |
+
"eval_steps_per_second": 4.168,
|
| 467 |
+
"eval_wer": 0.35488985786284616,
|
| 468 |
+
"step": 26244
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 28.0,
|
| 472 |
+
"grad_norm": 109.1835708618164,
|
| 473 |
+
"learning_rate": 0.00035581847279378143,
|
| 474 |
+
"loss": 528.5526,
|
| 475 |
+
"step": 27216
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 28.0,
|
| 479 |
+
"eval_cer": 0.10048619581837236,
|
| 480 |
+
"eval_loss": 223.02474975585938,
|
| 481 |
+
"eval_runtime": 51.9491,
|
| 482 |
+
"eval_samples_per_second": 33.244,
|
| 483 |
+
"eval_steps_per_second": 4.158,
|
| 484 |
+
"eval_wer": 0.3553995129962059,
|
| 485 |
+
"step": 27216
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 29.0,
|
| 489 |
+
"grad_norm": 356.6168518066406,
|
| 490 |
+
"learning_rate": 0.0003447073616826703,
|
| 491 |
+
"loss": 495.6612,
|
| 492 |
+
"step": 28188
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 29.0,
|
| 496 |
+
"eval_cer": 0.10109608809936949,
|
| 497 |
+
"eval_loss": 214.10452270507812,
|
| 498 |
+
"eval_runtime": 52.5972,
|
| 499 |
+
"eval_samples_per_second": 32.834,
|
| 500 |
+
"eval_steps_per_second": 4.107,
|
| 501 |
+
"eval_wer": 0.34214847952885213,
|
| 502 |
+
"step": 28188
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"epoch": 30.0,
|
| 506 |
+
"grad_norm": 93.94068908691406,
|
| 507 |
+
"learning_rate": 0.00033359625057155924,
|
| 508 |
+
"loss": 489.6183,
|
| 509 |
+
"step": 29160
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 30.0,
|
| 513 |
+
"eval_cer": 0.09630285017265965,
|
| 514 |
+
"eval_loss": 223.9063720703125,
|
| 515 |
+
"eval_runtime": 52.76,
|
| 516 |
+
"eval_samples_per_second": 32.733,
|
| 517 |
+
"eval_steps_per_second": 4.094,
|
| 518 |
+
"eval_wer": 0.33014326972082225,
|
| 519 |
+
"step": 29160
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 31.0,
|
| 523 |
+
"grad_norm": 254.2984619140625,
|
| 524 |
+
"learning_rate": 0.0003224851394604481,
|
| 525 |
+
"loss": 465.3683,
|
| 526 |
+
"step": 30132
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 31.0,
|
| 530 |
+
"eval_cer": 0.09759994502379439,
|
| 531 |
+
"eval_loss": 204.1656951904297,
|
| 532 |
+
"eval_runtime": 51.6234,
|
| 533 |
+
"eval_samples_per_second": 33.454,
|
| 534 |
+
"eval_steps_per_second": 4.184,
|
| 535 |
+
"eval_wer": 0.3407893991732261,
|
| 536 |
+
"step": 30132
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 32.0,
|
| 540 |
+
"grad_norm": 90.14813995361328,
|
| 541 |
+
"learning_rate": 0.0003113854595336077,
|
| 542 |
+
"loss": 447.7585,
|
| 543 |
+
"step": 31104
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 32.0,
|
| 547 |
+
"eval_cer": 0.0901867472984349,
|
| 548 |
+
"eval_loss": 204.1836395263672,
|
| 549 |
+
"eval_runtime": 52.1105,
|
| 550 |
+
"eval_samples_per_second": 33.141,
|
| 551 |
+
"eval_steps_per_second": 4.145,
|
| 552 |
+
"eval_wer": 0.31474035902372727,
|
| 553 |
+
"step": 31104
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 33.0,
|
| 557 |
+
"grad_norm": 284.7950134277344,
|
| 558 |
+
"learning_rate": 0.0003002743484224966,
|
| 559 |
+
"loss": 430.5009,
|
| 560 |
+
"step": 32076
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 33.0,
|
| 564 |
+
"eval_cer": 0.09058188877626402,
|
| 565 |
+
"eval_loss": 199.83998107910156,
|
| 566 |
+
"eval_runtime": 51.7349,
|
| 567 |
+
"eval_samples_per_second": 33.382,
|
| 568 |
+
"eval_steps_per_second": 4.175,
|
| 569 |
+
"eval_wer": 0.3149102440681805,
|
| 570 |
+
"step": 32076
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 34.0,
|
| 574 |
+
"grad_norm": 120.24885559082031,
|
| 575 |
+
"learning_rate": 0.00028916323731138544,
|
| 576 |
+
"loss": 423.8555,
|
| 577 |
+
"step": 33048
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 34.0,
|
| 581 |
+
"eval_cer": 0.08400192416719639,
|
| 582 |
+
"eval_loss": 188.04661560058594,
|
| 583 |
+
"eval_runtime": 52.0605,
|
| 584 |
+
"eval_samples_per_second": 33.173,
|
| 585 |
+
"eval_steps_per_second": 4.149,
|
| 586 |
+
"eval_wer": 0.29390112690412823,
|
| 587 |
+
"step": 33048
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 35.0,
|
| 591 |
+
"grad_norm": 108.36172485351562,
|
| 592 |
+
"learning_rate": 0.0002780521262002744,
|
| 593 |
+
"loss": 409.8573,
|
| 594 |
+
"step": 34020
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 35.0,
|
| 598 |
+
"eval_cer": 0.08067758173415568,
|
| 599 |
+
"eval_loss": 204.4093017578125,
|
| 600 |
+
"eval_runtime": 51.291,
|
| 601 |
+
"eval_samples_per_second": 33.671,
|
| 602 |
+
"eval_steps_per_second": 4.211,
|
| 603 |
+
"eval_wer": 0.2826887139702135,
|
| 604 |
+
"step": 34020
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 36.0,
|
| 608 |
+
"grad_norm": 993.7814331054688,
|
| 609 |
+
"learning_rate": 0.00026694101508916324,
|
| 610 |
+
"loss": 385.368,
|
| 611 |
+
"step": 34992
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 36.0,
|
| 615 |
+
"eval_cer": 0.08328036146855189,
|
| 616 |
+
"eval_loss": 187.58653259277344,
|
| 617 |
+
"eval_runtime": 51.8038,
|
| 618 |
+
"eval_samples_per_second": 33.337,
|
| 619 |
+
"eval_steps_per_second": 4.17,
|
| 620 |
+
"eval_wer": 0.29106970949657396,
|
| 621 |
+
"step": 34992
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 37.0,
|
| 625 |
+
"grad_norm": 117.57772827148438,
|
| 626 |
+
"learning_rate": 0.00025584133516232284,
|
| 627 |
+
"loss": 372.0869,
|
| 628 |
+
"step": 35964
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"epoch": 37.0,
|
| 632 |
+
"eval_cer": 0.07700104798391946,
|
| 633 |
+
"eval_loss": 181.96978759765625,
|
| 634 |
+
"eval_runtime": 52.5811,
|
| 635 |
+
"eval_samples_per_second": 32.844,
|
| 636 |
+
"eval_steps_per_second": 4.108,
|
| 637 |
+
"eval_wer": 0.27300526643637807,
|
| 638 |
+
"step": 35964
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 38.0,
|
| 642 |
+
"grad_norm": 16.713836669921875,
|
| 643 |
+
"learning_rate": 0.0002447302240512117,
|
| 644 |
+
"loss": 360.2271,
|
| 645 |
+
"step": 36936
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 38.0,
|
| 649 |
+
"eval_cer": 0.07677770714862474,
|
| 650 |
+
"eval_loss": 181.98971557617188,
|
| 651 |
+
"eval_runtime": 52.5466,
|
| 652 |
+
"eval_samples_per_second": 32.866,
|
| 653 |
+
"eval_steps_per_second": 4.111,
|
| 654 |
+
"eval_wer": 0.2687015119768956,
|
| 655 |
+
"step": 36936
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 39.0,
|
| 659 |
+
"grad_norm": 172.10801696777344,
|
| 660 |
+
"learning_rate": 0.0002336191129401006,
|
| 661 |
+
"loss": 340.9098,
|
| 662 |
+
"step": 37908
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 39.0,
|
| 666 |
+
"eval_cer": 0.07745631968663563,
|
| 667 |
+
"eval_loss": 188.7457733154297,
|
| 668 |
+
"eval_runtime": 52.2724,
|
| 669 |
+
"eval_samples_per_second": 33.038,
|
| 670 |
+
"eval_steps_per_second": 4.132,
|
| 671 |
+
"eval_wer": 0.26502066934707513,
|
| 672 |
+
"step": 37908
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 40.0,
|
| 676 |
+
"grad_norm": 556.4381713867188,
|
| 677 |
+
"learning_rate": 0.00022250800182898947,
|
| 678 |
+
"loss": 329.051,
|
| 679 |
+
"step": 38880
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 40.0,
|
| 683 |
+
"eval_cer": 0.07525727146219527,
|
| 684 |
+
"eval_loss": 165.11180114746094,
|
| 685 |
+
"eval_runtime": 51.7142,
|
| 686 |
+
"eval_samples_per_second": 33.395,
|
| 687 |
+
"eval_steps_per_second": 4.177,
|
| 688 |
+
"eval_wer": 0.26320856220624045,
|
| 689 |
+
"step": 38880
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 41.0,
|
| 693 |
+
"grad_norm": 9.856818199157715,
|
| 694 |
+
"learning_rate": 0.00021139689071787837,
|
| 695 |
+
"loss": 311.5036,
|
| 696 |
+
"step": 39852
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 41.0,
|
| 700 |
+
"eval_cer": 0.07304104317350146,
|
| 701 |
+
"eval_loss": 184.70675659179688,
|
| 702 |
+
"eval_runtime": 51.9748,
|
| 703 |
+
"eval_samples_per_second": 33.228,
|
| 704 |
+
"eval_steps_per_second": 4.156,
|
| 705 |
+
"eval_wer": 0.26309530550993826,
|
| 706 |
+
"step": 39852
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 42.0,
|
| 710 |
+
"grad_norm": 177.51455688476562,
|
| 711 |
+
"learning_rate": 0.00020028577960676727,
|
| 712 |
+
"loss": 300.7056,
|
| 713 |
+
"step": 40824
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 42.0,
|
| 717 |
+
"eval_cer": 0.07146906729431168,
|
| 718 |
+
"eval_loss": 169.64820861816406,
|
| 719 |
+
"eval_runtime": 70.852,
|
| 720 |
+
"eval_samples_per_second": 24.375,
|
| 721 |
+
"eval_steps_per_second": 3.049,
|
| 722 |
+
"eval_wer": 0.25029729882779317,
|
| 723 |
+
"step": 40824
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 43.0,
|
| 727 |
+
"grad_norm": 356.5669860839844,
|
| 728 |
+
"learning_rate": 0.00018918609967992684,
|
| 729 |
+
"loss": 293.3254,
|
| 730 |
+
"step": 41796
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 43.0,
|
| 734 |
+
"eval_cer": 0.0705413438246259,
|
| 735 |
+
"eval_loss": 156.13795471191406,
|
| 736 |
+
"eval_runtime": 52.1899,
|
| 737 |
+
"eval_samples_per_second": 33.091,
|
| 738 |
+
"eval_steps_per_second": 4.139,
|
| 739 |
+
"eval_wer": 0.24469109236083583,
|
| 740 |
+
"step": 41796
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 44.0,
|
| 744 |
+
"grad_norm": 59.47640609741211,
|
| 745 |
+
"learning_rate": 0.00017807498856881575,
|
| 746 |
+
"loss": 272.1289,
|
| 747 |
+
"step": 42768
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 44.0,
|
| 751 |
+
"eval_cer": 0.0682821653752985,
|
| 752 |
+
"eval_loss": 174.929931640625,
|
| 753 |
+
"eval_runtime": 52.4672,
|
| 754 |
+
"eval_samples_per_second": 32.916,
|
| 755 |
+
"eval_steps_per_second": 4.117,
|
| 756 |
+
"eval_wer": 0.2404439662495045,
|
| 757 |
+
"step": 42768
|
| 758 |
+
},
|
| 759 |
+
{
|
| 760 |
+
"epoch": 45.0,
|
| 761 |
+
"grad_norm": 62.990211486816406,
|
| 762 |
+
"learning_rate": 0.00016696387745770462,
|
| 763 |
+
"loss": 266.0369,
|
| 764 |
+
"step": 43740
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 45.0,
|
| 768 |
+
"eval_cer": 0.06732008177710584,
|
| 769 |
+
"eval_loss": 156.83799743652344,
|
| 770 |
+
"eval_runtime": 51.9519,
|
| 771 |
+
"eval_samples_per_second": 33.242,
|
| 772 |
+
"eval_steps_per_second": 4.158,
|
| 773 |
+
"eval_wer": 0.23681975196783508,
|
| 774 |
+
"step": 43740
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 46.0,
|
| 778 |
+
"grad_norm": 29.509681701660156,
|
| 779 |
+
"learning_rate": 0.0001558641975308642,
|
| 780 |
+
"loss": 251.5002,
|
| 781 |
+
"step": 44712
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 46.0,
|
| 785 |
+
"eval_cer": 0.06648684866081399,
|
| 786 |
+
"eval_loss": 170.41712951660156,
|
| 787 |
+
"eval_runtime": 51.371,
|
| 788 |
+
"eval_samples_per_second": 33.618,
|
| 789 |
+
"eval_steps_per_second": 4.205,
|
| 790 |
+
"eval_wer": 0.23512090152330256,
|
| 791 |
+
"step": 44712
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 47.0,
|
| 795 |
+
"grad_norm": 654.94677734375,
|
| 796 |
+
"learning_rate": 0.0001447530864197531,
|
| 797 |
+
"loss": 235.648,
|
| 798 |
+
"step": 45684
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 47.0,
|
| 802 |
+
"eval_cer": 0.06262133420378992,
|
| 803 |
+
"eval_loss": 159.82781982421875,
|
| 804 |
+
"eval_runtime": 51.498,
|
| 805 |
+
"eval_samples_per_second": 33.535,
|
| 806 |
+
"eval_steps_per_second": 4.194,
|
| 807 |
+
"eval_wer": 0.22271929327821507,
|
| 808 |
+
"step": 45684
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 48.0,
|
| 812 |
+
"grad_norm": 47.048065185546875,
|
| 813 |
+
"learning_rate": 0.00013364197530864197,
|
| 814 |
+
"loss": 229.8135,
|
| 815 |
+
"step": 46656
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 48.0,
|
| 819 |
+
"eval_cer": 0.06253543388252272,
|
| 820 |
+
"eval_loss": 163.2967071533203,
|
| 821 |
+
"eval_runtime": 55.2126,
|
| 822 |
+
"eval_samples_per_second": 31.279,
|
| 823 |
+
"eval_steps_per_second": 3.912,
|
| 824 |
+
"eval_wer": 0.21983124752250977,
|
| 825 |
+
"step": 46656
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 49.0,
|
| 829 |
+
"grad_norm": 149.44444274902344,
|
| 830 |
+
"learning_rate": 0.00012253086419753087,
|
| 831 |
+
"loss": 220.4877,
|
| 832 |
+
"step": 47628
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 49.0,
|
| 836 |
+
"eval_cer": 0.0611610287422475,
|
| 837 |
+
"eval_loss": 167.35101318359375,
|
| 838 |
+
"eval_runtime": 51.8387,
|
| 839 |
+
"eval_samples_per_second": 33.315,
|
| 840 |
+
"eval_steps_per_second": 4.167,
|
| 841 |
+
"eval_wer": 0.21382864261849482,
|
| 842 |
+
"step": 47628
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 50.0,
|
| 846 |
+
"grad_norm": 6.606735706329346,
|
| 847 |
+
"learning_rate": 0.00011143118427069045,
|
| 848 |
+
"loss": 210.467,
|
| 849 |
+
"step": 48600
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 50.0,
|
| 853 |
+
"eval_cer": 0.061470269898809424,
|
| 854 |
+
"eval_loss": 163.87408447265625,
|
| 855 |
+
"eval_runtime": 52.7221,
|
| 856 |
+
"eval_samples_per_second": 32.757,
|
| 857 |
+
"eval_steps_per_second": 4.097,
|
| 858 |
+
"eval_wer": 0.21377201427034373,
|
| 859 |
+
"step": 48600
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 51.0,
|
| 863 |
+
"grad_norm": 181.8941192626953,
|
| 864 |
+
"learning_rate": 0.00010032007315957933,
|
| 865 |
+
"loss": 208.0692,
|
| 866 |
+
"step": 49572
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 51.0,
|
| 870 |
+
"eval_cer": 0.05868709948975209,
|
| 871 |
+
"eval_loss": 159.9131622314453,
|
| 872 |
+
"eval_runtime": 52.7472,
|
| 873 |
+
"eval_samples_per_second": 32.741,
|
| 874 |
+
"eval_steps_per_second": 4.095,
|
| 875 |
+
"eval_wer": 0.20482473526247238,
|
| 876 |
+
"step": 49572
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 52.0,
|
| 880 |
+
"grad_norm": 153.64923095703125,
|
| 881 |
+
"learning_rate": 8.920896204846822e-05,
|
| 882 |
+
"loss": 188.9099,
|
| 883 |
+
"step": 50544
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 52.0,
|
| 887 |
+
"eval_cer": 0.05963200302369131,
|
| 888 |
+
"eval_loss": 160.80023193359375,
|
| 889 |
+
"eval_runtime": 52.5438,
|
| 890 |
+
"eval_samples_per_second": 32.868,
|
| 891 |
+
"eval_steps_per_second": 4.111,
|
| 892 |
+
"eval_wer": 0.20958151650716347,
|
| 893 |
+
"step": 50544
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 53.0,
|
| 897 |
+
"grad_norm": 438.6473388671875,
|
| 898 |
+
"learning_rate": 7.809785093735711e-05,
|
| 899 |
+
"loss": 183.2089,
|
| 900 |
+
"step": 51516
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 53.0,
|
| 904 |
+
"eval_cer": 0.05639356091191781,
|
| 905 |
+
"eval_loss": 154.5007781982422,
|
| 906 |
+
"eval_runtime": 53.724,
|
| 907 |
+
"eval_samples_per_second": 32.146,
|
| 908 |
+
"eval_steps_per_second": 4.021,
|
| 909 |
+
"eval_wer": 0.1979160767880401,
|
| 910 |
+
"step": 51516
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 54.0,
|
| 914 |
+
"grad_norm": 41.24275207519531,
|
| 915 |
+
"learning_rate": 6.698673982624601e-05,
|
| 916 |
+
"loss": 174.8595,
|
| 917 |
+
"step": 52488
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 54.0,
|
| 921 |
+
"eval_cer": 0.055955469273455086,
|
| 922 |
+
"eval_loss": 157.23744201660156,
|
| 923 |
+
"eval_runtime": 54.2259,
|
| 924 |
+
"eval_samples_per_second": 31.848,
|
| 925 |
+
"eval_steps_per_second": 3.983,
|
| 926 |
+
"eval_wer": 0.1963871113879608,
|
| 927 |
+
"step": 52488
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 55.0,
|
| 931 |
+
"grad_norm": 1312.384033203125,
|
| 932 |
+
"learning_rate": 5.587562871513489e-05,
|
| 933 |
+
"loss": 167.0076,
|
| 934 |
+
"step": 53460
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 55.0,
|
| 938 |
+
"eval_cer": 0.05501915577164258,
|
| 939 |
+
"eval_loss": 157.28213500976562,
|
| 940 |
+
"eval_runtime": 53.1558,
|
| 941 |
+
"eval_samples_per_second": 32.489,
|
| 942 |
+
"eval_steps_per_second": 4.064,
|
| 943 |
+
"eval_wer": 0.19180021518772297,
|
| 944 |
+
"step": 53460
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 56.0,
|
| 948 |
+
"grad_norm": 87.78731536865234,
|
| 949 |
+
"learning_rate": 4.476451760402378e-05,
|
| 950 |
+
"loss": 162.867,
|
| 951 |
+
"step": 54432
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 56.0,
|
| 955 |
+
"eval_cer": 0.05392822169154913,
|
| 956 |
+
"eval_loss": 158.17630004882812,
|
| 957 |
+
"eval_runtime": 53.3929,
|
| 958 |
+
"eval_samples_per_second": 32.345,
|
| 959 |
+
"eval_steps_per_second": 4.045,
|
| 960 |
+
"eval_wer": 0.185118070105895,
|
| 961 |
+
"step": 54432
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 57.0,
|
| 965 |
+
"grad_norm": 1.752543330192566,
|
| 966 |
+
"learning_rate": 3.366483767718336e-05,
|
| 967 |
+
"loss": 154.225,
|
| 968 |
+
"step": 55404
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 57.0,
|
| 972 |
+
"eval_cer": 0.053189478928651196,
|
| 973 |
+
"eval_loss": 157.72227478027344,
|
| 974 |
+
"eval_runtime": 52.67,
|
| 975 |
+
"eval_samples_per_second": 32.789,
|
| 976 |
+
"eval_steps_per_second": 4.101,
|
| 977 |
+
"eval_wer": 0.18290956452800272,
|
| 978 |
+
"step": 55404
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 58.0,
|
| 982 |
+
"grad_norm": 115.63265228271484,
|
| 983 |
+
"learning_rate": 2.2565157750342938e-05,
|
| 984 |
+
"loss": 149.2202,
|
| 985 |
+
"step": 56376
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 58.0,
|
| 989 |
+
"eval_cer": 0.05277715738656862,
|
| 990 |
+
"eval_loss": 159.25308227539062,
|
| 991 |
+
"eval_runtime": 52.3955,
|
| 992 |
+
"eval_samples_per_second": 32.961,
|
| 993 |
+
"eval_steps_per_second": 4.122,
|
| 994 |
+
"eval_wer": 0.18183362591313212,
|
| 995 |
+
"step": 56376
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 59.0,
|
| 999 |
+
"grad_norm": 131.82754516601562,
|
| 1000 |
+
"learning_rate": 1.1454046639231825e-05,
|
| 1001 |
+
"loss": 144.8521,
|
| 1002 |
+
"step": 57348
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 59.0,
|
| 1006 |
+
"eval_cer": 0.052193035201951656,
|
| 1007 |
+
"eval_loss": 158.10299682617188,
|
| 1008 |
+
"eval_runtime": 52.2821,
|
| 1009 |
+
"eval_samples_per_second": 33.032,
|
| 1010 |
+
"eval_steps_per_second": 4.131,
|
| 1011 |
+
"eval_wer": 0.1796251203352398,
|
| 1012 |
+
"step": 57348
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 60.0,
|
| 1016 |
+
"grad_norm": 105.3376693725586,
|
| 1017 |
+
"learning_rate": 3.4293552812071326e-07,
|
| 1018 |
+
"loss": 145.8063,
|
| 1019 |
+
"step": 58320
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 60.0,
|
| 1023 |
+
"eval_cer": 0.05190956414176989,
|
| 1024 |
+
"eval_loss": 156.37322998046875,
|
| 1025 |
+
"eval_runtime": 52.1099,
|
| 1026 |
+
"eval_samples_per_second": 33.141,
|
| 1027 |
+
"eval_steps_per_second": 4.145,
|
| 1028 |
+
"eval_wer": 0.1786624384166714,
|
| 1029 |
+
"step": 58320
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 60.0,
|
| 1033 |
+
"step": 58320,
|
| 1034 |
+
"total_flos": 4.979407221481362e+19,
|
| 1035 |
+
"train_loss": 513.291294528035,
|
| 1036 |
+
"train_runtime": 37892.4227,
|
| 1037 |
+
"train_samples_per_second": 24.607,
|
| 1038 |
+
"train_steps_per_second": 1.539
|
| 1039 |
+
}
|
| 1040 |
+
],
|
| 1041 |
+
"logging_steps": 500,
|
| 1042 |
+
"max_steps": 58320,
|
| 1043 |
+
"num_input_tokens_seen": 0,
|
| 1044 |
+
"num_train_epochs": 60,
|
| 1045 |
+
"save_steps": 500,
|
| 1046 |
+
"stateful_callbacks": {
|
| 1047 |
+
"TrainerControl": {
|
| 1048 |
+
"args": {
|
| 1049 |
+
"should_epoch_stop": false,
|
| 1050 |
+
"should_evaluate": false,
|
| 1051 |
+
"should_log": false,
|
| 1052 |
+
"should_save": true,
|
| 1053 |
+
"should_training_stop": true
|
| 1054 |
+
},
|
| 1055 |
+
"attributes": {}
|
| 1056 |
+
}
|
| 1057 |
+
},
|
| 1058 |
+
"total_flos": 4.979407221481362e+19,
|
| 1059 |
+
"train_batch_size": 16,
|
| 1060 |
+
"trial_name": null,
|
| 1061 |
+
"trial_params": null
|
| 1062 |
+
}
|