End of training
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.21,
|
| 3 |
+
"eval_loss": 0.1331787109375,
|
| 4 |
+
"eval_runtime": 4350.2148,
|
| 5 |
+
"eval_samples_per_second": 3.449,
|
| 6 |
+
"eval_steps_per_second": 0.216,
|
| 7 |
+
"eval_wer": 4.557596215181799,
|
| 8 |
+
"train_loss": 0.08706036122639974,
|
| 9 |
+
"train_runtime": 112219.1034,
|
| 10 |
+
"train_samples_per_second": 3.422,
|
| 11 |
+
"train_steps_per_second": 0.053
|
| 12 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.21,
|
| 3 |
+
"eval_loss": 0.1331787109375,
|
| 4 |
+
"eval_runtime": 4350.2148,
|
| 5 |
+
"eval_samples_per_second": 3.449,
|
| 6 |
+
"eval_steps_per_second": 0.216,
|
| 7 |
+
"eval_wer": 4.557596215181799
|
| 8 |
+
}
|
runs/Dec20_22-06-22_129-213-202-98/events.out.tfevents.1671690852.129-213-202-98.4032861.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f7c61f0f90b6a65b451e878beefe19b70b5332e07d3b024ec19155c98fd23e96
|
| 3 |
+
size 358
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.21,
|
| 3 |
+
"train_loss": 0.08706036122639974,
|
| 4 |
+
"train_runtime": 112219.1034,
|
| 5 |
+
"train_samples_per_second": 3.422,
|
| 6 |
+
"train_steps_per_second": 0.053
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1519 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 4.557596215181799,
|
| 3 |
+
"best_model_checkpoint": "./checkpoint-6000",
|
| 4 |
+
"epoch": 2.205,
|
| 5 |
+
"global_step": 6000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.0,
|
| 12 |
+
"learning_rate": 4.898977360288234e-06,
|
| 13 |
+
"loss": 0.3855,
|
| 14 |
+
"step": 25
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.01,
|
| 18 |
+
"learning_rate": 6.160712527409633e-06,
|
| 19 |
+
"loss": 0.16,
|
| 20 |
+
"step": 50
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.01,
|
| 24 |
+
"learning_rate": 6.85912902234906e-06,
|
| 25 |
+
"loss": 0.1358,
|
| 26 |
+
"step": 75
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.02,
|
| 30 |
+
"learning_rate": 7.344547104469332e-06,
|
| 31 |
+
"loss": 0.157,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.02,
|
| 36 |
+
"learning_rate": 7.716963756434345e-06,
|
| 37 |
+
"loss": 0.1538,
|
| 38 |
+
"step": 125
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.03,
|
| 42 |
+
"learning_rate": 8.019180844200955e-06,
|
| 43 |
+
"loss": 0.1557,
|
| 44 |
+
"step": 150
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.03,
|
| 48 |
+
"learning_rate": 8.27351214279797e-06,
|
| 49 |
+
"loss": 0.2211,
|
| 50 |
+
"step": 175
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.03,
|
| 54 |
+
"learning_rate": 8.49307723936858e-06,
|
| 55 |
+
"loss": 0.1959,
|
| 56 |
+
"step": 200
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.04,
|
| 60 |
+
"learning_rate": 8.686247975778677e-06,
|
| 61 |
+
"loss": 0.1804,
|
| 62 |
+
"step": 225
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.04,
|
| 66 |
+
"learning_rate": 8.858694625217149e-06,
|
| 67 |
+
"loss": 0.1335,
|
| 68 |
+
"step": 250
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.05,
|
| 72 |
+
"learning_rate": 9.014436199608479e-06,
|
| 73 |
+
"loss": 0.1477,
|
| 74 |
+
"step": 275
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.05,
|
| 78 |
+
"learning_rate": 9.156425255148058e-06,
|
| 79 |
+
"loss": 0.1371,
|
| 80 |
+
"step": 300
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.05,
|
| 84 |
+
"learning_rate": 9.28689473531776e-06,
|
| 85 |
+
"loss": 0.1334,
|
| 86 |
+
"step": 325
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.06,
|
| 90 |
+
"learning_rate": 9.407574351377137e-06,
|
| 91 |
+
"loss": 0.1238,
|
| 92 |
+
"step": 350
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.06,
|
| 96 |
+
"learning_rate": 9.519831289296397e-06,
|
| 97 |
+
"loss": 0.1206,
|
| 98 |
+
"step": 375
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.07,
|
| 102 |
+
"learning_rate": 9.624764935335318e-06,
|
| 103 |
+
"loss": 0.1167,
|
| 104 |
+
"step": 400
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.07,
|
| 108 |
+
"learning_rate": 9.723272550712454e-06,
|
| 109 |
+
"loss": 0.1192,
|
| 110 |
+
"step": 425
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.07,
|
| 114 |
+
"learning_rate": 9.812484046603779e-06,
|
| 115 |
+
"loss": 0.1158,
|
| 116 |
+
"step": 450
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.08,
|
| 120 |
+
"learning_rate": 9.900435550016748e-06,
|
| 121 |
+
"loss": 0.1409,
|
| 122 |
+
"step": 475
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.08,
|
| 126 |
+
"learning_rate": 9.98382788472848e-06,
|
| 127 |
+
"loss": 0.1188,
|
| 128 |
+
"step": 500
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.09,
|
| 132 |
+
"learning_rate": 9.965454545454546e-06,
|
| 133 |
+
"loss": 0.1257,
|
| 134 |
+
"step": 525
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.09,
|
| 138 |
+
"learning_rate": 9.920000000000002e-06,
|
| 139 |
+
"loss": 0.1328,
|
| 140 |
+
"step": 550
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.1,
|
| 144 |
+
"learning_rate": 9.874545454545455e-06,
|
| 145 |
+
"loss": 0.1327,
|
| 146 |
+
"step": 575
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.1,
|
| 150 |
+
"learning_rate": 9.82909090909091e-06,
|
| 151 |
+
"loss": 0.1511,
|
| 152 |
+
"step": 600
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.1,
|
| 156 |
+
"learning_rate": 9.783636363636365e-06,
|
| 157 |
+
"loss": 0.1401,
|
| 158 |
+
"step": 625
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.11,
|
| 162 |
+
"learning_rate": 9.738181818181818e-06,
|
| 163 |
+
"loss": 0.1285,
|
| 164 |
+
"step": 650
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.11,
|
| 168 |
+
"learning_rate": 9.692727272727274e-06,
|
| 169 |
+
"loss": 0.1363,
|
| 170 |
+
"step": 675
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.12,
|
| 174 |
+
"learning_rate": 9.64727272727273e-06,
|
| 175 |
+
"loss": 0.131,
|
| 176 |
+
"step": 700
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.12,
|
| 180 |
+
"learning_rate": 9.601818181818183e-06,
|
| 181 |
+
"loss": 0.112,
|
| 182 |
+
"step": 725
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.12,
|
| 186 |
+
"learning_rate": 9.556363636363637e-06,
|
| 187 |
+
"loss": 0.1205,
|
| 188 |
+
"step": 750
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.13,
|
| 192 |
+
"learning_rate": 9.510909090909092e-06,
|
| 193 |
+
"loss": 0.1215,
|
| 194 |
+
"step": 775
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.13,
|
| 198 |
+
"learning_rate": 9.465454545454546e-06,
|
| 199 |
+
"loss": 0.1366,
|
| 200 |
+
"step": 800
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.14,
|
| 204 |
+
"learning_rate": 9.42e-06,
|
| 205 |
+
"loss": 0.1123,
|
| 206 |
+
"step": 825
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.14,
|
| 210 |
+
"learning_rate": 9.374545454545455e-06,
|
| 211 |
+
"loss": 0.1121,
|
| 212 |
+
"step": 850
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.15,
|
| 216 |
+
"learning_rate": 9.32909090909091e-06,
|
| 217 |
+
"loss": 0.1142,
|
| 218 |
+
"step": 875
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.15,
|
| 222 |
+
"learning_rate": 9.283636363636364e-06,
|
| 223 |
+
"loss": 0.1252,
|
| 224 |
+
"step": 900
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.15,
|
| 228 |
+
"learning_rate": 9.23818181818182e-06,
|
| 229 |
+
"loss": 0.122,
|
| 230 |
+
"step": 925
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.16,
|
| 234 |
+
"learning_rate": 9.192727272727273e-06,
|
| 235 |
+
"loss": 0.1449,
|
| 236 |
+
"step": 950
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.16,
|
| 240 |
+
"learning_rate": 9.147272727272727e-06,
|
| 241 |
+
"loss": 0.1786,
|
| 242 |
+
"step": 975
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.17,
|
| 246 |
+
"learning_rate": 9.101818181818183e-06,
|
| 247 |
+
"loss": 0.1684,
|
| 248 |
+
"step": 1000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.17,
|
| 252 |
+
"eval_loss": 0.1619873046875,
|
| 253 |
+
"eval_runtime": 4324.26,
|
| 254 |
+
"eval_samples_per_second": 3.469,
|
| 255 |
+
"eval_steps_per_second": 0.217,
|
| 256 |
+
"eval_wer": 6.461965766532001,
|
| 257 |
+
"step": 1000
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 0.17,
|
| 261 |
+
"learning_rate": 9.056363636363638e-06,
|
| 262 |
+
"loss": 0.135,
|
| 263 |
+
"step": 1025
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.17,
|
| 267 |
+
"learning_rate": 9.010909090909092e-06,
|
| 268 |
+
"loss": 0.1217,
|
| 269 |
+
"step": 1050
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.18,
|
| 273 |
+
"learning_rate": 8.965454545454546e-06,
|
| 274 |
+
"loss": 0.11,
|
| 275 |
+
"step": 1075
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.18,
|
| 279 |
+
"learning_rate": 8.920000000000001e-06,
|
| 280 |
+
"loss": 0.1428,
|
| 281 |
+
"step": 1100
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 0.19,
|
| 285 |
+
"learning_rate": 8.874545454545456e-06,
|
| 286 |
+
"loss": 0.1382,
|
| 287 |
+
"step": 1125
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 0.19,
|
| 291 |
+
"learning_rate": 8.82909090909091e-06,
|
| 292 |
+
"loss": 0.1444,
|
| 293 |
+
"step": 1150
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 0.2,
|
| 297 |
+
"learning_rate": 8.783636363636364e-06,
|
| 298 |
+
"loss": 0.1476,
|
| 299 |
+
"step": 1175
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 0.2,
|
| 303 |
+
"learning_rate": 8.73818181818182e-06,
|
| 304 |
+
"loss": 0.1249,
|
| 305 |
+
"step": 1200
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.2,
|
| 309 |
+
"learning_rate": 8.692727272727273e-06,
|
| 310 |
+
"loss": 0.1512,
|
| 311 |
+
"step": 1225
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.21,
|
| 315 |
+
"learning_rate": 8.647272727272727e-06,
|
| 316 |
+
"loss": 0.1272,
|
| 317 |
+
"step": 1250
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.21,
|
| 321 |
+
"learning_rate": 8.601818181818182e-06,
|
| 322 |
+
"loss": 0.1376,
|
| 323 |
+
"step": 1275
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 0.22,
|
| 327 |
+
"learning_rate": 8.556363636363638e-06,
|
| 328 |
+
"loss": 0.1316,
|
| 329 |
+
"step": 1300
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 0.22,
|
| 333 |
+
"learning_rate": 8.510909090909092e-06,
|
| 334 |
+
"loss": 0.1233,
|
| 335 |
+
"step": 1325
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 0.23,
|
| 339 |
+
"learning_rate": 8.465454545454547e-06,
|
| 340 |
+
"loss": 0.1411,
|
| 341 |
+
"step": 1350
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 0.23,
|
| 345 |
+
"learning_rate": 8.42e-06,
|
| 346 |
+
"loss": 0.1325,
|
| 347 |
+
"step": 1375
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.23,
|
| 351 |
+
"learning_rate": 8.374545454545454e-06,
|
| 352 |
+
"loss": 0.1305,
|
| 353 |
+
"step": 1400
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.24,
|
| 357 |
+
"learning_rate": 8.32909090909091e-06,
|
| 358 |
+
"loss": 0.1339,
|
| 359 |
+
"step": 1425
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.24,
|
| 363 |
+
"learning_rate": 8.283636363636365e-06,
|
| 364 |
+
"loss": 0.1362,
|
| 365 |
+
"step": 1450
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 0.25,
|
| 369 |
+
"learning_rate": 8.238181818181819e-06,
|
| 370 |
+
"loss": 0.1251,
|
| 371 |
+
"step": 1475
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 0.25,
|
| 375 |
+
"learning_rate": 8.192727272727273e-06,
|
| 376 |
+
"loss": 0.1444,
|
| 377 |
+
"step": 1500
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.25,
|
| 381 |
+
"learning_rate": 8.147272727272728e-06,
|
| 382 |
+
"loss": 0.1391,
|
| 383 |
+
"step": 1525
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 0.26,
|
| 387 |
+
"learning_rate": 8.101818181818184e-06,
|
| 388 |
+
"loss": 0.1318,
|
| 389 |
+
"step": 1550
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.26,
|
| 393 |
+
"learning_rate": 8.056363636363636e-06,
|
| 394 |
+
"loss": 0.1303,
|
| 395 |
+
"step": 1575
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.27,
|
| 399 |
+
"learning_rate": 8.010909090909091e-06,
|
| 400 |
+
"loss": 0.1293,
|
| 401 |
+
"step": 1600
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.27,
|
| 405 |
+
"learning_rate": 7.965454545454547e-06,
|
| 406 |
+
"loss": 0.1238,
|
| 407 |
+
"step": 1625
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 0.28,
|
| 411 |
+
"learning_rate": 7.92e-06,
|
| 412 |
+
"loss": 0.1288,
|
| 413 |
+
"step": 1650
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.28,
|
| 417 |
+
"learning_rate": 7.874545454545454e-06,
|
| 418 |
+
"loss": 0.1144,
|
| 419 |
+
"step": 1675
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 0.28,
|
| 423 |
+
"learning_rate": 7.82909090909091e-06,
|
| 424 |
+
"loss": 0.1304,
|
| 425 |
+
"step": 1700
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 0.29,
|
| 429 |
+
"learning_rate": 7.783636363636365e-06,
|
| 430 |
+
"loss": 0.1139,
|
| 431 |
+
"step": 1725
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.29,
|
| 435 |
+
"learning_rate": 7.738181818181819e-06,
|
| 436 |
+
"loss": 0.1238,
|
| 437 |
+
"step": 1750
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.3,
|
| 441 |
+
"learning_rate": 7.692727272727273e-06,
|
| 442 |
+
"loss": 0.1416,
|
| 443 |
+
"step": 1775
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.3,
|
| 447 |
+
"learning_rate": 7.647272727272728e-06,
|
| 448 |
+
"loss": 0.1587,
|
| 449 |
+
"step": 1800
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 0.3,
|
| 453 |
+
"learning_rate": 7.6018181818181826e-06,
|
| 454 |
+
"loss": 0.1372,
|
| 455 |
+
"step": 1825
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 0.31,
|
| 459 |
+
"learning_rate": 7.556363636363637e-06,
|
| 460 |
+
"loss": 0.1307,
|
| 461 |
+
"step": 1850
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.31,
|
| 465 |
+
"learning_rate": 7.510909090909092e-06,
|
| 466 |
+
"loss": 0.1288,
|
| 467 |
+
"step": 1875
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 0.32,
|
| 471 |
+
"learning_rate": 7.4654545454545455e-06,
|
| 472 |
+
"loss": 0.1367,
|
| 473 |
+
"step": 1900
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 0.32,
|
| 477 |
+
"learning_rate": 7.420000000000001e-06,
|
| 478 |
+
"loss": 0.1222,
|
| 479 |
+
"step": 1925
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.33,
|
| 483 |
+
"learning_rate": 7.3745454545454556e-06,
|
| 484 |
+
"loss": 0.1054,
|
| 485 |
+
"step": 1950
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.33,
|
| 489 |
+
"learning_rate": 7.32909090909091e-06,
|
| 490 |
+
"loss": 0.1028,
|
| 491 |
+
"step": 1975
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.33,
|
| 495 |
+
"learning_rate": 7.283636363636364e-06,
|
| 496 |
+
"loss": 0.1174,
|
| 497 |
+
"step": 2000
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 0.33,
|
| 501 |
+
"eval_loss": 0.141845703125,
|
| 502 |
+
"eval_runtime": 4306.3621,
|
| 503 |
+
"eval_samples_per_second": 3.484,
|
| 504 |
+
"eval_steps_per_second": 0.218,
|
| 505 |
+
"eval_wer": 5.566260897299595,
|
| 506 |
+
"step": 2000
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.34,
|
| 510 |
+
"learning_rate": 7.2381818181818185e-06,
|
| 511 |
+
"loss": 0.1285,
|
| 512 |
+
"step": 2025
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 0.34,
|
| 516 |
+
"learning_rate": 7.192727272727273e-06,
|
| 517 |
+
"loss": 0.1318,
|
| 518 |
+
"step": 2050
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 0.35,
|
| 522 |
+
"learning_rate": 7.1472727272727285e-06,
|
| 523 |
+
"loss": 0.1123,
|
| 524 |
+
"step": 2075
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 0.35,
|
| 528 |
+
"learning_rate": 7.101818181818182e-06,
|
| 529 |
+
"loss": 0.1207,
|
| 530 |
+
"step": 2100
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 0.35,
|
| 534 |
+
"learning_rate": 7.056363636363637e-06,
|
| 535 |
+
"loss": 0.1016,
|
| 536 |
+
"step": 2125
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 0.36,
|
| 540 |
+
"learning_rate": 7.0109090909090915e-06,
|
| 541 |
+
"loss": 0.1027,
|
| 542 |
+
"step": 2150
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.36,
|
| 546 |
+
"learning_rate": 6.965454545454546e-06,
|
| 547 |
+
"loss": 0.1232,
|
| 548 |
+
"step": 2175
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.37,
|
| 552 |
+
"learning_rate": 6.92e-06,
|
| 553 |
+
"loss": 0.1082,
|
| 554 |
+
"step": 2200
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 0.37,
|
| 558 |
+
"learning_rate": 6.874545454545455e-06,
|
| 559 |
+
"loss": 0.0897,
|
| 560 |
+
"step": 2225
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 0.38,
|
| 564 |
+
"learning_rate": 6.82909090909091e-06,
|
| 565 |
+
"loss": 0.131,
|
| 566 |
+
"step": 2250
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 0.38,
|
| 570 |
+
"learning_rate": 6.7836363636363644e-06,
|
| 571 |
+
"loss": 0.1078,
|
| 572 |
+
"step": 2275
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 0.38,
|
| 576 |
+
"learning_rate": 6.738181818181819e-06,
|
| 577 |
+
"loss": 0.1435,
|
| 578 |
+
"step": 2300
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.39,
|
| 582 |
+
"learning_rate": 6.692727272727273e-06,
|
| 583 |
+
"loss": 0.127,
|
| 584 |
+
"step": 2325
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.39,
|
| 588 |
+
"learning_rate": 6.647272727272727e-06,
|
| 589 |
+
"loss": 0.1172,
|
| 590 |
+
"step": 2350
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.4,
|
| 594 |
+
"learning_rate": 6.601818181818183e-06,
|
| 595 |
+
"loss": 0.1097,
|
| 596 |
+
"step": 2375
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 1.0,
|
| 600 |
+
"learning_rate": 6.5563636363636374e-06,
|
| 601 |
+
"loss": 0.0885,
|
| 602 |
+
"step": 2400
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 1.01,
|
| 606 |
+
"learning_rate": 6.510909090909091e-06,
|
| 607 |
+
"loss": 0.0665,
|
| 608 |
+
"step": 2425
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 1.01,
|
| 612 |
+
"learning_rate": 6.465454545454546e-06,
|
| 613 |
+
"loss": 0.0648,
|
| 614 |
+
"step": 2450
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 1.01,
|
| 618 |
+
"learning_rate": 6.42e-06,
|
| 619 |
+
"loss": 0.0727,
|
| 620 |
+
"step": 2475
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 1.02,
|
| 624 |
+
"learning_rate": 6.374545454545456e-06,
|
| 625 |
+
"loss": 0.0684,
|
| 626 |
+
"step": 2500
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.02,
|
| 630 |
+
"learning_rate": 6.3290909090909096e-06,
|
| 631 |
+
"loss": 0.0646,
|
| 632 |
+
"step": 2525
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 1.03,
|
| 636 |
+
"learning_rate": 6.283636363636364e-06,
|
| 637 |
+
"loss": 0.0683,
|
| 638 |
+
"step": 2550
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 1.03,
|
| 642 |
+
"learning_rate": 6.238181818181819e-06,
|
| 643 |
+
"loss": 0.0664,
|
| 644 |
+
"step": 2575
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 1.04,
|
| 648 |
+
"learning_rate": 6.192727272727273e-06,
|
| 649 |
+
"loss": 0.0509,
|
| 650 |
+
"step": 2600
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.04,
|
| 654 |
+
"learning_rate": 6.147272727272727e-06,
|
| 655 |
+
"loss": 0.0534,
|
| 656 |
+
"step": 2625
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 1.04,
|
| 660 |
+
"learning_rate": 6.101818181818182e-06,
|
| 661 |
+
"loss": 0.0579,
|
| 662 |
+
"step": 2650
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 1.05,
|
| 666 |
+
"learning_rate": 6.056363636363637e-06,
|
| 667 |
+
"loss": 0.064,
|
| 668 |
+
"step": 2675
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 1.05,
|
| 672 |
+
"learning_rate": 6.010909090909092e-06,
|
| 673 |
+
"loss": 0.0574,
|
| 674 |
+
"step": 2700
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 1.06,
|
| 678 |
+
"learning_rate": 5.965454545454546e-06,
|
| 679 |
+
"loss": 0.0517,
|
| 680 |
+
"step": 2725
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 1.06,
|
| 684 |
+
"learning_rate": 5.92e-06,
|
| 685 |
+
"loss": 0.0569,
|
| 686 |
+
"step": 2750
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 1.06,
|
| 690 |
+
"learning_rate": 5.874545454545455e-06,
|
| 691 |
+
"loss": 0.06,
|
| 692 |
+
"step": 2775
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.07,
|
| 696 |
+
"learning_rate": 5.82909090909091e-06,
|
| 697 |
+
"loss": 0.0583,
|
| 698 |
+
"step": 2800
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 1.07,
|
| 702 |
+
"learning_rate": 5.783636363636365e-06,
|
| 703 |
+
"loss": 0.0695,
|
| 704 |
+
"step": 2825
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 1.08,
|
| 708 |
+
"learning_rate": 5.7381818181818185e-06,
|
| 709 |
+
"loss": 0.087,
|
| 710 |
+
"step": 2850
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 1.08,
|
| 714 |
+
"learning_rate": 5.692727272727273e-06,
|
| 715 |
+
"loss": 0.0846,
|
| 716 |
+
"step": 2875
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 1.09,
|
| 720 |
+
"learning_rate": 5.647272727272728e-06,
|
| 721 |
+
"loss": 0.0706,
|
| 722 |
+
"step": 2900
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 1.09,
|
| 726 |
+
"learning_rate": 5.601818181818183e-06,
|
| 727 |
+
"loss": 0.0616,
|
| 728 |
+
"step": 2925
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 1.09,
|
| 732 |
+
"learning_rate": 5.556363636363636e-06,
|
| 733 |
+
"loss": 0.0552,
|
| 734 |
+
"step": 2950
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 1.1,
|
| 738 |
+
"learning_rate": 5.5109090909090914e-06,
|
| 739 |
+
"loss": 0.0743,
|
| 740 |
+
"step": 2975
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 1.1,
|
| 744 |
+
"learning_rate": 5.465454545454546e-06,
|
| 745 |
+
"loss": 0.069,
|
| 746 |
+
"step": 3000
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 1.1,
|
| 750 |
+
"eval_loss": 0.1400146484375,
|
| 751 |
+
"eval_runtime": 4316.3895,
|
| 752 |
+
"eval_samples_per_second": 3.476,
|
| 753 |
+
"eval_steps_per_second": 0.217,
|
| 754 |
+
"eval_wer": 5.28651924303636,
|
| 755 |
+
"step": 3000
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 1.11,
|
| 759 |
+
"learning_rate": 5.420000000000001e-06,
|
| 760 |
+
"loss": 0.075,
|
| 761 |
+
"step": 3025
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 1.11,
|
| 765 |
+
"learning_rate": 5.374545454545454e-06,
|
| 766 |
+
"loss": 0.083,
|
| 767 |
+
"step": 3050
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 1.11,
|
| 771 |
+
"learning_rate": 5.329090909090909e-06,
|
| 772 |
+
"loss": 0.0801,
|
| 773 |
+
"step": 3075
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.12,
|
| 777 |
+
"learning_rate": 5.283636363636364e-06,
|
| 778 |
+
"loss": 0.0741,
|
| 779 |
+
"step": 3100
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 1.12,
|
| 783 |
+
"learning_rate": 5.238181818181819e-06,
|
| 784 |
+
"loss": 0.0639,
|
| 785 |
+
"step": 3125
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 1.13,
|
| 789 |
+
"learning_rate": 5.192727272727273e-06,
|
| 790 |
+
"loss": 0.0657,
|
| 791 |
+
"step": 3150
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 1.13,
|
| 795 |
+
"learning_rate": 5.147272727272727e-06,
|
| 796 |
+
"loss": 0.0681,
|
| 797 |
+
"step": 3175
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 1.14,
|
| 801 |
+
"learning_rate": 5.101818181818182e-06,
|
| 802 |
+
"loss": 0.0614,
|
| 803 |
+
"step": 3200
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 1.14,
|
| 807 |
+
"learning_rate": 5.056363636363637e-06,
|
| 808 |
+
"loss": 0.0505,
|
| 809 |
+
"step": 3225
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 1.14,
|
| 813 |
+
"learning_rate": 5.010909090909092e-06,
|
| 814 |
+
"loss": 0.0647,
|
| 815 |
+
"step": 3250
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 1.15,
|
| 819 |
+
"learning_rate": 4.965454545454546e-06,
|
| 820 |
+
"loss": 0.0693,
|
| 821 |
+
"step": 3275
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 1.15,
|
| 825 |
+
"learning_rate": 4.92e-06,
|
| 826 |
+
"loss": 0.0682,
|
| 827 |
+
"step": 3300
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 1.16,
|
| 831 |
+
"learning_rate": 4.8763636363636366e-06,
|
| 832 |
+
"loss": 0.0623,
|
| 833 |
+
"step": 3325
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 1.16,
|
| 837 |
+
"learning_rate": 4.830909090909091e-06,
|
| 838 |
+
"loss": 0.0647,
|
| 839 |
+
"step": 3350
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 1.17,
|
| 843 |
+
"learning_rate": 4.785454545454546e-06,
|
| 844 |
+
"loss": 0.0514,
|
| 845 |
+
"step": 3375
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 1.17,
|
| 849 |
+
"learning_rate": 4.74e-06,
|
| 850 |
+
"loss": 0.0563,
|
| 851 |
+
"step": 3400
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 1.17,
|
| 855 |
+
"learning_rate": 4.694545454545455e-06,
|
| 856 |
+
"loss": 0.0652,
|
| 857 |
+
"step": 3425
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 1.18,
|
| 861 |
+
"learning_rate": 4.6490909090909095e-06,
|
| 862 |
+
"loss": 0.0544,
|
| 863 |
+
"step": 3450
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 1.18,
|
| 867 |
+
"learning_rate": 4.603636363636364e-06,
|
| 868 |
+
"loss": 0.0494,
|
| 869 |
+
"step": 3475
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 1.19,
|
| 873 |
+
"learning_rate": 4.558181818181819e-06,
|
| 874 |
+
"loss": 0.0669,
|
| 875 |
+
"step": 3500
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 1.19,
|
| 879 |
+
"learning_rate": 4.512727272727273e-06,
|
| 880 |
+
"loss": 0.058,
|
| 881 |
+
"step": 3525
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 1.19,
|
| 885 |
+
"learning_rate": 4.467272727272728e-06,
|
| 886 |
+
"loss": 0.0721,
|
| 887 |
+
"step": 3550
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 1.2,
|
| 891 |
+
"learning_rate": 4.4218181818181825e-06,
|
| 892 |
+
"loss": 0.0708,
|
| 893 |
+
"step": 3575
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 1.2,
|
| 897 |
+
"learning_rate": 4.376363636363637e-06,
|
| 898 |
+
"loss": 0.0681,
|
| 899 |
+
"step": 3600
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 1.21,
|
| 903 |
+
"learning_rate": 4.330909090909091e-06,
|
| 904 |
+
"loss": 0.0606,
|
| 905 |
+
"step": 3625
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 1.21,
|
| 909 |
+
"learning_rate": 4.285454545454546e-06,
|
| 910 |
+
"loss": 0.0685,
|
| 911 |
+
"step": 3650
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 1.22,
|
| 915 |
+
"learning_rate": 4.24e-06,
|
| 916 |
+
"loss": 0.0748,
|
| 917 |
+
"step": 3675
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 1.22,
|
| 921 |
+
"learning_rate": 4.194545454545455e-06,
|
| 922 |
+
"loss": 0.0626,
|
| 923 |
+
"step": 3700
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 1.22,
|
| 927 |
+
"learning_rate": 4.149090909090909e-06,
|
| 928 |
+
"loss": 0.0745,
|
| 929 |
+
"step": 3725
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 1.23,
|
| 933 |
+
"learning_rate": 4.103636363636364e-06,
|
| 934 |
+
"loss": 0.0632,
|
| 935 |
+
"step": 3750
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 1.23,
|
| 939 |
+
"learning_rate": 4.0581818181818184e-06,
|
| 940 |
+
"loss": 0.0678,
|
| 941 |
+
"step": 3775
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 1.24,
|
| 945 |
+
"learning_rate": 4.012727272727273e-06,
|
| 946 |
+
"loss": 0.0652,
|
| 947 |
+
"step": 3800
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 1.24,
|
| 951 |
+
"learning_rate": 3.967272727272728e-06,
|
| 952 |
+
"loss": 0.0585,
|
| 953 |
+
"step": 3825
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 1.24,
|
| 957 |
+
"learning_rate": 3.921818181818182e-06,
|
| 958 |
+
"loss": 0.0689,
|
| 959 |
+
"step": 3850
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 1.25,
|
| 963 |
+
"learning_rate": 3.876363636363637e-06,
|
| 964 |
+
"loss": 0.0655,
|
| 965 |
+
"step": 3875
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 1.25,
|
| 969 |
+
"learning_rate": 3.830909090909091e-06,
|
| 970 |
+
"loss": 0.0673,
|
| 971 |
+
"step": 3900
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 1.26,
|
| 975 |
+
"learning_rate": 3.7854545454545456e-06,
|
| 976 |
+
"loss": 0.063,
|
| 977 |
+
"step": 3925
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 1.26,
|
| 981 |
+
"learning_rate": 3.74e-06,
|
| 982 |
+
"loss": 0.0649,
|
| 983 |
+
"step": 3950
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 1.27,
|
| 987 |
+
"learning_rate": 3.694545454545455e-06,
|
| 988 |
+
"loss": 0.0628,
|
| 989 |
+
"step": 3975
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 1.27,
|
| 993 |
+
"learning_rate": 3.6490909090909094e-06,
|
| 994 |
+
"loss": 0.0649,
|
| 995 |
+
"step": 4000
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 1.27,
|
| 999 |
+
"eval_loss": 0.1314697265625,
|
| 1000 |
+
"eval_runtime": 4287.8068,
|
| 1001 |
+
"eval_samples_per_second": 3.499,
|
| 1002 |
+
"eval_steps_per_second": 0.219,
|
| 1003 |
+
"eval_wer": 4.89315330640017,
|
| 1004 |
+
"step": 4000
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 1.27,
|
| 1008 |
+
"learning_rate": 3.603636363636364e-06,
|
| 1009 |
+
"loss": 0.0687,
|
| 1010 |
+
"step": 4025
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 1.28,
|
| 1014 |
+
"learning_rate": 3.5581818181818186e-06,
|
| 1015 |
+
"loss": 0.0693,
|
| 1016 |
+
"step": 4050
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 1.28,
|
| 1020 |
+
"learning_rate": 3.512727272727273e-06,
|
| 1021 |
+
"loss": 0.063,
|
| 1022 |
+
"step": 4075
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 1.29,
|
| 1026 |
+
"learning_rate": 3.4672727272727273e-06,
|
| 1027 |
+
"loss": 0.0594,
|
| 1028 |
+
"step": 4100
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 1.29,
|
| 1032 |
+
"learning_rate": 3.4218181818181824e-06,
|
| 1033 |
+
"loss": 0.0619,
|
| 1034 |
+
"step": 4125
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 1.29,
|
| 1038 |
+
"learning_rate": 3.3763636363636365e-06,
|
| 1039 |
+
"loss": 0.0607,
|
| 1040 |
+
"step": 4150
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 1.3,
|
| 1044 |
+
"learning_rate": 3.330909090909091e-06,
|
| 1045 |
+
"loss": 0.0579,
|
| 1046 |
+
"step": 4175
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 1.3,
|
| 1050 |
+
"learning_rate": 3.2872727272727274e-06,
|
| 1051 |
+
"loss": 0.0629,
|
| 1052 |
+
"step": 4200
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 1.31,
|
| 1056 |
+
"learning_rate": 3.241818181818182e-06,
|
| 1057 |
+
"loss": 0.0571,
|
| 1058 |
+
"step": 4225
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 1.31,
|
| 1062 |
+
"learning_rate": 3.196363636363637e-06,
|
| 1063 |
+
"loss": 0.0595,
|
| 1064 |
+
"step": 4250
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 1.31,
|
| 1068 |
+
"learning_rate": 3.150909090909091e-06,
|
| 1069 |
+
"loss": 0.0854,
|
| 1070 |
+
"step": 4275
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 1.32,
|
| 1074 |
+
"learning_rate": 3.1054545454545458e-06,
|
| 1075 |
+
"loss": 0.0728,
|
| 1076 |
+
"step": 4300
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 1.32,
|
| 1080 |
+
"learning_rate": 3.0600000000000003e-06,
|
| 1081 |
+
"loss": 0.0612,
|
| 1082 |
+
"step": 4325
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 1.33,
|
| 1086 |
+
"learning_rate": 3.014545454545455e-06,
|
| 1087 |
+
"loss": 0.0631,
|
| 1088 |
+
"step": 4350
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 1.33,
|
| 1092 |
+
"learning_rate": 2.969090909090909e-06,
|
| 1093 |
+
"loss": 0.0757,
|
| 1094 |
+
"step": 4375
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 1.34,
|
| 1098 |
+
"learning_rate": 2.9254545454545458e-06,
|
| 1099 |
+
"loss": 0.0728,
|
| 1100 |
+
"step": 4400
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 1.34,
|
| 1104 |
+
"learning_rate": 2.88e-06,
|
| 1105 |
+
"loss": 0.102,
|
| 1106 |
+
"step": 4425
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 1.34,
|
| 1110 |
+
"learning_rate": 2.834545454545455e-06,
|
| 1111 |
+
"loss": 0.0933,
|
| 1112 |
+
"step": 4450
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 1.35,
|
| 1116 |
+
"learning_rate": 2.789090909090909e-06,
|
| 1117 |
+
"loss": 0.1039,
|
| 1118 |
+
"step": 4475
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 1.35,
|
| 1122 |
+
"learning_rate": 2.7436363636363637e-06,
|
| 1123 |
+
"loss": 0.0669,
|
| 1124 |
+
"step": 4500
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 1.36,
|
| 1128 |
+
"learning_rate": 2.6981818181818188e-06,
|
| 1129 |
+
"loss": 0.0697,
|
| 1130 |
+
"step": 4525
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 1.36,
|
| 1134 |
+
"learning_rate": 2.652727272727273e-06,
|
| 1135 |
+
"loss": 0.0587,
|
| 1136 |
+
"step": 4550
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 1.36,
|
| 1140 |
+
"learning_rate": 2.6072727272727275e-06,
|
| 1141 |
+
"loss": 0.061,
|
| 1142 |
+
"step": 4575
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 1.37,
|
| 1146 |
+
"learning_rate": 2.561818181818182e-06,
|
| 1147 |
+
"loss": 0.0591,
|
| 1148 |
+
"step": 4600
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 1.37,
|
| 1152 |
+
"learning_rate": 2.5163636363636367e-06,
|
| 1153 |
+
"loss": 0.0572,
|
| 1154 |
+
"step": 4625
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 1.38,
|
| 1158 |
+
"learning_rate": 2.470909090909091e-06,
|
| 1159 |
+
"loss": 0.0584,
|
| 1160 |
+
"step": 4650
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 1.38,
|
| 1164 |
+
"learning_rate": 2.4254545454545455e-06,
|
| 1165 |
+
"loss": 0.0548,
|
| 1166 |
+
"step": 4675
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 1.39,
|
| 1170 |
+
"learning_rate": 2.38e-06,
|
| 1171 |
+
"loss": 0.0561,
|
| 1172 |
+
"step": 4700
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 1.39,
|
| 1176 |
+
"learning_rate": 2.3345454545454547e-06,
|
| 1177 |
+
"loss": 0.0608,
|
| 1178 |
+
"step": 4725
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 1.39,
|
| 1182 |
+
"learning_rate": 2.2890909090909093e-06,
|
| 1183 |
+
"loss": 0.0644,
|
| 1184 |
+
"step": 4750
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 2.0,
|
| 1188 |
+
"learning_rate": 2.243636363636364e-06,
|
| 1189 |
+
"loss": 0.0525,
|
| 1190 |
+
"step": 4775
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 2.0,
|
| 1194 |
+
"learning_rate": 2.1981818181818185e-06,
|
| 1195 |
+
"loss": 0.0494,
|
| 1196 |
+
"step": 4800
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 2.01,
|
| 1200 |
+
"learning_rate": 2.152727272727273e-06,
|
| 1201 |
+
"loss": 0.0504,
|
| 1202 |
+
"step": 4825
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 2.01,
|
| 1206 |
+
"learning_rate": 2.1072727272727277e-06,
|
| 1207 |
+
"loss": 0.0444,
|
| 1208 |
+
"step": 4850
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 2.02,
|
| 1212 |
+
"learning_rate": 2.0618181818181823e-06,
|
| 1213 |
+
"loss": 0.0381,
|
| 1214 |
+
"step": 4875
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 2.02,
|
| 1218 |
+
"learning_rate": 2.0163636363636364e-06,
|
| 1219 |
+
"loss": 0.0343,
|
| 1220 |
+
"step": 4900
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 2.03,
|
| 1224 |
+
"learning_rate": 1.970909090909091e-06,
|
| 1225 |
+
"loss": 0.0397,
|
| 1226 |
+
"step": 4925
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 2.03,
|
| 1230 |
+
"learning_rate": 1.9254545454545456e-06,
|
| 1231 |
+
"loss": 0.0392,
|
| 1232 |
+
"step": 4950
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 2.03,
|
| 1236 |
+
"learning_rate": 1.8800000000000002e-06,
|
| 1237 |
+
"loss": 0.0314,
|
| 1238 |
+
"step": 4975
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 2.04,
|
| 1242 |
+
"learning_rate": 1.8345454545454546e-06,
|
| 1243 |
+
"loss": 0.0334,
|
| 1244 |
+
"step": 5000
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 2.04,
|
| 1248 |
+
"eval_loss": 0.1368408203125,
|
| 1249 |
+
"eval_runtime": 4380.8288,
|
| 1250 |
+
"eval_samples_per_second": 3.425,
|
| 1251 |
+
"eval_steps_per_second": 0.214,
|
| 1252 |
+
"eval_wer": 4.684509887305975,
|
| 1253 |
+
"step": 5000
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 2.04,
|
| 1257 |
+
"learning_rate": 1.7890909090909092e-06,
|
| 1258 |
+
"loss": 0.0366,
|
| 1259 |
+
"step": 5025
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 2.05,
|
| 1263 |
+
"learning_rate": 1.7436363636363638e-06,
|
| 1264 |
+
"loss": 0.0393,
|
| 1265 |
+
"step": 5050
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 2.05,
|
| 1269 |
+
"learning_rate": 1.6981818181818182e-06,
|
| 1270 |
+
"loss": 0.0384,
|
| 1271 |
+
"step": 5075
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 2.06,
|
| 1275 |
+
"learning_rate": 1.6527272727272728e-06,
|
| 1276 |
+
"loss": 0.0374,
|
| 1277 |
+
"step": 5100
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 2.06,
|
| 1281 |
+
"learning_rate": 1.6072727272727274e-06,
|
| 1282 |
+
"loss": 0.0337,
|
| 1283 |
+
"step": 5125
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 2.06,
|
| 1287 |
+
"learning_rate": 1.5618181818181817e-06,
|
| 1288 |
+
"loss": 0.0307,
|
| 1289 |
+
"step": 5150
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 2.07,
|
| 1293 |
+
"learning_rate": 1.5163636363636363e-06,
|
| 1294 |
+
"loss": 0.0328,
|
| 1295 |
+
"step": 5175
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 2.07,
|
| 1299 |
+
"learning_rate": 1.470909090909091e-06,
|
| 1300 |
+
"loss": 0.0389,
|
| 1301 |
+
"step": 5200
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 2.08,
|
| 1305 |
+
"learning_rate": 1.4254545454545457e-06,
|
| 1306 |
+
"loss": 0.0289,
|
| 1307 |
+
"step": 5225
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 2.08,
|
| 1311 |
+
"learning_rate": 1.3800000000000001e-06,
|
| 1312 |
+
"loss": 0.0362,
|
| 1313 |
+
"step": 5250
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 2.08,
|
| 1317 |
+
"learning_rate": 1.3345454545454547e-06,
|
| 1318 |
+
"loss": 0.035,
|
| 1319 |
+
"step": 5275
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 2.09,
|
| 1323 |
+
"learning_rate": 1.2890909090909093e-06,
|
| 1324 |
+
"loss": 0.0381,
|
| 1325 |
+
"step": 5300
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 2.09,
|
| 1329 |
+
"learning_rate": 1.2436363636363637e-06,
|
| 1330 |
+
"loss": 0.0383,
|
| 1331 |
+
"step": 5325
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 2.1,
|
| 1335 |
+
"learning_rate": 1.1981818181818183e-06,
|
| 1336 |
+
"loss": 0.0501,
|
| 1337 |
+
"step": 5350
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 2.1,
|
| 1341 |
+
"learning_rate": 1.152727272727273e-06,
|
| 1342 |
+
"loss": 0.0354,
|
| 1343 |
+
"step": 5375
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 2.1,
|
| 1347 |
+
"learning_rate": 1.1072727272727275e-06,
|
| 1348 |
+
"loss": 0.0373,
|
| 1349 |
+
"step": 5400
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 2.11,
|
| 1353 |
+
"learning_rate": 1.0618181818181819e-06,
|
| 1354 |
+
"loss": 0.0354,
|
| 1355 |
+
"step": 5425
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 2.11,
|
| 1359 |
+
"learning_rate": 1.0163636363636365e-06,
|
| 1360 |
+
"loss": 0.0344,
|
| 1361 |
+
"step": 5450
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 2.12,
|
| 1365 |
+
"learning_rate": 9.70909090909091e-07,
|
| 1366 |
+
"loss": 0.034,
|
| 1367 |
+
"step": 5475
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 2.12,
|
| 1371 |
+
"learning_rate": 9.254545454545454e-07,
|
| 1372 |
+
"loss": 0.0364,
|
| 1373 |
+
"step": 5500
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 2.13,
|
| 1377 |
+
"learning_rate": 8.8e-07,
|
| 1378 |
+
"loss": 0.0399,
|
| 1379 |
+
"step": 5525
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 2.13,
|
| 1383 |
+
"learning_rate": 8.345454545454545e-07,
|
| 1384 |
+
"loss": 0.0355,
|
| 1385 |
+
"step": 5550
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 2.13,
|
| 1389 |
+
"learning_rate": 7.890909090909092e-07,
|
| 1390 |
+
"loss": 0.0372,
|
| 1391 |
+
"step": 5575
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 2.14,
|
| 1395 |
+
"learning_rate": 7.436363636363637e-07,
|
| 1396 |
+
"loss": 0.0284,
|
| 1397 |
+
"step": 5600
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 2.14,
|
| 1401 |
+
"learning_rate": 6.981818181818182e-07,
|
| 1402 |
+
"loss": 0.0312,
|
| 1403 |
+
"step": 5625
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 2.15,
|
| 1407 |
+
"learning_rate": 6.527272727272728e-07,
|
| 1408 |
+
"loss": 0.0296,
|
| 1409 |
+
"step": 5650
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 2.15,
|
| 1413 |
+
"learning_rate": 6.072727272727273e-07,
|
| 1414 |
+
"loss": 0.0349,
|
| 1415 |
+
"step": 5675
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 2.15,
|
| 1419 |
+
"learning_rate": 5.618181818181819e-07,
|
| 1420 |
+
"loss": 0.0316,
|
| 1421 |
+
"step": 5700
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 2.16,
|
| 1425 |
+
"learning_rate": 5.163636363636364e-07,
|
| 1426 |
+
"loss": 0.0305,
|
| 1427 |
+
"step": 5725
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 2.16,
|
| 1431 |
+
"learning_rate": 4.70909090909091e-07,
|
| 1432 |
+
"loss": 0.0289,
|
| 1433 |
+
"step": 5750
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 2.17,
|
| 1437 |
+
"learning_rate": 4.2545454545454547e-07,
|
| 1438 |
+
"loss": 0.0296,
|
| 1439 |
+
"step": 5775
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 2.17,
|
| 1443 |
+
"learning_rate": 3.8e-07,
|
| 1444 |
+
"loss": 0.0319,
|
| 1445 |
+
"step": 5800
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 2.18,
|
| 1449 |
+
"learning_rate": 3.3454545454545456e-07,
|
| 1450 |
+
"loss": 0.0325,
|
| 1451 |
+
"step": 5825
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 2.18,
|
| 1455 |
+
"learning_rate": 2.890909090909091e-07,
|
| 1456 |
+
"loss": 0.0442,
|
| 1457 |
+
"step": 5850
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 2.18,
|
| 1461 |
+
"learning_rate": 2.436363636363637e-07,
|
| 1462 |
+
"loss": 0.0448,
|
| 1463 |
+
"step": 5875
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 2.19,
|
| 1467 |
+
"learning_rate": 1.9818181818181819e-07,
|
| 1468 |
+
"loss": 0.0407,
|
| 1469 |
+
"step": 5900
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 2.19,
|
| 1473 |
+
"learning_rate": 1.5272727272727275e-07,
|
| 1474 |
+
"loss": 0.0386,
|
| 1475 |
+
"step": 5925
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 2.2,
|
| 1479 |
+
"learning_rate": 1.0727272727272728e-07,
|
| 1480 |
+
"loss": 0.0327,
|
| 1481 |
+
"step": 5950
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 2.2,
|
| 1485 |
+
"learning_rate": 6.181818181818183e-08,
|
| 1486 |
+
"loss": 0.0378,
|
| 1487 |
+
"step": 5975
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 2.21,
|
| 1491 |
+
"learning_rate": 1.6363636363636366e-08,
|
| 1492 |
+
"loss": 0.037,
|
| 1493 |
+
"step": 6000
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 2.21,
|
| 1497 |
+
"eval_loss": 0.1331787109375,
|
| 1498 |
+
"eval_runtime": 4331.8871,
|
| 1499 |
+
"eval_samples_per_second": 3.463,
|
| 1500 |
+
"eval_steps_per_second": 0.217,
|
| 1501 |
+
"eval_wer": 4.557596215181799,
|
| 1502 |
+
"step": 6000
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 2.21,
|
| 1506 |
+
"step": 6000,
|
| 1507 |
+
"total_flos": 8.151660428090115e+20,
|
| 1508 |
+
"train_loss": 0.08706036122639974,
|
| 1509 |
+
"train_runtime": 112219.1034,
|
| 1510 |
+
"train_samples_per_second": 3.422,
|
| 1511 |
+
"train_steps_per_second": 0.053
|
| 1512 |
+
}
|
| 1513 |
+
],
|
| 1514 |
+
"max_steps": 6000,
|
| 1515 |
+
"num_train_epochs": 9223372036854775807,
|
| 1516 |
+
"total_flos": 8.151660428090115e+20,
|
| 1517 |
+
"trial_name": null,
|
| 1518 |
+
"trial_params": null
|
| 1519 |
+
}
|