End of training
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 100.0,
|
| 3 |
+
"eval_loss": 6.5278449058532715,
|
| 4 |
+
"eval_runtime": 24.249,
|
| 5 |
+
"eval_samples": 493,
|
| 6 |
+
"eval_samples_per_second": 20.331,
|
| 7 |
+
"eval_steps_per_second": 10.186,
|
| 8 |
+
"eval_wer": 1.0,
|
| 9 |
+
"train_loss": 9.81739658203125,
|
| 10 |
+
"train_runtime": 23393.5837,
|
| 11 |
+
"train_samples": 6380,
|
| 12 |
+
"train_samples_per_second": 27.272,
|
| 13 |
+
"train_steps_per_second": 0.855
|
| 14 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 100.0,
|
| 3 |
+
"eval_loss": 6.5278449058532715,
|
| 4 |
+
"eval_runtime": 24.249,
|
| 5 |
+
"eval_samples": 493,
|
| 6 |
+
"eval_samples_per_second": 20.331,
|
| 7 |
+
"eval_steps_per_second": 10.186,
|
| 8 |
+
"eval_wer": 1.0
|
| 9 |
+
}
|
runs/Mar16_22-57-18_1b32746f485b/events.out.tfevents.1679004200.1b32746f485b.727724.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3a134e513a4caa1093ca285e39fce4c3553ee331d3e9225dba41021732eb4526
|
| 3 |
+
size 364
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 100.0,
|
| 3 |
+
"train_loss": 9.81739658203125,
|
| 4 |
+
"train_runtime": 23393.5837,
|
| 5 |
+
"train_samples": 6380,
|
| 6 |
+
"train_samples_per_second": 27.272,
|
| 7 |
+
"train_steps_per_second": 0.855
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1585 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 100.0,
|
| 5 |
+
"global_step": 20000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.5,
|
| 12 |
+
"learning_rate": 3.5624999999999998e-06,
|
| 13 |
+
"loss": 137.0348,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 1.0,
|
| 18 |
+
"learning_rate": 7.275e-06,
|
| 19 |
+
"loss": 106.075,
|
| 20 |
+
"step": 200
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 1.5,
|
| 24 |
+
"learning_rate": 1.1024999999999999e-05,
|
| 25 |
+
"loss": 95.1437,
|
| 26 |
+
"step": 300
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 2.0,
|
| 30 |
+
"learning_rate": 1.4775e-05,
|
| 31 |
+
"loss": 88.0386,
|
| 32 |
+
"step": 400
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 2.5,
|
| 36 |
+
"learning_rate": 1.8525e-05,
|
| 37 |
+
"loss": 82.628,
|
| 38 |
+
"step": 500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 2.5,
|
| 42 |
+
"eval_loss": 79.55873107910156,
|
| 43 |
+
"eval_runtime": 24.8027,
|
| 44 |
+
"eval_samples_per_second": 19.877,
|
| 45 |
+
"eval_steps_per_second": 9.959,
|
| 46 |
+
"eval_wer": 1.0,
|
| 47 |
+
"step": 500
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 3.0,
|
| 51 |
+
"learning_rate": 2.2274999999999996e-05,
|
| 52 |
+
"loss": 73.1146,
|
| 53 |
+
"step": 600
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 3.5,
|
| 57 |
+
"learning_rate": 2.6024999999999996e-05,
|
| 58 |
+
"loss": 62.7571,
|
| 59 |
+
"step": 700
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 4.0,
|
| 63 |
+
"learning_rate": 2.9775e-05,
|
| 64 |
+
"loss": 48.9782,
|
| 65 |
+
"step": 800
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 4.5,
|
| 69 |
+
"learning_rate": 3.3524999999999995e-05,
|
| 70 |
+
"loss": 33.2033,
|
| 71 |
+
"step": 900
|
| 72 |
+
},
|
| 73 |
+
{
|
| 74 |
+
"epoch": 5.0,
|
| 75 |
+
"learning_rate": 3.7275e-05,
|
| 76 |
+
"loss": 17.5635,
|
| 77 |
+
"step": 1000
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 5.0,
|
| 81 |
+
"eval_loss": 11.592885971069336,
|
| 82 |
+
"eval_runtime": 24.3352,
|
| 83 |
+
"eval_samples_per_second": 20.259,
|
| 84 |
+
"eval_steps_per_second": 10.15,
|
| 85 |
+
"eval_wer": 1.0,
|
| 86 |
+
"step": 1000
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 5.5,
|
| 90 |
+
"learning_rate": 4.1025e-05,
|
| 91 |
+
"loss": 8.6246,
|
| 92 |
+
"step": 1100
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 6.0,
|
| 96 |
+
"learning_rate": 4.4775e-05,
|
| 97 |
+
"loss": 6.5245,
|
| 98 |
+
"step": 1200
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 6.5,
|
| 102 |
+
"learning_rate": 4.8525e-05,
|
| 103 |
+
"loss": 6.4322,
|
| 104 |
+
"step": 1300
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 7.0,
|
| 108 |
+
"learning_rate": 5.227499999999999e-05,
|
| 109 |
+
"loss": 6.4273,
|
| 110 |
+
"step": 1400
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 7.5,
|
| 114 |
+
"learning_rate": 5.602499999999999e-05,
|
| 115 |
+
"loss": 6.4288,
|
| 116 |
+
"step": 1500
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 7.5,
|
| 120 |
+
"eval_loss": 6.447471618652344,
|
| 121 |
+
"eval_runtime": 24.3824,
|
| 122 |
+
"eval_samples_per_second": 20.219,
|
| 123 |
+
"eval_steps_per_second": 10.13,
|
| 124 |
+
"eval_wer": 1.0,
|
| 125 |
+
"step": 1500
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 8.0,
|
| 129 |
+
"learning_rate": 5.9774999999999996e-05,
|
| 130 |
+
"loss": 6.4087,
|
| 131 |
+
"step": 1600
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 8.5,
|
| 135 |
+
"learning_rate": 6.3525e-05,
|
| 136 |
+
"loss": 6.4099,
|
| 137 |
+
"step": 1700
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 9.0,
|
| 141 |
+
"learning_rate": 6.7275e-05,
|
| 142 |
+
"loss": 6.4216,
|
| 143 |
+
"step": 1800
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 9.5,
|
| 147 |
+
"learning_rate": 7.102499999999999e-05,
|
| 148 |
+
"loss": 6.4159,
|
| 149 |
+
"step": 1900
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 10.0,
|
| 153 |
+
"learning_rate": 7.477499999999999e-05,
|
| 154 |
+
"loss": 6.4092,
|
| 155 |
+
"step": 2000
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 10.0,
|
| 159 |
+
"eval_loss": 6.457908630371094,
|
| 160 |
+
"eval_runtime": 24.1035,
|
| 161 |
+
"eval_samples_per_second": 20.453,
|
| 162 |
+
"eval_steps_per_second": 10.247,
|
| 163 |
+
"eval_wer": 1.0,
|
| 164 |
+
"step": 2000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 10.5,
|
| 168 |
+
"learning_rate": 7.460833333333332e-05,
|
| 169 |
+
"loss": 6.4008,
|
| 170 |
+
"step": 2100
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 11.0,
|
| 174 |
+
"learning_rate": 7.419166666666666e-05,
|
| 175 |
+
"loss": 6.4257,
|
| 176 |
+
"step": 2200
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 11.5,
|
| 180 |
+
"learning_rate": 7.377499999999999e-05,
|
| 181 |
+
"loss": 6.4109,
|
| 182 |
+
"step": 2300
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 12.0,
|
| 186 |
+
"learning_rate": 7.335833333333333e-05,
|
| 187 |
+
"loss": 6.4135,
|
| 188 |
+
"step": 2400
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 12.5,
|
| 192 |
+
"learning_rate": 7.294166666666666e-05,
|
| 193 |
+
"loss": 6.3982,
|
| 194 |
+
"step": 2500
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 12.5,
|
| 198 |
+
"eval_loss": 6.466182708740234,
|
| 199 |
+
"eval_runtime": 24.3477,
|
| 200 |
+
"eval_samples_per_second": 20.248,
|
| 201 |
+
"eval_steps_per_second": 10.145,
|
| 202 |
+
"eval_wer": 1.0,
|
| 203 |
+
"step": 2500
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 13.0,
|
| 207 |
+
"learning_rate": 7.252499999999999e-05,
|
| 208 |
+
"loss": 6.4234,
|
| 209 |
+
"step": 2600
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 13.5,
|
| 213 |
+
"learning_rate": 7.210833333333333e-05,
|
| 214 |
+
"loss": 6.4333,
|
| 215 |
+
"step": 2700
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 14.0,
|
| 219 |
+
"learning_rate": 7.169166666666666e-05,
|
| 220 |
+
"loss": 6.3853,
|
| 221 |
+
"step": 2800
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 14.5,
|
| 225 |
+
"learning_rate": 7.1275e-05,
|
| 226 |
+
"loss": 6.4267,
|
| 227 |
+
"step": 2900
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 15.0,
|
| 231 |
+
"learning_rate": 7.085833333333333e-05,
|
| 232 |
+
"loss": 6.391,
|
| 233 |
+
"step": 3000
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 15.0,
|
| 237 |
+
"eval_loss": 6.465456008911133,
|
| 238 |
+
"eval_runtime": 24.1868,
|
| 239 |
+
"eval_samples_per_second": 20.383,
|
| 240 |
+
"eval_steps_per_second": 10.212,
|
| 241 |
+
"eval_wer": 1.0,
|
| 242 |
+
"step": 3000
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 15.5,
|
| 246 |
+
"learning_rate": 7.044166666666665e-05,
|
| 247 |
+
"loss": 6.4101,
|
| 248 |
+
"step": 3100
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 16.0,
|
| 252 |
+
"learning_rate": 7.0025e-05,
|
| 253 |
+
"loss": 6.4082,
|
| 254 |
+
"step": 3200
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 16.5,
|
| 258 |
+
"learning_rate": 6.960833333333332e-05,
|
| 259 |
+
"loss": 6.4155,
|
| 260 |
+
"step": 3300
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 17.0,
|
| 264 |
+
"learning_rate": 6.919166666666666e-05,
|
| 265 |
+
"loss": 6.4019,
|
| 266 |
+
"step": 3400
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 17.5,
|
| 270 |
+
"learning_rate": 6.877499999999999e-05,
|
| 271 |
+
"loss": 6.4097,
|
| 272 |
+
"step": 3500
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 17.5,
|
| 276 |
+
"eval_loss": 6.469109535217285,
|
| 277 |
+
"eval_runtime": 24.7207,
|
| 278 |
+
"eval_samples_per_second": 19.943,
|
| 279 |
+
"eval_steps_per_second": 9.992,
|
| 280 |
+
"eval_wer": 1.0,
|
| 281 |
+
"step": 3500
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 18.0,
|
| 285 |
+
"learning_rate": 6.835833333333332e-05,
|
| 286 |
+
"loss": 6.4073,
|
| 287 |
+
"step": 3600
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 18.5,
|
| 291 |
+
"learning_rate": 6.794166666666666e-05,
|
| 292 |
+
"loss": 6.4142,
|
| 293 |
+
"step": 3700
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 19.0,
|
| 297 |
+
"learning_rate": 6.752499999999999e-05,
|
| 298 |
+
"loss": 6.4013,
|
| 299 |
+
"step": 3800
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 19.5,
|
| 303 |
+
"learning_rate": 6.710833333333333e-05,
|
| 304 |
+
"loss": 6.4181,
|
| 305 |
+
"step": 3900
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 20.0,
|
| 309 |
+
"learning_rate": 6.669166666666666e-05,
|
| 310 |
+
"loss": 6.3986,
|
| 311 |
+
"step": 4000
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 20.0,
|
| 315 |
+
"eval_loss": 6.470158100128174,
|
| 316 |
+
"eval_runtime": 24.0989,
|
| 317 |
+
"eval_samples_per_second": 20.457,
|
| 318 |
+
"eval_steps_per_second": 10.249,
|
| 319 |
+
"eval_wer": 1.0,
|
| 320 |
+
"step": 4000
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 20.5,
|
| 324 |
+
"learning_rate": 6.6275e-05,
|
| 325 |
+
"loss": 6.4044,
|
| 326 |
+
"step": 4100
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 21.0,
|
| 330 |
+
"learning_rate": 6.585833333333333e-05,
|
| 331 |
+
"loss": 6.4134,
|
| 332 |
+
"step": 4200
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 21.5,
|
| 336 |
+
"learning_rate": 6.544166666666665e-05,
|
| 337 |
+
"loss": 6.4066,
|
| 338 |
+
"step": 4300
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 22.0,
|
| 342 |
+
"learning_rate": 6.5025e-05,
|
| 343 |
+
"loss": 6.4098,
|
| 344 |
+
"step": 4400
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 22.5,
|
| 348 |
+
"learning_rate": 6.460833333333332e-05,
|
| 349 |
+
"loss": 6.4069,
|
| 350 |
+
"step": 4500
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 22.5,
|
| 354 |
+
"eval_loss": 6.47609806060791,
|
| 355 |
+
"eval_runtime": 24.5428,
|
| 356 |
+
"eval_samples_per_second": 20.087,
|
| 357 |
+
"eval_steps_per_second": 10.064,
|
| 358 |
+
"eval_wer": 1.0,
|
| 359 |
+
"step": 4500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 23.0,
|
| 363 |
+
"learning_rate": 6.419166666666666e-05,
|
| 364 |
+
"loss": 6.4075,
|
| 365 |
+
"step": 4600
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 23.5,
|
| 369 |
+
"learning_rate": 6.377499999999999e-05,
|
| 370 |
+
"loss": 6.4069,
|
| 371 |
+
"step": 4700
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 24.0,
|
| 375 |
+
"learning_rate": 6.335833333333332e-05,
|
| 376 |
+
"loss": 6.407,
|
| 377 |
+
"step": 4800
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 24.5,
|
| 381 |
+
"learning_rate": 6.294166666666666e-05,
|
| 382 |
+
"loss": 6.3966,
|
| 383 |
+
"step": 4900
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 25.0,
|
| 387 |
+
"learning_rate": 6.252499999999999e-05,
|
| 388 |
+
"loss": 6.4158,
|
| 389 |
+
"step": 5000
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 25.0,
|
| 393 |
+
"eval_loss": 6.47500467300415,
|
| 394 |
+
"eval_runtime": 24.1455,
|
| 395 |
+
"eval_samples_per_second": 20.418,
|
| 396 |
+
"eval_steps_per_second": 10.23,
|
| 397 |
+
"eval_wer": 1.0,
|
| 398 |
+
"step": 5000
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 25.5,
|
| 402 |
+
"learning_rate": 6.210833333333333e-05,
|
| 403 |
+
"loss": 6.4081,
|
| 404 |
+
"step": 5100
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 26.0,
|
| 408 |
+
"learning_rate": 6.169166666666667e-05,
|
| 409 |
+
"loss": 6.4024,
|
| 410 |
+
"step": 5200
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 26.5,
|
| 414 |
+
"learning_rate": 6.127499999999999e-05,
|
| 415 |
+
"loss": 6.4122,
|
| 416 |
+
"step": 5300
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 27.0,
|
| 420 |
+
"learning_rate": 6.0858333333333326e-05,
|
| 421 |
+
"loss": 6.4011,
|
| 422 |
+
"step": 5400
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 27.5,
|
| 426 |
+
"learning_rate": 6.044166666666666e-05,
|
| 427 |
+
"loss": 6.4117,
|
| 428 |
+
"step": 5500
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 27.5,
|
| 432 |
+
"eval_loss": 6.48163366317749,
|
| 433 |
+
"eval_runtime": 24.2083,
|
| 434 |
+
"eval_samples_per_second": 20.365,
|
| 435 |
+
"eval_steps_per_second": 10.203,
|
| 436 |
+
"eval_wer": 1.0,
|
| 437 |
+
"step": 5500
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 28.0,
|
| 441 |
+
"learning_rate": 6.0024999999999995e-05,
|
| 442 |
+
"loss": 6.4013,
|
| 443 |
+
"step": 5600
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 28.5,
|
| 447 |
+
"learning_rate": 5.960833333333333e-05,
|
| 448 |
+
"loss": 6.4158,
|
| 449 |
+
"step": 5700
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 29.0,
|
| 453 |
+
"learning_rate": 5.9191666666666664e-05,
|
| 454 |
+
"loss": 6.3951,
|
| 455 |
+
"step": 5800
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 29.5,
|
| 459 |
+
"learning_rate": 5.877499999999999e-05,
|
| 460 |
+
"loss": 6.4047,
|
| 461 |
+
"step": 5900
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 30.0,
|
| 465 |
+
"learning_rate": 5.8358333333333327e-05,
|
| 466 |
+
"loss": 6.4086,
|
| 467 |
+
"step": 6000
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 30.0,
|
| 471 |
+
"eval_loss": 6.4805707931518555,
|
| 472 |
+
"eval_runtime": 24.395,
|
| 473 |
+
"eval_samples_per_second": 20.209,
|
| 474 |
+
"eval_steps_per_second": 10.125,
|
| 475 |
+
"eval_wer": 1.0,
|
| 476 |
+
"step": 6000
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 30.5,
|
| 480 |
+
"learning_rate": 5.794166666666666e-05,
|
| 481 |
+
"loss": 6.3924,
|
| 482 |
+
"step": 6100
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 31.0,
|
| 486 |
+
"learning_rate": 5.7524999999999996e-05,
|
| 487 |
+
"loss": 6.4216,
|
| 488 |
+
"step": 6200
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 31.5,
|
| 492 |
+
"learning_rate": 5.710833333333333e-05,
|
| 493 |
+
"loss": 6.3965,
|
| 494 |
+
"step": 6300
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 32.0,
|
| 498 |
+
"learning_rate": 5.6691666666666665e-05,
|
| 499 |
+
"loss": 6.4165,
|
| 500 |
+
"step": 6400
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 32.5,
|
| 504 |
+
"learning_rate": 5.627499999999999e-05,
|
| 505 |
+
"loss": 6.3992,
|
| 506 |
+
"step": 6500
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 32.5,
|
| 510 |
+
"eval_loss": 6.487205505371094,
|
| 511 |
+
"eval_runtime": 24.2919,
|
| 512 |
+
"eval_samples_per_second": 20.295,
|
| 513 |
+
"eval_steps_per_second": 10.168,
|
| 514 |
+
"eval_wer": 1.0,
|
| 515 |
+
"step": 6500
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 33.0,
|
| 519 |
+
"learning_rate": 5.585833333333333e-05,
|
| 520 |
+
"loss": 6.414,
|
| 521 |
+
"step": 6600
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 33.5,
|
| 525 |
+
"learning_rate": 5.544166666666666e-05,
|
| 526 |
+
"loss": 6.4138,
|
| 527 |
+
"step": 6700
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 34.0,
|
| 531 |
+
"learning_rate": 5.5024999999999996e-05,
|
| 532 |
+
"loss": 6.3973,
|
| 533 |
+
"step": 6800
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 34.5,
|
| 537 |
+
"learning_rate": 5.460833333333333e-05,
|
| 538 |
+
"loss": 6.4154,
|
| 539 |
+
"step": 6900
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 35.0,
|
| 543 |
+
"learning_rate": 5.419166666666666e-05,
|
| 544 |
+
"loss": 6.3946,
|
| 545 |
+
"step": 7000
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 35.0,
|
| 549 |
+
"eval_loss": 6.486586570739746,
|
| 550 |
+
"eval_runtime": 24.3253,
|
| 551 |
+
"eval_samples_per_second": 20.267,
|
| 552 |
+
"eval_steps_per_second": 10.154,
|
| 553 |
+
"eval_wer": 1.0,
|
| 554 |
+
"step": 7000
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 35.5,
|
| 558 |
+
"learning_rate": 5.377499999999999e-05,
|
| 559 |
+
"loss": 6.4203,
|
| 560 |
+
"step": 7100
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 36.0,
|
| 564 |
+
"learning_rate": 5.335833333333333e-05,
|
| 565 |
+
"loss": 6.3905,
|
| 566 |
+
"step": 7200
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 36.5,
|
| 570 |
+
"learning_rate": 5.294166666666666e-05,
|
| 571 |
+
"loss": 6.4104,
|
| 572 |
+
"step": 7300
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 37.0,
|
| 576 |
+
"learning_rate": 5.2524999999999996e-05,
|
| 577 |
+
"loss": 6.4002,
|
| 578 |
+
"step": 7400
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 37.5,
|
| 582 |
+
"learning_rate": 5.210833333333333e-05,
|
| 583 |
+
"loss": 6.4212,
|
| 584 |
+
"step": 7500
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 37.5,
|
| 588 |
+
"eval_loss": 6.489477634429932,
|
| 589 |
+
"eval_runtime": 24.1492,
|
| 590 |
+
"eval_samples_per_second": 20.415,
|
| 591 |
+
"eval_steps_per_second": 10.228,
|
| 592 |
+
"eval_wer": 1.0,
|
| 593 |
+
"step": 7500
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 38.0,
|
| 597 |
+
"learning_rate": 5.169166666666666e-05,
|
| 598 |
+
"loss": 6.3871,
|
| 599 |
+
"step": 7600
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 38.5,
|
| 603 |
+
"learning_rate": 5.127499999999999e-05,
|
| 604 |
+
"loss": 6.3982,
|
| 605 |
+
"step": 7700
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 39.0,
|
| 609 |
+
"learning_rate": 5.085833333333333e-05,
|
| 610 |
+
"loss": 6.4097,
|
| 611 |
+
"step": 7800
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 39.5,
|
| 615 |
+
"learning_rate": 5.044166666666666e-05,
|
| 616 |
+
"loss": 6.4015,
|
| 617 |
+
"step": 7900
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 40.0,
|
| 621 |
+
"learning_rate": 5.0024999999999996e-05,
|
| 622 |
+
"loss": 6.4051,
|
| 623 |
+
"step": 8000
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 40.0,
|
| 627 |
+
"eval_loss": 6.492602348327637,
|
| 628 |
+
"eval_runtime": 24.1509,
|
| 629 |
+
"eval_samples_per_second": 20.413,
|
| 630 |
+
"eval_steps_per_second": 10.227,
|
| 631 |
+
"eval_wer": 1.0,
|
| 632 |
+
"step": 8000
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 40.5,
|
| 636 |
+
"learning_rate": 4.9608333333333324e-05,
|
| 637 |
+
"loss": 6.3881,
|
| 638 |
+
"step": 8100
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 41.0,
|
| 642 |
+
"learning_rate": 4.919166666666666e-05,
|
| 643 |
+
"loss": 6.4196,
|
| 644 |
+
"step": 8200
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 41.5,
|
| 648 |
+
"learning_rate": 4.877499999999999e-05,
|
| 649 |
+
"loss": 6.4132,
|
| 650 |
+
"step": 8300
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 42.0,
|
| 654 |
+
"learning_rate": 4.835833333333333e-05,
|
| 655 |
+
"loss": 6.3967,
|
| 656 |
+
"step": 8400
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 42.5,
|
| 660 |
+
"learning_rate": 4.794166666666666e-05,
|
| 661 |
+
"loss": 6.398,
|
| 662 |
+
"step": 8500
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 42.5,
|
| 666 |
+
"eval_loss": 6.50146484375,
|
| 667 |
+
"eval_runtime": 24.2251,
|
| 668 |
+
"eval_samples_per_second": 20.351,
|
| 669 |
+
"eval_steps_per_second": 10.196,
|
| 670 |
+
"eval_wer": 1.0,
|
| 671 |
+
"step": 8500
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 43.0,
|
| 675 |
+
"learning_rate": 4.7524999999999996e-05,
|
| 676 |
+
"loss": 6.4101,
|
| 677 |
+
"step": 8600
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 43.5,
|
| 681 |
+
"learning_rate": 4.7108333333333324e-05,
|
| 682 |
+
"loss": 6.3947,
|
| 683 |
+
"step": 8700
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 44.0,
|
| 687 |
+
"learning_rate": 4.669166666666666e-05,
|
| 688 |
+
"loss": 6.4115,
|
| 689 |
+
"step": 8800
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 44.5,
|
| 693 |
+
"learning_rate": 4.627499999999999e-05,
|
| 694 |
+
"loss": 6.4088,
|
| 695 |
+
"step": 8900
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 45.0,
|
| 699 |
+
"learning_rate": 4.585833333333333e-05,
|
| 700 |
+
"loss": 6.3967,
|
| 701 |
+
"step": 9000
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 45.0,
|
| 705 |
+
"eval_loss": 6.496038913726807,
|
| 706 |
+
"eval_runtime": 24.4306,
|
| 707 |
+
"eval_samples_per_second": 20.18,
|
| 708 |
+
"eval_steps_per_second": 10.11,
|
| 709 |
+
"eval_wer": 1.0,
|
| 710 |
+
"step": 9000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 45.5,
|
| 714 |
+
"learning_rate": 4.544166666666666e-05,
|
| 715 |
+
"loss": 6.401,
|
| 716 |
+
"step": 9100
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 46.0,
|
| 720 |
+
"learning_rate": 4.5025e-05,
|
| 721 |
+
"loss": 6.4063,
|
| 722 |
+
"step": 9200
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 46.5,
|
| 726 |
+
"learning_rate": 4.4608333333333324e-05,
|
| 727 |
+
"loss": 6.4101,
|
| 728 |
+
"step": 9300
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 47.0,
|
| 732 |
+
"learning_rate": 4.419166666666666e-05,
|
| 733 |
+
"loss": 6.3957,
|
| 734 |
+
"step": 9400
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 47.5,
|
| 738 |
+
"learning_rate": 4.377499999999999e-05,
|
| 739 |
+
"loss": 6.4096,
|
| 740 |
+
"step": 9500
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 47.5,
|
| 744 |
+
"eval_loss": 6.500304222106934,
|
| 745 |
+
"eval_runtime": 24.1983,
|
| 746 |
+
"eval_samples_per_second": 20.373,
|
| 747 |
+
"eval_steps_per_second": 10.207,
|
| 748 |
+
"eval_wer": 1.0,
|
| 749 |
+
"step": 9500
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 48.0,
|
| 753 |
+
"learning_rate": 4.335833333333333e-05,
|
| 754 |
+
"loss": 6.3966,
|
| 755 |
+
"step": 9600
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 48.5,
|
| 759 |
+
"learning_rate": 4.294166666666667e-05,
|
| 760 |
+
"loss": 6.409,
|
| 761 |
+
"step": 9700
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 49.0,
|
| 765 |
+
"learning_rate": 4.252499999999999e-05,
|
| 766 |
+
"loss": 6.3979,
|
| 767 |
+
"step": 9800
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 49.5,
|
| 771 |
+
"learning_rate": 4.2108333333333325e-05,
|
| 772 |
+
"loss": 6.3985,
|
| 773 |
+
"step": 9900
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 50.0,
|
| 777 |
+
"learning_rate": 4.169166666666666e-05,
|
| 778 |
+
"loss": 6.4068,
|
| 779 |
+
"step": 10000
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 50.0,
|
| 783 |
+
"eval_loss": 6.502621650695801,
|
| 784 |
+
"eval_runtime": 24.1622,
|
| 785 |
+
"eval_samples_per_second": 20.404,
|
| 786 |
+
"eval_steps_per_second": 10.223,
|
| 787 |
+
"eval_wer": 1.0,
|
| 788 |
+
"step": 10000
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 50.5,
|
| 792 |
+
"learning_rate": 4.1274999999999994e-05,
|
| 793 |
+
"loss": 6.4059,
|
| 794 |
+
"step": 10100
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 51.0,
|
| 798 |
+
"learning_rate": 4.0858333333333335e-05,
|
| 799 |
+
"loss": 6.4012,
|
| 800 |
+
"step": 10200
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 51.5,
|
| 804 |
+
"learning_rate": 4.044166666666667e-05,
|
| 805 |
+
"loss": 6.4118,
|
| 806 |
+
"step": 10300
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 52.0,
|
| 810 |
+
"learning_rate": 4.002499999999999e-05,
|
| 811 |
+
"loss": 6.3924,
|
| 812 |
+
"step": 10400
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 52.5,
|
| 816 |
+
"learning_rate": 3.9608333333333325e-05,
|
| 817 |
+
"loss": 6.4062,
|
| 818 |
+
"step": 10500
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 52.5,
|
| 822 |
+
"eval_loss": 6.507063865661621,
|
| 823 |
+
"eval_runtime": 24.2765,
|
| 824 |
+
"eval_samples_per_second": 20.308,
|
| 825 |
+
"eval_steps_per_second": 10.174,
|
| 826 |
+
"eval_wer": 1.0,
|
| 827 |
+
"step": 10500
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 53.0,
|
| 831 |
+
"learning_rate": 3.9191666666666666e-05,
|
| 832 |
+
"loss": 6.3992,
|
| 833 |
+
"step": 10600
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 53.5,
|
| 837 |
+
"learning_rate": 3.8775e-05,
|
| 838 |
+
"loss": 6.4136,
|
| 839 |
+
"step": 10700
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 54.0,
|
| 843 |
+
"learning_rate": 3.8358333333333335e-05,
|
| 844 |
+
"loss": 6.3915,
|
| 845 |
+
"step": 10800
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 54.5,
|
| 849 |
+
"learning_rate": 3.794166666666667e-05,
|
| 850 |
+
"loss": 6.4096,
|
| 851 |
+
"step": 10900
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 55.0,
|
| 855 |
+
"learning_rate": 3.7525e-05,
|
| 856 |
+
"loss": 6.395,
|
| 857 |
+
"step": 11000
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 55.0,
|
| 861 |
+
"eval_loss": 6.506567001342773,
|
| 862 |
+
"eval_runtime": 24.2863,
|
| 863 |
+
"eval_samples_per_second": 20.3,
|
| 864 |
+
"eval_steps_per_second": 10.17,
|
| 865 |
+
"eval_wer": 1.0,
|
| 866 |
+
"step": 11000
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 55.5,
|
| 870 |
+
"learning_rate": 3.710833333333333e-05,
|
| 871 |
+
"loss": 6.4042,
|
| 872 |
+
"step": 11100
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 56.0,
|
| 876 |
+
"learning_rate": 3.6691666666666666e-05,
|
| 877 |
+
"loss": 6.402,
|
| 878 |
+
"step": 11200
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 56.5,
|
| 882 |
+
"learning_rate": 3.6275e-05,
|
| 883 |
+
"loss": 6.4056,
|
| 884 |
+
"step": 11300
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 57.0,
|
| 888 |
+
"learning_rate": 3.585833333333333e-05,
|
| 889 |
+
"loss": 6.4005,
|
| 890 |
+
"step": 11400
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 57.5,
|
| 894 |
+
"learning_rate": 3.544166666666666e-05,
|
| 895 |
+
"loss": 6.4079,
|
| 896 |
+
"step": 11500
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 57.5,
|
| 900 |
+
"eval_loss": 6.50927734375,
|
| 901 |
+
"eval_runtime": 24.3197,
|
| 902 |
+
"eval_samples_per_second": 20.272,
|
| 903 |
+
"eval_steps_per_second": 10.156,
|
| 904 |
+
"eval_wer": 1.0,
|
| 905 |
+
"step": 11500
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 58.0,
|
| 909 |
+
"learning_rate": 3.5025e-05,
|
| 910 |
+
"loss": 6.3957,
|
| 911 |
+
"step": 11600
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 58.5,
|
| 915 |
+
"learning_rate": 3.460833333333333e-05,
|
| 916 |
+
"loss": 6.4057,
|
| 917 |
+
"step": 11700
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 59.0,
|
| 921 |
+
"learning_rate": 3.4191666666666667e-05,
|
| 922 |
+
"loss": 6.3981,
|
| 923 |
+
"step": 11800
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 59.5,
|
| 927 |
+
"learning_rate": 3.3774999999999994e-05,
|
| 928 |
+
"loss": 6.394,
|
| 929 |
+
"step": 11900
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 60.0,
|
| 933 |
+
"learning_rate": 3.335833333333333e-05,
|
| 934 |
+
"loss": 6.411,
|
| 935 |
+
"step": 12000
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 60.0,
|
| 939 |
+
"eval_loss": 6.510580539703369,
|
| 940 |
+
"eval_runtime": 24.2221,
|
| 941 |
+
"eval_samples_per_second": 20.353,
|
| 942 |
+
"eval_steps_per_second": 10.197,
|
| 943 |
+
"eval_wer": 1.0,
|
| 944 |
+
"step": 12000
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 60.5,
|
| 948 |
+
"learning_rate": 3.294166666666666e-05,
|
| 949 |
+
"loss": 6.3902,
|
| 950 |
+
"step": 12100
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 61.0,
|
| 954 |
+
"learning_rate": 3.2525e-05,
|
| 955 |
+
"loss": 6.413,
|
| 956 |
+
"step": 12200
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 61.5,
|
| 960 |
+
"learning_rate": 3.210833333333333e-05,
|
| 961 |
+
"loss": 6.3998,
|
| 962 |
+
"step": 12300
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 62.0,
|
| 966 |
+
"learning_rate": 3.169166666666667e-05,
|
| 967 |
+
"loss": 6.404,
|
| 968 |
+
"step": 12400
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 62.5,
|
| 972 |
+
"learning_rate": 3.1274999999999994e-05,
|
| 973 |
+
"loss": 6.4023,
|
| 974 |
+
"step": 12500
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 62.5,
|
| 978 |
+
"eval_loss": 6.511165142059326,
|
| 979 |
+
"eval_runtime": 24.2209,
|
| 980 |
+
"eval_samples_per_second": 20.354,
|
| 981 |
+
"eval_steps_per_second": 10.198,
|
| 982 |
+
"eval_wer": 1.0,
|
| 983 |
+
"step": 12500
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 63.0,
|
| 987 |
+
"learning_rate": 3.085833333333333e-05,
|
| 988 |
+
"loss": 6.3992,
|
| 989 |
+
"step": 12600
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 63.5,
|
| 993 |
+
"learning_rate": 3.0441666666666667e-05,
|
| 994 |
+
"loss": 6.4093,
|
| 995 |
+
"step": 12700
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 64.0,
|
| 999 |
+
"learning_rate": 3.0024999999999995e-05,
|
| 1000 |
+
"loss": 6.3919,
|
| 1001 |
+
"step": 12800
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 64.5,
|
| 1005 |
+
"learning_rate": 2.9608333333333332e-05,
|
| 1006 |
+
"loss": 6.3972,
|
| 1007 |
+
"step": 12900
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 65.0,
|
| 1011 |
+
"learning_rate": 2.9191666666666667e-05,
|
| 1012 |
+
"loss": 6.4053,
|
| 1013 |
+
"step": 13000
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 65.0,
|
| 1017 |
+
"eval_loss": 6.51427698135376,
|
| 1018 |
+
"eval_runtime": 24.1929,
|
| 1019 |
+
"eval_samples_per_second": 20.378,
|
| 1020 |
+
"eval_steps_per_second": 10.21,
|
| 1021 |
+
"eval_wer": 1.0,
|
| 1022 |
+
"step": 13000
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 65.5,
|
| 1026 |
+
"learning_rate": 2.8774999999999998e-05,
|
| 1027 |
+
"loss": 6.3993,
|
| 1028 |
+
"step": 13100
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 66.0,
|
| 1032 |
+
"learning_rate": 2.8358333333333333e-05,
|
| 1033 |
+
"loss": 6.403,
|
| 1034 |
+
"step": 13200
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 66.5,
|
| 1038 |
+
"learning_rate": 2.7941666666666664e-05,
|
| 1039 |
+
"loss": 6.3973,
|
| 1040 |
+
"step": 13300
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 67.0,
|
| 1044 |
+
"learning_rate": 2.7524999999999998e-05,
|
| 1045 |
+
"loss": 6.405,
|
| 1046 |
+
"step": 13400
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 67.5,
|
| 1050 |
+
"learning_rate": 2.7108333333333333e-05,
|
| 1051 |
+
"loss": 6.4103,
|
| 1052 |
+
"step": 13500
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 67.5,
|
| 1056 |
+
"eval_loss": 6.517162322998047,
|
| 1057 |
+
"eval_runtime": 24.3081,
|
| 1058 |
+
"eval_samples_per_second": 20.281,
|
| 1059 |
+
"eval_steps_per_second": 10.161,
|
| 1060 |
+
"eval_wer": 1.0,
|
| 1061 |
+
"step": 13500
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 68.0,
|
| 1065 |
+
"learning_rate": 2.6691666666666664e-05,
|
| 1066 |
+
"loss": 6.3907,
|
| 1067 |
+
"step": 13600
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 68.5,
|
| 1071 |
+
"learning_rate": 2.6275e-05,
|
| 1072 |
+
"loss": 6.3932,
|
| 1073 |
+
"step": 13700
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 69.0,
|
| 1077 |
+
"learning_rate": 2.5858333333333333e-05,
|
| 1078 |
+
"loss": 6.4082,
|
| 1079 |
+
"step": 13800
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 69.5,
|
| 1083 |
+
"learning_rate": 2.5441666666666664e-05,
|
| 1084 |
+
"loss": 6.4119,
|
| 1085 |
+
"step": 13900
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 70.0,
|
| 1089 |
+
"learning_rate": 2.5025e-05,
|
| 1090 |
+
"loss": 6.3899,
|
| 1091 |
+
"step": 14000
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 70.0,
|
| 1095 |
+
"eval_loss": 6.518171310424805,
|
| 1096 |
+
"eval_runtime": 24.1998,
|
| 1097 |
+
"eval_samples_per_second": 20.372,
|
| 1098 |
+
"eval_steps_per_second": 10.207,
|
| 1099 |
+
"eval_wer": 1.0,
|
| 1100 |
+
"step": 14000
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 70.5,
|
| 1104 |
+
"learning_rate": 2.4608333333333333e-05,
|
| 1105 |
+
"loss": 6.4104,
|
| 1106 |
+
"step": 14100
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 71.0,
|
| 1110 |
+
"learning_rate": 2.4191666666666664e-05,
|
| 1111 |
+
"loss": 6.3909,
|
| 1112 |
+
"step": 14200
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 71.5,
|
| 1116 |
+
"learning_rate": 2.3775e-05,
|
| 1117 |
+
"loss": 6.3933,
|
| 1118 |
+
"step": 14300
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 72.0,
|
| 1122 |
+
"learning_rate": 2.3358333333333333e-05,
|
| 1123 |
+
"loss": 6.4065,
|
| 1124 |
+
"step": 14400
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 72.5,
|
| 1128 |
+
"learning_rate": 2.2941666666666664e-05,
|
| 1129 |
+
"loss": 6.4054,
|
| 1130 |
+
"step": 14500
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 72.5,
|
| 1134 |
+
"eval_loss": 6.519650936126709,
|
| 1135 |
+
"eval_runtime": 24.4435,
|
| 1136 |
+
"eval_samples_per_second": 20.169,
|
| 1137 |
+
"eval_steps_per_second": 10.105,
|
| 1138 |
+
"eval_wer": 1.0,
|
| 1139 |
+
"step": 14500
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 73.0,
|
| 1143 |
+
"learning_rate": 2.2525e-05,
|
| 1144 |
+
"loss": 6.3945,
|
| 1145 |
+
"step": 14600
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 73.5,
|
| 1149 |
+
"learning_rate": 2.2108333333333333e-05,
|
| 1150 |
+
"loss": 6.3972,
|
| 1151 |
+
"step": 14700
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 74.0,
|
| 1155 |
+
"learning_rate": 2.1691666666666664e-05,
|
| 1156 |
+
"loss": 6.4027,
|
| 1157 |
+
"step": 14800
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 74.5,
|
| 1161 |
+
"learning_rate": 2.1275e-05,
|
| 1162 |
+
"loss": 6.4082,
|
| 1163 |
+
"step": 14900
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 75.0,
|
| 1167 |
+
"learning_rate": 2.085833333333333e-05,
|
| 1168 |
+
"loss": 6.391,
|
| 1169 |
+
"step": 15000
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 75.0,
|
| 1173 |
+
"eval_loss": 6.51995849609375,
|
| 1174 |
+
"eval_runtime": 24.299,
|
| 1175 |
+
"eval_samples_per_second": 20.289,
|
| 1176 |
+
"eval_steps_per_second": 10.165,
|
| 1177 |
+
"eval_wer": 1.0,
|
| 1178 |
+
"step": 15000
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 75.5,
|
| 1182 |
+
"learning_rate": 2.0441666666666664e-05,
|
| 1183 |
+
"loss": 6.4077,
|
| 1184 |
+
"step": 15100
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 76.0,
|
| 1188 |
+
"learning_rate": 2.0025e-05,
|
| 1189 |
+
"loss": 6.3905,
|
| 1190 |
+
"step": 15200
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 76.5,
|
| 1194 |
+
"learning_rate": 1.960833333333333e-05,
|
| 1195 |
+
"loss": 6.3987,
|
| 1196 |
+
"step": 15300
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 77.0,
|
| 1200 |
+
"learning_rate": 1.9191666666666664e-05,
|
| 1201 |
+
"loss": 6.4004,
|
| 1202 |
+
"step": 15400
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 77.5,
|
| 1206 |
+
"learning_rate": 1.8775e-05,
|
| 1207 |
+
"loss": 6.3988,
|
| 1208 |
+
"step": 15500
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 77.5,
|
| 1212 |
+
"eval_loss": 6.522036075592041,
|
| 1213 |
+
"eval_runtime": 24.4796,
|
| 1214 |
+
"eval_samples_per_second": 20.139,
|
| 1215 |
+
"eval_steps_per_second": 10.09,
|
| 1216 |
+
"eval_wer": 1.0,
|
| 1217 |
+
"step": 15500
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 78.0,
|
| 1221 |
+
"learning_rate": 1.8358333333333333e-05,
|
| 1222 |
+
"loss": 6.3996,
|
| 1223 |
+
"step": 15600
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 78.5,
|
| 1227 |
+
"learning_rate": 1.7941666666666665e-05,
|
| 1228 |
+
"loss": 6.404,
|
| 1229 |
+
"step": 15700
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 79.0,
|
| 1233 |
+
"learning_rate": 1.7525e-05,
|
| 1234 |
+
"loss": 6.3918,
|
| 1235 |
+
"step": 15800
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 79.5,
|
| 1239 |
+
"learning_rate": 1.710833333333333e-05,
|
| 1240 |
+
"loss": 6.3915,
|
| 1241 |
+
"step": 15900
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 80.0,
|
| 1245 |
+
"learning_rate": 1.6691666666666665e-05,
|
| 1246 |
+
"loss": 6.4059,
|
| 1247 |
+
"step": 16000
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 80.0,
|
| 1251 |
+
"eval_loss": 6.522817611694336,
|
| 1252 |
+
"eval_runtime": 24.2749,
|
| 1253 |
+
"eval_samples_per_second": 20.309,
|
| 1254 |
+
"eval_steps_per_second": 10.175,
|
| 1255 |
+
"eval_wer": 1.0,
|
| 1256 |
+
"step": 16000
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 80.5,
|
| 1260 |
+
"learning_rate": 1.6275e-05,
|
| 1261 |
+
"loss": 6.3983,
|
| 1262 |
+
"step": 16100
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 81.0,
|
| 1266 |
+
"learning_rate": 1.585833333333333e-05,
|
| 1267 |
+
"loss": 6.3996,
|
| 1268 |
+
"step": 16200
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 81.5,
|
| 1272 |
+
"learning_rate": 1.5445833333333334e-05,
|
| 1273 |
+
"loss": 6.4038,
|
| 1274 |
+
"step": 16300
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 82.0,
|
| 1278 |
+
"learning_rate": 1.5029166666666665e-05,
|
| 1279 |
+
"loss": 6.3918,
|
| 1280 |
+
"step": 16400
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 82.5,
|
| 1284 |
+
"learning_rate": 1.4612499999999998e-05,
|
| 1285 |
+
"loss": 6.392,
|
| 1286 |
+
"step": 16500
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 82.5,
|
| 1290 |
+
"eval_loss": 6.523283004760742,
|
| 1291 |
+
"eval_runtime": 24.4538,
|
| 1292 |
+
"eval_samples_per_second": 20.16,
|
| 1293 |
+
"eval_steps_per_second": 10.101,
|
| 1294 |
+
"eval_wer": 1.0,
|
| 1295 |
+
"step": 16500
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 83.0,
|
| 1299 |
+
"learning_rate": 1.419583333333333e-05,
|
| 1300 |
+
"loss": 6.4043,
|
| 1301 |
+
"step": 16600
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 83.5,
|
| 1305 |
+
"learning_rate": 1.3779166666666667e-05,
|
| 1306 |
+
"loss": 6.3989,
|
| 1307 |
+
"step": 16700
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 84.0,
|
| 1311 |
+
"learning_rate": 1.33625e-05,
|
| 1312 |
+
"loss": 6.3956,
|
| 1313 |
+
"step": 16800
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 84.5,
|
| 1317 |
+
"learning_rate": 1.2945833333333332e-05,
|
| 1318 |
+
"loss": 6.403,
|
| 1319 |
+
"step": 16900
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 85.0,
|
| 1323 |
+
"learning_rate": 1.2529166666666667e-05,
|
| 1324 |
+
"loss": 6.3947,
|
| 1325 |
+
"step": 17000
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 85.0,
|
| 1329 |
+
"eval_loss": 6.525250434875488,
|
| 1330 |
+
"eval_runtime": 24.2494,
|
| 1331 |
+
"eval_samples_per_second": 20.33,
|
| 1332 |
+
"eval_steps_per_second": 10.186,
|
| 1333 |
+
"eval_wer": 1.0,
|
| 1334 |
+
"step": 17000
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 85.5,
|
| 1338 |
+
"learning_rate": 1.21125e-05,
|
| 1339 |
+
"loss": 6.4083,
|
| 1340 |
+
"step": 17100
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 86.0,
|
| 1344 |
+
"learning_rate": 1.1695833333333332e-05,
|
| 1345 |
+
"loss": 6.3874,
|
| 1346 |
+
"step": 17200
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 86.5,
|
| 1350 |
+
"learning_rate": 1.1279166666666665e-05,
|
| 1351 |
+
"loss": 6.3944,
|
| 1352 |
+
"step": 17300
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 87.0,
|
| 1356 |
+
"learning_rate": 1.08625e-05,
|
| 1357 |
+
"loss": 6.3974,
|
| 1358 |
+
"step": 17400
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 87.5,
|
| 1362 |
+
"learning_rate": 1.0445833333333333e-05,
|
| 1363 |
+
"loss": 6.3966,
|
| 1364 |
+
"step": 17500
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 87.5,
|
| 1368 |
+
"eval_loss": 6.525856971740723,
|
| 1369 |
+
"eval_runtime": 24.4696,
|
| 1370 |
+
"eval_samples_per_second": 20.147,
|
| 1371 |
+
"eval_steps_per_second": 10.094,
|
| 1372 |
+
"eval_wer": 1.0,
|
| 1373 |
+
"step": 17500
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 88.0,
|
| 1377 |
+
"learning_rate": 1.0029166666666665e-05,
|
| 1378 |
+
"loss": 6.397,
|
| 1379 |
+
"step": 17600
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 88.5,
|
| 1383 |
+
"learning_rate": 9.6125e-06,
|
| 1384 |
+
"loss": 6.4091,
|
| 1385 |
+
"step": 17700
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 89.0,
|
| 1389 |
+
"learning_rate": 9.195833333333333e-06,
|
| 1390 |
+
"loss": 6.3844,
|
| 1391 |
+
"step": 17800
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 89.5,
|
| 1395 |
+
"learning_rate": 8.779166666666665e-06,
|
| 1396 |
+
"loss": 6.4037,
|
| 1397 |
+
"step": 17900
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 90.0,
|
| 1401 |
+
"learning_rate": 8.3625e-06,
|
| 1402 |
+
"loss": 6.3905,
|
| 1403 |
+
"step": 18000
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 90.0,
|
| 1407 |
+
"eval_loss": 6.526439666748047,
|
| 1408 |
+
"eval_runtime": 24.3911,
|
| 1409 |
+
"eval_samples_per_second": 20.212,
|
| 1410 |
+
"eval_steps_per_second": 10.127,
|
| 1411 |
+
"eval_wer": 1.0,
|
| 1412 |
+
"step": 18000
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 90.5,
|
| 1416 |
+
"learning_rate": 7.945833333333333e-06,
|
| 1417 |
+
"loss": 6.4008,
|
| 1418 |
+
"step": 18100
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 91.0,
|
| 1422 |
+
"learning_rate": 7.529166666666666e-06,
|
| 1423 |
+
"loss": 6.3926,
|
| 1424 |
+
"step": 18200
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 91.5,
|
| 1428 |
+
"learning_rate": 7.116666666666666e-06,
|
| 1429 |
+
"loss": 6.3982,
|
| 1430 |
+
"step": 18300
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 92.0,
|
| 1434 |
+
"learning_rate": 6.699999999999999e-06,
|
| 1435 |
+
"loss": 6.3951,
|
| 1436 |
+
"step": 18400
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 92.5,
|
| 1440 |
+
"learning_rate": 6.283333333333333e-06,
|
| 1441 |
+
"loss": 6.4003,
|
| 1442 |
+
"step": 18500
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 92.5,
|
| 1446 |
+
"eval_loss": 6.527181148529053,
|
| 1447 |
+
"eval_runtime": 24.2624,
|
| 1448 |
+
"eval_samples_per_second": 20.32,
|
| 1449 |
+
"eval_steps_per_second": 10.18,
|
| 1450 |
+
"eval_wer": 1.0,
|
| 1451 |
+
"step": 18500
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 93.0,
|
| 1455 |
+
"learning_rate": 5.866666666666666e-06,
|
| 1456 |
+
"loss": 6.3932,
|
| 1457 |
+
"step": 18600
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 93.5,
|
| 1461 |
+
"learning_rate": 5.4499999999999995e-06,
|
| 1462 |
+
"loss": 6.4076,
|
| 1463 |
+
"step": 18700
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 94.0,
|
| 1467 |
+
"learning_rate": 5.033333333333332e-06,
|
| 1468 |
+
"loss": 6.3857,
|
| 1469 |
+
"step": 18800
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 94.5,
|
| 1473 |
+
"learning_rate": 4.616666666666667e-06,
|
| 1474 |
+
"loss": 6.4045,
|
| 1475 |
+
"step": 18900
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 95.0,
|
| 1479 |
+
"learning_rate": 4.2e-06,
|
| 1480 |
+
"loss": 6.3877,
|
| 1481 |
+
"step": 19000
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 95.0,
|
| 1485 |
+
"eval_loss": 6.527469635009766,
|
| 1486 |
+
"eval_runtime": 24.1993,
|
| 1487 |
+
"eval_samples_per_second": 20.372,
|
| 1488 |
+
"eval_steps_per_second": 10.207,
|
| 1489 |
+
"eval_wer": 1.0,
|
| 1490 |
+
"step": 19000
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 95.5,
|
| 1494 |
+
"learning_rate": 3.783333333333333e-06,
|
| 1495 |
+
"loss": 6.4,
|
| 1496 |
+
"step": 19100
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 96.0,
|
| 1500 |
+
"learning_rate": 3.3666666666666665e-06,
|
| 1501 |
+
"loss": 6.3918,
|
| 1502 |
+
"step": 19200
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 96.5,
|
| 1506 |
+
"learning_rate": 2.9499999999999997e-06,
|
| 1507 |
+
"loss": 6.4001,
|
| 1508 |
+
"step": 19300
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 97.0,
|
| 1512 |
+
"learning_rate": 2.533333333333333e-06,
|
| 1513 |
+
"loss": 6.3935,
|
| 1514 |
+
"step": 19400
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 97.5,
|
| 1518 |
+
"learning_rate": 2.1166666666666666e-06,
|
| 1519 |
+
"loss": 6.3903,
|
| 1520 |
+
"step": 19500
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 97.5,
|
| 1524 |
+
"eval_loss": 6.527697563171387,
|
| 1525 |
+
"eval_runtime": 24.3683,
|
| 1526 |
+
"eval_samples_per_second": 20.231,
|
| 1527 |
+
"eval_steps_per_second": 10.136,
|
| 1528 |
+
"eval_wer": 1.0,
|
| 1529 |
+
"step": 19500
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 98.0,
|
| 1533 |
+
"learning_rate": 1.7e-06,
|
| 1534 |
+
"loss": 6.4019,
|
| 1535 |
+
"step": 19600
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 98.5,
|
| 1539 |
+
"learning_rate": 1.2833333333333333e-06,
|
| 1540 |
+
"loss": 6.4,
|
| 1541 |
+
"step": 19700
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 99.0,
|
| 1545 |
+
"learning_rate": 8.666666666666666e-07,
|
| 1546 |
+
"loss": 6.3896,
|
| 1547 |
+
"step": 19800
|
| 1548 |
+
},
|
| 1549 |
+
{
|
| 1550 |
+
"epoch": 99.5,
|
| 1551 |
+
"learning_rate": 4.5e-07,
|
| 1552 |
+
"loss": 6.3972,
|
| 1553 |
+
"step": 19900
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 100.0,
|
| 1557 |
+
"learning_rate": 3.3333333333333334e-08,
|
| 1558 |
+
"loss": 6.3944,
|
| 1559 |
+
"step": 20000
|
| 1560 |
+
},
|
| 1561 |
+
{
|
| 1562 |
+
"epoch": 100.0,
|
| 1563 |
+
"eval_loss": 6.5278449058532715,
|
| 1564 |
+
"eval_runtime": 24.1079,
|
| 1565 |
+
"eval_samples_per_second": 20.45,
|
| 1566 |
+
"eval_steps_per_second": 10.246,
|
| 1567 |
+
"eval_wer": 1.0,
|
| 1568 |
+
"step": 20000
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 100.0,
|
| 1572 |
+
"step": 20000,
|
| 1573 |
+
"total_flos": 2.051320516956903e+19,
|
| 1574 |
+
"train_loss": 9.81739658203125,
|
| 1575 |
+
"train_runtime": 23393.5837,
|
| 1576 |
+
"train_samples_per_second": 27.272,
|
| 1577 |
+
"train_steps_per_second": 0.855
|
| 1578 |
+
}
|
| 1579 |
+
],
|
| 1580 |
+
"max_steps": 20000,
|
| 1581 |
+
"num_train_epochs": 100,
|
| 1582 |
+
"total_flos": 2.051320516956903e+19,
|
| 1583 |
+
"trial_name": null,
|
| 1584 |
+
"trial_params": null
|
| 1585 |
+
}
|