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: 0.
|
| 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: 0.0285
|
| 20 |
+
- Wer: 0.0215
|
| 21 |
+
- Cer: 0.0044
|
| 22 |
|
| 23 |
## Model description
|
| 24 |
|
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"eval_cer": 0.004363984932059734,
|
| 4 |
+
"eval_loss": 0.028477508574724197,
|
| 5 |
+
"eval_runtime": 126.9135,
|
| 6 |
+
"eval_samples": 1727,
|
| 7 |
+
"eval_samples_per_second": 13.608,
|
| 8 |
+
"eval_steps_per_second": 1.702,
|
| 9 |
+
"eval_wer": 0.02150478495215048,
|
| 10 |
+
"total_flos": 4.963160913067481e+19,
|
| 11 |
+
"train_loss": 0.2655376970686213,
|
| 12 |
+
"train_runtime": 66419.3143,
|
| 13 |
+
"train_samples": 15540,
|
| 14 |
+
"train_samples_per_second": 14.038,
|
| 15 |
+
"train_steps_per_second": 0.878
|
| 16 |
+
}
|
eval--in domain set_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"eval_cer": 0.004363984932059734,
|
| 4 |
+
"eval_loss": 0.028477508574724197,
|
| 5 |
+
"eval_runtime": 126.9135,
|
| 6 |
+
"eval_samples": 1727,
|
| 7 |
+
"eval_samples_per_second": 13.608,
|
| 8 |
+
"eval_steps_per_second": 1.702,
|
| 9 |
+
"eval_wer": 0.02150478495215048
|
| 10 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"total_flos": 4.963160913067481e+19,
|
| 4 |
+
"train_loss": 0.2655376970686213,
|
| 5 |
+
"train_runtime": 66419.3143,
|
| 6 |
+
"train_samples": 15540,
|
| 7 |
+
"train_samples_per_second": 14.038,
|
| 8 |
+
"train_steps_per_second": 0.878
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1062 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.02128478715212848,
|
| 3 |
+
"best_model_checkpoint": "/scratch/elec/puhe/p/palp3/sami_ASR/base_model_output/wav2vec2-base-sami-22k-finetuned/widv/widv_2dips_validate/60epoch/outputs/checkpoint-972",
|
| 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": 3.892906904220581,
|
| 14 |
+
"learning_rate": 3.326474622770919e-05,
|
| 15 |
+
"loss": 0.1722,
|
| 16 |
+
"step": 972
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 1.0,
|
| 20 |
+
"eval_cer": 0.004321942688012915,
|
| 21 |
+
"eval_loss": 0.028580639511346817,
|
| 22 |
+
"eval_runtime": 106.6617,
|
| 23 |
+
"eval_samples_per_second": 16.191,
|
| 24 |
+
"eval_steps_per_second": 2.025,
|
| 25 |
+
"eval_wer": 0.02128478715212848,
|
| 26 |
+
"step": 972
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 2.0,
|
| 30 |
+
"grad_norm": 38.92696762084961,
|
| 31 |
+
"learning_rate": 6.656378600823046e-05,
|
| 32 |
+
"loss": 0.1455,
|
| 33 |
+
"step": 1944
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 2.0,
|
| 37 |
+
"eval_cer": 0.007718956006995829,
|
| 38 |
+
"eval_loss": 0.048419274389743805,
|
| 39 |
+
"eval_runtime": 110.2333,
|
| 40 |
+
"eval_samples_per_second": 15.667,
|
| 41 |
+
"eval_steps_per_second": 1.959,
|
| 42 |
+
"eval_wer": 0.038444615553844465,
|
| 43 |
+
"step": 1944
|
| 44 |
+
},
|
| 45 |
+
{
|
| 46 |
+
"epoch": 3.0,
|
| 47 |
+
"grad_norm": 29.349037170410156,
|
| 48 |
+
"learning_rate": 9.989711934156378e-05,
|
| 49 |
+
"loss": 0.1681,
|
| 50 |
+
"step": 2916
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 3.0,
|
| 54 |
+
"eval_cer": 0.014504574196152293,
|
| 55 |
+
"eval_loss": 0.06561160087585449,
|
| 56 |
+
"eval_runtime": 111.977,
|
| 57 |
+
"eval_samples_per_second": 15.423,
|
| 58 |
+
"eval_steps_per_second": 1.929,
|
| 59 |
+
"eval_wer": 0.07149928500714993,
|
| 60 |
+
"step": 2916
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 4.0,
|
| 64 |
+
"grad_norm": 46.283809661865234,
|
| 65 |
+
"learning_rate": 0.0001332304526748971,
|
| 66 |
+
"loss": 0.192,
|
| 67 |
+
"step": 3888
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 4.0,
|
| 71 |
+
"eval_cer": 0.023434346831696488,
|
| 72 |
+
"eval_loss": 0.11780054122209549,
|
| 73 |
+
"eval_runtime": 112.6302,
|
| 74 |
+
"eval_samples_per_second": 15.333,
|
| 75 |
+
"eval_steps_per_second": 1.918,
|
| 76 |
+
"eval_wer": 0.11054889451105489,
|
| 77 |
+
"step": 3888
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 5.0,
|
| 81 |
+
"grad_norm": 10.867298126220703,
|
| 82 |
+
"learning_rate": 0.00016652949245541838,
|
| 83 |
+
"loss": 0.2199,
|
| 84 |
+
"step": 4860
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 5.0,
|
| 88 |
+
"eval_cer": 0.04258879321942688,
|
| 89 |
+
"eval_loss": 0.14914190769195557,
|
| 90 |
+
"eval_runtime": 112.5685,
|
| 91 |
+
"eval_samples_per_second": 15.342,
|
| 92 |
+
"eval_steps_per_second": 1.919,
|
| 93 |
+
"eval_wer": 0.1903530964690353,
|
| 94 |
+
"step": 4860
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 6.0,
|
| 98 |
+
"grad_norm": 13.334542274475098,
|
| 99 |
+
"learning_rate": 0.00019986282578875173,
|
| 100 |
+
"loss": 0.2408,
|
| 101 |
+
"step": 5832
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 6.0,
|
| 105 |
+
"eval_cer": 0.050391833714516346,
|
| 106 |
+
"eval_loss": 0.1710040420293808,
|
| 107 |
+
"eval_runtime": 111.2348,
|
| 108 |
+
"eval_samples_per_second": 15.526,
|
| 109 |
+
"eval_steps_per_second": 1.942,
|
| 110 |
+
"eval_wer": 0.2053679463205368,
|
| 111 |
+
"step": 5832
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 7.0,
|
| 115 |
+
"grad_norm": 5.9388203620910645,
|
| 116 |
+
"learning_rate": 0.00023319615912208505,
|
| 117 |
+
"loss": 0.2761,
|
| 118 |
+
"step": 6804
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 7.0,
|
| 122 |
+
"eval_cer": 0.06069218350598682,
|
| 123 |
+
"eval_loss": 0.24052941799163818,
|
| 124 |
+
"eval_runtime": 116.3136,
|
| 125 |
+
"eval_samples_per_second": 14.848,
|
| 126 |
+
"eval_steps_per_second": 1.857,
|
| 127 |
+
"eval_wer": 0.23979760202397976,
|
| 128 |
+
"step": 6804
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 8.0,
|
| 132 |
+
"grad_norm": 2.925326347351074,
|
| 133 |
+
"learning_rate": 0.0002665294924554184,
|
| 134 |
+
"loss": 0.3111,
|
| 135 |
+
"step": 7776
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 8.0,
|
| 139 |
+
"eval_cer": 0.0968737387326786,
|
| 140 |
+
"eval_loss": 0.2559922933578491,
|
| 141 |
+
"eval_runtime": 114.6573,
|
| 142 |
+
"eval_samples_per_second": 15.062,
|
| 143 |
+
"eval_steps_per_second": 1.884,
|
| 144 |
+
"eval_wer": 0.3893961060389396,
|
| 145 |
+
"step": 7776
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 9.0,
|
| 149 |
+
"grad_norm": 13.22968578338623,
|
| 150 |
+
"learning_rate": 0.00029982853223593965,
|
| 151 |
+
"loss": 0.3474,
|
| 152 |
+
"step": 8748
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 9.0,
|
| 156 |
+
"eval_cer": 0.12339398627741155,
|
| 157 |
+
"eval_loss": 0.32267919182777405,
|
| 158 |
+
"eval_runtime": 112.5775,
|
| 159 |
+
"eval_samples_per_second": 15.341,
|
| 160 |
+
"eval_steps_per_second": 1.919,
|
| 161 |
+
"eval_wer": 0.41964580354196457,
|
| 162 |
+
"step": 8748
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 10.0,
|
| 166 |
+
"grad_norm": 20.562284469604492,
|
| 167 |
+
"learning_rate": 0.00033309327846364886,
|
| 168 |
+
"loss": 0.3899,
|
| 169 |
+
"step": 9720
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 10.0,
|
| 173 |
+
"eval_cer": 0.1137579039418808,
|
| 174 |
+
"eval_loss": 0.3252728581428528,
|
| 175 |
+
"eval_runtime": 112.1566,
|
| 176 |
+
"eval_samples_per_second": 15.398,
|
| 177 |
+
"eval_steps_per_second": 1.926,
|
| 178 |
+
"eval_wer": 0.40589594104058957,
|
| 179 |
+
"step": 9720
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 11.0,
|
| 183 |
+
"grad_norm": 14.398542404174805,
|
| 184 |
+
"learning_rate": 0.00036642661179698215,
|
| 185 |
+
"loss": 0.4153,
|
| 186 |
+
"step": 10692
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 11.0,
|
| 190 |
+
"eval_cer": 0.1411274048163595,
|
| 191 |
+
"eval_loss": 0.3555086553096771,
|
| 192 |
+
"eval_runtime": 112.1279,
|
| 193 |
+
"eval_samples_per_second": 15.402,
|
| 194 |
+
"eval_steps_per_second": 1.926,
|
| 195 |
+
"eval_wer": 0.488120118798812,
|
| 196 |
+
"step": 10692
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 12.0,
|
| 200 |
+
"grad_norm": 24.424081802368164,
|
| 201 |
+
"learning_rate": 0.0003997599451303155,
|
| 202 |
+
"loss": 0.4633,
|
| 203 |
+
"step": 11664
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 12.0,
|
| 207 |
+
"eval_cer": 0.15536290865061214,
|
| 208 |
+
"eval_loss": 0.4050753712654114,
|
| 209 |
+
"eval_runtime": 113.3569,
|
| 210 |
+
"eval_samples_per_second": 15.235,
|
| 211 |
+
"eval_steps_per_second": 1.905,
|
| 212 |
+
"eval_wer": 0.5147948520514795,
|
| 213 |
+
"step": 11664
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 13.0,
|
| 217 |
+
"grad_norm": 53.538700103759766,
|
| 218 |
+
"learning_rate": 0.0004330932784636488,
|
| 219 |
+
"loss": 0.5276,
|
| 220 |
+
"step": 12636
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 13.0,
|
| 224 |
+
"eval_cer": 0.16010527377909323,
|
| 225 |
+
"eval_loss": 0.48709672689437866,
|
| 226 |
+
"eval_runtime": 110.0153,
|
| 227 |
+
"eval_samples_per_second": 15.698,
|
| 228 |
+
"eval_steps_per_second": 1.963,
|
| 229 |
+
"eval_wer": 0.511439885601144,
|
| 230 |
+
"step": 12636
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 14.0,
|
| 234 |
+
"grad_norm": 12.373359680175781,
|
| 235 |
+
"learning_rate": 0.0004663923182441701,
|
| 236 |
+
"loss": 0.5434,
|
| 237 |
+
"step": 13608
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 14.0,
|
| 241 |
+
"eval_cer": 0.20774754473294765,
|
| 242 |
+
"eval_loss": 0.5271967053413391,
|
| 243 |
+
"eval_runtime": 111.471,
|
| 244 |
+
"eval_samples_per_second": 15.493,
|
| 245 |
+
"eval_steps_per_second": 1.938,
|
| 246 |
+
"eval_wer": 0.6464635353646464,
|
| 247 |
+
"step": 13608
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 15.0,
|
| 251 |
+
"grad_norm": 7.079484939575195,
|
| 252 |
+
"learning_rate": 0.0004997256515775034,
|
| 253 |
+
"loss": 0.555,
|
| 254 |
+
"step": 14580
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 15.0,
|
| 258 |
+
"eval_cer": 0.22033499260056505,
|
| 259 |
+
"eval_loss": 0.5461715459823608,
|
| 260 |
+
"eval_runtime": 110.8347,
|
| 261 |
+
"eval_samples_per_second": 15.582,
|
| 262 |
+
"eval_steps_per_second": 1.949,
|
| 263 |
+
"eval_wer": 0.6470135298647014,
|
| 264 |
+
"step": 14580
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 16.0,
|
| 268 |
+
"grad_norm": 8.357855796813965,
|
| 269 |
+
"learning_rate": 0.0004889803383630545,
|
| 270 |
+
"loss": 0.5475,
|
| 271 |
+
"step": 15552
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 16.0,
|
| 275 |
+
"eval_cer": 0.18553242297860892,
|
| 276 |
+
"eval_loss": 0.4817240834236145,
|
| 277 |
+
"eval_runtime": 111.6706,
|
| 278 |
+
"eval_samples_per_second": 15.465,
|
| 279 |
+
"eval_steps_per_second": 1.934,
|
| 280 |
+
"eval_wer": 0.5891541084589154,
|
| 281 |
+
"step": 15552
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 17.0,
|
| 285 |
+
"grad_norm": 21.565086364746094,
|
| 286 |
+
"learning_rate": 0.0004778692272519433,
|
| 287 |
+
"loss": 0.5442,
|
| 288 |
+
"step": 16524
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 17.0,
|
| 292 |
+
"eval_cer": 0.1834975783667429,
|
| 293 |
+
"eval_loss": 0.4537910521030426,
|
| 294 |
+
"eval_runtime": 112.4474,
|
| 295 |
+
"eval_samples_per_second": 15.358,
|
| 296 |
+
"eval_steps_per_second": 1.921,
|
| 297 |
+
"eval_wer": 0.5851391486085139,
|
| 298 |
+
"step": 16524
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 18.0,
|
| 302 |
+
"grad_norm": 37.555049896240234,
|
| 303 |
+
"learning_rate": 0.00046681527206218564,
|
| 304 |
+
"loss": 0.8933,
|
| 305 |
+
"step": 17496
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 18.0,
|
| 309 |
+
"eval_cer": 0.16757197632180815,
|
| 310 |
+
"eval_loss": 0.4261943995952606,
|
| 311 |
+
"eval_runtime": 111.8475,
|
| 312 |
+
"eval_samples_per_second": 15.441,
|
| 313 |
+
"eval_steps_per_second": 1.931,
|
| 314 |
+
"eval_wer": 0.5166648333516665,
|
| 315 |
+
"step": 17496
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 19.0,
|
| 319 |
+
"grad_norm": 5.985906600952148,
|
| 320 |
+
"learning_rate": 0.00045570416095107454,
|
| 321 |
+
"loss": 0.5099,
|
| 322 |
+
"step": 18468
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 19.0,
|
| 326 |
+
"eval_cer": 0.14852683976859948,
|
| 327 |
+
"eval_loss": 0.3809051215648651,
|
| 328 |
+
"eval_runtime": 111.8392,
|
| 329 |
+
"eval_samples_per_second": 15.442,
|
| 330 |
+
"eval_steps_per_second": 1.931,
|
| 331 |
+
"eval_wer": 0.4778352216477835,
|
| 332 |
+
"step": 18468
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 20.0,
|
| 336 |
+
"grad_norm": 29.389266967773438,
|
| 337 |
+
"learning_rate": 0.0004445930498399634,
|
| 338 |
+
"loss": 0.5529,
|
| 339 |
+
"step": 19440
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 20.0,
|
| 343 |
+
"eval_cer": 0.15105778286021795,
|
| 344 |
+
"eval_loss": 0.3896856904029846,
|
| 345 |
+
"eval_runtime": 111.1566,
|
| 346 |
+
"eval_samples_per_second": 15.537,
|
| 347 |
+
"eval_steps_per_second": 1.943,
|
| 348 |
+
"eval_wer": 0.4954350456495435,
|
| 349 |
+
"step": 19440
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 21.0,
|
| 353 |
+
"grad_norm": 10.6626615524292,
|
| 354 |
+
"learning_rate": 0.00043348193872885235,
|
| 355 |
+
"loss": 0.4446,
|
| 356 |
+
"step": 20412
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 21.0,
|
| 360 |
+
"eval_cer": 0.13650275797120948,
|
| 361 |
+
"eval_loss": 0.3874945044517517,
|
| 362 |
+
"eval_runtime": 111.1711,
|
| 363 |
+
"eval_samples_per_second": 15.535,
|
| 364 |
+
"eval_steps_per_second": 1.943,
|
| 365 |
+
"eval_wer": 0.45072049279507204,
|
| 366 |
+
"step": 20412
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 22.0,
|
| 370 |
+
"grad_norm": 9.340583801269531,
|
| 371 |
+
"learning_rate": 0.0004223708276177412,
|
| 372 |
+
"loss": 0.4376,
|
| 373 |
+
"step": 21384
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 22.0,
|
| 377 |
+
"eval_cer": 0.14034541907708867,
|
| 378 |
+
"eval_loss": 0.3486928343772888,
|
| 379 |
+
"eval_runtime": 110.597,
|
| 380 |
+
"eval_samples_per_second": 15.615,
|
| 381 |
+
"eval_steps_per_second": 1.953,
|
| 382 |
+
"eval_wer": 0.45919040809591904,
|
| 383 |
+
"step": 21384
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 23.0,
|
| 387 |
+
"grad_norm": 16.277769088745117,
|
| 388 |
+
"learning_rate": 0.0004112597165066301,
|
| 389 |
+
"loss": 0.4044,
|
| 390 |
+
"step": 22356
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 23.0,
|
| 394 |
+
"eval_cer": 0.123427620072649,
|
| 395 |
+
"eval_loss": 0.3339575231075287,
|
| 396 |
+
"eval_runtime": 111.0272,
|
| 397 |
+
"eval_samples_per_second": 15.555,
|
| 398 |
+
"eval_steps_per_second": 1.945,
|
| 399 |
+
"eval_wer": 0.4145858541414586,
|
| 400 |
+
"step": 22356
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 24.0,
|
| 404 |
+
"grad_norm": 7.630257606506348,
|
| 405 |
+
"learning_rate": 0.000400148605395519,
|
| 406 |
+
"loss": 0.4067,
|
| 407 |
+
"step": 23328
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 24.0,
|
| 411 |
+
"eval_cer": 0.12230929638100363,
|
| 412 |
+
"eval_loss": 0.3253650963306427,
|
| 413 |
+
"eval_runtime": 111.0464,
|
| 414 |
+
"eval_samples_per_second": 15.552,
|
| 415 |
+
"eval_steps_per_second": 1.945,
|
| 416 |
+
"eval_wer": 0.4068859311406886,
|
| 417 |
+
"step": 23328
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 25.0,
|
| 421 |
+
"grad_norm": 7.71258020401001,
|
| 422 |
+
"learning_rate": 0.0003890374942844079,
|
| 423 |
+
"loss": 0.3696,
|
| 424 |
+
"step": 24300
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 25.0,
|
| 428 |
+
"eval_cer": 0.1160954527108839,
|
| 429 |
+
"eval_loss": 0.3158465027809143,
|
| 430 |
+
"eval_runtime": 110.7966,
|
| 431 |
+
"eval_samples_per_second": 15.587,
|
| 432 |
+
"eval_steps_per_second": 1.95,
|
| 433 |
+
"eval_wer": 0.3937960620393796,
|
| 434 |
+
"step": 24300
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 26.0,
|
| 438 |
+
"grad_norm": 15.249846458435059,
|
| 439 |
+
"learning_rate": 0.00037792638317329675,
|
| 440 |
+
"loss": 0.3417,
|
| 441 |
+
"step": 25272
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 26.0,
|
| 445 |
+
"eval_cer": 0.11476691779900444,
|
| 446 |
+
"eval_loss": 0.304636687040329,
|
| 447 |
+
"eval_runtime": 109.8939,
|
| 448 |
+
"eval_samples_per_second": 15.715,
|
| 449 |
+
"eval_steps_per_second": 1.966,
|
| 450 |
+
"eval_wer": 0.3980860191398086,
|
| 451 |
+
"step": 25272
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 27.0,
|
| 455 |
+
"grad_norm": 5.0151777267456055,
|
| 456 |
+
"learning_rate": 0.00036681527206218565,
|
| 457 |
+
"loss": 0.3381,
|
| 458 |
+
"step": 26244
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 27.0,
|
| 462 |
+
"eval_cer": 0.11050383425265707,
|
| 463 |
+
"eval_loss": 0.291059285402298,
|
| 464 |
+
"eval_runtime": 106.5162,
|
| 465 |
+
"eval_samples_per_second": 16.214,
|
| 466 |
+
"eval_steps_per_second": 2.028,
|
| 467 |
+
"eval_wer": 0.3753162468375316,
|
| 468 |
+
"step": 26244
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 28.0,
|
| 472 |
+
"grad_norm": 9.235563278198242,
|
| 473 |
+
"learning_rate": 0.0003557155921353452,
|
| 474 |
+
"loss": 0.3187,
|
| 475 |
+
"step": 27216
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 28.0,
|
| 479 |
+
"eval_cer": 0.1189627337548769,
|
| 480 |
+
"eval_loss": 0.30689507722854614,
|
| 481 |
+
"eval_runtime": 108.4021,
|
| 482 |
+
"eval_samples_per_second": 15.931,
|
| 483 |
+
"eval_steps_per_second": 1.993,
|
| 484 |
+
"eval_wer": 0.38648113518864813,
|
| 485 |
+
"step": 27216
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 29.0,
|
| 489 |
+
"grad_norm": 4.142171859741211,
|
| 490 |
+
"learning_rate": 0.0003446159122085048,
|
| 491 |
+
"loss": 0.3016,
|
| 492 |
+
"step": 28188
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 29.0,
|
| 496 |
+
"eval_cer": 0.1005734562087986,
|
| 497 |
+
"eval_loss": 0.2522578537464142,
|
| 498 |
+
"eval_runtime": 105.9527,
|
| 499 |
+
"eval_samples_per_second": 16.3,
|
| 500 |
+
"eval_steps_per_second": 2.039,
|
| 501 |
+
"eval_wer": 0.35249147508524914,
|
| 502 |
+
"step": 28188
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"epoch": 30.0,
|
| 506 |
+
"grad_norm": 8.742466926574707,
|
| 507 |
+
"learning_rate": 0.0003335048010973937,
|
| 508 |
+
"loss": 0.2837,
|
| 509 |
+
"step": 29160
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 30.0,
|
| 513 |
+
"eval_cer": 0.09587313332436433,
|
| 514 |
+
"eval_loss": 0.24332503974437714,
|
| 515 |
+
"eval_runtime": 106.9612,
|
| 516 |
+
"eval_samples_per_second": 16.146,
|
| 517 |
+
"eval_steps_per_second": 2.019,
|
| 518 |
+
"eval_wer": 0.32939170608293916,
|
| 519 |
+
"step": 29160
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 31.0,
|
| 523 |
+
"grad_norm": 6.368826866149902,
|
| 524 |
+
"learning_rate": 0.0003223936899862826,
|
| 525 |
+
"loss": 0.2622,
|
| 526 |
+
"step": 30132
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 31.0,
|
| 530 |
+
"eval_cer": 0.09076920489708058,
|
| 531 |
+
"eval_loss": 0.23585936427116394,
|
| 532 |
+
"eval_runtime": 108.5575,
|
| 533 |
+
"eval_samples_per_second": 15.909,
|
| 534 |
+
"eval_steps_per_second": 1.99,
|
| 535 |
+
"eval_wer": 0.3172368276317237,
|
| 536 |
+
"step": 30132
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 32.0,
|
| 540 |
+
"grad_norm": 0.5108231902122498,
|
| 541 |
+
"learning_rate": 0.00031128257887517145,
|
| 542 |
+
"loss": 0.2584,
|
| 543 |
+
"step": 31104
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 32.0,
|
| 547 |
+
"eval_cer": 0.0962935557648325,
|
| 548 |
+
"eval_loss": 0.24952448904514313,
|
| 549 |
+
"eval_runtime": 107.2589,
|
| 550 |
+
"eval_samples_per_second": 16.101,
|
| 551 |
+
"eval_steps_per_second": 2.014,
|
| 552 |
+
"eval_wer": 0.32686173138268615,
|
| 553 |
+
"step": 31104
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 33.0,
|
| 557 |
+
"grad_norm": 6.757967948913574,
|
| 558 |
+
"learning_rate": 0.00030017146776406035,
|
| 559 |
+
"loss": 0.239,
|
| 560 |
+
"step": 32076
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 33.0,
|
| 564 |
+
"eval_cer": 0.09224068343871923,
|
| 565 |
+
"eval_loss": 0.24214725196361542,
|
| 566 |
+
"eval_runtime": 108.1862,
|
| 567 |
+
"eval_samples_per_second": 15.963,
|
| 568 |
+
"eval_steps_per_second": 1.997,
|
| 569 |
+
"eval_wer": 0.3187768122318777,
|
| 570 |
+
"step": 32076
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 34.0,
|
| 574 |
+
"grad_norm": 3.6875059604644775,
|
| 575 |
+
"learning_rate": 0.00028906035665294925,
|
| 576 |
+
"loss": 0.2275,
|
| 577 |
+
"step": 33048
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 34.0,
|
| 581 |
+
"eval_cer": 0.0917361765101574,
|
| 582 |
+
"eval_loss": 0.22530558705329895,
|
| 583 |
+
"eval_runtime": 107.8021,
|
| 584 |
+
"eval_samples_per_second": 16.02,
|
| 585 |
+
"eval_steps_per_second": 2.004,
|
| 586 |
+
"eval_wer": 0.31239687603123967,
|
| 587 |
+
"step": 33048
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 35.0,
|
| 591 |
+
"grad_norm": 11.401963233947754,
|
| 592 |
+
"learning_rate": 0.00027794924554183816,
|
| 593 |
+
"loss": 0.2216,
|
| 594 |
+
"step": 34020
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 35.0,
|
| 598 |
+
"eval_cer": 0.0827643616305664,
|
| 599 |
+
"eval_loss": 0.21883530914783478,
|
| 600 |
+
"eval_runtime": 109.0635,
|
| 601 |
+
"eval_samples_per_second": 15.835,
|
| 602 |
+
"eval_steps_per_second": 1.98,
|
| 603 |
+
"eval_wer": 0.2876471235287647,
|
| 604 |
+
"step": 34020
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 36.0,
|
| 608 |
+
"grad_norm": 14.137062072753906,
|
| 609 |
+
"learning_rate": 0.00026684956561499776,
|
| 610 |
+
"loss": 0.2078,
|
| 611 |
+
"step": 34992
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 36.0,
|
| 615 |
+
"eval_cer": 0.08280640387461322,
|
| 616 |
+
"eval_loss": 0.22628778219223022,
|
| 617 |
+
"eval_runtime": 107.0828,
|
| 618 |
+
"eval_samples_per_second": 16.128,
|
| 619 |
+
"eval_steps_per_second": 2.017,
|
| 620 |
+
"eval_wer": 0.2967220327796722,
|
| 621 |
+
"step": 34992
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 37.0,
|
| 625 |
+
"grad_norm": 4.607612133026123,
|
| 626 |
+
"learning_rate": 0.0002557498856881573,
|
| 627 |
+
"loss": 0.1993,
|
| 628 |
+
"step": 35964
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"epoch": 37.0,
|
| 632 |
+
"eval_cer": 0.08237757298533567,
|
| 633 |
+
"eval_loss": 0.21688050031661987,
|
| 634 |
+
"eval_runtime": 108.3947,
|
| 635 |
+
"eval_samples_per_second": 15.933,
|
| 636 |
+
"eval_steps_per_second": 1.993,
|
| 637 |
+
"eval_wer": 0.28731712682873173,
|
| 638 |
+
"step": 35964
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 38.0,
|
| 642 |
+
"grad_norm": 0.11699577420949936,
|
| 643 |
+
"learning_rate": 0.0002446387745770462,
|
| 644 |
+
"loss": 0.1846,
|
| 645 |
+
"step": 36936
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 38.0,
|
| 649 |
+
"eval_cer": 0.07778655993542312,
|
| 650 |
+
"eval_loss": 0.2135714888572693,
|
| 651 |
+
"eval_runtime": 107.5828,
|
| 652 |
+
"eval_samples_per_second": 16.053,
|
| 653 |
+
"eval_steps_per_second": 2.008,
|
| 654 |
+
"eval_wer": 0.27351226487735125,
|
| 655 |
+
"step": 36936
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 39.0,
|
| 659 |
+
"grad_norm": 7.092660903930664,
|
| 660 |
+
"learning_rate": 0.00023353909465020575,
|
| 661 |
+
"loss": 0.1797,
|
| 662 |
+
"step": 37908
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 39.0,
|
| 666 |
+
"eval_cer": 0.07624781380330957,
|
| 667 |
+
"eval_loss": 0.20947343111038208,
|
| 668 |
+
"eval_runtime": 109.3587,
|
| 669 |
+
"eval_samples_per_second": 15.792,
|
| 670 |
+
"eval_steps_per_second": 1.975,
|
| 671 |
+
"eval_wer": 0.26245737542624575,
|
| 672 |
+
"step": 37908
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 40.0,
|
| 676 |
+
"grad_norm": 0.5255852341651917,
|
| 677 |
+
"learning_rate": 0.00022243941472336535,
|
| 678 |
+
"loss": 0.182,
|
| 679 |
+
"step": 38880
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 40.0,
|
| 683 |
+
"eval_cer": 0.07417933539620611,
|
| 684 |
+
"eval_loss": 0.2156570702791214,
|
| 685 |
+
"eval_runtime": 108.6791,
|
| 686 |
+
"eval_samples_per_second": 15.891,
|
| 687 |
+
"eval_steps_per_second": 1.988,
|
| 688 |
+
"eval_wer": 0.25805741942580573,
|
| 689 |
+
"step": 38880
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 41.0,
|
| 693 |
+
"grad_norm": 11.525845527648926,
|
| 694 |
+
"learning_rate": 0.00021132830361225425,
|
| 695 |
+
"loss": 0.1633,
|
| 696 |
+
"step": 39852
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 41.0,
|
| 700 |
+
"eval_cer": 0.07259854702004574,
|
| 701 |
+
"eval_loss": 0.1871972680091858,
|
| 702 |
+
"eval_runtime": 108.4995,
|
| 703 |
+
"eval_samples_per_second": 15.917,
|
| 704 |
+
"eval_steps_per_second": 1.991,
|
| 705 |
+
"eval_wer": 0.2525574744252557,
|
| 706 |
+
"step": 39852
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 42.0,
|
| 710 |
+
"grad_norm": 5.1948628425598145,
|
| 711 |
+
"learning_rate": 0.0002002286236854138,
|
| 712 |
+
"loss": 0.1674,
|
| 713 |
+
"step": 40824
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 42.0,
|
| 717 |
+
"eval_cer": 0.07172406834387192,
|
| 718 |
+
"eval_loss": 0.1886238008737564,
|
| 719 |
+
"eval_runtime": 108.7161,
|
| 720 |
+
"eval_samples_per_second": 15.885,
|
| 721 |
+
"eval_steps_per_second": 1.987,
|
| 722 |
+
"eval_wer": 0.24705752942470574,
|
| 723 |
+
"step": 40824
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 43.0,
|
| 727 |
+
"grad_norm": 35.27286911010742,
|
| 728 |
+
"learning_rate": 0.0001891175125743027,
|
| 729 |
+
"loss": 0.1598,
|
| 730 |
+
"step": 41796
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 43.0,
|
| 734 |
+
"eval_cer": 0.07292647652361092,
|
| 735 |
+
"eval_loss": 0.21099978685379028,
|
| 736 |
+
"eval_runtime": 108.7091,
|
| 737 |
+
"eval_samples_per_second": 15.886,
|
| 738 |
+
"eval_steps_per_second": 1.987,
|
| 739 |
+
"eval_wer": 0.2572874271257287,
|
| 740 |
+
"step": 41796
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 44.0,
|
| 744 |
+
"grad_norm": 5.230064868927002,
|
| 745 |
+
"learning_rate": 0.00017804069501600368,
|
| 746 |
+
"loss": 0.1485,
|
| 747 |
+
"step": 42768
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 44.0,
|
| 751 |
+
"eval_cer": 0.06976489977129019,
|
| 752 |
+
"eval_loss": 0.2113291323184967,
|
| 753 |
+
"eval_runtime": 108.5915,
|
| 754 |
+
"eval_samples_per_second": 15.904,
|
| 755 |
+
"eval_steps_per_second": 1.989,
|
| 756 |
+
"eval_wer": 0.24128258717412826,
|
| 757 |
+
"step": 42768
|
| 758 |
+
},
|
| 759 |
+
{
|
| 760 |
+
"epoch": 45.0,
|
| 761 |
+
"grad_norm": 7.1028571128845215,
|
| 762 |
+
"learning_rate": 0.00016692958390489255,
|
| 763 |
+
"loss": 0.143,
|
| 764 |
+
"step": 43740
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 45.0,
|
| 768 |
+
"eval_cer": 0.06891564644154446,
|
| 769 |
+
"eval_loss": 0.19445079565048218,
|
| 770 |
+
"eval_runtime": 108.1917,
|
| 771 |
+
"eval_samples_per_second": 15.962,
|
| 772 |
+
"eval_steps_per_second": 1.996,
|
| 773 |
+
"eval_wer": 0.23352766472335276,
|
| 774 |
+
"step": 43740
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 46.0,
|
| 778 |
+
"grad_norm": 91.1045150756836,
|
| 779 |
+
"learning_rate": 0.00015581847279378145,
|
| 780 |
+
"loss": 0.1297,
|
| 781 |
+
"step": 44712
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 46.0,
|
| 785 |
+
"eval_cer": 0.06398829543925737,
|
| 786 |
+
"eval_loss": 0.19453474879264832,
|
| 787 |
+
"eval_runtime": 107.8084,
|
| 788 |
+
"eval_samples_per_second": 16.019,
|
| 789 |
+
"eval_steps_per_second": 2.004,
|
| 790 |
+
"eval_wer": 0.22219777802221977,
|
| 791 |
+
"step": 44712
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 47.0,
|
| 795 |
+
"grad_norm": 1.2814885377883911,
|
| 796 |
+
"learning_rate": 0.00014470736168267033,
|
| 797 |
+
"loss": 0.1243,
|
| 798 |
+
"step": 45684
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 47.0,
|
| 802 |
+
"eval_cer": 0.06283633795237455,
|
| 803 |
+
"eval_loss": 0.1877937912940979,
|
| 804 |
+
"eval_runtime": 104.4978,
|
| 805 |
+
"eval_samples_per_second": 16.527,
|
| 806 |
+
"eval_steps_per_second": 2.067,
|
| 807 |
+
"eval_wer": 0.22170278297217028,
|
| 808 |
+
"step": 45684
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 48.0,
|
| 812 |
+
"grad_norm": 16.771364212036133,
|
| 813 |
+
"learning_rate": 0.00013359625057155923,
|
| 814 |
+
"loss": 0.1153,
|
| 815 |
+
"step": 46656
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 48.0,
|
| 819 |
+
"eval_cer": 0.06144053545002018,
|
| 820 |
+
"eval_loss": 0.19347365200519562,
|
| 821 |
+
"eval_runtime": 104.1447,
|
| 822 |
+
"eval_samples_per_second": 16.583,
|
| 823 |
+
"eval_steps_per_second": 2.074,
|
| 824 |
+
"eval_wer": 0.2165328346716533,
|
| 825 |
+
"step": 46656
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 49.0,
|
| 829 |
+
"grad_norm": 0.7877097129821777,
|
| 830 |
+
"learning_rate": 0.0001224851394604481,
|
| 831 |
+
"loss": 0.1115,
|
| 832 |
+
"step": 47628
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 49.0,
|
| 836 |
+
"eval_cer": 0.060372662451231,
|
| 837 |
+
"eval_loss": 0.1975562423467636,
|
| 838 |
+
"eval_runtime": 116.3496,
|
| 839 |
+
"eval_samples_per_second": 14.843,
|
| 840 |
+
"eval_steps_per_second": 1.856,
|
| 841 |
+
"eval_wer": 0.2114728852711473,
|
| 842 |
+
"step": 47628
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 50.0,
|
| 846 |
+
"grad_norm": 4.4330010414123535,
|
| 847 |
+
"learning_rate": 0.00011137402834933699,
|
| 848 |
+
"loss": 0.1012,
|
| 849 |
+
"step": 48600
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 50.0,
|
| 853 |
+
"eval_cer": 0.060086775191712634,
|
| 854 |
+
"eval_loss": 0.18160125613212585,
|
| 855 |
+
"eval_runtime": 104.8388,
|
| 856 |
+
"eval_samples_per_second": 16.473,
|
| 857 |
+
"eval_steps_per_second": 2.06,
|
| 858 |
+
"eval_wer": 0.2102078979210208,
|
| 859 |
+
"step": 48600
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 51.0,
|
| 863 |
+
"grad_norm": 7.805563449859619,
|
| 864 |
+
"learning_rate": 0.00010026291723822588,
|
| 865 |
+
"loss": 0.1002,
|
| 866 |
+
"step": 49572
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 51.0,
|
| 870 |
+
"eval_cer": 0.05822850800484327,
|
| 871 |
+
"eval_loss": 0.17819127440452576,
|
| 872 |
+
"eval_runtime": 104.3273,
|
| 873 |
+
"eval_samples_per_second": 16.554,
|
| 874 |
+
"eval_steps_per_second": 2.07,
|
| 875 |
+
"eval_wer": 0.2041029589704103,
|
| 876 |
+
"step": 49572
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 52.0,
|
| 880 |
+
"grad_norm": 0.6484061479568481,
|
| 881 |
+
"learning_rate": 8.916323731138547e-05,
|
| 882 |
+
"loss": 0.0917,
|
| 883 |
+
"step": 50544
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 52.0,
|
| 887 |
+
"eval_cer": 0.05881709942149872,
|
| 888 |
+
"eval_loss": 0.20612208545207977,
|
| 889 |
+
"eval_runtime": 104.2003,
|
| 890 |
+
"eval_samples_per_second": 16.574,
|
| 891 |
+
"eval_steps_per_second": 2.073,
|
| 892 |
+
"eval_wer": 0.20146298537014629,
|
| 893 |
+
"step": 50544
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 53.0,
|
| 897 |
+
"grad_norm": 1.1902787685394287,
|
| 898 |
+
"learning_rate": 7.806355738454504e-05,
|
| 899 |
+
"loss": 0.086,
|
| 900 |
+
"step": 51516
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 53.0,
|
| 904 |
+
"eval_cer": 0.0578501278084219,
|
| 905 |
+
"eval_loss": 0.17770683765411377,
|
| 906 |
+
"eval_runtime": 106.2062,
|
| 907 |
+
"eval_samples_per_second": 16.261,
|
| 908 |
+
"eval_steps_per_second": 2.034,
|
| 909 |
+
"eval_wer": 0.1965680343196568,
|
| 910 |
+
"step": 51516
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 54.0,
|
| 914 |
+
"grad_norm": 11.53394889831543,
|
| 915 |
+
"learning_rate": 6.695244627343393e-05,
|
| 916 |
+
"loss": 0.0769,
|
| 917 |
+
"step": 52488
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 54.0,
|
| 921 |
+
"eval_cer": 0.055024889008475714,
|
| 922 |
+
"eval_loss": 0.18718810379505157,
|
| 923 |
+
"eval_runtime": 110.6146,
|
| 924 |
+
"eval_samples_per_second": 15.613,
|
| 925 |
+
"eval_steps_per_second": 1.953,
|
| 926 |
+
"eval_wer": 0.1888681113188868,
|
| 927 |
+
"step": 52488
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 55.0,
|
| 931 |
+
"grad_norm": 0.4634610712528229,
|
| 932 |
+
"learning_rate": 5.585276634659351e-05,
|
| 933 |
+
"loss": 0.0765,
|
| 934 |
+
"step": 53460
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 55.0,
|
| 938 |
+
"eval_cer": 0.053923382214449075,
|
| 939 |
+
"eval_loss": 0.19234400987625122,
|
| 940 |
+
"eval_runtime": 109.7879,
|
| 941 |
+
"eval_samples_per_second": 15.73,
|
| 942 |
+
"eval_steps_per_second": 1.967,
|
| 943 |
+
"eval_wer": 0.1879881201187988,
|
| 944 |
+
"step": 53460
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 56.0,
|
| 948 |
+
"grad_norm": 0.03393052890896797,
|
| 949 |
+
"learning_rate": 4.47416552354824e-05,
|
| 950 |
+
"loss": 0.0673,
|
| 951 |
+
"step": 54432
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 56.0,
|
| 955 |
+
"eval_cer": 0.05345250908112471,
|
| 956 |
+
"eval_loss": 0.18786019086837769,
|
| 957 |
+
"eval_runtime": 109.5798,
|
| 958 |
+
"eval_samples_per_second": 15.76,
|
| 959 |
+
"eval_steps_per_second": 1.971,
|
| 960 |
+
"eval_wer": 0.1819931800681993,
|
| 961 |
+
"step": 54432
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 57.0,
|
| 965 |
+
"grad_norm": 1.5471312999725342,
|
| 966 |
+
"learning_rate": 3.363054412437128e-05,
|
| 967 |
+
"loss": 0.0646,
|
| 968 |
+
"step": 55404
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 57.0,
|
| 972 |
+
"eval_cer": 0.0534104668370779,
|
| 973 |
+
"eval_loss": 0.20105193555355072,
|
| 974 |
+
"eval_runtime": 120.726,
|
| 975 |
+
"eval_samples_per_second": 14.305,
|
| 976 |
+
"eval_steps_per_second": 1.789,
|
| 977 |
+
"eval_wer": 0.1819931800681993,
|
| 978 |
+
"step": 55404
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 58.0,
|
| 982 |
+
"grad_norm": 6.701435565948486,
|
| 983 |
+
"learning_rate": 2.2519433013260172e-05,
|
| 984 |
+
"loss": 0.0636,
|
| 985 |
+
"step": 56376
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 58.0,
|
| 989 |
+
"eval_cer": 0.05278824162518499,
|
| 990 |
+
"eval_loss": 0.19523786008358002,
|
| 991 |
+
"eval_runtime": 109.9847,
|
| 992 |
+
"eval_samples_per_second": 15.702,
|
| 993 |
+
"eval_steps_per_second": 1.964,
|
| 994 |
+
"eval_wer": 0.18001319986800132,
|
| 995 |
+
"step": 56376
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 59.0,
|
| 999 |
+
"grad_norm": 3.7623682022094727,
|
| 1000 |
+
"learning_rate": 1.1408321902149062e-05,
|
| 1001 |
+
"loss": 0.0618,
|
| 1002 |
+
"step": 57348
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 59.0,
|
| 1006 |
+
"eval_cer": 0.05197262209067671,
|
| 1007 |
+
"eval_loss": 0.20214368402957916,
|
| 1008 |
+
"eval_runtime": 110.2129,
|
| 1009 |
+
"eval_samples_per_second": 15.67,
|
| 1010 |
+
"eval_steps_per_second": 1.96,
|
| 1011 |
+
"eval_wer": 0.17753822461775381,
|
| 1012 |
+
"step": 57348
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 60.0,
|
| 1016 |
+
"grad_norm": 5.769583225250244,
|
| 1017 |
+
"learning_rate": 2.9721079103795153e-07,
|
| 1018 |
+
"loss": 0.0557,
|
| 1019 |
+
"step": 58320
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 60.0,
|
| 1023 |
+
"eval_cer": 0.05120745324902462,
|
| 1024 |
+
"eval_loss": 0.19957400858402252,
|
| 1025 |
+
"eval_runtime": 110.7577,
|
| 1026 |
+
"eval_samples_per_second": 15.593,
|
| 1027 |
+
"eval_steps_per_second": 1.95,
|
| 1028 |
+
"eval_wer": 0.1745132548674513,
|
| 1029 |
+
"step": 58320
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 60.0,
|
| 1033 |
+
"step": 58320,
|
| 1034 |
+
"total_flos": 4.963160913067481e+19,
|
| 1035 |
+
"train_loss": 0.2655376970686213,
|
| 1036 |
+
"train_runtime": 66419.3143,
|
| 1037 |
+
"train_samples_per_second": 14.038,
|
| 1038 |
+
"train_steps_per_second": 0.878
|
| 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": 4.963160913067481e+19,
|
| 1059 |
+
"train_batch_size": 16,
|
| 1060 |
+
"trial_name": null,
|
| 1061 |
+
"trial_params": null
|
| 1062 |
+
}
|