large_sami_22k_finetuned_pseudolabels
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: 223.1608
|
| 20 |
+
- Wer: 0.3813
|
| 21 |
+
- Cer: 0.1186
|
| 22 |
|
| 23 |
## Model description
|
| 24 |
|
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"eval_cer": 0.11855950205009139,
|
| 4 |
+
"eval_loss": 223.1608123779297,
|
| 5 |
+
"eval_runtime": 25.7763,
|
| 6 |
+
"eval_samples": 890,
|
| 7 |
+
"eval_samples_per_second": 34.528,
|
| 8 |
+
"eval_steps_per_second": 4.345,
|
| 9 |
+
"eval_wer": 0.3813131313131313,
|
| 10 |
+
"total_flos": 1.8440987587856836e+20,
|
| 11 |
+
"train_loss": 306.16538580246913,
|
| 12 |
+
"train_runtime": 47401.1605,
|
| 13 |
+
"train_samples": 17267,
|
| 14 |
+
"train_samples_per_second": 21.856,
|
| 15 |
+
"train_steps_per_second": 1.367
|
| 16 |
+
}
|
eval--1hr test set_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"eval_cer": 0.11855950205009139,
|
| 4 |
+
"eval_loss": 223.1608123779297,
|
| 5 |
+
"eval_runtime": 25.7763,
|
| 6 |
+
"eval_samples": 890,
|
| 7 |
+
"eval_samples_per_second": 34.528,
|
| 8 |
+
"eval_steps_per_second": 4.345,
|
| 9 |
+
"eval_wer": 0.3813131313131313
|
| 10 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"total_flos": 1.8440987587856836e+20,
|
| 4 |
+
"train_loss": 306.16538580246913,
|
| 5 |
+
"train_runtime": 47401.1605,
|
| 6 |
+
"train_samples": 17267,
|
| 7 |
+
"train_samples_per_second": 21.856,
|
| 8 |
+
"train_steps_per_second": 1.367
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1062 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.3817866161616162,
|
| 3 |
+
"best_model_checkpoint": "/scratch/elec/puhe/p/palp3/sami_ASR/large_model_output/large-sami-22k/outputs/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": 843.546142578125,
|
| 14 |
+
"learning_rate": 3.277777777777778e-05,
|
| 15 |
+
"loss": 3985.5875,
|
| 16 |
+
"step": 1080
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 1.0,
|
| 20 |
+
"eval_cer": 0.16660080027663884,
|
| 21 |
+
"eval_loss": 283.68731689453125,
|
| 22 |
+
"eval_runtime": 26.5521,
|
| 23 |
+
"eval_samples_per_second": 33.519,
|
| 24 |
+
"eval_steps_per_second": 4.218,
|
| 25 |
+
"eval_wer": 0.5932765151515151,
|
| 26 |
+
"step": 1080
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 2.0,
|
| 30 |
+
"grad_norm": 152.15969848632812,
|
| 31 |
+
"learning_rate": 6.611111111111111e-05,
|
| 32 |
+
"loss": 665.3671,
|
| 33 |
+
"step": 2160
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 2.0,
|
| 37 |
+
"eval_cer": 0.11860890184261226,
|
| 38 |
+
"eval_loss": 223.17312622070312,
|
| 39 |
+
"eval_runtime": 26.9667,
|
| 40 |
+
"eval_samples_per_second": 33.004,
|
| 41 |
+
"eval_steps_per_second": 4.153,
|
| 42 |
+
"eval_wer": 0.3817866161616162,
|
| 43 |
+
"step": 2160
|
| 44 |
+
},
|
| 45 |
+
{
|
| 46 |
+
"epoch": 3.0,
|
| 47 |
+
"grad_norm": 69.1477279663086,
|
| 48 |
+
"learning_rate": 9.944444444444444e-05,
|
| 49 |
+
"loss": 481.9958,
|
| 50 |
+
"step": 3240
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 3.0,
|
| 54 |
+
"eval_cer": 0.1353801314034481,
|
| 55 |
+
"eval_loss": 223.3375244140625,
|
| 56 |
+
"eval_runtime": 26.3476,
|
| 57 |
+
"eval_samples_per_second": 33.779,
|
| 58 |
+
"eval_steps_per_second": 4.251,
|
| 59 |
+
"eval_wer": 0.4375,
|
| 60 |
+
"step": 3240
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 4.0,
|
| 64 |
+
"grad_norm": 252.46255493164062,
|
| 65 |
+
"learning_rate": 0.00013277777777777778,
|
| 66 |
+
"loss": 442.5208,
|
| 67 |
+
"step": 4320
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 4.0,
|
| 71 |
+
"eval_cer": 0.14442029343476756,
|
| 72 |
+
"eval_loss": 236.09030151367188,
|
| 73 |
+
"eval_runtime": 26.3227,
|
| 74 |
+
"eval_samples_per_second": 33.811,
|
| 75 |
+
"eval_steps_per_second": 4.255,
|
| 76 |
+
"eval_wer": 0.42329545454545453,
|
| 77 |
+
"step": 4320
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 5.0,
|
| 81 |
+
"grad_norm": 23.017539978027344,
|
| 82 |
+
"learning_rate": 0.0001661111111111111,
|
| 83 |
+
"loss": 421.2154,
|
| 84 |
+
"step": 5400
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 5.0,
|
| 88 |
+
"eval_cer": 0.15635034332855802,
|
| 89 |
+
"eval_loss": 304.9386901855469,
|
| 90 |
+
"eval_runtime": 27.9275,
|
| 91 |
+
"eval_samples_per_second": 31.868,
|
| 92 |
+
"eval_steps_per_second": 4.01,
|
| 93 |
+
"eval_wer": 0.46259469696969696,
|
| 94 |
+
"step": 5400
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 6.0,
|
| 98 |
+
"grad_norm": 154.2694549560547,
|
| 99 |
+
"learning_rate": 0.00019944444444444445,
|
| 100 |
+
"loss": 431.8639,
|
| 101 |
+
"step": 6480
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 6.0,
|
| 105 |
+
"eval_cer": 0.17932124685076323,
|
| 106 |
+
"eval_loss": 298.5758056640625,
|
| 107 |
+
"eval_runtime": 26.8301,
|
| 108 |
+
"eval_samples_per_second": 33.172,
|
| 109 |
+
"eval_steps_per_second": 4.174,
|
| 110 |
+
"eval_wer": 0.5198863636363636,
|
| 111 |
+
"step": 6480
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 7.0,
|
| 115 |
+
"grad_norm": 252.90023803710938,
|
| 116 |
+
"learning_rate": 0.0002327777777777778,
|
| 117 |
+
"loss": 444.0917,
|
| 118 |
+
"step": 7560
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 7.0,
|
| 122 |
+
"eval_cer": 0.198809465000247,
|
| 123 |
+
"eval_loss": 347.2078857421875,
|
| 124 |
+
"eval_runtime": 26.7772,
|
| 125 |
+
"eval_samples_per_second": 33.237,
|
| 126 |
+
"eval_steps_per_second": 4.183,
|
| 127 |
+
"eval_wer": 0.55697601010101,
|
| 128 |
+
"step": 7560
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 8.0,
|
| 132 |
+
"grad_norm": 205.64749145507812,
|
| 133 |
+
"learning_rate": 0.0002661111111111111,
|
| 134 |
+
"loss": 451.4575,
|
| 135 |
+
"step": 8640
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 8.0,
|
| 139 |
+
"eval_cer": 0.20920812132589042,
|
| 140 |
+
"eval_loss": 375.0052795410156,
|
| 141 |
+
"eval_runtime": 26.6502,
|
| 142 |
+
"eval_samples_per_second": 33.396,
|
| 143 |
+
"eval_steps_per_second": 4.203,
|
| 144 |
+
"eval_wer": 0.5730744949494949,
|
| 145 |
+
"step": 8640
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 9.0,
|
| 149 |
+
"grad_norm": 186.66598510742188,
|
| 150 |
+
"learning_rate": 0.00029944444444444446,
|
| 151 |
+
"loss": 445.6943,
|
| 152 |
+
"step": 9720
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 9.0,
|
| 156 |
+
"eval_cer": 0.24057698957664378,
|
| 157 |
+
"eval_loss": 360.6094665527344,
|
| 158 |
+
"eval_runtime": 26.557,
|
| 159 |
+
"eval_samples_per_second": 33.513,
|
| 160 |
+
"eval_steps_per_second": 4.217,
|
| 161 |
+
"eval_wer": 0.6164772727272727,
|
| 162 |
+
"step": 9720
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 10.0,
|
| 166 |
+
"grad_norm": 148.4665985107422,
|
| 167 |
+
"learning_rate": 0.0003327777777777778,
|
| 168 |
+
"loss": 457.5267,
|
| 169 |
+
"step": 10800
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 10.0,
|
| 173 |
+
"eval_cer": 0.252210640715309,
|
| 174 |
+
"eval_loss": 413.63214111328125,
|
| 175 |
+
"eval_runtime": 27.5043,
|
| 176 |
+
"eval_samples_per_second": 32.359,
|
| 177 |
+
"eval_steps_per_second": 4.072,
|
| 178 |
+
"eval_wer": 0.6510416666666666,
|
| 179 |
+
"step": 10800
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 11.0,
|
| 183 |
+
"grad_norm": 83.87371063232422,
|
| 184 |
+
"learning_rate": 0.00036608024691358026,
|
| 185 |
+
"loss": 468.0921,
|
| 186 |
+
"step": 11880
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 11.0,
|
| 190 |
+
"eval_cer": 0.2446771723558761,
|
| 191 |
+
"eval_loss": 405.5586853027344,
|
| 192 |
+
"eval_runtime": 26.521,
|
| 193 |
+
"eval_samples_per_second": 33.558,
|
| 194 |
+
"eval_steps_per_second": 4.223,
|
| 195 |
+
"eval_wer": 0.6422032828282829,
|
| 196 |
+
"step": 11880
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 12.0,
|
| 200 |
+
"grad_norm": 115.64622497558594,
|
| 201 |
+
"learning_rate": 0.0003994135802469136,
|
| 202 |
+
"loss": 469.1014,
|
| 203 |
+
"step": 12960
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 12.0,
|
| 207 |
+
"eval_cer": 0.24956775181544238,
|
| 208 |
+
"eval_loss": 423.7670593261719,
|
| 209 |
+
"eval_runtime": 27.3392,
|
| 210 |
+
"eval_samples_per_second": 32.554,
|
| 211 |
+
"eval_steps_per_second": 4.097,
|
| 212 |
+
"eval_wer": 0.6369949494949495,
|
| 213 |
+
"step": 12960
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 13.0,
|
| 217 |
+
"grad_norm": 154.60157775878906,
|
| 218 |
+
"learning_rate": 0.0004327469135802469,
|
| 219 |
+
"loss": 472.7028,
|
| 220 |
+
"step": 14040
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 13.0,
|
| 224 |
+
"eval_cer": 0.25208714123400683,
|
| 225 |
+
"eval_loss": 476.42498779296875,
|
| 226 |
+
"eval_runtime": 26.8242,
|
| 227 |
+
"eval_samples_per_second": 33.179,
|
| 228 |
+
"eval_steps_per_second": 4.175,
|
| 229 |
+
"eval_wer": 0.6583017676767676,
|
| 230 |
+
"step": 14040
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 14.0,
|
| 234 |
+
"grad_norm": 35.594207763671875,
|
| 235 |
+
"learning_rate": 0.00046608024691358025,
|
| 236 |
+
"loss": 477.4817,
|
| 237 |
+
"step": 15120
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 14.0,
|
| 241 |
+
"eval_cer": 0.2522847404040903,
|
| 242 |
+
"eval_loss": 431.7320861816406,
|
| 243 |
+
"eval_runtime": 25.8162,
|
| 244 |
+
"eval_samples_per_second": 34.474,
|
| 245 |
+
"eval_steps_per_second": 4.338,
|
| 246 |
+
"eval_wer": 0.6496212121212122,
|
| 247 |
+
"step": 15120
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 15.0,
|
| 251 |
+
"grad_norm": 5.936054706573486,
|
| 252 |
+
"learning_rate": 0.0004994135802469135,
|
| 253 |
+
"loss": 476.1046,
|
| 254 |
+
"step": 16200
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 15.0,
|
| 258 |
+
"eval_cer": 0.2686113718322383,
|
| 259 |
+
"eval_loss": 456.8628234863281,
|
| 260 |
+
"eval_runtime": 27.4231,
|
| 261 |
+
"eval_samples_per_second": 32.454,
|
| 262 |
+
"eval_steps_per_second": 4.084,
|
| 263 |
+
"eval_wer": 0.6734532828282829,
|
| 264 |
+
"step": 16200
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 16.0,
|
| 268 |
+
"grad_norm": 66.7186279296875,
|
| 269 |
+
"learning_rate": 0.0004890843621399176,
|
| 270 |
+
"loss": 478.12,
|
| 271 |
+
"step": 17280
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 16.0,
|
| 275 |
+
"eval_cer": 0.25418663241614387,
|
| 276 |
+
"eval_loss": 477.70703125,
|
| 277 |
+
"eval_runtime": 27.501,
|
| 278 |
+
"eval_samples_per_second": 32.362,
|
| 279 |
+
"eval_steps_per_second": 4.073,
|
| 280 |
+
"eval_wer": 0.6666666666666666,
|
| 281 |
+
"step": 17280
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 17.0,
|
| 285 |
+
"grad_norm": 5.836469650268555,
|
| 286 |
+
"learning_rate": 0.0004779938271604938,
|
| 287 |
+
"loss": 439.3815,
|
| 288 |
+
"step": 18360
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 17.0,
|
| 292 |
+
"eval_cer": 0.25999110803734626,
|
| 293 |
+
"eval_loss": 441.1052551269531,
|
| 294 |
+
"eval_runtime": 26.8497,
|
| 295 |
+
"eval_samples_per_second": 33.147,
|
| 296 |
+
"eval_steps_per_second": 4.171,
|
| 297 |
+
"eval_wer": 0.6576704545454546,
|
| 298 |
+
"step": 18360
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 18.0,
|
| 302 |
+
"grad_norm": 4.255939483642578,
|
| 303 |
+
"learning_rate": 0.0004668827160493827,
|
| 304 |
+
"loss": 406.8238,
|
| 305 |
+
"step": 19440
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 18.0,
|
| 309 |
+
"eval_cer": 0.24961715160796324,
|
| 310 |
+
"eval_loss": 417.4708557128906,
|
| 311 |
+
"eval_runtime": 26.667,
|
| 312 |
+
"eval_samples_per_second": 33.375,
|
| 313 |
+
"eval_steps_per_second": 4.2,
|
| 314 |
+
"eval_wer": 0.6792929292929293,
|
| 315 |
+
"step": 19440
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 19.0,
|
| 319 |
+
"grad_norm": 0.6231774687767029,
|
| 320 |
+
"learning_rate": 0.0004557716049382716,
|
| 321 |
+
"loss": 381.221,
|
| 322 |
+
"step": 20520
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 19.0,
|
| 326 |
+
"eval_cer": 0.23459961468161833,
|
| 327 |
+
"eval_loss": 385.214599609375,
|
| 328 |
+
"eval_runtime": 27.8465,
|
| 329 |
+
"eval_samples_per_second": 31.961,
|
| 330 |
+
"eval_steps_per_second": 4.022,
|
| 331 |
+
"eval_wer": 0.6339962121212122,
|
| 332 |
+
"step": 20520
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 20.0,
|
| 336 |
+
"grad_norm": 7.669593811035156,
|
| 337 |
+
"learning_rate": 0.0004446604938271605,
|
| 338 |
+
"loss": 349.7026,
|
| 339 |
+
"step": 21600
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 20.0,
|
| 343 |
+
"eval_cer": 0.24514647038482437,
|
| 344 |
+
"eval_loss": 417.8040771484375,
|
| 345 |
+
"eval_runtime": 26.5289,
|
| 346 |
+
"eval_samples_per_second": 33.548,
|
| 347 |
+
"eval_steps_per_second": 4.222,
|
| 348 |
+
"eval_wer": 0.6319444444444444,
|
| 349 |
+
"step": 21600
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 21.0,
|
| 353 |
+
"grad_norm": 45.017032623291016,
|
| 354 |
+
"learning_rate": 0.000433559670781893,
|
| 355 |
+
"loss": 336.5142,
|
| 356 |
+
"step": 22680
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 21.0,
|
| 360 |
+
"eval_cer": 0.2432198784765104,
|
| 361 |
+
"eval_loss": 446.5249328613281,
|
| 362 |
+
"eval_runtime": 27.1849,
|
| 363 |
+
"eval_samples_per_second": 32.739,
|
| 364 |
+
"eval_steps_per_second": 4.12,
|
| 365 |
+
"eval_wer": 0.6699810606060606,
|
| 366 |
+
"step": 22680
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 22.0,
|
| 370 |
+
"grad_norm": 43.4493522644043,
|
| 371 |
+
"learning_rate": 0.0004224485596707819,
|
| 372 |
+
"loss": 316.0003,
|
| 373 |
+
"step": 23760
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 22.0,
|
| 377 |
+
"eval_cer": 0.22963493553327077,
|
| 378 |
+
"eval_loss": 414.8999938964844,
|
| 379 |
+
"eval_runtime": 26.6715,
|
| 380 |
+
"eval_samples_per_second": 33.369,
|
| 381 |
+
"eval_steps_per_second": 4.199,
|
| 382 |
+
"eval_wer": 0.657354797979798,
|
| 383 |
+
"step": 23760
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 23.0,
|
| 387 |
+
"grad_norm": 74.57106018066406,
|
| 388 |
+
"learning_rate": 0.0004113374485596708,
|
| 389 |
+
"loss": 301.4025,
|
| 390 |
+
"step": 24840
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 23.0,
|
| 394 |
+
"eval_cer": 0.22943733636318728,
|
| 395 |
+
"eval_loss": 435.7232360839844,
|
| 396 |
+
"eval_runtime": 27.1302,
|
| 397 |
+
"eval_samples_per_second": 32.805,
|
| 398 |
+
"eval_steps_per_second": 4.128,
|
| 399 |
+
"eval_wer": 0.6242108585858586,
|
| 400 |
+
"step": 24840
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 24.0,
|
| 404 |
+
"grad_norm": 99.78678894042969,
|
| 405 |
+
"learning_rate": 0.0004002263374485597,
|
| 406 |
+
"loss": 276.272,
|
| 407 |
+
"step": 25920
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 24.0,
|
| 411 |
+
"eval_cer": 0.22733784518105024,
|
| 412 |
+
"eval_loss": 447.37799072265625,
|
| 413 |
+
"eval_runtime": 26.3157,
|
| 414 |
+
"eval_samples_per_second": 33.82,
|
| 415 |
+
"eval_steps_per_second": 4.256,
|
| 416 |
+
"eval_wer": 0.6107954545454546,
|
| 417 |
+
"step": 25920
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 25.0,
|
| 421 |
+
"grad_norm": 39.30922317504883,
|
| 422 |
+
"learning_rate": 0.0003891255144032922,
|
| 423 |
+
"loss": 267.6378,
|
| 424 |
+
"step": 27000
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 25.0,
|
| 428 |
+
"eval_cer": 0.24299757941016648,
|
| 429 |
+
"eval_loss": 412.04449462890625,
|
| 430 |
+
"eval_runtime": 26.2307,
|
| 431 |
+
"eval_samples_per_second": 33.93,
|
| 432 |
+
"eval_steps_per_second": 4.27,
|
| 433 |
+
"eval_wer": 0.6068497474747475,
|
| 434 |
+
"step": 27000
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 26.0,
|
| 438 |
+
"grad_norm": 0.1310397982597351,
|
| 439 |
+
"learning_rate": 0.00037801440329218104,
|
| 440 |
+
"loss": 249.7945,
|
| 441 |
+
"step": 28080
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 26.0,
|
| 445 |
+
"eval_cer": 0.23860099787580893,
|
| 446 |
+
"eval_loss": 411.8222351074219,
|
| 447 |
+
"eval_runtime": 26.2906,
|
| 448 |
+
"eval_samples_per_second": 33.852,
|
| 449 |
+
"eval_steps_per_second": 4.26,
|
| 450 |
+
"eval_wer": 0.6144255050505051,
|
| 451 |
+
"step": 28080
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 27.0,
|
| 455 |
+
"grad_norm": 26.45827293395996,
|
| 456 |
+
"learning_rate": 0.0003669135802469136,
|
| 457 |
+
"loss": 238.2421,
|
| 458 |
+
"step": 29160
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 27.0,
|
| 462 |
+
"eval_cer": 0.2365015066936719,
|
| 463 |
+
"eval_loss": 403.0203552246094,
|
| 464 |
+
"eval_runtime": 26.1924,
|
| 465 |
+
"eval_samples_per_second": 33.979,
|
| 466 |
+
"eval_steps_per_second": 4.276,
|
| 467 |
+
"eval_wer": 0.6283143939393939,
|
| 468 |
+
"step": 29160
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 28.0,
|
| 472 |
+
"grad_norm": 3.8607170581817627,
|
| 473 |
+
"learning_rate": 0.0003558024691358025,
|
| 474 |
+
"loss": 226.9799,
|
| 475 |
+
"step": 30240
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 28.0,
|
| 479 |
+
"eval_cer": 0.2248925554512671,
|
| 480 |
+
"eval_loss": 399.3008728027344,
|
| 481 |
+
"eval_runtime": 28.3266,
|
| 482 |
+
"eval_samples_per_second": 31.419,
|
| 483 |
+
"eval_steps_per_second": 3.954,
|
| 484 |
+
"eval_wer": 0.6117424242424242,
|
| 485 |
+
"step": 30240
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 29.0,
|
| 489 |
+
"grad_norm": 0.44324901700019836,
|
| 490 |
+
"learning_rate": 0.0003446913580246914,
|
| 491 |
+
"loss": 209.7066,
|
| 492 |
+
"step": 31320
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 29.0,
|
| 496 |
+
"eval_cer": 0.22968433532579163,
|
| 497 |
+
"eval_loss": 418.8338623046875,
|
| 498 |
+
"eval_runtime": 26.3962,
|
| 499 |
+
"eval_samples_per_second": 33.717,
|
| 500 |
+
"eval_steps_per_second": 4.243,
|
| 501 |
+
"eval_wer": 0.5924873737373737,
|
| 502 |
+
"step": 31320
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"epoch": 30.0,
|
| 506 |
+
"grad_norm": 52.79179000854492,
|
| 507 |
+
"learning_rate": 0.00033359053497942386,
|
| 508 |
+
"loss": 200.3202,
|
| 509 |
+
"step": 32400
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 30.0,
|
| 513 |
+
"eval_cer": 0.2221261670700983,
|
| 514 |
+
"eval_loss": 411.5075988769531,
|
| 515 |
+
"eval_runtime": 26.3529,
|
| 516 |
+
"eval_samples_per_second": 33.772,
|
| 517 |
+
"eval_steps_per_second": 4.25,
|
| 518 |
+
"eval_wer": 0.6036931818181818,
|
| 519 |
+
"step": 32400
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 31.0,
|
| 523 |
+
"grad_norm": 19.727815628051758,
|
| 524 |
+
"learning_rate": 0.0003224897119341564,
|
| 525 |
+
"loss": 193.5312,
|
| 526 |
+
"step": 33480
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 31.0,
|
| 530 |
+
"eval_cer": 0.21987847651039866,
|
| 531 |
+
"eval_loss": 421.6947937011719,
|
| 532 |
+
"eval_runtime": 26.3853,
|
| 533 |
+
"eval_samples_per_second": 33.731,
|
| 534 |
+
"eval_steps_per_second": 4.245,
|
| 535 |
+
"eval_wer": 0.5924873737373737,
|
| 536 |
+
"step": 33480
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 32.0,
|
| 540 |
+
"grad_norm": 11.547778129577637,
|
| 541 |
+
"learning_rate": 0.00031137860082304525,
|
| 542 |
+
"loss": 186.5759,
|
| 543 |
+
"step": 34560
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 32.0,
|
| 547 |
+
"eval_cer": 0.22017487526552387,
|
| 548 |
+
"eval_loss": 403.4676818847656,
|
| 549 |
+
"eval_runtime": 26.3332,
|
| 550 |
+
"eval_samples_per_second": 33.798,
|
| 551 |
+
"eval_steps_per_second": 4.253,
|
| 552 |
+
"eval_wer": 0.5838068181818182,
|
| 553 |
+
"step": 34560
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 33.0,
|
| 557 |
+
"grad_norm": 26.520191192626953,
|
| 558 |
+
"learning_rate": 0.0003002674897119342,
|
| 559 |
+
"loss": 176.4434,
|
| 560 |
+
"step": 35640
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 33.0,
|
| 564 |
+
"eval_cer": 0.22563355233908017,
|
| 565 |
+
"eval_loss": 449.66937255859375,
|
| 566 |
+
"eval_runtime": 28.5656,
|
| 567 |
+
"eval_samples_per_second": 31.156,
|
| 568 |
+
"eval_steps_per_second": 3.921,
|
| 569 |
+
"eval_wer": 0.5984848484848485,
|
| 570 |
+
"step": 35640
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 34.0,
|
| 574 |
+
"grad_norm": 44.24443435668945,
|
| 575 |
+
"learning_rate": 0.00028915637860082305,
|
| 576 |
+
"loss": 164.8203,
|
| 577 |
+
"step": 36720
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 34.0,
|
| 581 |
+
"eval_cer": 0.2237810601195475,
|
| 582 |
+
"eval_loss": 373.5114440917969,
|
| 583 |
+
"eval_runtime": 26.1741,
|
| 584 |
+
"eval_samples_per_second": 34.003,
|
| 585 |
+
"eval_steps_per_second": 4.279,
|
| 586 |
+
"eval_wer": 0.5935921717171717,
|
| 587 |
+
"step": 36720
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 35.0,
|
| 591 |
+
"grad_norm": 263.10101318359375,
|
| 592 |
+
"learning_rate": 0.00027804526748971195,
|
| 593 |
+
"loss": 153.8149,
|
| 594 |
+
"step": 37800
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 35.0,
|
| 598 |
+
"eval_cer": 0.21960677765153386,
|
| 599 |
+
"eval_loss": 387.5101013183594,
|
| 600 |
+
"eval_runtime": 25.9639,
|
| 601 |
+
"eval_samples_per_second": 34.278,
|
| 602 |
+
"eval_steps_per_second": 4.314,
|
| 603 |
+
"eval_wer": 0.5713383838383839,
|
| 604 |
+
"step": 37800
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 36.0,
|
| 608 |
+
"grad_norm": 24.568132400512695,
|
| 609 |
+
"learning_rate": 0.00026694444444444443,
|
| 610 |
+
"loss": 148.2955,
|
| 611 |
+
"step": 38880
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 36.0,
|
| 615 |
+
"eval_cer": 0.22192856790001483,
|
| 616 |
+
"eval_loss": 422.5177307128906,
|
| 617 |
+
"eval_runtime": 25.9731,
|
| 618 |
+
"eval_samples_per_second": 34.266,
|
| 619 |
+
"eval_steps_per_second": 4.312,
|
| 620 |
+
"eval_wer": 0.58364898989899,
|
| 621 |
+
"step": 38880
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 37.0,
|
| 625 |
+
"grad_norm": 10.271224975585938,
|
| 626 |
+
"learning_rate": 0.00025583333333333334,
|
| 627 |
+
"loss": 139.206,
|
| 628 |
+
"step": 39960
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"epoch": 37.0,
|
| 632 |
+
"eval_cer": 0.22331176209059922,
|
| 633 |
+
"eval_loss": 424.9959411621094,
|
| 634 |
+
"eval_runtime": 25.8486,
|
| 635 |
+
"eval_samples_per_second": 34.431,
|
| 636 |
+
"eval_steps_per_second": 4.333,
|
| 637 |
+
"eval_wer": 0.58364898989899,
|
| 638 |
+
"step": 39960
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 38.0,
|
| 642 |
+
"grad_norm": 0.18810251355171204,
|
| 643 |
+
"learning_rate": 0.00024472222222222224,
|
| 644 |
+
"loss": 137.3851,
|
| 645 |
+
"step": 41040
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 38.0,
|
| 649 |
+
"eval_cer": 0.21884108086746035,
|
| 650 |
+
"eval_loss": 428.62481689453125,
|
| 651 |
+
"eval_runtime": 29.4353,
|
| 652 |
+
"eval_samples_per_second": 30.236,
|
| 653 |
+
"eval_steps_per_second": 3.805,
|
| 654 |
+
"eval_wer": 0.5863320707070707,
|
| 655 |
+
"step": 41040
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 39.0,
|
| 659 |
+
"grad_norm": 126.07501220703125,
|
| 660 |
+
"learning_rate": 0.00023362139917695475,
|
| 661 |
+
"loss": 128.706,
|
| 662 |
+
"step": 42120
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 39.0,
|
| 666 |
+
"eval_cer": 0.21730968729931335,
|
| 667 |
+
"eval_loss": 420.81134033203125,
|
| 668 |
+
"eval_runtime": 27.0888,
|
| 669 |
+
"eval_samples_per_second": 32.855,
|
| 670 |
+
"eval_steps_per_second": 4.135,
|
| 671 |
+
"eval_wer": 0.5759154040404041,
|
| 672 |
+
"step": 42120
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 40.0,
|
| 676 |
+
"grad_norm": 3.7092888355255127,
|
| 677 |
+
"learning_rate": 0.00022251028806584362,
|
| 678 |
+
"loss": 122.1991,
|
| 679 |
+
"step": 43200
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 40.0,
|
| 683 |
+
"eval_cer": 0.2217803685224522,
|
| 684 |
+
"eval_loss": 449.3416748046875,
|
| 685 |
+
"eval_runtime": 28.1156,
|
| 686 |
+
"eval_samples_per_second": 31.655,
|
| 687 |
+
"eval_steps_per_second": 3.984,
|
| 688 |
+
"eval_wer": 0.579229797979798,
|
| 689 |
+
"step": 43200
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 41.0,
|
| 693 |
+
"grad_norm": 53.470237731933594,
|
| 694 |
+
"learning_rate": 0.00021140946502057616,
|
| 695 |
+
"loss": 114.7006,
|
| 696 |
+
"step": 44280
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 41.0,
|
| 700 |
+
"eval_cer": 0.22318826260929703,
|
| 701 |
+
"eval_loss": 472.12640380859375,
|
| 702 |
+
"eval_runtime": 27.1058,
|
| 703 |
+
"eval_samples_per_second": 32.834,
|
| 704 |
+
"eval_steps_per_second": 4.132,
|
| 705 |
+
"eval_wer": 0.5743371212121212,
|
| 706 |
+
"step": 44280
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 42.0,
|
| 710 |
+
"grad_norm": 6.227956295013428,
|
| 711 |
+
"learning_rate": 0.00020030864197530864,
|
| 712 |
+
"loss": 112.9208,
|
| 713 |
+
"step": 45360
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 42.0,
|
| 717 |
+
"eval_cer": 0.21343180358642494,
|
| 718 |
+
"eval_loss": 484.6367492675781,
|
| 719 |
+
"eval_runtime": 27.3265,
|
| 720 |
+
"eval_samples_per_second": 32.569,
|
| 721 |
+
"eval_steps_per_second": 4.099,
|
| 722 |
+
"eval_wer": 0.5869633838383839,
|
| 723 |
+
"step": 45360
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 43.0,
|
| 727 |
+
"grad_norm": 45.22465896606445,
|
| 728 |
+
"learning_rate": 0.00018919753086419752,
|
| 729 |
+
"loss": 104.3684,
|
| 730 |
+
"step": 46440
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 43.0,
|
| 734 |
+
"eval_cer": 0.20782492713530604,
|
| 735 |
+
"eval_loss": 475.1777648925781,
|
| 736 |
+
"eval_runtime": 30.3439,
|
| 737 |
+
"eval_samples_per_second": 29.33,
|
| 738 |
+
"eval_steps_per_second": 3.691,
|
| 739 |
+
"eval_wer": 0.5617108585858586,
|
| 740 |
+
"step": 46440
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 44.0,
|
| 744 |
+
"grad_norm": 4.431763648986816,
|
| 745 |
+
"learning_rate": 0.00017808641975308642,
|
| 746 |
+
"loss": 101.3417,
|
| 747 |
+
"step": 47520
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 44.0,
|
| 751 |
+
"eval_cer": 0.2081460257866917,
|
| 752 |
+
"eval_loss": 438.15020751953125,
|
| 753 |
+
"eval_runtime": 34.0126,
|
| 754 |
+
"eval_samples_per_second": 26.167,
|
| 755 |
+
"eval_steps_per_second": 3.293,
|
| 756 |
+
"eval_wer": 0.5639204545454546,
|
| 757 |
+
"step": 47520
|
| 758 |
+
},
|
| 759 |
+
{
|
| 760 |
+
"epoch": 45.0,
|
| 761 |
+
"grad_norm": 27.047523498535156,
|
| 762 |
+
"learning_rate": 0.0001669753086419753,
|
| 763 |
+
"loss": 93.9768,
|
| 764 |
+
"step": 48600
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 45.0,
|
| 768 |
+
"eval_cer": 0.20668873190732598,
|
| 769 |
+
"eval_loss": 450.406494140625,
|
| 770 |
+
"eval_runtime": 26.339,
|
| 771 |
+
"eval_samples_per_second": 33.79,
|
| 772 |
+
"eval_steps_per_second": 4.252,
|
| 773 |
+
"eval_wer": 0.5582386363636364,
|
| 774 |
+
"step": 48600
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 46.0,
|
| 778 |
+
"grad_norm": 2.549964666366577,
|
| 779 |
+
"learning_rate": 0.00015587448559670783,
|
| 780 |
+
"loss": 95.7108,
|
| 781 |
+
"step": 49680
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 46.0,
|
| 785 |
+
"eval_cer": 0.2114064120930692,
|
| 786 |
+
"eval_loss": 471.1705627441406,
|
| 787 |
+
"eval_runtime": 26.695,
|
| 788 |
+
"eval_samples_per_second": 33.34,
|
| 789 |
+
"eval_steps_per_second": 4.196,
|
| 790 |
+
"eval_wer": 0.5645517676767676,
|
| 791 |
+
"step": 49680
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 47.0,
|
| 795 |
+
"grad_norm": 37.24876022338867,
|
| 796 |
+
"learning_rate": 0.0001447633744855967,
|
| 797 |
+
"loss": 85.0469,
|
| 798 |
+
"step": 50760
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 47.0,
|
| 802 |
+
"eval_cer": 0.20777552734278515,
|
| 803 |
+
"eval_loss": 466.9778747558594,
|
| 804 |
+
"eval_runtime": 27.1859,
|
| 805 |
+
"eval_samples_per_second": 32.738,
|
| 806 |
+
"eval_steps_per_second": 4.12,
|
| 807 |
+
"eval_wer": 0.561395202020202,
|
| 808 |
+
"step": 50760
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 48.0,
|
| 812 |
+
"grad_norm": 0.1757580041885376,
|
| 813 |
+
"learning_rate": 0.0001336625514403292,
|
| 814 |
+
"loss": 86.9859,
|
| 815 |
+
"step": 51840
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 48.0,
|
| 819 |
+
"eval_cer": 0.20693573086993033,
|
| 820 |
+
"eval_loss": 457.32208251953125,
|
| 821 |
+
"eval_runtime": 26.898,
|
| 822 |
+
"eval_samples_per_second": 33.088,
|
| 823 |
+
"eval_steps_per_second": 4.164,
|
| 824 |
+
"eval_wer": 0.5639204545454546,
|
| 825 |
+
"step": 51840
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 49.0,
|
| 829 |
+
"grad_norm": 33.84082794189453,
|
| 830 |
+
"learning_rate": 0.00012255144032921811,
|
| 831 |
+
"loss": 82.6203,
|
| 832 |
+
"step": 52920
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 49.0,
|
| 836 |
+
"eval_cer": 0.2059971348120338,
|
| 837 |
+
"eval_loss": 467.1033630371094,
|
| 838 |
+
"eval_runtime": 26.7011,
|
| 839 |
+
"eval_samples_per_second": 33.332,
|
| 840 |
+
"eval_steps_per_second": 4.195,
|
| 841 |
+
"eval_wer": 0.5558712121212122,
|
| 842 |
+
"step": 52920
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 50.0,
|
| 846 |
+
"grad_norm": 0.03401001915335655,
|
| 847 |
+
"learning_rate": 0.000111440329218107,
|
| 848 |
+
"loss": 76.6465,
|
| 849 |
+
"step": 54000
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 50.0,
|
| 853 |
+
"eval_cer": 0.2074544286913995,
|
| 854 |
+
"eval_loss": 488.0355224609375,
|
| 855 |
+
"eval_runtime": 25.9988,
|
| 856 |
+
"eval_samples_per_second": 34.232,
|
| 857 |
+
"eval_steps_per_second": 4.308,
|
| 858 |
+
"eval_wer": 0.5606060606060606,
|
| 859 |
+
"step": 54000
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 51.0,
|
| 863 |
+
"grad_norm": 0.004368220455944538,
|
| 864 |
+
"learning_rate": 0.00010032921810699588,
|
| 865 |
+
"loss": 72.2897,
|
| 866 |
+
"step": 55080
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 51.0,
|
| 870 |
+
"eval_cer": 0.20172405275897842,
|
| 871 |
+
"eval_loss": 506.8006286621094,
|
| 872 |
+
"eval_runtime": 26.3849,
|
| 873 |
+
"eval_samples_per_second": 33.731,
|
| 874 |
+
"eval_steps_per_second": 4.245,
|
| 875 |
+
"eval_wer": 0.5528724747474747,
|
| 876 |
+
"step": 55080
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 52.0,
|
| 880 |
+
"grad_norm": 0.492847204208374,
|
| 881 |
+
"learning_rate": 8.921810699588477e-05,
|
| 882 |
+
"loss": 66.3217,
|
| 883 |
+
"step": 56160
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 52.0,
|
| 887 |
+
"eval_cer": 0.20658993232228426,
|
| 888 |
+
"eval_loss": 513.1781616210938,
|
| 889 |
+
"eval_runtime": 26.4834,
|
| 890 |
+
"eval_samples_per_second": 33.606,
|
| 891 |
+
"eval_steps_per_second": 4.229,
|
| 892 |
+
"eval_wer": 0.5503472222222222,
|
| 893 |
+
"step": 56160
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 53.0,
|
| 897 |
+
"grad_norm": 216.27200317382812,
|
| 898 |
+
"learning_rate": 7.811728395061729e-05,
|
| 899 |
+
"loss": 63.3613,
|
| 900 |
+
"step": 57240
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 53.0,
|
| 904 |
+
"eval_cer": 0.20547843699056464,
|
| 905 |
+
"eval_loss": 509.8887023925781,
|
| 906 |
+
"eval_runtime": 26.8112,
|
| 907 |
+
"eval_samples_per_second": 33.195,
|
| 908 |
+
"eval_steps_per_second": 4.177,
|
| 909 |
+
"eval_wer": 0.5571338383838383,
|
| 910 |
+
"step": 57240
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 54.0,
|
| 914 |
+
"grad_norm": 82.88473510742188,
|
| 915 |
+
"learning_rate": 6.700617283950618e-05,
|
| 916 |
+
"loss": 63.2432,
|
| 917 |
+
"step": 58320
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 54.0,
|
| 921 |
+
"eval_cer": 0.20933162080719261,
|
| 922 |
+
"eval_loss": 532.0458374023438,
|
| 923 |
+
"eval_runtime": 27.9984,
|
| 924 |
+
"eval_samples_per_second": 31.788,
|
| 925 |
+
"eval_steps_per_second": 4.0,
|
| 926 |
+
"eval_wer": 0.5602904040404041,
|
| 927 |
+
"step": 58320
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 55.0,
|
| 931 |
+
"grad_norm": 167.6044464111328,
|
| 932 |
+
"learning_rate": 5.5895061728395066e-05,
|
| 933 |
+
"loss": 58.8891,
|
| 934 |
+
"step": 59400
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 55.0,
|
| 938 |
+
"eval_cer": 0.20617003408585685,
|
| 939 |
+
"eval_loss": 529.6932983398438,
|
| 940 |
+
"eval_runtime": 26.5553,
|
| 941 |
+
"eval_samples_per_second": 33.515,
|
| 942 |
+
"eval_steps_per_second": 4.218,
|
| 943 |
+
"eval_wer": 0.5525568181818182,
|
| 944 |
+
"step": 59400
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 56.0,
|
| 948 |
+
"grad_norm": 0.0747949555516243,
|
| 949 |
+
"learning_rate": 4.4794238683127575e-05,
|
| 950 |
+
"loss": 64.8119,
|
| 951 |
+
"step": 60480
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 56.0,
|
| 955 |
+
"eval_cer": 0.2025391493355728,
|
| 956 |
+
"eval_loss": 530.6396484375,
|
| 957 |
+
"eval_runtime": 26.4411,
|
| 958 |
+
"eval_samples_per_second": 33.66,
|
| 959 |
+
"eval_steps_per_second": 4.236,
|
| 960 |
+
"eval_wer": 0.5465593434343434,
|
| 961 |
+
"step": 60480
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 57.0,
|
| 965 |
+
"grad_norm": 111.74144744873047,
|
| 966 |
+
"learning_rate": 3.3683127572016464e-05,
|
| 967 |
+
"loss": 53.2317,
|
| 968 |
+
"step": 61560
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 57.0,
|
| 972 |
+
"eval_cer": 0.19937756261423703,
|
| 973 |
+
"eval_loss": 548.7178955078125,
|
| 974 |
+
"eval_runtime": 26.8929,
|
| 975 |
+
"eval_samples_per_second": 33.094,
|
| 976 |
+
"eval_steps_per_second": 4.165,
|
| 977 |
+
"eval_wer": 0.5415088383838383,
|
| 978 |
+
"step": 61560
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 58.0,
|
| 982 |
+
"grad_norm": 0.0003476064885035157,
|
| 983 |
+
"learning_rate": 2.257201646090535e-05,
|
| 984 |
+
"loss": 52.8465,
|
| 985 |
+
"step": 62640
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 58.0,
|
| 989 |
+
"eval_cer": 0.20071135701230056,
|
| 990 |
+
"eval_loss": 547.544677734375,
|
| 991 |
+
"eval_runtime": 26.8853,
|
| 992 |
+
"eval_samples_per_second": 33.104,
|
| 993 |
+
"eval_steps_per_second": 4.166,
|
| 994 |
+
"eval_wer": 0.5399305555555556,
|
| 995 |
+
"step": 62640
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 59.0,
|
| 999 |
+
"grad_norm": 331.07489013671875,
|
| 1000 |
+
"learning_rate": 1.147119341563786e-05,
|
| 1001 |
+
"loss": 50.4224,
|
| 1002 |
+
"step": 63720
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 59.0,
|
| 1006 |
+
"eval_cer": 0.20066195721977967,
|
| 1007 |
+
"eval_loss": 545.9059448242188,
|
| 1008 |
+
"eval_runtime": 26.3111,
|
| 1009 |
+
"eval_samples_per_second": 33.826,
|
| 1010 |
+
"eval_steps_per_second": 4.257,
|
| 1011 |
+
"eval_wer": 0.5426136363636364,
|
| 1012 |
+
"step": 63720
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 60.0,
|
| 1016 |
+
"grad_norm": 0.002585458569228649,
|
| 1017 |
+
"learning_rate": 3.60082304526749e-07,
|
| 1018 |
+
"loss": 50.2867,
|
| 1019 |
+
"step": 64800
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 60.0,
|
| 1023 |
+
"eval_cer": 0.19982216074692485,
|
| 1024 |
+
"eval_loss": 551.5419921875,
|
| 1025 |
+
"eval_runtime": 26.3733,
|
| 1026 |
+
"eval_samples_per_second": 33.746,
|
| 1027 |
+
"eval_steps_per_second": 4.247,
|
| 1028 |
+
"eval_wer": 0.5422979797979798,
|
| 1029 |
+
"step": 64800
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 60.0,
|
| 1033 |
+
"step": 64800,
|
| 1034 |
+
"total_flos": 1.8440987587856836e+20,
|
| 1035 |
+
"train_loss": 306.16538580246913,
|
| 1036 |
+
"train_runtime": 47401.1605,
|
| 1037 |
+
"train_samples_per_second": 21.856,
|
| 1038 |
+
"train_steps_per_second": 1.367
|
| 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": 1.8440987587856836e+20,
|
| 1059 |
+
"train_batch_size": 16,
|
| 1060 |
+
"trial_name": null,
|
| 1061 |
+
"trial_params": null
|
| 1062 |
+
}
|