End of training
Browse files- README.md +3 -3
- all_results.json +16 -0
- eval--1hr test 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:
|
| 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: 272.0352
|
| 20 |
+
- Wer: 0.4500
|
| 21 |
+
- Cer: 0.1412
|
| 22 |
|
| 23 |
## Model description
|
| 24 |
|
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"eval_cer": 0.1411846070246505,
|
| 4 |
+
"eval_loss": 272.0352478027344,
|
| 5 |
+
"eval_runtime": 22.8227,
|
| 6 |
+
"eval_samples": 890,
|
| 7 |
+
"eval_samples_per_second": 38.996,
|
| 8 |
+
"eval_steps_per_second": 4.907,
|
| 9 |
+
"eval_wer": 0.44996843434343436,
|
| 10 |
+
"total_flos": 5.518391525676315e+19,
|
| 11 |
+
"train_loss": 505.59943335262346,
|
| 12 |
+
"train_runtime": 39665.0417,
|
| 13 |
+
"train_samples": 17267,
|
| 14 |
+
"train_samples_per_second": 26.119,
|
| 15 |
+
"train_steps_per_second": 1.634
|
| 16 |
+
}
|
eval--1hr test set_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"eval_cer": 0.1411846070246505,
|
| 4 |
+
"eval_loss": 272.0352478027344,
|
| 5 |
+
"eval_runtime": 22.8227,
|
| 6 |
+
"eval_samples": 890,
|
| 7 |
+
"eval_samples_per_second": 38.996,
|
| 8 |
+
"eval_steps_per_second": 4.907,
|
| 9 |
+
"eval_wer": 0.44996843434343436
|
| 10 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"total_flos": 5.518391525676315e+19,
|
| 4 |
+
"train_loss": 505.59943335262346,
|
| 5 |
+
"train_runtime": 39665.0417,
|
| 6 |
+
"train_samples": 17267,
|
| 7 |
+
"train_samples_per_second": 26.119,
|
| 8 |
+
"train_steps_per_second": 1.634
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1062 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.45059974747474746,
|
| 3 |
+
"best_model_checkpoint": "/scratch/elec/puhe/p/palp3/sami_ASR/base_model_output/wav2vec2-base-sami-22k/outputs/v2/checkpoint-2160",
|
| 4 |
+
"epoch": 60.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 64800,
|
| 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": 72.47711944580078,
|
| 14 |
+
"learning_rate": 3.280864197530864e-05,
|
| 15 |
+
"loss": 3988.287,
|
| 16 |
+
"step": 1080
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 1.0,
|
| 20 |
+
"eval_cer": 0.17220767672775775,
|
| 21 |
+
"eval_loss": 303.8312072753906,
|
| 22 |
+
"eval_runtime": 23.5046,
|
| 23 |
+
"eval_samples_per_second": 37.865,
|
| 24 |
+
"eval_steps_per_second": 4.765,
|
| 25 |
+
"eval_wer": 0.5741792929292929,
|
| 26 |
+
"step": 1080
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 2.0,
|
| 30 |
+
"grad_norm": 781.6192626953125,
|
| 31 |
+
"learning_rate": 6.614197530864198e-05,
|
| 32 |
+
"loss": 674.1274,
|
| 33 |
+
"step": 2160
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 2.0,
|
| 37 |
+
"eval_cer": 0.14150570567603615,
|
| 38 |
+
"eval_loss": 270.2951965332031,
|
| 39 |
+
"eval_runtime": 22.9653,
|
| 40 |
+
"eval_samples_per_second": 38.754,
|
| 41 |
+
"eval_steps_per_second": 4.877,
|
| 42 |
+
"eval_wer": 0.45059974747474746,
|
| 43 |
+
"step": 2160
|
| 44 |
+
},
|
| 45 |
+
{
|
| 46 |
+
"epoch": 3.0,
|
| 47 |
+
"grad_norm": 70.77464294433594,
|
| 48 |
+
"learning_rate": 9.94753086419753e-05,
|
| 49 |
+
"loss": 517.9397,
|
| 50 |
+
"step": 3240
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 3.0,
|
| 54 |
+
"eval_cer": 0.1630440152151361,
|
| 55 |
+
"eval_loss": 339.81317138671875,
|
| 56 |
+
"eval_runtime": 23.0091,
|
| 57 |
+
"eval_samples_per_second": 38.68,
|
| 58 |
+
"eval_steps_per_second": 4.868,
|
| 59 |
+
"eval_wer": 0.4930555555555556,
|
| 60 |
+
"step": 3240
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 4.0,
|
| 64 |
+
"grad_norm": 499.06243896484375,
|
| 65 |
+
"learning_rate": 0.00013280864197530865,
|
| 66 |
+
"loss": 478.4345,
|
| 67 |
+
"step": 4320
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 4.0,
|
| 71 |
+
"eval_cer": 0.16909548979894284,
|
| 72 |
+
"eval_loss": 292.6837463378906,
|
| 73 |
+
"eval_runtime": 22.9042,
|
| 74 |
+
"eval_samples_per_second": 38.857,
|
| 75 |
+
"eval_steps_per_second": 4.89,
|
| 76 |
+
"eval_wer": 0.4895833333333333,
|
| 77 |
+
"step": 4320
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 5.0,
|
| 81 |
+
"grad_norm": 550.7621459960938,
|
| 82 |
+
"learning_rate": 0.00016614197530864197,
|
| 83 |
+
"loss": 476.3436,
|
| 84 |
+
"step": 5400
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 5.0,
|
| 88 |
+
"eval_cer": 0.17993874425727413,
|
| 89 |
+
"eval_loss": 331.53070068359375,
|
| 90 |
+
"eval_runtime": 22.969,
|
| 91 |
+
"eval_samples_per_second": 38.748,
|
| 92 |
+
"eval_steps_per_second": 4.876,
|
| 93 |
+
"eval_wer": 0.509469696969697,
|
| 94 |
+
"step": 5400
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 6.0,
|
| 98 |
+
"grad_norm": 407.94061279296875,
|
| 99 |
+
"learning_rate": 0.0001994753086419753,
|
| 100 |
+
"loss": 507.3865,
|
| 101 |
+
"step": 6480
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 6.0,
|
| 105 |
+
"eval_cer": 0.20058785753099836,
|
| 106 |
+
"eval_loss": 359.2405090332031,
|
| 107 |
+
"eval_runtime": 23.949,
|
| 108 |
+
"eval_samples_per_second": 37.162,
|
| 109 |
+
"eval_steps_per_second": 4.677,
|
| 110 |
+
"eval_wer": 0.5620265151515151,
|
| 111 |
+
"step": 6480
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 7.0,
|
| 115 |
+
"grad_norm": 248.67550659179688,
|
| 116 |
+
"learning_rate": 0.00023280864197530864,
|
| 117 |
+
"loss": 511.676,
|
| 118 |
+
"step": 7560
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 7.0,
|
| 122 |
+
"eval_cer": 0.2163710912414168,
|
| 123 |
+
"eval_loss": 406.4503479003906,
|
| 124 |
+
"eval_runtime": 25.894,
|
| 125 |
+
"eval_samples_per_second": 34.371,
|
| 126 |
+
"eval_steps_per_second": 4.325,
|
| 127 |
+
"eval_wer": 0.5880681818181818,
|
| 128 |
+
"step": 7560
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 8.0,
|
| 132 |
+
"grad_norm": 398.6786193847656,
|
| 133 |
+
"learning_rate": 0.00026614197530864197,
|
| 134 |
+
"loss": 546.8947,
|
| 135 |
+
"step": 8640
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 8.0,
|
| 139 |
+
"eval_cer": 0.2145185990218841,
|
| 140 |
+
"eval_loss": 367.28216552734375,
|
| 141 |
+
"eval_runtime": 24.1336,
|
| 142 |
+
"eval_samples_per_second": 36.878,
|
| 143 |
+
"eval_steps_per_second": 4.641,
|
| 144 |
+
"eval_wer": 0.5834911616161617,
|
| 145 |
+
"step": 8640
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 9.0,
|
| 149 |
+
"grad_norm": 268.1892395019531,
|
| 150 |
+
"learning_rate": 0.0002994753086419753,
|
| 151 |
+
"loss": 578.7039,
|
| 152 |
+
"step": 9720
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 9.0,
|
| 156 |
+
"eval_cer": 0.25860791384676185,
|
| 157 |
+
"eval_loss": 441.93701171875,
|
| 158 |
+
"eval_runtime": 24.1196,
|
| 159 |
+
"eval_samples_per_second": 36.899,
|
| 160 |
+
"eval_steps_per_second": 4.644,
|
| 161 |
+
"eval_wer": 0.6559343434343434,
|
| 162 |
+
"step": 9720
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 10.0,
|
| 166 |
+
"grad_norm": 191.7284393310547,
|
| 167 |
+
"learning_rate": 0.0003328086419753086,
|
| 168 |
+
"loss": 610.5788,
|
| 169 |
+
"step": 10800
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 10.0,
|
| 173 |
+
"eval_cer": 0.2608803043027219,
|
| 174 |
+
"eval_loss": 454.6888427734375,
|
| 175 |
+
"eval_runtime": 24.3544,
|
| 176 |
+
"eval_samples_per_second": 36.544,
|
| 177 |
+
"eval_steps_per_second": 4.599,
|
| 178 |
+
"eval_wer": 0.6856060606060606,
|
| 179 |
+
"step": 10800
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 11.0,
|
| 183 |
+
"grad_norm": 265.8825988769531,
|
| 184 |
+
"learning_rate": 0.00036614197530864196,
|
| 185 |
+
"loss": 658.402,
|
| 186 |
+
"step": 11880
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 11.0,
|
| 190 |
+
"eval_cer": 0.30240082991651435,
|
| 191 |
+
"eval_loss": 472.7101745605469,
|
| 192 |
+
"eval_runtime": 24.2238,
|
| 193 |
+
"eval_samples_per_second": 36.741,
|
| 194 |
+
"eval_steps_per_second": 4.624,
|
| 195 |
+
"eval_wer": 0.7457386363636364,
|
| 196 |
+
"step": 11880
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 12.0,
|
| 200 |
+
"grad_norm": 260.43603515625,
|
| 201 |
+
"learning_rate": 0.00039944444444444446,
|
| 202 |
+
"loss": 692.8005,
|
| 203 |
+
"step": 12960
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 12.0,
|
| 207 |
+
"eval_cer": 0.2891369856246604,
|
| 208 |
+
"eval_loss": 474.2547607421875,
|
| 209 |
+
"eval_runtime": 33.8391,
|
| 210 |
+
"eval_samples_per_second": 26.301,
|
| 211 |
+
"eval_steps_per_second": 3.31,
|
| 212 |
+
"eval_wer": 0.6993371212121212,
|
| 213 |
+
"step": 12960
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 13.0,
|
| 217 |
+
"grad_norm": 305.93817138671875,
|
| 218 |
+
"learning_rate": 0.00043277777777777775,
|
| 219 |
+
"loss": 726.6378,
|
| 220 |
+
"step": 14040
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 13.0,
|
| 224 |
+
"eval_cer": 0.2942251642543101,
|
| 225 |
+
"eval_loss": 476.57232666015625,
|
| 226 |
+
"eval_runtime": 25.299,
|
| 227 |
+
"eval_samples_per_second": 35.179,
|
| 228 |
+
"eval_steps_per_second": 4.427,
|
| 229 |
+
"eval_wer": 0.7154356060606061,
|
| 230 |
+
"step": 14040
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 14.0,
|
| 234 |
+
"grad_norm": 124.7826919555664,
|
| 235 |
+
"learning_rate": 0.00046608024691358025,
|
| 236 |
+
"loss": 775.3879,
|
| 237 |
+
"step": 15120
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 14.0,
|
| 241 |
+
"eval_cer": 0.2997579410166477,
|
| 242 |
+
"eval_loss": 469.1170959472656,
|
| 243 |
+
"eval_runtime": 24.1494,
|
| 244 |
+
"eval_samples_per_second": 36.854,
|
| 245 |
+
"eval_steps_per_second": 4.638,
|
| 246 |
+
"eval_wer": 0.7359532828282829,
|
| 247 |
+
"step": 15120
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 15.0,
|
| 251 |
+
"grad_norm": 18.523788452148438,
|
| 252 |
+
"learning_rate": 0.0004994135802469135,
|
| 253 |
+
"loss": 803.9686,
|
| 254 |
+
"step": 16200
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 15.0,
|
| 258 |
+
"eval_cer": 0.30311712690806697,
|
| 259 |
+
"eval_loss": 503.41363525390625,
|
| 260 |
+
"eval_runtime": 24.0216,
|
| 261 |
+
"eval_samples_per_second": 37.05,
|
| 262 |
+
"eval_steps_per_second": 4.662,
|
| 263 |
+
"eval_wer": 0.7706755050505051,
|
| 264 |
+
"step": 16200
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 16.0,
|
| 268 |
+
"grad_norm": 951.2265014648438,
|
| 269 |
+
"learning_rate": 0.0004891049382716049,
|
| 270 |
+
"loss": 818.0579,
|
| 271 |
+
"step": 17280
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 16.0,
|
| 275 |
+
"eval_cer": 0.31324408437484563,
|
| 276 |
+
"eval_loss": 544.9780883789062,
|
| 277 |
+
"eval_runtime": 24.4651,
|
| 278 |
+
"eval_samples_per_second": 36.378,
|
| 279 |
+
"eval_steps_per_second": 4.578,
|
| 280 |
+
"eval_wer": 0.7586805555555556,
|
| 281 |
+
"step": 17280
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 17.0,
|
| 285 |
+
"grad_norm": 137.27139282226562,
|
| 286 |
+
"learning_rate": 0.0004779938271604938,
|
| 287 |
+
"loss": 808.2149,
|
| 288 |
+
"step": 18360
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 17.0,
|
| 292 |
+
"eval_cer": 0.30148693375487823,
|
| 293 |
+
"eval_loss": 493.0830078125,
|
| 294 |
+
"eval_runtime": 23.7982,
|
| 295 |
+
"eval_samples_per_second": 37.398,
|
| 296 |
+
"eval_steps_per_second": 4.706,
|
| 297 |
+
"eval_wer": 0.7395833333333334,
|
| 298 |
+
"step": 18360
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 18.0,
|
| 302 |
+
"grad_norm": 212.45159912109375,
|
| 303 |
+
"learning_rate": 0.00046689300411522635,
|
| 304 |
+
"loss": 767.5317,
|
| 305 |
+
"step": 19440
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 18.0,
|
| 309 |
+
"eval_cer": 0.3046238205799536,
|
| 310 |
+
"eval_loss": 527.8341064453125,
|
| 311 |
+
"eval_runtime": 25.0889,
|
| 312 |
+
"eval_samples_per_second": 35.474,
|
| 313 |
+
"eval_steps_per_second": 4.464,
|
| 314 |
+
"eval_wer": 0.7296401515151515,
|
| 315 |
+
"step": 19440
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 19.0,
|
| 319 |
+
"grad_norm": 351.9056701660156,
|
| 320 |
+
"learning_rate": 0.0004557818930041152,
|
| 321 |
+
"loss": 739.8194,
|
| 322 |
+
"step": 20520
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 19.0,
|
| 326 |
+
"eval_cer": 0.30847700439658154,
|
| 327 |
+
"eval_loss": 500.3179016113281,
|
| 328 |
+
"eval_runtime": 24.165,
|
| 329 |
+
"eval_samples_per_second": 36.83,
|
| 330 |
+
"eval_steps_per_second": 4.635,
|
| 331 |
+
"eval_wer": 0.7558396464646465,
|
| 332 |
+
"step": 20520
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 20.0,
|
| 336 |
+
"grad_norm": 279.2209167480469,
|
| 337 |
+
"learning_rate": 0.00044467078189300416,
|
| 338 |
+
"loss": 716.691,
|
| 339 |
+
"step": 21600
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 20.0,
|
| 343 |
+
"eval_cer": 0.2984241466185842,
|
| 344 |
+
"eval_loss": 545.5074462890625,
|
| 345 |
+
"eval_runtime": 23.6519,
|
| 346 |
+
"eval_samples_per_second": 37.629,
|
| 347 |
+
"eval_steps_per_second": 4.735,
|
| 348 |
+
"eval_wer": 0.7234848484848485,
|
| 349 |
+
"step": 21600
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 21.0,
|
| 353 |
+
"grad_norm": 1185.030029296875,
|
| 354 |
+
"learning_rate": 0.00043356995884773664,
|
| 355 |
+
"loss": 682.661,
|
| 356 |
+
"step": 22680
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 21.0,
|
| 360 |
+
"eval_cer": 0.29002618189003604,
|
| 361 |
+
"eval_loss": 516.1239013671875,
|
| 362 |
+
"eval_runtime": 23.2998,
|
| 363 |
+
"eval_samples_per_second": 38.198,
|
| 364 |
+
"eval_steps_per_second": 4.807,
|
| 365 |
+
"eval_wer": 0.751104797979798,
|
| 366 |
+
"step": 22680
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 22.0,
|
| 370 |
+
"grad_norm": 240.5199737548828,
|
| 371 |
+
"learning_rate": 0.00042245884773662554,
|
| 372 |
+
"loss": 657.0491,
|
| 373 |
+
"step": 23760
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 22.0,
|
| 377 |
+
"eval_cer": 0.2775774341747765,
|
| 378 |
+
"eval_loss": 549.7003784179688,
|
| 379 |
+
"eval_runtime": 24.7473,
|
| 380 |
+
"eval_samples_per_second": 35.964,
|
| 381 |
+
"eval_steps_per_second": 4.526,
|
| 382 |
+
"eval_wer": 0.6968118686868687,
|
| 383 |
+
"step": 23760
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 23.0,
|
| 387 |
+
"grad_norm": 62.15650939941406,
|
| 388 |
+
"learning_rate": 0.0004113477366255144,
|
| 389 |
+
"loss": 629.1355,
|
| 390 |
+
"step": 24840
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 23.0,
|
| 394 |
+
"eval_cer": 0.28081312058489355,
|
| 395 |
+
"eval_loss": 500.3792724609375,
|
| 396 |
+
"eval_runtime": 23.2668,
|
| 397 |
+
"eval_samples_per_second": 38.252,
|
| 398 |
+
"eval_steps_per_second": 4.814,
|
| 399 |
+
"eval_wer": 0.6974431818181818,
|
| 400 |
+
"step": 24840
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 24.0,
|
| 404 |
+
"grad_norm": 132.52377319335938,
|
| 405 |
+
"learning_rate": 0.00040023662551440334,
|
| 406 |
+
"loss": 607.2812,
|
| 407 |
+
"step": 25920
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 24.0,
|
| 411 |
+
"eval_cer": 0.26996986612656226,
|
| 412 |
+
"eval_loss": 528.1495971679688,
|
| 413 |
+
"eval_runtime": 24.3686,
|
| 414 |
+
"eval_samples_per_second": 36.522,
|
| 415 |
+
"eval_steps_per_second": 4.596,
|
| 416 |
+
"eval_wer": 0.695864898989899,
|
| 417 |
+
"step": 25920
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 25.0,
|
| 421 |
+
"grad_norm": 199.2706756591797,
|
| 422 |
+
"learning_rate": 0.0003891255144032922,
|
| 423 |
+
"loss": 595.4605,
|
| 424 |
+
"step": 27000
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 25.0,
|
| 428 |
+
"eval_cer": 0.2833819097959789,
|
| 429 |
+
"eval_loss": 495.3538513183594,
|
| 430 |
+
"eval_runtime": 23.1173,
|
| 431 |
+
"eval_samples_per_second": 38.499,
|
| 432 |
+
"eval_steps_per_second": 4.845,
|
| 433 |
+
"eval_wer": 0.7015467171717171,
|
| 434 |
+
"step": 27000
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 26.0,
|
| 438 |
+
"grad_norm": 902.6602172851562,
|
| 439 |
+
"learning_rate": 0.0003780246913580247,
|
| 440 |
+
"loss": 555.9978,
|
| 441 |
+
"step": 28080
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 26.0,
|
| 445 |
+
"eval_cer": 0.2781702316850269,
|
| 446 |
+
"eval_loss": 500.2841491699219,
|
| 447 |
+
"eval_runtime": 23.4702,
|
| 448 |
+
"eval_samples_per_second": 37.92,
|
| 449 |
+
"eval_steps_per_second": 4.772,
|
| 450 |
+
"eval_wer": 0.7070707070707071,
|
| 451 |
+
"step": 28080
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 27.0,
|
| 455 |
+
"grad_norm": 215.98854064941406,
|
| 456 |
+
"learning_rate": 0.0003669135802469136,
|
| 457 |
+
"loss": 544.9409,
|
| 458 |
+
"step": 29160
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 27.0,
|
| 462 |
+
"eval_cer": 0.2840488069950106,
|
| 463 |
+
"eval_loss": 476.8066711425781,
|
| 464 |
+
"eval_runtime": 23.3103,
|
| 465 |
+
"eval_samples_per_second": 38.181,
|
| 466 |
+
"eval_steps_per_second": 4.805,
|
| 467 |
+
"eval_wer": 0.7075441919191919,
|
| 468 |
+
"step": 29160
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 28.0,
|
| 472 |
+
"grad_norm": 149.82615661621094,
|
| 473 |
+
"learning_rate": 0.0003558024691358025,
|
| 474 |
+
"loss": 517.4491,
|
| 475 |
+
"step": 30240
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 28.0,
|
| 479 |
+
"eval_cer": 0.2703403645704688,
|
| 480 |
+
"eval_loss": 513.64892578125,
|
| 481 |
+
"eval_runtime": 23.5863,
|
| 482 |
+
"eval_samples_per_second": 37.734,
|
| 483 |
+
"eval_steps_per_second": 4.749,
|
| 484 |
+
"eval_wer": 0.6824494949494949,
|
| 485 |
+
"step": 30240
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 29.0,
|
| 489 |
+
"grad_norm": 158.89735412597656,
|
| 490 |
+
"learning_rate": 0.000344701646090535,
|
| 491 |
+
"loss": 502.3091,
|
| 492 |
+
"step": 31320
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 29.0,
|
| 496 |
+
"eval_cer": 0.26236229807834804,
|
| 497 |
+
"eval_loss": 450.82098388671875,
|
| 498 |
+
"eval_runtime": 25.6603,
|
| 499 |
+
"eval_samples_per_second": 34.684,
|
| 500 |
+
"eval_steps_per_second": 4.365,
|
| 501 |
+
"eval_wer": 0.6879734848484849,
|
| 502 |
+
"step": 31320
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"epoch": 30.0,
|
| 506 |
+
"grad_norm": 51.25532913208008,
|
| 507 |
+
"learning_rate": 0.00033359053497942386,
|
| 508 |
+
"loss": 477.324,
|
| 509 |
+
"step": 32400
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 30.0,
|
| 513 |
+
"eval_cer": 0.2615719013980141,
|
| 514 |
+
"eval_loss": 469.6161804199219,
|
| 515 |
+
"eval_runtime": 23.2095,
|
| 516 |
+
"eval_samples_per_second": 38.346,
|
| 517 |
+
"eval_steps_per_second": 4.826,
|
| 518 |
+
"eval_wer": 0.65625,
|
| 519 |
+
"step": 32400
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 31.0,
|
| 523 |
+
"grad_norm": 70.39710235595703,
|
| 524 |
+
"learning_rate": 0.00032247942386831276,
|
| 525 |
+
"loss": 461.2854,
|
| 526 |
+
"step": 33480
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 31.0,
|
| 530 |
+
"eval_cer": 0.24838215679494147,
|
| 531 |
+
"eval_loss": 480.2809753417969,
|
| 532 |
+
"eval_runtime": 25.0487,
|
| 533 |
+
"eval_samples_per_second": 35.531,
|
| 534 |
+
"eval_steps_per_second": 4.471,
|
| 535 |
+
"eval_wer": 0.6639835858585859,
|
| 536 |
+
"step": 33480
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 32.0,
|
| 540 |
+
"grad_norm": 117.32414245605469,
|
| 541 |
+
"learning_rate": 0.00031136831275720167,
|
| 542 |
+
"loss": 452.682,
|
| 543 |
+
"step": 34560
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 32.0,
|
| 547 |
+
"eval_cer": 0.26517808625203776,
|
| 548 |
+
"eval_loss": 477.9761962890625,
|
| 549 |
+
"eval_runtime": 23.3634,
|
| 550 |
+
"eval_samples_per_second": 38.094,
|
| 551 |
+
"eval_steps_per_second": 4.794,
|
| 552 |
+
"eval_wer": 0.6638257575757576,
|
| 553 |
+
"step": 34560
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 33.0,
|
| 557 |
+
"grad_norm": 3.261843204498291,
|
| 558 |
+
"learning_rate": 0.00030025720164609057,
|
| 559 |
+
"loss": 424.353,
|
| 560 |
+
"step": 35640
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 33.0,
|
| 564 |
+
"eval_cer": 0.25198834164896505,
|
| 565 |
+
"eval_loss": 444.65106201171875,
|
| 566 |
+
"eval_runtime": 23.2973,
|
| 567 |
+
"eval_samples_per_second": 38.202,
|
| 568 |
+
"eval_steps_per_second": 4.807,
|
| 569 |
+
"eval_wer": 0.6532512626262627,
|
| 570 |
+
"step": 35640
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 34.0,
|
| 574 |
+
"grad_norm": 66.93054962158203,
|
| 575 |
+
"learning_rate": 0.00028915637860082305,
|
| 576 |
+
"loss": 417.6179,
|
| 577 |
+
"step": 36720
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 34.0,
|
| 581 |
+
"eval_cer": 0.25255643926295507,
|
| 582 |
+
"eval_loss": 412.5328674316406,
|
| 583 |
+
"eval_runtime": 23.4814,
|
| 584 |
+
"eval_samples_per_second": 37.902,
|
| 585 |
+
"eval_steps_per_second": 4.77,
|
| 586 |
+
"eval_wer": 0.6504103535353535,
|
| 587 |
+
"step": 36720
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 35.0,
|
| 591 |
+
"grad_norm": 53.988014221191406,
|
| 592 |
+
"learning_rate": 0.00027804526748971195,
|
| 593 |
+
"loss": 389.705,
|
| 594 |
+
"step": 37800
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 35.0,
|
| 598 |
+
"eval_cer": 0.26332559403250505,
|
| 599 |
+
"eval_loss": 485.3770446777344,
|
| 600 |
+
"eval_runtime": 24.4441,
|
| 601 |
+
"eval_samples_per_second": 36.41,
|
| 602 |
+
"eval_steps_per_second": 4.582,
|
| 603 |
+
"eval_wer": 0.6744002525252525,
|
| 604 |
+
"step": 37800
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 36.0,
|
| 608 |
+
"grad_norm": 106.55242156982422,
|
| 609 |
+
"learning_rate": 0.00026694444444444443,
|
| 610 |
+
"loss": 375.7767,
|
| 611 |
+
"step": 38880
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 36.0,
|
| 615 |
+
"eval_cer": 0.2663883811687991,
|
| 616 |
+
"eval_loss": 467.3829345703125,
|
| 617 |
+
"eval_runtime": 25.0973,
|
| 618 |
+
"eval_samples_per_second": 35.462,
|
| 619 |
+
"eval_steps_per_second": 4.463,
|
| 620 |
+
"eval_wer": 0.6474116161616161,
|
| 621 |
+
"step": 38880
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 37.0,
|
| 625 |
+
"grad_norm": 42.98028564453125,
|
| 626 |
+
"learning_rate": 0.00025583333333333334,
|
| 627 |
+
"loss": 361.8829,
|
| 628 |
+
"step": 39960
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"epoch": 37.0,
|
| 632 |
+
"eval_cer": 0.2516672429975794,
|
| 633 |
+
"eval_loss": 469.9674377441406,
|
| 634 |
+
"eval_runtime": 23.3905,
|
| 635 |
+
"eval_samples_per_second": 38.05,
|
| 636 |
+
"eval_steps_per_second": 4.788,
|
| 637 |
+
"eval_wer": 0.631155303030303,
|
| 638 |
+
"step": 39960
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 38.0,
|
| 642 |
+
"grad_norm": 509.82037353515625,
|
| 643 |
+
"learning_rate": 0.00024472222222222224,
|
| 644 |
+
"loss": 352.311,
|
| 645 |
+
"step": 41040
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 38.0,
|
| 649 |
+
"eval_cer": 0.2544830311712691,
|
| 650 |
+
"eval_loss": 457.0285339355469,
|
| 651 |
+
"eval_runtime": 23.3717,
|
| 652 |
+
"eval_samples_per_second": 38.08,
|
| 653 |
+
"eval_steps_per_second": 4.792,
|
| 654 |
+
"eval_wer": 0.6494633838383839,
|
| 655 |
+
"step": 41040
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 39.0,
|
| 659 |
+
"grad_norm": 173.2177276611328,
|
| 660 |
+
"learning_rate": 0.00023363168724279835,
|
| 661 |
+
"loss": 340.1846,
|
| 662 |
+
"step": 42120
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 39.0,
|
| 666 |
+
"eval_cer": 0.24615916613150224,
|
| 667 |
+
"eval_loss": 463.1924743652344,
|
| 668 |
+
"eval_runtime": 23.316,
|
| 669 |
+
"eval_samples_per_second": 38.171,
|
| 670 |
+
"eval_steps_per_second": 4.804,
|
| 671 |
+
"eval_wer": 0.634469696969697,
|
| 672 |
+
"step": 42120
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 40.0,
|
| 676 |
+
"grad_norm": 435.5245361328125,
|
| 677 |
+
"learning_rate": 0.00022252057613168726,
|
| 678 |
+
"loss": 323.3272,
|
| 679 |
+
"step": 43200
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 40.0,
|
| 683 |
+
"eval_cer": 0.23939139455614286,
|
| 684 |
+
"eval_loss": 421.072509765625,
|
| 685 |
+
"eval_runtime": 23.2659,
|
| 686 |
+
"eval_samples_per_second": 38.253,
|
| 687 |
+
"eval_steps_per_second": 4.814,
|
| 688 |
+
"eval_wer": 0.6171085858585859,
|
| 689 |
+
"step": 43200
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 41.0,
|
| 693 |
+
"grad_norm": 286.759765625,
|
| 694 |
+
"learning_rate": 0.00021141975308641977,
|
| 695 |
+
"loss": 312.6201,
|
| 696 |
+
"step": 44280
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 41.0,
|
| 700 |
+
"eval_cer": 0.23835399891320455,
|
| 701 |
+
"eval_loss": 443.36468505859375,
|
| 702 |
+
"eval_runtime": 23.0069,
|
| 703 |
+
"eval_samples_per_second": 38.684,
|
| 704 |
+
"eval_steps_per_second": 4.868,
|
| 705 |
+
"eval_wer": 0.6201073232323232,
|
| 706 |
+
"step": 44280
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 42.0,
|
| 710 |
+
"grad_norm": 305.650390625,
|
| 711 |
+
"learning_rate": 0.00020030864197530864,
|
| 712 |
+
"loss": 301.6251,
|
| 713 |
+
"step": 45360
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 42.0,
|
| 717 |
+
"eval_cer": 0.23499481302178532,
|
| 718 |
+
"eval_loss": 429.3775634765625,
|
| 719 |
+
"eval_runtime": 25.784,
|
| 720 |
+
"eval_samples_per_second": 34.517,
|
| 721 |
+
"eval_steps_per_second": 4.344,
|
| 722 |
+
"eval_wer": 0.610479797979798,
|
| 723 |
+
"step": 45360
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 43.0,
|
| 727 |
+
"grad_norm": 1.2366968393325806,
|
| 728 |
+
"learning_rate": 0.00018919753086419752,
|
| 729 |
+
"loss": 284.7902,
|
| 730 |
+
"step": 46440
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 43.0,
|
| 734 |
+
"eval_cer": 0.23210492515931433,
|
| 735 |
+
"eval_loss": 466.15533447265625,
|
| 736 |
+
"eval_runtime": 22.7914,
|
| 737 |
+
"eval_samples_per_second": 39.05,
|
| 738 |
+
"eval_steps_per_second": 4.914,
|
| 739 |
+
"eval_wer": 0.602114898989899,
|
| 740 |
+
"step": 46440
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 44.0,
|
| 744 |
+
"grad_norm": 194.98329162597656,
|
| 745 |
+
"learning_rate": 0.00017809670781893002,
|
| 746 |
+
"loss": 279.8459,
|
| 747 |
+
"step": 47520
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 44.0,
|
| 751 |
+
"eval_cer": 0.23185792619670997,
|
| 752 |
+
"eval_loss": 487.2147521972656,
|
| 753 |
+
"eval_runtime": 23.4301,
|
| 754 |
+
"eval_samples_per_second": 37.985,
|
| 755 |
+
"eval_steps_per_second": 4.78,
|
| 756 |
+
"eval_wer": 0.6161616161616161,
|
| 757 |
+
"step": 47520
|
| 758 |
+
},
|
| 759 |
+
{
|
| 760 |
+
"epoch": 45.0,
|
| 761 |
+
"grad_norm": 80.29993438720703,
|
| 762 |
+
"learning_rate": 0.00016698559670781893,
|
| 763 |
+
"loss": 260.5616,
|
| 764 |
+
"step": 48600
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 45.0,
|
| 768 |
+
"eval_cer": 0.2306476312799486,
|
| 769 |
+
"eval_loss": 445.4757080078125,
|
| 770 |
+
"eval_runtime": 23.0166,
|
| 771 |
+
"eval_samples_per_second": 38.668,
|
| 772 |
+
"eval_steps_per_second": 4.866,
|
| 773 |
+
"eval_wer": 0.6022727272727273,
|
| 774 |
+
"step": 48600
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 46.0,
|
| 778 |
+
"grad_norm": 64.79682159423828,
|
| 779 |
+
"learning_rate": 0.00015587448559670783,
|
| 780 |
+
"loss": 254.3347,
|
| 781 |
+
"step": 49680
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 46.0,
|
| 785 |
+
"eval_cer": 0.2392184952823198,
|
| 786 |
+
"eval_loss": 439.6965026855469,
|
| 787 |
+
"eval_runtime": 22.7648,
|
| 788 |
+
"eval_samples_per_second": 39.095,
|
| 789 |
+
"eval_steps_per_second": 4.92,
|
| 790 |
+
"eval_wer": 0.6054292929292929,
|
| 791 |
+
"step": 49680
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 47.0,
|
| 795 |
+
"grad_norm": 124.6217269897461,
|
| 796 |
+
"learning_rate": 0.0001447633744855967,
|
| 797 |
+
"loss": 244.043,
|
| 798 |
+
"step": 50760
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 47.0,
|
| 802 |
+
"eval_cer": 0.23166032702662648,
|
| 803 |
+
"eval_loss": 459.5867919921875,
|
| 804 |
+
"eval_runtime": 22.7729,
|
| 805 |
+
"eval_samples_per_second": 39.081,
|
| 806 |
+
"eval_steps_per_second": 4.918,
|
| 807 |
+
"eval_wer": 0.5885416666666666,
|
| 808 |
+
"step": 50760
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 48.0,
|
| 812 |
+
"grad_norm": 3.2691023349761963,
|
| 813 |
+
"learning_rate": 0.0001336522633744856,
|
| 814 |
+
"loss": 227.4755,
|
| 815 |
+
"step": 51840
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 48.0,
|
| 819 |
+
"eval_cer": 0.2307711307612508,
|
| 820 |
+
"eval_loss": 492.8036804199219,
|
| 821 |
+
"eval_runtime": 24.8308,
|
| 822 |
+
"eval_samples_per_second": 35.843,
|
| 823 |
+
"eval_steps_per_second": 4.511,
|
| 824 |
+
"eval_wer": 0.6002209595959596,
|
| 825 |
+
"step": 51840
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 49.0,
|
| 829 |
+
"grad_norm": 4.506575107574463,
|
| 830 |
+
"learning_rate": 0.00012254115226337448,
|
| 831 |
+
"loss": 216.7,
|
| 832 |
+
"step": 52920
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 49.0,
|
| 836 |
+
"eval_cer": 0.2282764412389468,
|
| 837 |
+
"eval_loss": 452.6693115234375,
|
| 838 |
+
"eval_runtime": 23.3003,
|
| 839 |
+
"eval_samples_per_second": 38.197,
|
| 840 |
+
"eval_steps_per_second": 4.807,
|
| 841 |
+
"eval_wer": 0.5934343434343434,
|
| 842 |
+
"step": 52920
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 50.0,
|
| 846 |
+
"grad_norm": 296.6214294433594,
|
| 847 |
+
"learning_rate": 0.00011143004115226338,
|
| 848 |
+
"loss": 211.8976,
|
| 849 |
+
"step": 54000
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 50.0,
|
| 853 |
+
"eval_cer": 0.22879513906041596,
|
| 854 |
+
"eval_loss": 482.38861083984375,
|
| 855 |
+
"eval_runtime": 23.149,
|
| 856 |
+
"eval_samples_per_second": 38.447,
|
| 857 |
+
"eval_steps_per_second": 4.838,
|
| 858 |
+
"eval_wer": 0.5946969696969697,
|
| 859 |
+
"step": 54000
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 51.0,
|
| 863 |
+
"grad_norm": 4.26457405090332,
|
| 864 |
+
"learning_rate": 0.00010031893004115227,
|
| 865 |
+
"loss": 202.0287,
|
| 866 |
+
"step": 55080
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 51.0,
|
| 870 |
+
"eval_cer": 0.2352665118806501,
|
| 871 |
+
"eval_loss": 475.8258361816406,
|
| 872 |
+
"eval_runtime": 22.9599,
|
| 873 |
+
"eval_samples_per_second": 38.763,
|
| 874 |
+
"eval_steps_per_second": 4.878,
|
| 875 |
+
"eval_wer": 0.6052714646464646,
|
| 876 |
+
"step": 55080
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 52.0,
|
| 880 |
+
"grad_norm": 203.13275146484375,
|
| 881 |
+
"learning_rate": 8.920781893004115e-05,
|
| 882 |
+
"loss": 186.2731,
|
| 883 |
+
"step": 56160
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 52.0,
|
| 887 |
+
"eval_cer": 0.23109222941263646,
|
| 888 |
+
"eval_loss": 465.3925476074219,
|
| 889 |
+
"eval_runtime": 24.1264,
|
| 890 |
+
"eval_samples_per_second": 36.889,
|
| 891 |
+
"eval_steps_per_second": 4.642,
|
| 892 |
+
"eval_wer": 0.5907512626262627,
|
| 893 |
+
"step": 56160
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 53.0,
|
| 897 |
+
"grad_norm": 295.8645935058594,
|
| 898 |
+
"learning_rate": 7.809670781893003e-05,
|
| 899 |
+
"loss": 187.1888,
|
| 900 |
+
"step": 57240
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 53.0,
|
| 904 |
+
"eval_cer": 0.2247443560737045,
|
| 905 |
+
"eval_loss": 459.6522216796875,
|
| 906 |
+
"eval_runtime": 23.8259,
|
| 907 |
+
"eval_samples_per_second": 37.354,
|
| 908 |
+
"eval_steps_per_second": 4.701,
|
| 909 |
+
"eval_wer": 0.5890151515151515,
|
| 910 |
+
"step": 57240
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 54.0,
|
| 914 |
+
"grad_norm": 2880.46630859375,
|
| 915 |
+
"learning_rate": 6.700617283950618e-05,
|
| 916 |
+
"loss": 179.0453,
|
| 917 |
+
"step": 58320
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 54.0,
|
| 921 |
+
"eval_cer": 0.22434915773353753,
|
| 922 |
+
"eval_loss": 473.7304382324219,
|
| 923 |
+
"eval_runtime": 23.3348,
|
| 924 |
+
"eval_samples_per_second": 38.14,
|
| 925 |
+
"eval_steps_per_second": 4.8,
|
| 926 |
+
"eval_wer": 0.5789141414141414,
|
| 927 |
+
"step": 58320
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 55.0,
|
| 931 |
+
"grad_norm": 42.78361129760742,
|
| 932 |
+
"learning_rate": 5.5895061728395066e-05,
|
| 933 |
+
"loss": 165.2614,
|
| 934 |
+
"step": 59400
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 55.0,
|
| 938 |
+
"eval_cer": 0.22375636022328707,
|
| 939 |
+
"eval_loss": 453.96917724609375,
|
| 940 |
+
"eval_runtime": 26.3542,
|
| 941 |
+
"eval_samples_per_second": 33.771,
|
| 942 |
+
"eval_steps_per_second": 4.25,
|
| 943 |
+
"eval_wer": 0.5787563131313131,
|
| 944 |
+
"step": 59400
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 56.0,
|
| 948 |
+
"grad_norm": 187.145263671875,
|
| 949 |
+
"learning_rate": 4.478395061728395e-05,
|
| 950 |
+
"loss": 160.4416,
|
| 951 |
+
"step": 60480
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 56.0,
|
| 955 |
+
"eval_cer": 0.22118757101220174,
|
| 956 |
+
"eval_loss": 474.8050537109375,
|
| 957 |
+
"eval_runtime": 23.2203,
|
| 958 |
+
"eval_samples_per_second": 38.329,
|
| 959 |
+
"eval_steps_per_second": 4.823,
|
| 960 |
+
"eval_wer": 0.5732323232323232,
|
| 961 |
+
"step": 60480
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 57.0,
|
| 965 |
+
"grad_norm": 83.02323150634766,
|
| 966 |
+
"learning_rate": 3.367283950617284e-05,
|
| 967 |
+
"loss": 153.8781,
|
| 968 |
+
"step": 61560
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 57.0,
|
| 972 |
+
"eval_cer": 0.22015017536926346,
|
| 973 |
+
"eval_loss": 478.45806884765625,
|
| 974 |
+
"eval_runtime": 23.1723,
|
| 975 |
+
"eval_samples_per_second": 38.408,
|
| 976 |
+
"eval_steps_per_second": 4.833,
|
| 977 |
+
"eval_wer": 0.5729166666666666,
|
| 978 |
+
"step": 61560
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 58.0,
|
| 982 |
+
"grad_norm": 0.9642492532730103,
|
| 983 |
+
"learning_rate": 2.2582304526748973e-05,
|
| 984 |
+
"loss": 151.1706,
|
| 985 |
+
"step": 62640
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 58.0,
|
| 989 |
+
"eval_cer": 0.21960677765153386,
|
| 990 |
+
"eval_loss": 467.0158386230469,
|
| 991 |
+
"eval_runtime": 23.0956,
|
| 992 |
+
"eval_samples_per_second": 38.535,
|
| 993 |
+
"eval_steps_per_second": 4.849,
|
| 994 |
+
"eval_wer": 0.5688131313131313,
|
| 995 |
+
"step": 62640
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 59.0,
|
| 999 |
+
"grad_norm": 721.9008178710938,
|
| 1000 |
+
"learning_rate": 1.147119341563786e-05,
|
| 1001 |
+
"loss": 147.0876,
|
| 1002 |
+
"step": 63720
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 59.0,
|
| 1006 |
+
"eval_cer": 0.21600059279751024,
|
| 1007 |
+
"eval_loss": 474.2251892089844,
|
| 1008 |
+
"eval_runtime": 22.9681,
|
| 1009 |
+
"eval_samples_per_second": 38.749,
|
| 1010 |
+
"eval_steps_per_second": 4.876,
|
| 1011 |
+
"eval_wer": 0.5602904040404041,
|
| 1012 |
+
"step": 63720
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 60.0,
|
| 1016 |
+
"grad_norm": 8.912408828735352,
|
| 1017 |
+
"learning_rate": 3.60082304526749e-07,
|
| 1018 |
+
"loss": 143.0797,
|
| 1019 |
+
"step": 64800
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 60.0,
|
| 1023 |
+
"eval_cer": 0.2167909894778442,
|
| 1024 |
+
"eval_loss": 469.5599060058594,
|
| 1025 |
+
"eval_runtime": 23.6755,
|
| 1026 |
+
"eval_samples_per_second": 37.592,
|
| 1027 |
+
"eval_steps_per_second": 4.731,
|
| 1028 |
+
"eval_wer": 0.5640782828282829,
|
| 1029 |
+
"step": 64800
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 60.0,
|
| 1033 |
+
"step": 64800,
|
| 1034 |
+
"total_flos": 5.518391525676315e+19,
|
| 1035 |
+
"train_loss": 505.59943335262346,
|
| 1036 |
+
"train_runtime": 39665.0417,
|
| 1037 |
+
"train_samples_per_second": 26.119,
|
| 1038 |
+
"train_steps_per_second": 1.634
|
| 1039 |
+
}
|
| 1040 |
+
],
|
| 1041 |
+
"logging_steps": 500,
|
| 1042 |
+
"max_steps": 64800,
|
| 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": 5.518391525676315e+19,
|
| 1059 |
+
"train_batch_size": 16,
|
| 1060 |
+
"trial_name": null,
|
| 1061 |
+
"trial_params": null
|
| 1062 |
+
}
|