End of training
Browse files- all_results.json +13 -0
- eval_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +3775 -0
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 50.0,
|
| 3 |
+
"eval_accuracy": 0.9305555555555556,
|
| 4 |
+
"eval_loss": 0.2058059573173523,
|
| 5 |
+
"eval_runtime": 1.6811,
|
| 6 |
+
"eval_samples_per_second": 171.312,
|
| 7 |
+
"eval_steps_per_second": 2.974,
|
| 8 |
+
"total_flos": 1.0035482482400256e+19,
|
| 9 |
+
"train_loss": 0.24394690781493078,
|
| 10 |
+
"train_runtime": 2413.5397,
|
| 11 |
+
"train_samples_per_second": 53.656,
|
| 12 |
+
"train_steps_per_second": 0.228
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 50.0,
|
| 3 |
+
"eval_accuracy": 0.9305555555555556,
|
| 4 |
+
"eval_loss": 0.2058059573173523,
|
| 5 |
+
"eval_runtime": 1.6811,
|
| 6 |
+
"eval_samples_per_second": 171.312,
|
| 7 |
+
"eval_steps_per_second": 2.974
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 50.0,
|
| 3 |
+
"total_flos": 1.0035482482400256e+19,
|
| 4 |
+
"train_loss": 0.24394690781493078,
|
| 5 |
+
"train_runtime": 2413.5397,
|
| 6 |
+
"train_samples_per_second": 53.656,
|
| 7 |
+
"train_steps_per_second": 0.228
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3775 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9305555555555556,
|
| 3 |
+
"best_model_checkpoint": "vit-base-patch16-224-Soybean_11-46/checkpoint-473",
|
| 4 |
+
"epoch": 50.0,
|
| 5 |
+
"global_step": 550,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.09,
|
| 12 |
+
"learning_rate": 9.09090909090909e-07,
|
| 13 |
+
"loss": 1.677,
|
| 14 |
+
"step": 1
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.18,
|
| 18 |
+
"learning_rate": 1.818181818181818e-06,
|
| 19 |
+
"loss": 1.663,
|
| 20 |
+
"step": 2
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.27,
|
| 24 |
+
"learning_rate": 2.7272727272727272e-06,
|
| 25 |
+
"loss": 1.6847,
|
| 26 |
+
"step": 3
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.36,
|
| 30 |
+
"learning_rate": 3.636363636363636e-06,
|
| 31 |
+
"loss": 1.7074,
|
| 32 |
+
"step": 4
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.45,
|
| 36 |
+
"learning_rate": 4.5454545454545455e-06,
|
| 37 |
+
"loss": 1.6866,
|
| 38 |
+
"step": 5
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.55,
|
| 42 |
+
"learning_rate": 5.4545454545454545e-06,
|
| 43 |
+
"loss": 1.571,
|
| 44 |
+
"step": 6
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.64,
|
| 48 |
+
"learning_rate": 6.363636363636363e-06,
|
| 49 |
+
"loss": 1.5806,
|
| 50 |
+
"step": 7
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.73,
|
| 54 |
+
"learning_rate": 7.272727272727272e-06,
|
| 55 |
+
"loss": 1.5372,
|
| 56 |
+
"step": 8
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.82,
|
| 60 |
+
"learning_rate": 8.181818181818183e-06,
|
| 61 |
+
"loss": 1.51,
|
| 62 |
+
"step": 9
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.91,
|
| 66 |
+
"learning_rate": 9.090909090909091e-06,
|
| 67 |
+
"loss": 1.4441,
|
| 68 |
+
"step": 10
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 1.0,
|
| 72 |
+
"learning_rate": 1e-05,
|
| 73 |
+
"loss": 1.3661,
|
| 74 |
+
"step": 11
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 1.0,
|
| 78 |
+
"eval_accuracy": 0.5069444444444444,
|
| 79 |
+
"eval_loss": 1.3697991371154785,
|
| 80 |
+
"eval_runtime": 1.9108,
|
| 81 |
+
"eval_samples_per_second": 150.721,
|
| 82 |
+
"eval_steps_per_second": 2.617,
|
| 83 |
+
"step": 11
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 1.09,
|
| 87 |
+
"learning_rate": 1.0909090909090909e-05,
|
| 88 |
+
"loss": 1.386,
|
| 89 |
+
"step": 12
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 1.18,
|
| 93 |
+
"learning_rate": 1.1818181818181819e-05,
|
| 94 |
+
"loss": 1.4072,
|
| 95 |
+
"step": 13
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 1.27,
|
| 99 |
+
"learning_rate": 1.2727272727272727e-05,
|
| 100 |
+
"loss": 1.3612,
|
| 101 |
+
"step": 14
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 1.36,
|
| 105 |
+
"learning_rate": 1.3636363636363637e-05,
|
| 106 |
+
"loss": 1.2649,
|
| 107 |
+
"step": 15
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 1.45,
|
| 111 |
+
"learning_rate": 1.4545454545454545e-05,
|
| 112 |
+
"loss": 1.2608,
|
| 113 |
+
"step": 16
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 1.55,
|
| 117 |
+
"learning_rate": 1.5454545454545454e-05,
|
| 118 |
+
"loss": 1.2141,
|
| 119 |
+
"step": 17
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.64,
|
| 123 |
+
"learning_rate": 1.6363636363636366e-05,
|
| 124 |
+
"loss": 1.1708,
|
| 125 |
+
"step": 18
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 1.73,
|
| 129 |
+
"learning_rate": 1.7272727272727274e-05,
|
| 130 |
+
"loss": 1.1569,
|
| 131 |
+
"step": 19
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 1.82,
|
| 135 |
+
"learning_rate": 1.8181818181818182e-05,
|
| 136 |
+
"loss": 1.0257,
|
| 137 |
+
"step": 20
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.91,
|
| 141 |
+
"learning_rate": 1.9090909090909094e-05,
|
| 142 |
+
"loss": 1.0738,
|
| 143 |
+
"step": 21
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 2.0,
|
| 147 |
+
"learning_rate": 2e-05,
|
| 148 |
+
"loss": 0.9979,
|
| 149 |
+
"step": 22
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 2.0,
|
| 153 |
+
"eval_accuracy": 0.6631944444444444,
|
| 154 |
+
"eval_loss": 0.9816781282424927,
|
| 155 |
+
"eval_runtime": 1.5585,
|
| 156 |
+
"eval_samples_per_second": 184.794,
|
| 157 |
+
"eval_steps_per_second": 3.208,
|
| 158 |
+
"step": 22
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 2.09,
|
| 162 |
+
"learning_rate": 2.090909090909091e-05,
|
| 163 |
+
"loss": 0.9297,
|
| 164 |
+
"step": 23
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 2.18,
|
| 168 |
+
"learning_rate": 2.1818181818181818e-05,
|
| 169 |
+
"loss": 1.0059,
|
| 170 |
+
"step": 24
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 2.27,
|
| 174 |
+
"learning_rate": 2.272727272727273e-05,
|
| 175 |
+
"loss": 0.9702,
|
| 176 |
+
"step": 25
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 2.36,
|
| 180 |
+
"learning_rate": 2.3636363636363637e-05,
|
| 181 |
+
"loss": 0.9763,
|
| 182 |
+
"step": 26
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 2.45,
|
| 186 |
+
"learning_rate": 2.4545454545454545e-05,
|
| 187 |
+
"loss": 0.9261,
|
| 188 |
+
"step": 27
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 2.55,
|
| 192 |
+
"learning_rate": 2.5454545454545454e-05,
|
| 193 |
+
"loss": 0.8262,
|
| 194 |
+
"step": 28
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 2.64,
|
| 198 |
+
"learning_rate": 2.636363636363636e-05,
|
| 199 |
+
"loss": 0.9163,
|
| 200 |
+
"step": 29
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 2.73,
|
| 204 |
+
"learning_rate": 2.7272727272727273e-05,
|
| 205 |
+
"loss": 0.8509,
|
| 206 |
+
"step": 30
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 2.82,
|
| 210 |
+
"learning_rate": 2.818181818181818e-05,
|
| 211 |
+
"loss": 0.9219,
|
| 212 |
+
"step": 31
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 2.91,
|
| 216 |
+
"learning_rate": 2.909090909090909e-05,
|
| 217 |
+
"loss": 0.7396,
|
| 218 |
+
"step": 32
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 3.0,
|
| 222 |
+
"learning_rate": 3e-05,
|
| 223 |
+
"loss": 0.6746,
|
| 224 |
+
"step": 33
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 3.0,
|
| 228 |
+
"eval_accuracy": 0.7395833333333334,
|
| 229 |
+
"eval_loss": 0.7423216700553894,
|
| 230 |
+
"eval_runtime": 1.6283,
|
| 231 |
+
"eval_samples_per_second": 176.87,
|
| 232 |
+
"eval_steps_per_second": 3.071,
|
| 233 |
+
"step": 33
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 3.09,
|
| 237 |
+
"learning_rate": 3.090909090909091e-05,
|
| 238 |
+
"loss": 0.8067,
|
| 239 |
+
"step": 34
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 3.18,
|
| 243 |
+
"learning_rate": 3.181818181818182e-05,
|
| 244 |
+
"loss": 0.7232,
|
| 245 |
+
"step": 35
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 3.27,
|
| 249 |
+
"learning_rate": 3.272727272727273e-05,
|
| 250 |
+
"loss": 0.7808,
|
| 251 |
+
"step": 36
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 3.36,
|
| 255 |
+
"learning_rate": 3.3636363636363636e-05,
|
| 256 |
+
"loss": 0.6977,
|
| 257 |
+
"step": 37
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 3.45,
|
| 261 |
+
"learning_rate": 3.454545454545455e-05,
|
| 262 |
+
"loss": 0.777,
|
| 263 |
+
"step": 38
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 3.55,
|
| 267 |
+
"learning_rate": 3.545454545454546e-05,
|
| 268 |
+
"loss": 0.6499,
|
| 269 |
+
"step": 39
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 3.64,
|
| 273 |
+
"learning_rate": 3.6363636363636364e-05,
|
| 274 |
+
"loss": 0.773,
|
| 275 |
+
"step": 40
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 3.73,
|
| 279 |
+
"learning_rate": 3.7272727272727276e-05,
|
| 280 |
+
"loss": 0.6632,
|
| 281 |
+
"step": 41
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 3.82,
|
| 285 |
+
"learning_rate": 3.818181818181819e-05,
|
| 286 |
+
"loss": 0.6182,
|
| 287 |
+
"step": 42
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 3.91,
|
| 291 |
+
"learning_rate": 3.909090909090909e-05,
|
| 292 |
+
"loss": 0.6828,
|
| 293 |
+
"step": 43
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 4.0,
|
| 297 |
+
"learning_rate": 4e-05,
|
| 298 |
+
"loss": 0.6364,
|
| 299 |
+
"step": 44
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 4.0,
|
| 303 |
+
"eval_accuracy": 0.7569444444444444,
|
| 304 |
+
"eval_loss": 0.6075392365455627,
|
| 305 |
+
"eval_runtime": 1.563,
|
| 306 |
+
"eval_samples_per_second": 184.265,
|
| 307 |
+
"eval_steps_per_second": 3.199,
|
| 308 |
+
"step": 44
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 4.09,
|
| 312 |
+
"learning_rate": 4.0909090909090915e-05,
|
| 313 |
+
"loss": 0.5537,
|
| 314 |
+
"step": 45
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 4.18,
|
| 318 |
+
"learning_rate": 4.181818181818182e-05,
|
| 319 |
+
"loss": 0.5685,
|
| 320 |
+
"step": 46
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 4.27,
|
| 324 |
+
"learning_rate": 4.2727272727272724e-05,
|
| 325 |
+
"loss": 0.6091,
|
| 326 |
+
"step": 47
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 4.36,
|
| 330 |
+
"learning_rate": 4.3636363636363636e-05,
|
| 331 |
+
"loss": 0.617,
|
| 332 |
+
"step": 48
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 4.45,
|
| 336 |
+
"learning_rate": 4.454545454545455e-05,
|
| 337 |
+
"loss": 0.6308,
|
| 338 |
+
"step": 49
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 4.55,
|
| 342 |
+
"learning_rate": 4.545454545454546e-05,
|
| 343 |
+
"loss": 0.6292,
|
| 344 |
+
"step": 50
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 4.64,
|
| 348 |
+
"learning_rate": 4.636363636363636e-05,
|
| 349 |
+
"loss": 0.5864,
|
| 350 |
+
"step": 51
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 4.73,
|
| 354 |
+
"learning_rate": 4.7272727272727275e-05,
|
| 355 |
+
"loss": 0.6432,
|
| 356 |
+
"step": 52
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 4.82,
|
| 360 |
+
"learning_rate": 4.8181818181818186e-05,
|
| 361 |
+
"loss": 0.5012,
|
| 362 |
+
"step": 53
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 4.91,
|
| 366 |
+
"learning_rate": 4.909090909090909e-05,
|
| 367 |
+
"loss": 0.5659,
|
| 368 |
+
"step": 54
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 5.0,
|
| 372 |
+
"learning_rate": 5e-05,
|
| 373 |
+
"loss": 0.5425,
|
| 374 |
+
"step": 55
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 5.0,
|
| 378 |
+
"eval_accuracy": 0.7951388888888888,
|
| 379 |
+
"eval_loss": 0.5500215291976929,
|
| 380 |
+
"eval_runtime": 1.5702,
|
| 381 |
+
"eval_samples_per_second": 183.416,
|
| 382 |
+
"eval_steps_per_second": 3.184,
|
| 383 |
+
"step": 55
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 5.09,
|
| 387 |
+
"learning_rate": 4.98989898989899e-05,
|
| 388 |
+
"loss": 0.5388,
|
| 389 |
+
"step": 56
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 5.18,
|
| 393 |
+
"learning_rate": 4.97979797979798e-05,
|
| 394 |
+
"loss": 0.474,
|
| 395 |
+
"step": 57
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 5.27,
|
| 399 |
+
"learning_rate": 4.9696969696969694e-05,
|
| 400 |
+
"loss": 0.5091,
|
| 401 |
+
"step": 58
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 5.36,
|
| 405 |
+
"learning_rate": 4.9595959595959594e-05,
|
| 406 |
+
"loss": 0.4634,
|
| 407 |
+
"step": 59
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 5.45,
|
| 411 |
+
"learning_rate": 4.94949494949495e-05,
|
| 412 |
+
"loss": 0.5044,
|
| 413 |
+
"step": 60
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 5.55,
|
| 417 |
+
"learning_rate": 4.93939393939394e-05,
|
| 418 |
+
"loss": 0.4715,
|
| 419 |
+
"step": 61
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 5.64,
|
| 423 |
+
"learning_rate": 4.92929292929293e-05,
|
| 424 |
+
"loss": 0.4705,
|
| 425 |
+
"step": 62
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 5.73,
|
| 429 |
+
"learning_rate": 4.919191919191919e-05,
|
| 430 |
+
"loss": 0.4706,
|
| 431 |
+
"step": 63
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 5.82,
|
| 435 |
+
"learning_rate": 4.909090909090909e-05,
|
| 436 |
+
"loss": 0.4945,
|
| 437 |
+
"step": 64
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 5.91,
|
| 441 |
+
"learning_rate": 4.898989898989899e-05,
|
| 442 |
+
"loss": 0.5611,
|
| 443 |
+
"step": 65
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 6.0,
|
| 447 |
+
"learning_rate": 4.888888888888889e-05,
|
| 448 |
+
"loss": 0.5001,
|
| 449 |
+
"step": 66
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 6.0,
|
| 453 |
+
"eval_accuracy": 0.8159722222222222,
|
| 454 |
+
"eval_loss": 0.4882892072200775,
|
| 455 |
+
"eval_runtime": 1.6356,
|
| 456 |
+
"eval_samples_per_second": 176.082,
|
| 457 |
+
"eval_steps_per_second": 3.057,
|
| 458 |
+
"step": 66
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 6.09,
|
| 462 |
+
"learning_rate": 4.878787878787879e-05,
|
| 463 |
+
"loss": 0.3538,
|
| 464 |
+
"step": 67
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 6.18,
|
| 468 |
+
"learning_rate": 4.868686868686869e-05,
|
| 469 |
+
"loss": 0.4787,
|
| 470 |
+
"step": 68
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 6.27,
|
| 474 |
+
"learning_rate": 4.858585858585859e-05,
|
| 475 |
+
"loss": 0.4181,
|
| 476 |
+
"step": 69
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 6.36,
|
| 480 |
+
"learning_rate": 4.848484848484849e-05,
|
| 481 |
+
"loss": 0.451,
|
| 482 |
+
"step": 70
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 6.45,
|
| 486 |
+
"learning_rate": 4.838383838383839e-05,
|
| 487 |
+
"loss": 0.4455,
|
| 488 |
+
"step": 71
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 6.55,
|
| 492 |
+
"learning_rate": 4.828282828282829e-05,
|
| 493 |
+
"loss": 0.4275,
|
| 494 |
+
"step": 72
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 6.64,
|
| 498 |
+
"learning_rate": 4.8181818181818186e-05,
|
| 499 |
+
"loss": 0.4366,
|
| 500 |
+
"step": 73
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 6.73,
|
| 504 |
+
"learning_rate": 4.808080808080808e-05,
|
| 505 |
+
"loss": 0.4176,
|
| 506 |
+
"step": 74
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 6.82,
|
| 510 |
+
"learning_rate": 4.797979797979798e-05,
|
| 511 |
+
"loss": 0.4344,
|
| 512 |
+
"step": 75
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 6.91,
|
| 516 |
+
"learning_rate": 4.787878787878788e-05,
|
| 517 |
+
"loss": 0.4482,
|
| 518 |
+
"step": 76
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 7.0,
|
| 522 |
+
"learning_rate": 4.7777777777777784e-05,
|
| 523 |
+
"loss": 0.3519,
|
| 524 |
+
"step": 77
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 7.0,
|
| 528 |
+
"eval_accuracy": 0.8263888888888888,
|
| 529 |
+
"eval_loss": 0.4539467394351959,
|
| 530 |
+
"eval_runtime": 1.5682,
|
| 531 |
+
"eval_samples_per_second": 183.653,
|
| 532 |
+
"eval_steps_per_second": 3.188,
|
| 533 |
+
"step": 77
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 7.09,
|
| 537 |
+
"learning_rate": 4.7676767676767684e-05,
|
| 538 |
+
"loss": 0.42,
|
| 539 |
+
"step": 78
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 7.18,
|
| 543 |
+
"learning_rate": 4.7575757575757576e-05,
|
| 544 |
+
"loss": 0.4017,
|
| 545 |
+
"step": 79
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 7.27,
|
| 549 |
+
"learning_rate": 4.7474747474747476e-05,
|
| 550 |
+
"loss": 0.3604,
|
| 551 |
+
"step": 80
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 7.36,
|
| 555 |
+
"learning_rate": 4.7373737373737375e-05,
|
| 556 |
+
"loss": 0.4075,
|
| 557 |
+
"step": 81
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 7.45,
|
| 561 |
+
"learning_rate": 4.7272727272727275e-05,
|
| 562 |
+
"loss": 0.399,
|
| 563 |
+
"step": 82
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 7.55,
|
| 567 |
+
"learning_rate": 4.7171717171717174e-05,
|
| 568 |
+
"loss": 0.335,
|
| 569 |
+
"step": 83
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 7.64,
|
| 573 |
+
"learning_rate": 4.7070707070707074e-05,
|
| 574 |
+
"loss": 0.297,
|
| 575 |
+
"step": 84
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 7.73,
|
| 579 |
+
"learning_rate": 4.696969696969697e-05,
|
| 580 |
+
"loss": 0.3727,
|
| 581 |
+
"step": 85
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 7.82,
|
| 585 |
+
"learning_rate": 4.686868686868687e-05,
|
| 586 |
+
"loss": 0.3886,
|
| 587 |
+
"step": 86
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 7.91,
|
| 591 |
+
"learning_rate": 4.676767676767677e-05,
|
| 592 |
+
"loss": 0.4117,
|
| 593 |
+
"step": 87
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 8.0,
|
| 597 |
+
"learning_rate": 4.666666666666667e-05,
|
| 598 |
+
"loss": 0.4421,
|
| 599 |
+
"step": 88
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 8.0,
|
| 603 |
+
"eval_accuracy": 0.8194444444444444,
|
| 604 |
+
"eval_loss": 0.4483272135257721,
|
| 605 |
+
"eval_runtime": 1.5685,
|
| 606 |
+
"eval_samples_per_second": 183.62,
|
| 607 |
+
"eval_steps_per_second": 3.188,
|
| 608 |
+
"step": 88
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 8.09,
|
| 612 |
+
"learning_rate": 4.656565656565657e-05,
|
| 613 |
+
"loss": 0.334,
|
| 614 |
+
"step": 89
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 8.18,
|
| 618 |
+
"learning_rate": 4.6464646464646464e-05,
|
| 619 |
+
"loss": 0.3811,
|
| 620 |
+
"step": 90
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 8.27,
|
| 624 |
+
"learning_rate": 4.636363636363636e-05,
|
| 625 |
+
"loss": 0.3582,
|
| 626 |
+
"step": 91
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 8.36,
|
| 630 |
+
"learning_rate": 4.626262626262626e-05,
|
| 631 |
+
"loss": 0.381,
|
| 632 |
+
"step": 92
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 8.45,
|
| 636 |
+
"learning_rate": 4.616161616161616e-05,
|
| 637 |
+
"loss": 0.3449,
|
| 638 |
+
"step": 93
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 8.55,
|
| 642 |
+
"learning_rate": 4.606060606060607e-05,
|
| 643 |
+
"loss": 0.3465,
|
| 644 |
+
"step": 94
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 8.64,
|
| 648 |
+
"learning_rate": 4.595959595959596e-05,
|
| 649 |
+
"loss": 0.3676,
|
| 650 |
+
"step": 95
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 8.73,
|
| 654 |
+
"learning_rate": 4.585858585858586e-05,
|
| 655 |
+
"loss": 0.3616,
|
| 656 |
+
"step": 96
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 8.82,
|
| 660 |
+
"learning_rate": 4.575757575757576e-05,
|
| 661 |
+
"loss": 0.3392,
|
| 662 |
+
"step": 97
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 8.91,
|
| 666 |
+
"learning_rate": 4.565656565656566e-05,
|
| 667 |
+
"loss": 0.3374,
|
| 668 |
+
"step": 98
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 9.0,
|
| 672 |
+
"learning_rate": 4.555555555555556e-05,
|
| 673 |
+
"loss": 0.3207,
|
| 674 |
+
"step": 99
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 9.0,
|
| 678 |
+
"eval_accuracy": 0.84375,
|
| 679 |
+
"eval_loss": 0.3785366117954254,
|
| 680 |
+
"eval_runtime": 1.5932,
|
| 681 |
+
"eval_samples_per_second": 180.771,
|
| 682 |
+
"eval_steps_per_second": 3.138,
|
| 683 |
+
"step": 99
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 9.09,
|
| 687 |
+
"learning_rate": 4.545454545454546e-05,
|
| 688 |
+
"loss": 0.3319,
|
| 689 |
+
"step": 100
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 9.18,
|
| 693 |
+
"learning_rate": 4.535353535353535e-05,
|
| 694 |
+
"loss": 0.3582,
|
| 695 |
+
"step": 101
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 9.27,
|
| 699 |
+
"learning_rate": 4.525252525252526e-05,
|
| 700 |
+
"loss": 0.2702,
|
| 701 |
+
"step": 102
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 9.36,
|
| 705 |
+
"learning_rate": 4.515151515151516e-05,
|
| 706 |
+
"loss": 0.3054,
|
| 707 |
+
"step": 103
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 9.45,
|
| 711 |
+
"learning_rate": 4.5050505050505056e-05,
|
| 712 |
+
"loss": 0.3041,
|
| 713 |
+
"step": 104
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 9.55,
|
| 717 |
+
"learning_rate": 4.494949494949495e-05,
|
| 718 |
+
"loss": 0.2398,
|
| 719 |
+
"step": 105
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 9.64,
|
| 723 |
+
"learning_rate": 4.484848484848485e-05,
|
| 724 |
+
"loss": 0.338,
|
| 725 |
+
"step": 106
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 9.73,
|
| 729 |
+
"learning_rate": 4.474747474747475e-05,
|
| 730 |
+
"loss": 0.2486,
|
| 731 |
+
"step": 107
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 9.82,
|
| 735 |
+
"learning_rate": 4.464646464646465e-05,
|
| 736 |
+
"loss": 0.279,
|
| 737 |
+
"step": 108
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 9.91,
|
| 741 |
+
"learning_rate": 4.454545454545455e-05,
|
| 742 |
+
"loss": 0.3784,
|
| 743 |
+
"step": 109
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 10.0,
|
| 747 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 748 |
+
"loss": 0.3682,
|
| 749 |
+
"step": 110
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 10.0,
|
| 753 |
+
"eval_accuracy": 0.8645833333333334,
|
| 754 |
+
"eval_loss": 0.3384600281715393,
|
| 755 |
+
"eval_runtime": 1.5672,
|
| 756 |
+
"eval_samples_per_second": 183.763,
|
| 757 |
+
"eval_steps_per_second": 3.19,
|
| 758 |
+
"step": 110
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 10.09,
|
| 762 |
+
"learning_rate": 4.4343434343434346e-05,
|
| 763 |
+
"loss": 0.3374,
|
| 764 |
+
"step": 111
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 10.18,
|
| 768 |
+
"learning_rate": 4.4242424242424246e-05,
|
| 769 |
+
"loss": 0.3114,
|
| 770 |
+
"step": 112
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 10.27,
|
| 774 |
+
"learning_rate": 4.4141414141414145e-05,
|
| 775 |
+
"loss": 0.3062,
|
| 776 |
+
"step": 113
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 10.36,
|
| 780 |
+
"learning_rate": 4.4040404040404044e-05,
|
| 781 |
+
"loss": 0.3199,
|
| 782 |
+
"step": 114
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 10.45,
|
| 786 |
+
"learning_rate": 4.3939393939393944e-05,
|
| 787 |
+
"loss": 0.2839,
|
| 788 |
+
"step": 115
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 10.55,
|
| 792 |
+
"learning_rate": 4.383838383838384e-05,
|
| 793 |
+
"loss": 0.2866,
|
| 794 |
+
"step": 116
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 10.64,
|
| 798 |
+
"learning_rate": 4.3737373737373736e-05,
|
| 799 |
+
"loss": 0.2628,
|
| 800 |
+
"step": 117
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 10.73,
|
| 804 |
+
"learning_rate": 4.3636363636363636e-05,
|
| 805 |
+
"loss": 0.3463,
|
| 806 |
+
"step": 118
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 10.82,
|
| 810 |
+
"learning_rate": 4.3535353535353535e-05,
|
| 811 |
+
"loss": 0.263,
|
| 812 |
+
"step": 119
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 10.91,
|
| 816 |
+
"learning_rate": 4.343434343434344e-05,
|
| 817 |
+
"loss": 0.3273,
|
| 818 |
+
"step": 120
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 11.0,
|
| 822 |
+
"learning_rate": 4.3333333333333334e-05,
|
| 823 |
+
"loss": 0.2642,
|
| 824 |
+
"step": 121
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 11.0,
|
| 828 |
+
"eval_accuracy": 0.8402777777777778,
|
| 829 |
+
"eval_loss": 0.38269615173339844,
|
| 830 |
+
"eval_runtime": 1.622,
|
| 831 |
+
"eval_samples_per_second": 177.56,
|
| 832 |
+
"eval_steps_per_second": 3.083,
|
| 833 |
+
"step": 121
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 11.09,
|
| 837 |
+
"learning_rate": 4.3232323232323234e-05,
|
| 838 |
+
"loss": 0.2527,
|
| 839 |
+
"step": 122
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 11.18,
|
| 843 |
+
"learning_rate": 4.313131313131313e-05,
|
| 844 |
+
"loss": 0.2828,
|
| 845 |
+
"step": 123
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 11.27,
|
| 849 |
+
"learning_rate": 4.303030303030303e-05,
|
| 850 |
+
"loss": 0.2746,
|
| 851 |
+
"step": 124
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 11.36,
|
| 855 |
+
"learning_rate": 4.292929292929293e-05,
|
| 856 |
+
"loss": 0.2746,
|
| 857 |
+
"step": 125
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 11.45,
|
| 861 |
+
"learning_rate": 4.282828282828283e-05,
|
| 862 |
+
"loss": 0.2531,
|
| 863 |
+
"step": 126
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 11.55,
|
| 867 |
+
"learning_rate": 4.2727272727272724e-05,
|
| 868 |
+
"loss": 0.3099,
|
| 869 |
+
"step": 127
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 11.64,
|
| 873 |
+
"learning_rate": 4.262626262626263e-05,
|
| 874 |
+
"loss": 0.2967,
|
| 875 |
+
"step": 128
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 11.73,
|
| 879 |
+
"learning_rate": 4.252525252525253e-05,
|
| 880 |
+
"loss": 0.2549,
|
| 881 |
+
"step": 129
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 11.82,
|
| 885 |
+
"learning_rate": 4.242424242424243e-05,
|
| 886 |
+
"loss": 0.278,
|
| 887 |
+
"step": 130
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 11.91,
|
| 891 |
+
"learning_rate": 4.232323232323233e-05,
|
| 892 |
+
"loss": 0.2509,
|
| 893 |
+
"step": 131
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 12.0,
|
| 897 |
+
"learning_rate": 4.222222222222222e-05,
|
| 898 |
+
"loss": 0.3444,
|
| 899 |
+
"step": 132
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 12.0,
|
| 903 |
+
"eval_accuracy": 0.8506944444444444,
|
| 904 |
+
"eval_loss": 0.346247136592865,
|
| 905 |
+
"eval_runtime": 1.5635,
|
| 906 |
+
"eval_samples_per_second": 184.201,
|
| 907 |
+
"eval_steps_per_second": 3.198,
|
| 908 |
+
"step": 132
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 12.09,
|
| 912 |
+
"learning_rate": 4.212121212121212e-05,
|
| 913 |
+
"loss": 0.2451,
|
| 914 |
+
"step": 133
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 12.18,
|
| 918 |
+
"learning_rate": 4.202020202020202e-05,
|
| 919 |
+
"loss": 0.2623,
|
| 920 |
+
"step": 134
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 12.27,
|
| 924 |
+
"learning_rate": 4.191919191919192e-05,
|
| 925 |
+
"loss": 0.226,
|
| 926 |
+
"step": 135
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 12.36,
|
| 930 |
+
"learning_rate": 4.181818181818182e-05,
|
| 931 |
+
"loss": 0.2158,
|
| 932 |
+
"step": 136
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 12.45,
|
| 936 |
+
"learning_rate": 4.171717171717172e-05,
|
| 937 |
+
"loss": 0.2471,
|
| 938 |
+
"step": 137
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 12.55,
|
| 942 |
+
"learning_rate": 4.161616161616162e-05,
|
| 943 |
+
"loss": 0.249,
|
| 944 |
+
"step": 138
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 12.64,
|
| 948 |
+
"learning_rate": 4.151515151515152e-05,
|
| 949 |
+
"loss": 0.2439,
|
| 950 |
+
"step": 139
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 12.73,
|
| 954 |
+
"learning_rate": 4.141414141414142e-05,
|
| 955 |
+
"loss": 0.284,
|
| 956 |
+
"step": 140
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 12.82,
|
| 960 |
+
"learning_rate": 4.131313131313132e-05,
|
| 961 |
+
"loss": 0.2701,
|
| 962 |
+
"step": 141
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 12.91,
|
| 966 |
+
"learning_rate": 4.1212121212121216e-05,
|
| 967 |
+
"loss": 0.3189,
|
| 968 |
+
"step": 142
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 13.0,
|
| 972 |
+
"learning_rate": 4.111111111111111e-05,
|
| 973 |
+
"loss": 0.2423,
|
| 974 |
+
"step": 143
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 13.0,
|
| 978 |
+
"eval_accuracy": 0.8680555555555556,
|
| 979 |
+
"eval_loss": 0.317039430141449,
|
| 980 |
+
"eval_runtime": 1.5869,
|
| 981 |
+
"eval_samples_per_second": 181.49,
|
| 982 |
+
"eval_steps_per_second": 3.151,
|
| 983 |
+
"step": 143
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 13.09,
|
| 987 |
+
"learning_rate": 4.101010101010101e-05,
|
| 988 |
+
"loss": 0.2665,
|
| 989 |
+
"step": 144
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 13.18,
|
| 993 |
+
"learning_rate": 4.0909090909090915e-05,
|
| 994 |
+
"loss": 0.2167,
|
| 995 |
+
"step": 145
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 13.27,
|
| 999 |
+
"learning_rate": 4.0808080808080814e-05,
|
| 1000 |
+
"loss": 0.2306,
|
| 1001 |
+
"step": 146
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 13.36,
|
| 1005 |
+
"learning_rate": 4.070707070707071e-05,
|
| 1006 |
+
"loss": 0.2633,
|
| 1007 |
+
"step": 147
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 13.45,
|
| 1011 |
+
"learning_rate": 4.0606060606060606e-05,
|
| 1012 |
+
"loss": 0.2385,
|
| 1013 |
+
"step": 148
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 13.55,
|
| 1017 |
+
"learning_rate": 4.0505050505050506e-05,
|
| 1018 |
+
"loss": 0.3078,
|
| 1019 |
+
"step": 149
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 13.64,
|
| 1023 |
+
"learning_rate": 4.0404040404040405e-05,
|
| 1024 |
+
"loss": 0.2579,
|
| 1025 |
+
"step": 150
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 13.73,
|
| 1029 |
+
"learning_rate": 4.0303030303030305e-05,
|
| 1030 |
+
"loss": 0.2388,
|
| 1031 |
+
"step": 151
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 13.82,
|
| 1035 |
+
"learning_rate": 4.0202020202020204e-05,
|
| 1036 |
+
"loss": 0.2028,
|
| 1037 |
+
"step": 152
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 13.91,
|
| 1041 |
+
"learning_rate": 4.01010101010101e-05,
|
| 1042 |
+
"loss": 0.2058,
|
| 1043 |
+
"step": 153
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 14.0,
|
| 1047 |
+
"learning_rate": 4e-05,
|
| 1048 |
+
"loss": 0.3168,
|
| 1049 |
+
"step": 154
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 14.0,
|
| 1053 |
+
"eval_accuracy": 0.8715277777777778,
|
| 1054 |
+
"eval_loss": 0.3167605996131897,
|
| 1055 |
+
"eval_runtime": 1.5799,
|
| 1056 |
+
"eval_samples_per_second": 182.284,
|
| 1057 |
+
"eval_steps_per_second": 3.165,
|
| 1058 |
+
"step": 154
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 14.09,
|
| 1062 |
+
"learning_rate": 3.98989898989899e-05,
|
| 1063 |
+
"loss": 0.1858,
|
| 1064 |
+
"step": 155
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 14.18,
|
| 1068 |
+
"learning_rate": 3.97979797979798e-05,
|
| 1069 |
+
"loss": 0.251,
|
| 1070 |
+
"step": 156
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 14.27,
|
| 1074 |
+
"learning_rate": 3.96969696969697e-05,
|
| 1075 |
+
"loss": 0.1862,
|
| 1076 |
+
"step": 157
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 14.36,
|
| 1080 |
+
"learning_rate": 3.9595959595959594e-05,
|
| 1081 |
+
"loss": 0.2788,
|
| 1082 |
+
"step": 158
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 14.45,
|
| 1086 |
+
"learning_rate": 3.9494949494949494e-05,
|
| 1087 |
+
"loss": 0.1804,
|
| 1088 |
+
"step": 159
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 14.55,
|
| 1092 |
+
"learning_rate": 3.939393939393939e-05,
|
| 1093 |
+
"loss": 0.2121,
|
| 1094 |
+
"step": 160
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 14.64,
|
| 1098 |
+
"learning_rate": 3.929292929292929e-05,
|
| 1099 |
+
"loss": 0.2029,
|
| 1100 |
+
"step": 161
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 14.73,
|
| 1104 |
+
"learning_rate": 3.91919191919192e-05,
|
| 1105 |
+
"loss": 0.2308,
|
| 1106 |
+
"step": 162
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 14.82,
|
| 1110 |
+
"learning_rate": 3.909090909090909e-05,
|
| 1111 |
+
"loss": 0.2396,
|
| 1112 |
+
"step": 163
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 14.91,
|
| 1116 |
+
"learning_rate": 3.898989898989899e-05,
|
| 1117 |
+
"loss": 0.2234,
|
| 1118 |
+
"step": 164
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 15.0,
|
| 1122 |
+
"learning_rate": 3.888888888888889e-05,
|
| 1123 |
+
"loss": 0.2781,
|
| 1124 |
+
"step": 165
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 15.0,
|
| 1128 |
+
"eval_accuracy": 0.8333333333333334,
|
| 1129 |
+
"eval_loss": 0.33227282762527466,
|
| 1130 |
+
"eval_runtime": 1.5652,
|
| 1131 |
+
"eval_samples_per_second": 184.003,
|
| 1132 |
+
"eval_steps_per_second": 3.194,
|
| 1133 |
+
"step": 165
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 15.09,
|
| 1137 |
+
"learning_rate": 3.878787878787879e-05,
|
| 1138 |
+
"loss": 0.2058,
|
| 1139 |
+
"step": 166
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 15.18,
|
| 1143 |
+
"learning_rate": 3.868686868686869e-05,
|
| 1144 |
+
"loss": 0.2155,
|
| 1145 |
+
"step": 167
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 15.27,
|
| 1149 |
+
"learning_rate": 3.858585858585859e-05,
|
| 1150 |
+
"loss": 0.2097,
|
| 1151 |
+
"step": 168
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 15.36,
|
| 1155 |
+
"learning_rate": 3.848484848484848e-05,
|
| 1156 |
+
"loss": 0.2237,
|
| 1157 |
+
"step": 169
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 15.45,
|
| 1161 |
+
"learning_rate": 3.838383838383838e-05,
|
| 1162 |
+
"loss": 0.2715,
|
| 1163 |
+
"step": 170
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 15.55,
|
| 1167 |
+
"learning_rate": 3.828282828282829e-05,
|
| 1168 |
+
"loss": 0.2293,
|
| 1169 |
+
"step": 171
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 15.64,
|
| 1173 |
+
"learning_rate": 3.818181818181819e-05,
|
| 1174 |
+
"loss": 0.2285,
|
| 1175 |
+
"step": 172
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 15.73,
|
| 1179 |
+
"learning_rate": 3.8080808080808087e-05,
|
| 1180 |
+
"loss": 0.2612,
|
| 1181 |
+
"step": 173
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 15.82,
|
| 1185 |
+
"learning_rate": 3.797979797979798e-05,
|
| 1186 |
+
"loss": 0.196,
|
| 1187 |
+
"step": 174
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 15.91,
|
| 1191 |
+
"learning_rate": 3.787878787878788e-05,
|
| 1192 |
+
"loss": 0.2298,
|
| 1193 |
+
"step": 175
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 16.0,
|
| 1197 |
+
"learning_rate": 3.777777777777778e-05,
|
| 1198 |
+
"loss": 0.2411,
|
| 1199 |
+
"step": 176
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 16.0,
|
| 1203 |
+
"eval_accuracy": 0.8715277777777778,
|
| 1204 |
+
"eval_loss": 0.31996577978134155,
|
| 1205 |
+
"eval_runtime": 1.5652,
|
| 1206 |
+
"eval_samples_per_second": 184.008,
|
| 1207 |
+
"eval_steps_per_second": 3.195,
|
| 1208 |
+
"step": 176
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 16.09,
|
| 1212 |
+
"learning_rate": 3.767676767676768e-05,
|
| 1213 |
+
"loss": 0.1726,
|
| 1214 |
+
"step": 177
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 16.18,
|
| 1218 |
+
"learning_rate": 3.757575757575758e-05,
|
| 1219 |
+
"loss": 0.2112,
|
| 1220 |
+
"step": 178
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 16.27,
|
| 1224 |
+
"learning_rate": 3.747474747474748e-05,
|
| 1225 |
+
"loss": 0.1964,
|
| 1226 |
+
"step": 179
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 16.36,
|
| 1230 |
+
"learning_rate": 3.7373737373737376e-05,
|
| 1231 |
+
"loss": 0.1967,
|
| 1232 |
+
"step": 180
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 16.45,
|
| 1236 |
+
"learning_rate": 3.7272727272727276e-05,
|
| 1237 |
+
"loss": 0.2126,
|
| 1238 |
+
"step": 181
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 16.55,
|
| 1242 |
+
"learning_rate": 3.7171717171717175e-05,
|
| 1243 |
+
"loss": 0.1554,
|
| 1244 |
+
"step": 182
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 16.64,
|
| 1248 |
+
"learning_rate": 3.7070707070707075e-05,
|
| 1249 |
+
"loss": 0.1857,
|
| 1250 |
+
"step": 183
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 16.73,
|
| 1254 |
+
"learning_rate": 3.6969696969696974e-05,
|
| 1255 |
+
"loss": 0.211,
|
| 1256 |
+
"step": 184
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 16.82,
|
| 1260 |
+
"learning_rate": 3.686868686868687e-05,
|
| 1261 |
+
"loss": 0.2358,
|
| 1262 |
+
"step": 185
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 16.91,
|
| 1266 |
+
"learning_rate": 3.6767676767676766e-05,
|
| 1267 |
+
"loss": 0.1626,
|
| 1268 |
+
"step": 186
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 17.0,
|
| 1272 |
+
"learning_rate": 3.6666666666666666e-05,
|
| 1273 |
+
"loss": 0.2276,
|
| 1274 |
+
"step": 187
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 17.0,
|
| 1278 |
+
"eval_accuracy": 0.875,
|
| 1279 |
+
"eval_loss": 0.32961538434028625,
|
| 1280 |
+
"eval_runtime": 1.6093,
|
| 1281 |
+
"eval_samples_per_second": 178.958,
|
| 1282 |
+
"eval_steps_per_second": 3.107,
|
| 1283 |
+
"step": 187
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 17.09,
|
| 1287 |
+
"learning_rate": 3.656565656565657e-05,
|
| 1288 |
+
"loss": 0.1915,
|
| 1289 |
+
"step": 188
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 17.18,
|
| 1293 |
+
"learning_rate": 3.6464646464646465e-05,
|
| 1294 |
+
"loss": 0.1826,
|
| 1295 |
+
"step": 189
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 17.27,
|
| 1299 |
+
"learning_rate": 3.6363636363636364e-05,
|
| 1300 |
+
"loss": 0.2162,
|
| 1301 |
+
"step": 190
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 17.36,
|
| 1305 |
+
"learning_rate": 3.6262626262626264e-05,
|
| 1306 |
+
"loss": 0.1755,
|
| 1307 |
+
"step": 191
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 17.45,
|
| 1311 |
+
"learning_rate": 3.616161616161616e-05,
|
| 1312 |
+
"loss": 0.24,
|
| 1313 |
+
"step": 192
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 17.55,
|
| 1317 |
+
"learning_rate": 3.606060606060606e-05,
|
| 1318 |
+
"loss": 0.1746,
|
| 1319 |
+
"step": 193
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 17.64,
|
| 1323 |
+
"learning_rate": 3.595959595959596e-05,
|
| 1324 |
+
"loss": 0.1443,
|
| 1325 |
+
"step": 194
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 17.73,
|
| 1329 |
+
"learning_rate": 3.5858585858585855e-05,
|
| 1330 |
+
"loss": 0.189,
|
| 1331 |
+
"step": 195
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 17.82,
|
| 1335 |
+
"learning_rate": 3.575757575757576e-05,
|
| 1336 |
+
"loss": 0.1735,
|
| 1337 |
+
"step": 196
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 17.91,
|
| 1341 |
+
"learning_rate": 3.565656565656566e-05,
|
| 1342 |
+
"loss": 0.1383,
|
| 1343 |
+
"step": 197
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 18.0,
|
| 1347 |
+
"learning_rate": 3.555555555555556e-05,
|
| 1348 |
+
"loss": 0.192,
|
| 1349 |
+
"step": 198
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 18.0,
|
| 1353 |
+
"eval_accuracy": 0.8854166666666666,
|
| 1354 |
+
"eval_loss": 0.3118571639060974,
|
| 1355 |
+
"eval_runtime": 1.5666,
|
| 1356 |
+
"eval_samples_per_second": 183.832,
|
| 1357 |
+
"eval_steps_per_second": 3.192,
|
| 1358 |
+
"step": 198
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 18.09,
|
| 1362 |
+
"learning_rate": 3.545454545454546e-05,
|
| 1363 |
+
"loss": 0.2313,
|
| 1364 |
+
"step": 199
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 18.18,
|
| 1368 |
+
"learning_rate": 3.535353535353535e-05,
|
| 1369 |
+
"loss": 0.2429,
|
| 1370 |
+
"step": 200
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 18.27,
|
| 1374 |
+
"learning_rate": 3.525252525252525e-05,
|
| 1375 |
+
"loss": 0.1745,
|
| 1376 |
+
"step": 201
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 18.36,
|
| 1380 |
+
"learning_rate": 3.515151515151515e-05,
|
| 1381 |
+
"loss": 0.189,
|
| 1382 |
+
"step": 202
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 18.45,
|
| 1386 |
+
"learning_rate": 3.505050505050505e-05,
|
| 1387 |
+
"loss": 0.1607,
|
| 1388 |
+
"step": 203
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 18.55,
|
| 1392 |
+
"learning_rate": 3.494949494949495e-05,
|
| 1393 |
+
"loss": 0.2268,
|
| 1394 |
+
"step": 204
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 18.64,
|
| 1398 |
+
"learning_rate": 3.484848484848485e-05,
|
| 1399 |
+
"loss": 0.1263,
|
| 1400 |
+
"step": 205
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 18.73,
|
| 1404 |
+
"learning_rate": 3.474747474747475e-05,
|
| 1405 |
+
"loss": 0.2372,
|
| 1406 |
+
"step": 206
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 18.82,
|
| 1410 |
+
"learning_rate": 3.464646464646465e-05,
|
| 1411 |
+
"loss": 0.1604,
|
| 1412 |
+
"step": 207
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 18.91,
|
| 1416 |
+
"learning_rate": 3.454545454545455e-05,
|
| 1417 |
+
"loss": 0.2622,
|
| 1418 |
+
"step": 208
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 19.0,
|
| 1422 |
+
"learning_rate": 3.444444444444445e-05,
|
| 1423 |
+
"loss": 0.1612,
|
| 1424 |
+
"step": 209
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 19.0,
|
| 1428 |
+
"eval_accuracy": 0.875,
|
| 1429 |
+
"eval_loss": 0.36473166942596436,
|
| 1430 |
+
"eval_runtime": 1.5912,
|
| 1431 |
+
"eval_samples_per_second": 180.99,
|
| 1432 |
+
"eval_steps_per_second": 3.142,
|
| 1433 |
+
"step": 209
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 19.09,
|
| 1437 |
+
"learning_rate": 3.434343434343435e-05,
|
| 1438 |
+
"loss": 0.2007,
|
| 1439 |
+
"step": 210
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 19.18,
|
| 1443 |
+
"learning_rate": 3.424242424242424e-05,
|
| 1444 |
+
"loss": 0.1945,
|
| 1445 |
+
"step": 211
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 19.27,
|
| 1449 |
+
"learning_rate": 3.414141414141414e-05,
|
| 1450 |
+
"loss": 0.1654,
|
| 1451 |
+
"step": 212
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 19.36,
|
| 1455 |
+
"learning_rate": 3.4040404040404045e-05,
|
| 1456 |
+
"loss": 0.141,
|
| 1457 |
+
"step": 213
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 19.45,
|
| 1461 |
+
"learning_rate": 3.3939393939393945e-05,
|
| 1462 |
+
"loss": 0.2012,
|
| 1463 |
+
"step": 214
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 19.55,
|
| 1467 |
+
"learning_rate": 3.3838383838383844e-05,
|
| 1468 |
+
"loss": 0.1692,
|
| 1469 |
+
"step": 215
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 19.64,
|
| 1473 |
+
"learning_rate": 3.373737373737374e-05,
|
| 1474 |
+
"loss": 0.163,
|
| 1475 |
+
"step": 216
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 19.73,
|
| 1479 |
+
"learning_rate": 3.3636363636363636e-05,
|
| 1480 |
+
"loss": 0.2155,
|
| 1481 |
+
"step": 217
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 19.82,
|
| 1485 |
+
"learning_rate": 3.3535353535353536e-05,
|
| 1486 |
+
"loss": 0.1732,
|
| 1487 |
+
"step": 218
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 19.91,
|
| 1491 |
+
"learning_rate": 3.3434343434343435e-05,
|
| 1492 |
+
"loss": 0.1395,
|
| 1493 |
+
"step": 219
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 20.0,
|
| 1497 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 1498 |
+
"loss": 0.1084,
|
| 1499 |
+
"step": 220
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 20.0,
|
| 1503 |
+
"eval_accuracy": 0.8993055555555556,
|
| 1504 |
+
"eval_loss": 0.2640966773033142,
|
| 1505 |
+
"eval_runtime": 1.5715,
|
| 1506 |
+
"eval_samples_per_second": 183.268,
|
| 1507 |
+
"eval_steps_per_second": 3.182,
|
| 1508 |
+
"step": 220
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 20.09,
|
| 1512 |
+
"learning_rate": 3.3232323232323234e-05,
|
| 1513 |
+
"loss": 0.138,
|
| 1514 |
+
"step": 221
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 20.18,
|
| 1518 |
+
"learning_rate": 3.3131313131313134e-05,
|
| 1519 |
+
"loss": 0.1625,
|
| 1520 |
+
"step": 222
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 20.27,
|
| 1524 |
+
"learning_rate": 3.303030303030303e-05,
|
| 1525 |
+
"loss": 0.1838,
|
| 1526 |
+
"step": 223
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 20.36,
|
| 1530 |
+
"learning_rate": 3.292929292929293e-05,
|
| 1531 |
+
"loss": 0.1558,
|
| 1532 |
+
"step": 224
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 20.45,
|
| 1536 |
+
"learning_rate": 3.282828282828283e-05,
|
| 1537 |
+
"loss": 0.2134,
|
| 1538 |
+
"step": 225
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 20.55,
|
| 1542 |
+
"learning_rate": 3.272727272727273e-05,
|
| 1543 |
+
"loss": 0.1694,
|
| 1544 |
+
"step": 226
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 20.64,
|
| 1548 |
+
"learning_rate": 3.2626262626262624e-05,
|
| 1549 |
+
"loss": 0.2243,
|
| 1550 |
+
"step": 227
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 20.73,
|
| 1554 |
+
"learning_rate": 3.2525252525252524e-05,
|
| 1555 |
+
"loss": 0.219,
|
| 1556 |
+
"step": 228
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 20.82,
|
| 1560 |
+
"learning_rate": 3.2424242424242423e-05,
|
| 1561 |
+
"loss": 0.1761,
|
| 1562 |
+
"step": 229
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 20.91,
|
| 1566 |
+
"learning_rate": 3.232323232323233e-05,
|
| 1567 |
+
"loss": 0.1231,
|
| 1568 |
+
"step": 230
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 21.0,
|
| 1572 |
+
"learning_rate": 3.222222222222223e-05,
|
| 1573 |
+
"loss": 0.2099,
|
| 1574 |
+
"step": 231
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 21.0,
|
| 1578 |
+
"eval_accuracy": 0.8958333333333334,
|
| 1579 |
+
"eval_loss": 0.2806669771671295,
|
| 1580 |
+
"eval_runtime": 1.5627,
|
| 1581 |
+
"eval_samples_per_second": 184.299,
|
| 1582 |
+
"eval_steps_per_second": 3.2,
|
| 1583 |
+
"step": 231
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 21.09,
|
| 1587 |
+
"learning_rate": 3.212121212121212e-05,
|
| 1588 |
+
"loss": 0.1831,
|
| 1589 |
+
"step": 232
|
| 1590 |
+
},
|
| 1591 |
+
{
|
| 1592 |
+
"epoch": 21.18,
|
| 1593 |
+
"learning_rate": 3.202020202020202e-05,
|
| 1594 |
+
"loss": 0.123,
|
| 1595 |
+
"step": 233
|
| 1596 |
+
},
|
| 1597 |
+
{
|
| 1598 |
+
"epoch": 21.27,
|
| 1599 |
+
"learning_rate": 3.191919191919192e-05,
|
| 1600 |
+
"loss": 0.168,
|
| 1601 |
+
"step": 234
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 21.36,
|
| 1605 |
+
"learning_rate": 3.181818181818182e-05,
|
| 1606 |
+
"loss": 0.1516,
|
| 1607 |
+
"step": 235
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 21.45,
|
| 1611 |
+
"learning_rate": 3.171717171717172e-05,
|
| 1612 |
+
"loss": 0.1424,
|
| 1613 |
+
"step": 236
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 21.55,
|
| 1617 |
+
"learning_rate": 3.161616161616161e-05,
|
| 1618 |
+
"loss": 0.1368,
|
| 1619 |
+
"step": 237
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 21.64,
|
| 1623 |
+
"learning_rate": 3.151515151515151e-05,
|
| 1624 |
+
"loss": 0.234,
|
| 1625 |
+
"step": 238
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 21.73,
|
| 1629 |
+
"learning_rate": 3.141414141414142e-05,
|
| 1630 |
+
"loss": 0.1563,
|
| 1631 |
+
"step": 239
|
| 1632 |
+
},
|
| 1633 |
+
{
|
| 1634 |
+
"epoch": 21.82,
|
| 1635 |
+
"learning_rate": 3.131313131313132e-05,
|
| 1636 |
+
"loss": 0.1288,
|
| 1637 |
+
"step": 240
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 21.91,
|
| 1641 |
+
"learning_rate": 3.121212121212122e-05,
|
| 1642 |
+
"loss": 0.1547,
|
| 1643 |
+
"step": 241
|
| 1644 |
+
},
|
| 1645 |
+
{
|
| 1646 |
+
"epoch": 22.0,
|
| 1647 |
+
"learning_rate": 3.111111111111111e-05,
|
| 1648 |
+
"loss": 0.1666,
|
| 1649 |
+
"step": 242
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 22.0,
|
| 1653 |
+
"eval_accuracy": 0.9097222222222222,
|
| 1654 |
+
"eval_loss": 0.25945279002189636,
|
| 1655 |
+
"eval_runtime": 1.5674,
|
| 1656 |
+
"eval_samples_per_second": 183.74,
|
| 1657 |
+
"eval_steps_per_second": 3.19,
|
| 1658 |
+
"step": 242
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 22.09,
|
| 1662 |
+
"learning_rate": 3.101010101010101e-05,
|
| 1663 |
+
"loss": 0.1237,
|
| 1664 |
+
"step": 243
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 22.18,
|
| 1668 |
+
"learning_rate": 3.090909090909091e-05,
|
| 1669 |
+
"loss": 0.1206,
|
| 1670 |
+
"step": 244
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 22.27,
|
| 1674 |
+
"learning_rate": 3.080808080808081e-05,
|
| 1675 |
+
"loss": 0.1572,
|
| 1676 |
+
"step": 245
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 22.36,
|
| 1680 |
+
"learning_rate": 3.070707070707071e-05,
|
| 1681 |
+
"loss": 0.2006,
|
| 1682 |
+
"step": 246
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 22.45,
|
| 1686 |
+
"learning_rate": 3.060606060606061e-05,
|
| 1687 |
+
"loss": 0.191,
|
| 1688 |
+
"step": 247
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 22.55,
|
| 1692 |
+
"learning_rate": 3.050505050505051e-05,
|
| 1693 |
+
"loss": 0.1191,
|
| 1694 |
+
"step": 248
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 22.64,
|
| 1698 |
+
"learning_rate": 3.0404040404040406e-05,
|
| 1699 |
+
"loss": 0.1462,
|
| 1700 |
+
"step": 249
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 22.73,
|
| 1704 |
+
"learning_rate": 3.0303030303030306e-05,
|
| 1705 |
+
"loss": 0.1679,
|
| 1706 |
+
"step": 250
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 22.82,
|
| 1710 |
+
"learning_rate": 3.0202020202020205e-05,
|
| 1711 |
+
"loss": 0.2127,
|
| 1712 |
+
"step": 251
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 22.91,
|
| 1716 |
+
"learning_rate": 3.01010101010101e-05,
|
| 1717 |
+
"loss": 0.1359,
|
| 1718 |
+
"step": 252
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 23.0,
|
| 1722 |
+
"learning_rate": 3e-05,
|
| 1723 |
+
"loss": 0.1355,
|
| 1724 |
+
"step": 253
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 23.0,
|
| 1728 |
+
"eval_accuracy": 0.8923611111111112,
|
| 1729 |
+
"eval_loss": 0.2734811007976532,
|
| 1730 |
+
"eval_runtime": 1.5633,
|
| 1731 |
+
"eval_samples_per_second": 184.23,
|
| 1732 |
+
"eval_steps_per_second": 3.198,
|
| 1733 |
+
"step": 253
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 23.09,
|
| 1737 |
+
"learning_rate": 2.98989898989899e-05,
|
| 1738 |
+
"loss": 0.1139,
|
| 1739 |
+
"step": 254
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 23.18,
|
| 1743 |
+
"learning_rate": 2.9797979797979796e-05,
|
| 1744 |
+
"loss": 0.1299,
|
| 1745 |
+
"step": 255
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 23.27,
|
| 1749 |
+
"learning_rate": 2.96969696969697e-05,
|
| 1750 |
+
"loss": 0.1732,
|
| 1751 |
+
"step": 256
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 23.36,
|
| 1755 |
+
"learning_rate": 2.95959595959596e-05,
|
| 1756 |
+
"loss": 0.1417,
|
| 1757 |
+
"step": 257
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 23.45,
|
| 1761 |
+
"learning_rate": 2.9494949494949498e-05,
|
| 1762 |
+
"loss": 0.1588,
|
| 1763 |
+
"step": 258
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 23.55,
|
| 1767 |
+
"learning_rate": 2.9393939393939394e-05,
|
| 1768 |
+
"loss": 0.1159,
|
| 1769 |
+
"step": 259
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 23.64,
|
| 1773 |
+
"learning_rate": 2.9292929292929294e-05,
|
| 1774 |
+
"loss": 0.1636,
|
| 1775 |
+
"step": 260
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 23.73,
|
| 1779 |
+
"learning_rate": 2.9191919191919193e-05,
|
| 1780 |
+
"loss": 0.1375,
|
| 1781 |
+
"step": 261
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 23.82,
|
| 1785 |
+
"learning_rate": 2.909090909090909e-05,
|
| 1786 |
+
"loss": 0.128,
|
| 1787 |
+
"step": 262
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 23.91,
|
| 1791 |
+
"learning_rate": 2.898989898989899e-05,
|
| 1792 |
+
"loss": 0.1347,
|
| 1793 |
+
"step": 263
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 24.0,
|
| 1797 |
+
"learning_rate": 2.8888888888888888e-05,
|
| 1798 |
+
"loss": 0.1165,
|
| 1799 |
+
"step": 264
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 24.0,
|
| 1803 |
+
"eval_accuracy": 0.8784722222222222,
|
| 1804 |
+
"eval_loss": 0.3238440155982971,
|
| 1805 |
+
"eval_runtime": 1.6248,
|
| 1806 |
+
"eval_samples_per_second": 177.253,
|
| 1807 |
+
"eval_steps_per_second": 3.077,
|
| 1808 |
+
"step": 264
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 24.09,
|
| 1812 |
+
"learning_rate": 2.878787878787879e-05,
|
| 1813 |
+
"loss": 0.108,
|
| 1814 |
+
"step": 265
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 24.18,
|
| 1818 |
+
"learning_rate": 2.868686868686869e-05,
|
| 1819 |
+
"loss": 0.1276,
|
| 1820 |
+
"step": 266
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 24.27,
|
| 1824 |
+
"learning_rate": 2.8585858585858587e-05,
|
| 1825 |
+
"loss": 0.1893,
|
| 1826 |
+
"step": 267
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 24.36,
|
| 1830 |
+
"learning_rate": 2.8484848484848486e-05,
|
| 1831 |
+
"loss": 0.1291,
|
| 1832 |
+
"step": 268
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 24.45,
|
| 1836 |
+
"learning_rate": 2.8383838383838386e-05,
|
| 1837 |
+
"loss": 0.1547,
|
| 1838 |
+
"step": 269
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 24.55,
|
| 1842 |
+
"learning_rate": 2.8282828282828282e-05,
|
| 1843 |
+
"loss": 0.1697,
|
| 1844 |
+
"step": 270
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 24.64,
|
| 1848 |
+
"learning_rate": 2.818181818181818e-05,
|
| 1849 |
+
"loss": 0.1465,
|
| 1850 |
+
"step": 271
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 24.73,
|
| 1854 |
+
"learning_rate": 2.808080808080808e-05,
|
| 1855 |
+
"loss": 0.1422,
|
| 1856 |
+
"step": 272
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 24.82,
|
| 1860 |
+
"learning_rate": 2.7979797979797984e-05,
|
| 1861 |
+
"loss": 0.165,
|
| 1862 |
+
"step": 273
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 24.91,
|
| 1866 |
+
"learning_rate": 2.7878787878787883e-05,
|
| 1867 |
+
"loss": 0.1262,
|
| 1868 |
+
"step": 274
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 25.0,
|
| 1872 |
+
"learning_rate": 2.777777777777778e-05,
|
| 1873 |
+
"loss": 0.112,
|
| 1874 |
+
"step": 275
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 25.0,
|
| 1878 |
+
"eval_accuracy": 0.8888888888888888,
|
| 1879 |
+
"eval_loss": 0.3065586984157562,
|
| 1880 |
+
"eval_runtime": 1.5997,
|
| 1881 |
+
"eval_samples_per_second": 180.037,
|
| 1882 |
+
"eval_steps_per_second": 3.126,
|
| 1883 |
+
"step": 275
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 25.09,
|
| 1887 |
+
"learning_rate": 2.767676767676768e-05,
|
| 1888 |
+
"loss": 0.1132,
|
| 1889 |
+
"step": 276
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 25.18,
|
| 1893 |
+
"learning_rate": 2.7575757575757578e-05,
|
| 1894 |
+
"loss": 0.1252,
|
| 1895 |
+
"step": 277
|
| 1896 |
+
},
|
| 1897 |
+
{
|
| 1898 |
+
"epoch": 25.27,
|
| 1899 |
+
"learning_rate": 2.7474747474747474e-05,
|
| 1900 |
+
"loss": 0.1023,
|
| 1901 |
+
"step": 278
|
| 1902 |
+
},
|
| 1903 |
+
{
|
| 1904 |
+
"epoch": 25.36,
|
| 1905 |
+
"learning_rate": 2.7373737373737374e-05,
|
| 1906 |
+
"loss": 0.1254,
|
| 1907 |
+
"step": 279
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 25.45,
|
| 1911 |
+
"learning_rate": 2.7272727272727273e-05,
|
| 1912 |
+
"loss": 0.1575,
|
| 1913 |
+
"step": 280
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 25.55,
|
| 1917 |
+
"learning_rate": 2.717171717171717e-05,
|
| 1918 |
+
"loss": 0.1072,
|
| 1919 |
+
"step": 281
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 25.64,
|
| 1923 |
+
"learning_rate": 2.7070707070707075e-05,
|
| 1924 |
+
"loss": 0.135,
|
| 1925 |
+
"step": 282
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 25.73,
|
| 1929 |
+
"learning_rate": 2.696969696969697e-05,
|
| 1930 |
+
"loss": 0.1851,
|
| 1931 |
+
"step": 283
|
| 1932 |
+
},
|
| 1933 |
+
{
|
| 1934 |
+
"epoch": 25.82,
|
| 1935 |
+
"learning_rate": 2.686868686868687e-05,
|
| 1936 |
+
"loss": 0.1454,
|
| 1937 |
+
"step": 284
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 25.91,
|
| 1941 |
+
"learning_rate": 2.676767676767677e-05,
|
| 1942 |
+
"loss": 0.1629,
|
| 1943 |
+
"step": 285
|
| 1944 |
+
},
|
| 1945 |
+
{
|
| 1946 |
+
"epoch": 26.0,
|
| 1947 |
+
"learning_rate": 2.6666666666666667e-05,
|
| 1948 |
+
"loss": 0.1191,
|
| 1949 |
+
"step": 286
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 26.0,
|
| 1953 |
+
"eval_accuracy": 0.90625,
|
| 1954 |
+
"eval_loss": 0.24274842441082,
|
| 1955 |
+
"eval_runtime": 1.5782,
|
| 1956 |
+
"eval_samples_per_second": 182.491,
|
| 1957 |
+
"eval_steps_per_second": 3.168,
|
| 1958 |
+
"step": 286
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 26.09,
|
| 1962 |
+
"learning_rate": 2.6565656565656566e-05,
|
| 1963 |
+
"loss": 0.1726,
|
| 1964 |
+
"step": 287
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 26.18,
|
| 1968 |
+
"learning_rate": 2.6464646464646466e-05,
|
| 1969 |
+
"loss": 0.1017,
|
| 1970 |
+
"step": 288
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 26.27,
|
| 1974 |
+
"learning_rate": 2.636363636363636e-05,
|
| 1975 |
+
"loss": 0.0857,
|
| 1976 |
+
"step": 289
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 26.36,
|
| 1980 |
+
"learning_rate": 2.6262626262626268e-05,
|
| 1981 |
+
"loss": 0.1463,
|
| 1982 |
+
"step": 290
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 26.45,
|
| 1986 |
+
"learning_rate": 2.6161616161616164e-05,
|
| 1987 |
+
"loss": 0.1513,
|
| 1988 |
+
"step": 291
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 26.55,
|
| 1992 |
+
"learning_rate": 2.6060606060606063e-05,
|
| 1993 |
+
"loss": 0.0975,
|
| 1994 |
+
"step": 292
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 26.64,
|
| 1998 |
+
"learning_rate": 2.5959595959595963e-05,
|
| 1999 |
+
"loss": 0.1405,
|
| 2000 |
+
"step": 293
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 26.73,
|
| 2004 |
+
"learning_rate": 2.585858585858586e-05,
|
| 2005 |
+
"loss": 0.1236,
|
| 2006 |
+
"step": 294
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 26.82,
|
| 2010 |
+
"learning_rate": 2.575757575757576e-05,
|
| 2011 |
+
"loss": 0.1641,
|
| 2012 |
+
"step": 295
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 26.91,
|
| 2016 |
+
"learning_rate": 2.5656565656565658e-05,
|
| 2017 |
+
"loss": 0.1164,
|
| 2018 |
+
"step": 296
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 27.0,
|
| 2022 |
+
"learning_rate": 2.5555555555555554e-05,
|
| 2023 |
+
"loss": 0.1293,
|
| 2024 |
+
"step": 297
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 27.0,
|
| 2028 |
+
"eval_accuracy": 0.9201388888888888,
|
| 2029 |
+
"eval_loss": 0.25357797741889954,
|
| 2030 |
+
"eval_runtime": 1.574,
|
| 2031 |
+
"eval_samples_per_second": 182.979,
|
| 2032 |
+
"eval_steps_per_second": 3.177,
|
| 2033 |
+
"step": 297
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 27.09,
|
| 2037 |
+
"learning_rate": 2.5454545454545454e-05,
|
| 2038 |
+
"loss": 0.1206,
|
| 2039 |
+
"step": 298
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 27.18,
|
| 2043 |
+
"learning_rate": 2.5353535353535356e-05,
|
| 2044 |
+
"loss": 0.1533,
|
| 2045 |
+
"step": 299
|
| 2046 |
+
},
|
| 2047 |
+
{
|
| 2048 |
+
"epoch": 27.27,
|
| 2049 |
+
"learning_rate": 2.5252525252525256e-05,
|
| 2050 |
+
"loss": 0.1269,
|
| 2051 |
+
"step": 300
|
| 2052 |
+
},
|
| 2053 |
+
{
|
| 2054 |
+
"epoch": 27.36,
|
| 2055 |
+
"learning_rate": 2.5151515151515155e-05,
|
| 2056 |
+
"loss": 0.1208,
|
| 2057 |
+
"step": 301
|
| 2058 |
+
},
|
| 2059 |
+
{
|
| 2060 |
+
"epoch": 27.45,
|
| 2061 |
+
"learning_rate": 2.505050505050505e-05,
|
| 2062 |
+
"loss": 0.1069,
|
| 2063 |
+
"step": 302
|
| 2064 |
+
},
|
| 2065 |
+
{
|
| 2066 |
+
"epoch": 27.55,
|
| 2067 |
+
"learning_rate": 2.494949494949495e-05,
|
| 2068 |
+
"loss": 0.1034,
|
| 2069 |
+
"step": 303
|
| 2070 |
+
},
|
| 2071 |
+
{
|
| 2072 |
+
"epoch": 27.64,
|
| 2073 |
+
"learning_rate": 2.4848484848484847e-05,
|
| 2074 |
+
"loss": 0.087,
|
| 2075 |
+
"step": 304
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 27.73,
|
| 2079 |
+
"learning_rate": 2.474747474747475e-05,
|
| 2080 |
+
"loss": 0.0929,
|
| 2081 |
+
"step": 305
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 27.82,
|
| 2085 |
+
"learning_rate": 2.464646464646465e-05,
|
| 2086 |
+
"loss": 0.1334,
|
| 2087 |
+
"step": 306
|
| 2088 |
+
},
|
| 2089 |
+
{
|
| 2090 |
+
"epoch": 27.91,
|
| 2091 |
+
"learning_rate": 2.4545454545454545e-05,
|
| 2092 |
+
"loss": 0.112,
|
| 2093 |
+
"step": 307
|
| 2094 |
+
},
|
| 2095 |
+
{
|
| 2096 |
+
"epoch": 28.0,
|
| 2097 |
+
"learning_rate": 2.4444444444444445e-05,
|
| 2098 |
+
"loss": 0.2932,
|
| 2099 |
+
"step": 308
|
| 2100 |
+
},
|
| 2101 |
+
{
|
| 2102 |
+
"epoch": 28.0,
|
| 2103 |
+
"eval_accuracy": 0.8923611111111112,
|
| 2104 |
+
"eval_loss": 0.27067822217941284,
|
| 2105 |
+
"eval_runtime": 1.5637,
|
| 2106 |
+
"eval_samples_per_second": 184.181,
|
| 2107 |
+
"eval_steps_per_second": 3.198,
|
| 2108 |
+
"step": 308
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 28.09,
|
| 2112 |
+
"learning_rate": 2.4343434343434344e-05,
|
| 2113 |
+
"loss": 0.1466,
|
| 2114 |
+
"step": 309
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 28.18,
|
| 2118 |
+
"learning_rate": 2.4242424242424244e-05,
|
| 2119 |
+
"loss": 0.1069,
|
| 2120 |
+
"step": 310
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 28.27,
|
| 2124 |
+
"learning_rate": 2.4141414141414143e-05,
|
| 2125 |
+
"loss": 0.0743,
|
| 2126 |
+
"step": 311
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 28.36,
|
| 2130 |
+
"learning_rate": 2.404040404040404e-05,
|
| 2131 |
+
"loss": 0.0703,
|
| 2132 |
+
"step": 312
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 28.45,
|
| 2136 |
+
"learning_rate": 2.393939393939394e-05,
|
| 2137 |
+
"loss": 0.1335,
|
| 2138 |
+
"step": 313
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 28.55,
|
| 2142 |
+
"learning_rate": 2.3838383838383842e-05,
|
| 2143 |
+
"loss": 0.1313,
|
| 2144 |
+
"step": 314
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 28.64,
|
| 2148 |
+
"learning_rate": 2.3737373737373738e-05,
|
| 2149 |
+
"loss": 0.1224,
|
| 2150 |
+
"step": 315
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 28.73,
|
| 2154 |
+
"learning_rate": 2.3636363636363637e-05,
|
| 2155 |
+
"loss": 0.1197,
|
| 2156 |
+
"step": 316
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 28.82,
|
| 2160 |
+
"learning_rate": 2.3535353535353537e-05,
|
| 2161 |
+
"loss": 0.1188,
|
| 2162 |
+
"step": 317
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 28.91,
|
| 2166 |
+
"learning_rate": 2.3434343434343436e-05,
|
| 2167 |
+
"loss": 0.1575,
|
| 2168 |
+
"step": 318
|
| 2169 |
+
},
|
| 2170 |
+
{
|
| 2171 |
+
"epoch": 29.0,
|
| 2172 |
+
"learning_rate": 2.3333333333333336e-05,
|
| 2173 |
+
"loss": 0.0918,
|
| 2174 |
+
"step": 319
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 29.0,
|
| 2178 |
+
"eval_accuracy": 0.8923611111111112,
|
| 2179 |
+
"eval_loss": 0.26875555515289307,
|
| 2180 |
+
"eval_runtime": 1.5673,
|
| 2181 |
+
"eval_samples_per_second": 183.752,
|
| 2182 |
+
"eval_steps_per_second": 3.19,
|
| 2183 |
+
"step": 319
|
| 2184 |
+
},
|
| 2185 |
+
{
|
| 2186 |
+
"epoch": 29.09,
|
| 2187 |
+
"learning_rate": 2.3232323232323232e-05,
|
| 2188 |
+
"loss": 0.0987,
|
| 2189 |
+
"step": 320
|
| 2190 |
+
},
|
| 2191 |
+
{
|
| 2192 |
+
"epoch": 29.18,
|
| 2193 |
+
"learning_rate": 2.313131313131313e-05,
|
| 2194 |
+
"loss": 0.1075,
|
| 2195 |
+
"step": 321
|
| 2196 |
+
},
|
| 2197 |
+
{
|
| 2198 |
+
"epoch": 29.27,
|
| 2199 |
+
"learning_rate": 2.3030303030303034e-05,
|
| 2200 |
+
"loss": 0.1035,
|
| 2201 |
+
"step": 322
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 29.36,
|
| 2205 |
+
"learning_rate": 2.292929292929293e-05,
|
| 2206 |
+
"loss": 0.1098,
|
| 2207 |
+
"step": 323
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 29.45,
|
| 2211 |
+
"learning_rate": 2.282828282828283e-05,
|
| 2212 |
+
"loss": 0.0967,
|
| 2213 |
+
"step": 324
|
| 2214 |
+
},
|
| 2215 |
+
{
|
| 2216 |
+
"epoch": 29.55,
|
| 2217 |
+
"learning_rate": 2.272727272727273e-05,
|
| 2218 |
+
"loss": 0.1303,
|
| 2219 |
+
"step": 325
|
| 2220 |
+
},
|
| 2221 |
+
{
|
| 2222 |
+
"epoch": 29.64,
|
| 2223 |
+
"learning_rate": 2.262626262626263e-05,
|
| 2224 |
+
"loss": 0.1276,
|
| 2225 |
+
"step": 326
|
| 2226 |
+
},
|
| 2227 |
+
{
|
| 2228 |
+
"epoch": 29.73,
|
| 2229 |
+
"learning_rate": 2.2525252525252528e-05,
|
| 2230 |
+
"loss": 0.1105,
|
| 2231 |
+
"step": 327
|
| 2232 |
+
},
|
| 2233 |
+
{
|
| 2234 |
+
"epoch": 29.82,
|
| 2235 |
+
"learning_rate": 2.2424242424242424e-05,
|
| 2236 |
+
"loss": 0.1067,
|
| 2237 |
+
"step": 328
|
| 2238 |
+
},
|
| 2239 |
+
{
|
| 2240 |
+
"epoch": 29.91,
|
| 2241 |
+
"learning_rate": 2.2323232323232324e-05,
|
| 2242 |
+
"loss": 0.1119,
|
| 2243 |
+
"step": 329
|
| 2244 |
+
},
|
| 2245 |
+
{
|
| 2246 |
+
"epoch": 30.0,
|
| 2247 |
+
"learning_rate": 2.2222222222222223e-05,
|
| 2248 |
+
"loss": 0.1529,
|
| 2249 |
+
"step": 330
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 30.0,
|
| 2253 |
+
"eval_accuracy": 0.8888888888888888,
|
| 2254 |
+
"eval_loss": 0.2714914083480835,
|
| 2255 |
+
"eval_runtime": 1.5703,
|
| 2256 |
+
"eval_samples_per_second": 183.402,
|
| 2257 |
+
"eval_steps_per_second": 3.184,
|
| 2258 |
+
"step": 330
|
| 2259 |
+
},
|
| 2260 |
+
{
|
| 2261 |
+
"epoch": 30.09,
|
| 2262 |
+
"learning_rate": 2.2121212121212123e-05,
|
| 2263 |
+
"loss": 0.1097,
|
| 2264 |
+
"step": 331
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 30.18,
|
| 2268 |
+
"learning_rate": 2.2020202020202022e-05,
|
| 2269 |
+
"loss": 0.0645,
|
| 2270 |
+
"step": 332
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 30.27,
|
| 2274 |
+
"learning_rate": 2.191919191919192e-05,
|
| 2275 |
+
"loss": 0.098,
|
| 2276 |
+
"step": 333
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 30.36,
|
| 2280 |
+
"learning_rate": 2.1818181818181818e-05,
|
| 2281 |
+
"loss": 0.1134,
|
| 2282 |
+
"step": 334
|
| 2283 |
+
},
|
| 2284 |
+
{
|
| 2285 |
+
"epoch": 30.45,
|
| 2286 |
+
"learning_rate": 2.171717171717172e-05,
|
| 2287 |
+
"loss": 0.1161,
|
| 2288 |
+
"step": 335
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 30.55,
|
| 2292 |
+
"learning_rate": 2.1616161616161617e-05,
|
| 2293 |
+
"loss": 0.1452,
|
| 2294 |
+
"step": 336
|
| 2295 |
+
},
|
| 2296 |
+
{
|
| 2297 |
+
"epoch": 30.64,
|
| 2298 |
+
"learning_rate": 2.1515151515151516e-05,
|
| 2299 |
+
"loss": 0.0576,
|
| 2300 |
+
"step": 337
|
| 2301 |
+
},
|
| 2302 |
+
{
|
| 2303 |
+
"epoch": 30.73,
|
| 2304 |
+
"learning_rate": 2.1414141414141416e-05,
|
| 2305 |
+
"loss": 0.1293,
|
| 2306 |
+
"step": 338
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 30.82,
|
| 2310 |
+
"learning_rate": 2.1313131313131315e-05,
|
| 2311 |
+
"loss": 0.1267,
|
| 2312 |
+
"step": 339
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 30.91,
|
| 2316 |
+
"learning_rate": 2.1212121212121215e-05,
|
| 2317 |
+
"loss": 0.1049,
|
| 2318 |
+
"step": 340
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 31.0,
|
| 2322 |
+
"learning_rate": 2.111111111111111e-05,
|
| 2323 |
+
"loss": 0.227,
|
| 2324 |
+
"step": 341
|
| 2325 |
+
},
|
| 2326 |
+
{
|
| 2327 |
+
"epoch": 31.0,
|
| 2328 |
+
"eval_accuracy": 0.9027777777777778,
|
| 2329 |
+
"eval_loss": 0.2664012014865875,
|
| 2330 |
+
"eval_runtime": 1.5768,
|
| 2331 |
+
"eval_samples_per_second": 182.651,
|
| 2332 |
+
"eval_steps_per_second": 3.171,
|
| 2333 |
+
"step": 341
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 31.09,
|
| 2337 |
+
"learning_rate": 2.101010101010101e-05,
|
| 2338 |
+
"loss": 0.1239,
|
| 2339 |
+
"step": 342
|
| 2340 |
+
},
|
| 2341 |
+
{
|
| 2342 |
+
"epoch": 31.18,
|
| 2343 |
+
"learning_rate": 2.090909090909091e-05,
|
| 2344 |
+
"loss": 0.0908,
|
| 2345 |
+
"step": 343
|
| 2346 |
+
},
|
| 2347 |
+
{
|
| 2348 |
+
"epoch": 31.27,
|
| 2349 |
+
"learning_rate": 2.080808080808081e-05,
|
| 2350 |
+
"loss": 0.1132,
|
| 2351 |
+
"step": 344
|
| 2352 |
+
},
|
| 2353 |
+
{
|
| 2354 |
+
"epoch": 31.36,
|
| 2355 |
+
"learning_rate": 2.070707070707071e-05,
|
| 2356 |
+
"loss": 0.0946,
|
| 2357 |
+
"step": 345
|
| 2358 |
+
},
|
| 2359 |
+
{
|
| 2360 |
+
"epoch": 31.45,
|
| 2361 |
+
"learning_rate": 2.0606060606060608e-05,
|
| 2362 |
+
"loss": 0.093,
|
| 2363 |
+
"step": 346
|
| 2364 |
+
},
|
| 2365 |
+
{
|
| 2366 |
+
"epoch": 31.55,
|
| 2367 |
+
"learning_rate": 2.0505050505050504e-05,
|
| 2368 |
+
"loss": 0.1207,
|
| 2369 |
+
"step": 347
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 31.64,
|
| 2373 |
+
"learning_rate": 2.0404040404040407e-05,
|
| 2374 |
+
"loss": 0.111,
|
| 2375 |
+
"step": 348
|
| 2376 |
+
},
|
| 2377 |
+
{
|
| 2378 |
+
"epoch": 31.73,
|
| 2379 |
+
"learning_rate": 2.0303030303030303e-05,
|
| 2380 |
+
"loss": 0.1069,
|
| 2381 |
+
"step": 349
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 31.82,
|
| 2385 |
+
"learning_rate": 2.0202020202020203e-05,
|
| 2386 |
+
"loss": 0.0982,
|
| 2387 |
+
"step": 350
|
| 2388 |
+
},
|
| 2389 |
+
{
|
| 2390 |
+
"epoch": 31.91,
|
| 2391 |
+
"learning_rate": 2.0101010101010102e-05,
|
| 2392 |
+
"loss": 0.158,
|
| 2393 |
+
"step": 351
|
| 2394 |
+
},
|
| 2395 |
+
{
|
| 2396 |
+
"epoch": 32.0,
|
| 2397 |
+
"learning_rate": 2e-05,
|
| 2398 |
+
"loss": 0.1044,
|
| 2399 |
+
"step": 352
|
| 2400 |
+
},
|
| 2401 |
+
{
|
| 2402 |
+
"epoch": 32.0,
|
| 2403 |
+
"eval_accuracy": 0.8993055555555556,
|
| 2404 |
+
"eval_loss": 0.2808922529220581,
|
| 2405 |
+
"eval_runtime": 1.5749,
|
| 2406 |
+
"eval_samples_per_second": 182.87,
|
| 2407 |
+
"eval_steps_per_second": 3.175,
|
| 2408 |
+
"step": 352
|
| 2409 |
+
},
|
| 2410 |
+
{
|
| 2411 |
+
"epoch": 32.09,
|
| 2412 |
+
"learning_rate": 1.98989898989899e-05,
|
| 2413 |
+
"loss": 0.0917,
|
| 2414 |
+
"step": 353
|
| 2415 |
+
},
|
| 2416 |
+
{
|
| 2417 |
+
"epoch": 32.18,
|
| 2418 |
+
"learning_rate": 1.9797979797979797e-05,
|
| 2419 |
+
"loss": 0.0979,
|
| 2420 |
+
"step": 354
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 32.27,
|
| 2424 |
+
"learning_rate": 1.9696969696969697e-05,
|
| 2425 |
+
"loss": 0.111,
|
| 2426 |
+
"step": 355
|
| 2427 |
+
},
|
| 2428 |
+
{
|
| 2429 |
+
"epoch": 32.36,
|
| 2430 |
+
"learning_rate": 1.95959595959596e-05,
|
| 2431 |
+
"loss": 0.121,
|
| 2432 |
+
"step": 356
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 32.45,
|
| 2436 |
+
"learning_rate": 1.9494949494949496e-05,
|
| 2437 |
+
"loss": 0.1075,
|
| 2438 |
+
"step": 357
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 32.55,
|
| 2442 |
+
"learning_rate": 1.9393939393939395e-05,
|
| 2443 |
+
"loss": 0.1152,
|
| 2444 |
+
"step": 358
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 32.64,
|
| 2448 |
+
"learning_rate": 1.9292929292929295e-05,
|
| 2449 |
+
"loss": 0.0991,
|
| 2450 |
+
"step": 359
|
| 2451 |
+
},
|
| 2452 |
+
{
|
| 2453 |
+
"epoch": 32.73,
|
| 2454 |
+
"learning_rate": 1.919191919191919e-05,
|
| 2455 |
+
"loss": 0.1117,
|
| 2456 |
+
"step": 360
|
| 2457 |
+
},
|
| 2458 |
+
{
|
| 2459 |
+
"epoch": 32.82,
|
| 2460 |
+
"learning_rate": 1.9090909090909094e-05,
|
| 2461 |
+
"loss": 0.1275,
|
| 2462 |
+
"step": 361
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 32.91,
|
| 2466 |
+
"learning_rate": 1.898989898989899e-05,
|
| 2467 |
+
"loss": 0.1033,
|
| 2468 |
+
"step": 362
|
| 2469 |
+
},
|
| 2470 |
+
{
|
| 2471 |
+
"epoch": 33.0,
|
| 2472 |
+
"learning_rate": 1.888888888888889e-05,
|
| 2473 |
+
"loss": 0.0894,
|
| 2474 |
+
"step": 363
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 33.0,
|
| 2478 |
+
"eval_accuracy": 0.8923611111111112,
|
| 2479 |
+
"eval_loss": 0.2863243818283081,
|
| 2480 |
+
"eval_runtime": 1.5629,
|
| 2481 |
+
"eval_samples_per_second": 184.271,
|
| 2482 |
+
"eval_steps_per_second": 3.199,
|
| 2483 |
+
"step": 363
|
| 2484 |
+
},
|
| 2485 |
+
{
|
| 2486 |
+
"epoch": 33.09,
|
| 2487 |
+
"learning_rate": 1.878787878787879e-05,
|
| 2488 |
+
"loss": 0.0979,
|
| 2489 |
+
"step": 364
|
| 2490 |
+
},
|
| 2491 |
+
{
|
| 2492 |
+
"epoch": 33.18,
|
| 2493 |
+
"learning_rate": 1.8686868686868688e-05,
|
| 2494 |
+
"loss": 0.076,
|
| 2495 |
+
"step": 365
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 33.27,
|
| 2499 |
+
"learning_rate": 1.8585858585858588e-05,
|
| 2500 |
+
"loss": 0.0744,
|
| 2501 |
+
"step": 366
|
| 2502 |
+
},
|
| 2503 |
+
{
|
| 2504 |
+
"epoch": 33.36,
|
| 2505 |
+
"learning_rate": 1.8484848484848487e-05,
|
| 2506 |
+
"loss": 0.0952,
|
| 2507 |
+
"step": 367
|
| 2508 |
+
},
|
| 2509 |
+
{
|
| 2510 |
+
"epoch": 33.45,
|
| 2511 |
+
"learning_rate": 1.8383838383838383e-05,
|
| 2512 |
+
"loss": 0.1057,
|
| 2513 |
+
"step": 368
|
| 2514 |
+
},
|
| 2515 |
+
{
|
| 2516 |
+
"epoch": 33.55,
|
| 2517 |
+
"learning_rate": 1.8282828282828286e-05,
|
| 2518 |
+
"loss": 0.0713,
|
| 2519 |
+
"step": 369
|
| 2520 |
+
},
|
| 2521 |
+
{
|
| 2522 |
+
"epoch": 33.64,
|
| 2523 |
+
"learning_rate": 1.8181818181818182e-05,
|
| 2524 |
+
"loss": 0.0957,
|
| 2525 |
+
"step": 370
|
| 2526 |
+
},
|
| 2527 |
+
{
|
| 2528 |
+
"epoch": 33.73,
|
| 2529 |
+
"learning_rate": 1.808080808080808e-05,
|
| 2530 |
+
"loss": 0.0771,
|
| 2531 |
+
"step": 371
|
| 2532 |
+
},
|
| 2533 |
+
{
|
| 2534 |
+
"epoch": 33.82,
|
| 2535 |
+
"learning_rate": 1.797979797979798e-05,
|
| 2536 |
+
"loss": 0.111,
|
| 2537 |
+
"step": 372
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 33.91,
|
| 2541 |
+
"learning_rate": 1.787878787878788e-05,
|
| 2542 |
+
"loss": 0.1235,
|
| 2543 |
+
"step": 373
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 34.0,
|
| 2547 |
+
"learning_rate": 1.777777777777778e-05,
|
| 2548 |
+
"loss": 0.0566,
|
| 2549 |
+
"step": 374
|
| 2550 |
+
},
|
| 2551 |
+
{
|
| 2552 |
+
"epoch": 34.0,
|
| 2553 |
+
"eval_accuracy": 0.9201388888888888,
|
| 2554 |
+
"eval_loss": 0.24742072820663452,
|
| 2555 |
+
"eval_runtime": 1.5761,
|
| 2556 |
+
"eval_samples_per_second": 182.728,
|
| 2557 |
+
"eval_steps_per_second": 3.172,
|
| 2558 |
+
"step": 374
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 34.09,
|
| 2562 |
+
"learning_rate": 1.7676767676767676e-05,
|
| 2563 |
+
"loss": 0.0865,
|
| 2564 |
+
"step": 375
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 34.18,
|
| 2568 |
+
"learning_rate": 1.7575757575757576e-05,
|
| 2569 |
+
"loss": 0.1036,
|
| 2570 |
+
"step": 376
|
| 2571 |
+
},
|
| 2572 |
+
{
|
| 2573 |
+
"epoch": 34.27,
|
| 2574 |
+
"learning_rate": 1.7474747474747475e-05,
|
| 2575 |
+
"loss": 0.0638,
|
| 2576 |
+
"step": 377
|
| 2577 |
+
},
|
| 2578 |
+
{
|
| 2579 |
+
"epoch": 34.36,
|
| 2580 |
+
"learning_rate": 1.7373737373737375e-05,
|
| 2581 |
+
"loss": 0.0975,
|
| 2582 |
+
"step": 378
|
| 2583 |
+
},
|
| 2584 |
+
{
|
| 2585 |
+
"epoch": 34.45,
|
| 2586 |
+
"learning_rate": 1.7272727272727274e-05,
|
| 2587 |
+
"loss": 0.0933,
|
| 2588 |
+
"step": 379
|
| 2589 |
+
},
|
| 2590 |
+
{
|
| 2591 |
+
"epoch": 34.55,
|
| 2592 |
+
"learning_rate": 1.7171717171717173e-05,
|
| 2593 |
+
"loss": 0.1113,
|
| 2594 |
+
"step": 380
|
| 2595 |
+
},
|
| 2596 |
+
{
|
| 2597 |
+
"epoch": 34.64,
|
| 2598 |
+
"learning_rate": 1.707070707070707e-05,
|
| 2599 |
+
"loss": 0.0718,
|
| 2600 |
+
"step": 381
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 34.73,
|
| 2604 |
+
"learning_rate": 1.6969696969696972e-05,
|
| 2605 |
+
"loss": 0.1104,
|
| 2606 |
+
"step": 382
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 34.82,
|
| 2610 |
+
"learning_rate": 1.686868686868687e-05,
|
| 2611 |
+
"loss": 0.0873,
|
| 2612 |
+
"step": 383
|
| 2613 |
+
},
|
| 2614 |
+
{
|
| 2615 |
+
"epoch": 34.91,
|
| 2616 |
+
"learning_rate": 1.6767676767676768e-05,
|
| 2617 |
+
"loss": 0.1197,
|
| 2618 |
+
"step": 384
|
| 2619 |
+
},
|
| 2620 |
+
{
|
| 2621 |
+
"epoch": 35.0,
|
| 2622 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 2623 |
+
"loss": 0.0915,
|
| 2624 |
+
"step": 385
|
| 2625 |
+
},
|
| 2626 |
+
{
|
| 2627 |
+
"epoch": 35.0,
|
| 2628 |
+
"eval_accuracy": 0.9097222222222222,
|
| 2629 |
+
"eval_loss": 0.2427683174610138,
|
| 2630 |
+
"eval_runtime": 1.5702,
|
| 2631 |
+
"eval_samples_per_second": 183.413,
|
| 2632 |
+
"eval_steps_per_second": 3.184,
|
| 2633 |
+
"step": 385
|
| 2634 |
+
},
|
| 2635 |
+
{
|
| 2636 |
+
"epoch": 35.09,
|
| 2637 |
+
"learning_rate": 1.6565656565656567e-05,
|
| 2638 |
+
"loss": 0.1235,
|
| 2639 |
+
"step": 386
|
| 2640 |
+
},
|
| 2641 |
+
{
|
| 2642 |
+
"epoch": 35.18,
|
| 2643 |
+
"learning_rate": 1.6464646464646466e-05,
|
| 2644 |
+
"loss": 0.0871,
|
| 2645 |
+
"step": 387
|
| 2646 |
+
},
|
| 2647 |
+
{
|
| 2648 |
+
"epoch": 35.27,
|
| 2649 |
+
"learning_rate": 1.6363636363636366e-05,
|
| 2650 |
+
"loss": 0.0765,
|
| 2651 |
+
"step": 388
|
| 2652 |
+
},
|
| 2653 |
+
{
|
| 2654 |
+
"epoch": 35.36,
|
| 2655 |
+
"learning_rate": 1.6262626262626262e-05,
|
| 2656 |
+
"loss": 0.0927,
|
| 2657 |
+
"step": 389
|
| 2658 |
+
},
|
| 2659 |
+
{
|
| 2660 |
+
"epoch": 35.45,
|
| 2661 |
+
"learning_rate": 1.6161616161616165e-05,
|
| 2662 |
+
"loss": 0.1012,
|
| 2663 |
+
"step": 390
|
| 2664 |
+
},
|
| 2665 |
+
{
|
| 2666 |
+
"epoch": 35.55,
|
| 2667 |
+
"learning_rate": 1.606060606060606e-05,
|
| 2668 |
+
"loss": 0.084,
|
| 2669 |
+
"step": 391
|
| 2670 |
+
},
|
| 2671 |
+
{
|
| 2672 |
+
"epoch": 35.64,
|
| 2673 |
+
"learning_rate": 1.595959595959596e-05,
|
| 2674 |
+
"loss": 0.0936,
|
| 2675 |
+
"step": 392
|
| 2676 |
+
},
|
| 2677 |
+
{
|
| 2678 |
+
"epoch": 35.73,
|
| 2679 |
+
"learning_rate": 1.585858585858586e-05,
|
| 2680 |
+
"loss": 0.128,
|
| 2681 |
+
"step": 393
|
| 2682 |
+
},
|
| 2683 |
+
{
|
| 2684 |
+
"epoch": 35.82,
|
| 2685 |
+
"learning_rate": 1.5757575757575756e-05,
|
| 2686 |
+
"loss": 0.0969,
|
| 2687 |
+
"step": 394
|
| 2688 |
+
},
|
| 2689 |
+
{
|
| 2690 |
+
"epoch": 35.91,
|
| 2691 |
+
"learning_rate": 1.565656565656566e-05,
|
| 2692 |
+
"loss": 0.0794,
|
| 2693 |
+
"step": 395
|
| 2694 |
+
},
|
| 2695 |
+
{
|
| 2696 |
+
"epoch": 36.0,
|
| 2697 |
+
"learning_rate": 1.5555555555555555e-05,
|
| 2698 |
+
"loss": 0.1136,
|
| 2699 |
+
"step": 396
|
| 2700 |
+
},
|
| 2701 |
+
{
|
| 2702 |
+
"epoch": 36.0,
|
| 2703 |
+
"eval_accuracy": 0.9097222222222222,
|
| 2704 |
+
"eval_loss": 0.25445806980133057,
|
| 2705 |
+
"eval_runtime": 1.5637,
|
| 2706 |
+
"eval_samples_per_second": 184.175,
|
| 2707 |
+
"eval_steps_per_second": 3.197,
|
| 2708 |
+
"step": 396
|
| 2709 |
+
},
|
| 2710 |
+
{
|
| 2711 |
+
"epoch": 36.09,
|
| 2712 |
+
"learning_rate": 1.5454545454545454e-05,
|
| 2713 |
+
"loss": 0.0435,
|
| 2714 |
+
"step": 397
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 36.18,
|
| 2718 |
+
"learning_rate": 1.5353535353535354e-05,
|
| 2719 |
+
"loss": 0.0474,
|
| 2720 |
+
"step": 398
|
| 2721 |
+
},
|
| 2722 |
+
{
|
| 2723 |
+
"epoch": 36.27,
|
| 2724 |
+
"learning_rate": 1.5252525252525255e-05,
|
| 2725 |
+
"loss": 0.1082,
|
| 2726 |
+
"step": 399
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 36.36,
|
| 2730 |
+
"learning_rate": 1.5151515151515153e-05,
|
| 2731 |
+
"loss": 0.0613,
|
| 2732 |
+
"step": 400
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 36.45,
|
| 2736 |
+
"learning_rate": 1.505050505050505e-05,
|
| 2737 |
+
"loss": 0.0681,
|
| 2738 |
+
"step": 401
|
| 2739 |
+
},
|
| 2740 |
+
{
|
| 2741 |
+
"epoch": 36.55,
|
| 2742 |
+
"learning_rate": 1.494949494949495e-05,
|
| 2743 |
+
"loss": 0.0777,
|
| 2744 |
+
"step": 402
|
| 2745 |
+
},
|
| 2746 |
+
{
|
| 2747 |
+
"epoch": 36.64,
|
| 2748 |
+
"learning_rate": 1.484848484848485e-05,
|
| 2749 |
+
"loss": 0.0706,
|
| 2750 |
+
"step": 403
|
| 2751 |
+
},
|
| 2752 |
+
{
|
| 2753 |
+
"epoch": 36.73,
|
| 2754 |
+
"learning_rate": 1.4747474747474749e-05,
|
| 2755 |
+
"loss": 0.0696,
|
| 2756 |
+
"step": 404
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 36.82,
|
| 2760 |
+
"learning_rate": 1.4646464646464647e-05,
|
| 2761 |
+
"loss": 0.0689,
|
| 2762 |
+
"step": 405
|
| 2763 |
+
},
|
| 2764 |
+
{
|
| 2765 |
+
"epoch": 36.91,
|
| 2766 |
+
"learning_rate": 1.4545454545454545e-05,
|
| 2767 |
+
"loss": 0.0978,
|
| 2768 |
+
"step": 406
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 37.0,
|
| 2772 |
+
"learning_rate": 1.4444444444444444e-05,
|
| 2773 |
+
"loss": 0.0947,
|
| 2774 |
+
"step": 407
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 37.0,
|
| 2778 |
+
"eval_accuracy": 0.9097222222222222,
|
| 2779 |
+
"eval_loss": 0.2598806619644165,
|
| 2780 |
+
"eval_runtime": 1.5727,
|
| 2781 |
+
"eval_samples_per_second": 183.13,
|
| 2782 |
+
"eval_steps_per_second": 3.179,
|
| 2783 |
+
"step": 407
|
| 2784 |
+
},
|
| 2785 |
+
{
|
| 2786 |
+
"epoch": 37.09,
|
| 2787 |
+
"learning_rate": 1.4343434343434345e-05,
|
| 2788 |
+
"loss": 0.05,
|
| 2789 |
+
"step": 408
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 37.18,
|
| 2793 |
+
"learning_rate": 1.4242424242424243e-05,
|
| 2794 |
+
"loss": 0.1237,
|
| 2795 |
+
"step": 409
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 37.27,
|
| 2799 |
+
"learning_rate": 1.4141414141414141e-05,
|
| 2800 |
+
"loss": 0.0491,
|
| 2801 |
+
"step": 410
|
| 2802 |
+
},
|
| 2803 |
+
{
|
| 2804 |
+
"epoch": 37.36,
|
| 2805 |
+
"learning_rate": 1.404040404040404e-05,
|
| 2806 |
+
"loss": 0.0882,
|
| 2807 |
+
"step": 411
|
| 2808 |
+
},
|
| 2809 |
+
{
|
| 2810 |
+
"epoch": 37.45,
|
| 2811 |
+
"learning_rate": 1.3939393939393942e-05,
|
| 2812 |
+
"loss": 0.0765,
|
| 2813 |
+
"step": 412
|
| 2814 |
+
},
|
| 2815 |
+
{
|
| 2816 |
+
"epoch": 37.55,
|
| 2817 |
+
"learning_rate": 1.383838383838384e-05,
|
| 2818 |
+
"loss": 0.073,
|
| 2819 |
+
"step": 413
|
| 2820 |
+
},
|
| 2821 |
+
{
|
| 2822 |
+
"epoch": 37.64,
|
| 2823 |
+
"learning_rate": 1.3737373737373737e-05,
|
| 2824 |
+
"loss": 0.0728,
|
| 2825 |
+
"step": 414
|
| 2826 |
+
},
|
| 2827 |
+
{
|
| 2828 |
+
"epoch": 37.73,
|
| 2829 |
+
"learning_rate": 1.3636363636363637e-05,
|
| 2830 |
+
"loss": 0.075,
|
| 2831 |
+
"step": 415
|
| 2832 |
+
},
|
| 2833 |
+
{
|
| 2834 |
+
"epoch": 37.82,
|
| 2835 |
+
"learning_rate": 1.3535353535353538e-05,
|
| 2836 |
+
"loss": 0.0682,
|
| 2837 |
+
"step": 416
|
| 2838 |
+
},
|
| 2839 |
+
{
|
| 2840 |
+
"epoch": 37.91,
|
| 2841 |
+
"learning_rate": 1.3434343434343436e-05,
|
| 2842 |
+
"loss": 0.0675,
|
| 2843 |
+
"step": 417
|
| 2844 |
+
},
|
| 2845 |
+
{
|
| 2846 |
+
"epoch": 38.0,
|
| 2847 |
+
"learning_rate": 1.3333333333333333e-05,
|
| 2848 |
+
"loss": 0.1012,
|
| 2849 |
+
"step": 418
|
| 2850 |
+
},
|
| 2851 |
+
{
|
| 2852 |
+
"epoch": 38.0,
|
| 2853 |
+
"eval_accuracy": 0.9166666666666666,
|
| 2854 |
+
"eval_loss": 0.24541756510734558,
|
| 2855 |
+
"eval_runtime": 1.5701,
|
| 2856 |
+
"eval_samples_per_second": 183.432,
|
| 2857 |
+
"eval_steps_per_second": 3.185,
|
| 2858 |
+
"step": 418
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 38.09,
|
| 2862 |
+
"learning_rate": 1.3232323232323233e-05,
|
| 2863 |
+
"loss": 0.0642,
|
| 2864 |
+
"step": 419
|
| 2865 |
+
},
|
| 2866 |
+
{
|
| 2867 |
+
"epoch": 38.18,
|
| 2868 |
+
"learning_rate": 1.3131313131313134e-05,
|
| 2869 |
+
"loss": 0.0701,
|
| 2870 |
+
"step": 420
|
| 2871 |
+
},
|
| 2872 |
+
{
|
| 2873 |
+
"epoch": 38.27,
|
| 2874 |
+
"learning_rate": 1.3030303030303032e-05,
|
| 2875 |
+
"loss": 0.0675,
|
| 2876 |
+
"step": 421
|
| 2877 |
+
},
|
| 2878 |
+
{
|
| 2879 |
+
"epoch": 38.36,
|
| 2880 |
+
"learning_rate": 1.292929292929293e-05,
|
| 2881 |
+
"loss": 0.0763,
|
| 2882 |
+
"step": 422
|
| 2883 |
+
},
|
| 2884 |
+
{
|
| 2885 |
+
"epoch": 38.45,
|
| 2886 |
+
"learning_rate": 1.2828282828282829e-05,
|
| 2887 |
+
"loss": 0.0664,
|
| 2888 |
+
"step": 423
|
| 2889 |
+
},
|
| 2890 |
+
{
|
| 2891 |
+
"epoch": 38.55,
|
| 2892 |
+
"learning_rate": 1.2727272727272727e-05,
|
| 2893 |
+
"loss": 0.1222,
|
| 2894 |
+
"step": 424
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 38.64,
|
| 2898 |
+
"learning_rate": 1.2626262626262628e-05,
|
| 2899 |
+
"loss": 0.072,
|
| 2900 |
+
"step": 425
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 38.73,
|
| 2904 |
+
"learning_rate": 1.2525252525252526e-05,
|
| 2905 |
+
"loss": 0.0877,
|
| 2906 |
+
"step": 426
|
| 2907 |
+
},
|
| 2908 |
+
{
|
| 2909 |
+
"epoch": 38.82,
|
| 2910 |
+
"learning_rate": 1.2424242424242424e-05,
|
| 2911 |
+
"loss": 0.1098,
|
| 2912 |
+
"step": 427
|
| 2913 |
+
},
|
| 2914 |
+
{
|
| 2915 |
+
"epoch": 38.91,
|
| 2916 |
+
"learning_rate": 1.2323232323232325e-05,
|
| 2917 |
+
"loss": 0.0774,
|
| 2918 |
+
"step": 428
|
| 2919 |
+
},
|
| 2920 |
+
{
|
| 2921 |
+
"epoch": 39.0,
|
| 2922 |
+
"learning_rate": 1.2222222222222222e-05,
|
| 2923 |
+
"loss": 0.0465,
|
| 2924 |
+
"step": 429
|
| 2925 |
+
},
|
| 2926 |
+
{
|
| 2927 |
+
"epoch": 39.0,
|
| 2928 |
+
"eval_accuracy": 0.9201388888888888,
|
| 2929 |
+
"eval_loss": 0.24348551034927368,
|
| 2930 |
+
"eval_runtime": 1.5838,
|
| 2931 |
+
"eval_samples_per_second": 181.842,
|
| 2932 |
+
"eval_steps_per_second": 3.157,
|
| 2933 |
+
"step": 429
|
| 2934 |
+
},
|
| 2935 |
+
{
|
| 2936 |
+
"epoch": 39.09,
|
| 2937 |
+
"learning_rate": 1.2121212121212122e-05,
|
| 2938 |
+
"loss": 0.0474,
|
| 2939 |
+
"step": 430
|
| 2940 |
+
},
|
| 2941 |
+
{
|
| 2942 |
+
"epoch": 39.18,
|
| 2943 |
+
"learning_rate": 1.202020202020202e-05,
|
| 2944 |
+
"loss": 0.0535,
|
| 2945 |
+
"step": 431
|
| 2946 |
+
},
|
| 2947 |
+
{
|
| 2948 |
+
"epoch": 39.27,
|
| 2949 |
+
"learning_rate": 1.1919191919191921e-05,
|
| 2950 |
+
"loss": 0.1069,
|
| 2951 |
+
"step": 432
|
| 2952 |
+
},
|
| 2953 |
+
{
|
| 2954 |
+
"epoch": 39.36,
|
| 2955 |
+
"learning_rate": 1.1818181818181819e-05,
|
| 2956 |
+
"loss": 0.0872,
|
| 2957 |
+
"step": 433
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 39.45,
|
| 2961 |
+
"learning_rate": 1.1717171717171718e-05,
|
| 2962 |
+
"loss": 0.0852,
|
| 2963 |
+
"step": 434
|
| 2964 |
+
},
|
| 2965 |
+
{
|
| 2966 |
+
"epoch": 39.55,
|
| 2967 |
+
"learning_rate": 1.1616161616161616e-05,
|
| 2968 |
+
"loss": 0.0503,
|
| 2969 |
+
"step": 435
|
| 2970 |
+
},
|
| 2971 |
+
{
|
| 2972 |
+
"epoch": 39.64,
|
| 2973 |
+
"learning_rate": 1.1515151515151517e-05,
|
| 2974 |
+
"loss": 0.0746,
|
| 2975 |
+
"step": 436
|
| 2976 |
+
},
|
| 2977 |
+
{
|
| 2978 |
+
"epoch": 39.73,
|
| 2979 |
+
"learning_rate": 1.1414141414141415e-05,
|
| 2980 |
+
"loss": 0.0866,
|
| 2981 |
+
"step": 437
|
| 2982 |
+
},
|
| 2983 |
+
{
|
| 2984 |
+
"epoch": 39.82,
|
| 2985 |
+
"learning_rate": 1.1313131313131314e-05,
|
| 2986 |
+
"loss": 0.1031,
|
| 2987 |
+
"step": 438
|
| 2988 |
+
},
|
| 2989 |
+
{
|
| 2990 |
+
"epoch": 39.91,
|
| 2991 |
+
"learning_rate": 1.1212121212121212e-05,
|
| 2992 |
+
"loss": 0.075,
|
| 2993 |
+
"step": 439
|
| 2994 |
+
},
|
| 2995 |
+
{
|
| 2996 |
+
"epoch": 40.0,
|
| 2997 |
+
"learning_rate": 1.1111111111111112e-05,
|
| 2998 |
+
"loss": 0.0299,
|
| 2999 |
+
"step": 440
|
| 3000 |
+
},
|
| 3001 |
+
{
|
| 3002 |
+
"epoch": 40.0,
|
| 3003 |
+
"eval_accuracy": 0.90625,
|
| 3004 |
+
"eval_loss": 0.25319477915763855,
|
| 3005 |
+
"eval_runtime": 1.5767,
|
| 3006 |
+
"eval_samples_per_second": 182.661,
|
| 3007 |
+
"eval_steps_per_second": 3.171,
|
| 3008 |
+
"step": 440
|
| 3009 |
+
},
|
| 3010 |
+
{
|
| 3011 |
+
"epoch": 40.09,
|
| 3012 |
+
"learning_rate": 1.1010101010101011e-05,
|
| 3013 |
+
"loss": 0.0958,
|
| 3014 |
+
"step": 441
|
| 3015 |
+
},
|
| 3016 |
+
{
|
| 3017 |
+
"epoch": 40.18,
|
| 3018 |
+
"learning_rate": 1.0909090909090909e-05,
|
| 3019 |
+
"loss": 0.0964,
|
| 3020 |
+
"step": 442
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 40.27,
|
| 3024 |
+
"learning_rate": 1.0808080808080808e-05,
|
| 3025 |
+
"loss": 0.0524,
|
| 3026 |
+
"step": 443
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 40.36,
|
| 3030 |
+
"learning_rate": 1.0707070707070708e-05,
|
| 3031 |
+
"loss": 0.0487,
|
| 3032 |
+
"step": 444
|
| 3033 |
+
},
|
| 3034 |
+
{
|
| 3035 |
+
"epoch": 40.45,
|
| 3036 |
+
"learning_rate": 1.0606060606060607e-05,
|
| 3037 |
+
"loss": 0.0801,
|
| 3038 |
+
"step": 445
|
| 3039 |
+
},
|
| 3040 |
+
{
|
| 3041 |
+
"epoch": 40.55,
|
| 3042 |
+
"learning_rate": 1.0505050505050505e-05,
|
| 3043 |
+
"loss": 0.1102,
|
| 3044 |
+
"step": 446
|
| 3045 |
+
},
|
| 3046 |
+
{
|
| 3047 |
+
"epoch": 40.64,
|
| 3048 |
+
"learning_rate": 1.0404040404040405e-05,
|
| 3049 |
+
"loss": 0.0523,
|
| 3050 |
+
"step": 447
|
| 3051 |
+
},
|
| 3052 |
+
{
|
| 3053 |
+
"epoch": 40.73,
|
| 3054 |
+
"learning_rate": 1.0303030303030304e-05,
|
| 3055 |
+
"loss": 0.0825,
|
| 3056 |
+
"step": 448
|
| 3057 |
+
},
|
| 3058 |
+
{
|
| 3059 |
+
"epoch": 40.82,
|
| 3060 |
+
"learning_rate": 1.0202020202020204e-05,
|
| 3061 |
+
"loss": 0.0664,
|
| 3062 |
+
"step": 449
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 40.91,
|
| 3066 |
+
"learning_rate": 1.0101010101010101e-05,
|
| 3067 |
+
"loss": 0.0621,
|
| 3068 |
+
"step": 450
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 41.0,
|
| 3072 |
+
"learning_rate": 1e-05,
|
| 3073 |
+
"loss": 0.0311,
|
| 3074 |
+
"step": 451
|
| 3075 |
+
},
|
| 3076 |
+
{
|
| 3077 |
+
"epoch": 41.0,
|
| 3078 |
+
"eval_accuracy": 0.9270833333333334,
|
| 3079 |
+
"eval_loss": 0.2298324704170227,
|
| 3080 |
+
"eval_runtime": 1.577,
|
| 3081 |
+
"eval_samples_per_second": 182.629,
|
| 3082 |
+
"eval_steps_per_second": 3.171,
|
| 3083 |
+
"step": 451
|
| 3084 |
+
},
|
| 3085 |
+
{
|
| 3086 |
+
"epoch": 41.09,
|
| 3087 |
+
"learning_rate": 9.898989898989899e-06,
|
| 3088 |
+
"loss": 0.0482,
|
| 3089 |
+
"step": 452
|
| 3090 |
+
},
|
| 3091 |
+
{
|
| 3092 |
+
"epoch": 41.18,
|
| 3093 |
+
"learning_rate": 9.7979797979798e-06,
|
| 3094 |
+
"loss": 0.0874,
|
| 3095 |
+
"step": 453
|
| 3096 |
+
},
|
| 3097 |
+
{
|
| 3098 |
+
"epoch": 41.27,
|
| 3099 |
+
"learning_rate": 9.696969696969698e-06,
|
| 3100 |
+
"loss": 0.0582,
|
| 3101 |
+
"step": 454
|
| 3102 |
+
},
|
| 3103 |
+
{
|
| 3104 |
+
"epoch": 41.36,
|
| 3105 |
+
"learning_rate": 9.595959595959595e-06,
|
| 3106 |
+
"loss": 0.0519,
|
| 3107 |
+
"step": 455
|
| 3108 |
+
},
|
| 3109 |
+
{
|
| 3110 |
+
"epoch": 41.45,
|
| 3111 |
+
"learning_rate": 9.494949494949495e-06,
|
| 3112 |
+
"loss": 0.1067,
|
| 3113 |
+
"step": 456
|
| 3114 |
+
},
|
| 3115 |
+
{
|
| 3116 |
+
"epoch": 41.55,
|
| 3117 |
+
"learning_rate": 9.393939393939394e-06,
|
| 3118 |
+
"loss": 0.0855,
|
| 3119 |
+
"step": 457
|
| 3120 |
+
},
|
| 3121 |
+
{
|
| 3122 |
+
"epoch": 41.64,
|
| 3123 |
+
"learning_rate": 9.292929292929294e-06,
|
| 3124 |
+
"loss": 0.0874,
|
| 3125 |
+
"step": 458
|
| 3126 |
+
},
|
| 3127 |
+
{
|
| 3128 |
+
"epoch": 41.73,
|
| 3129 |
+
"learning_rate": 9.191919191919192e-06,
|
| 3130 |
+
"loss": 0.0601,
|
| 3131 |
+
"step": 459
|
| 3132 |
+
},
|
| 3133 |
+
{
|
| 3134 |
+
"epoch": 41.82,
|
| 3135 |
+
"learning_rate": 9.090909090909091e-06,
|
| 3136 |
+
"loss": 0.0572,
|
| 3137 |
+
"step": 460
|
| 3138 |
+
},
|
| 3139 |
+
{
|
| 3140 |
+
"epoch": 41.91,
|
| 3141 |
+
"learning_rate": 8.98989898989899e-06,
|
| 3142 |
+
"loss": 0.0762,
|
| 3143 |
+
"step": 461
|
| 3144 |
+
},
|
| 3145 |
+
{
|
| 3146 |
+
"epoch": 42.0,
|
| 3147 |
+
"learning_rate": 8.88888888888889e-06,
|
| 3148 |
+
"loss": 0.0796,
|
| 3149 |
+
"step": 462
|
| 3150 |
+
},
|
| 3151 |
+
{
|
| 3152 |
+
"epoch": 42.0,
|
| 3153 |
+
"eval_accuracy": 0.9166666666666666,
|
| 3154 |
+
"eval_loss": 0.24221903085708618,
|
| 3155 |
+
"eval_runtime": 1.5621,
|
| 3156 |
+
"eval_samples_per_second": 184.366,
|
| 3157 |
+
"eval_steps_per_second": 3.201,
|
| 3158 |
+
"step": 462
|
| 3159 |
+
},
|
| 3160 |
+
{
|
| 3161 |
+
"epoch": 42.09,
|
| 3162 |
+
"learning_rate": 8.787878787878788e-06,
|
| 3163 |
+
"loss": 0.0636,
|
| 3164 |
+
"step": 463
|
| 3165 |
+
},
|
| 3166 |
+
{
|
| 3167 |
+
"epoch": 42.18,
|
| 3168 |
+
"learning_rate": 8.686868686868687e-06,
|
| 3169 |
+
"loss": 0.0522,
|
| 3170 |
+
"step": 464
|
| 3171 |
+
},
|
| 3172 |
+
{
|
| 3173 |
+
"epoch": 42.27,
|
| 3174 |
+
"learning_rate": 8.585858585858587e-06,
|
| 3175 |
+
"loss": 0.0725,
|
| 3176 |
+
"step": 465
|
| 3177 |
+
},
|
| 3178 |
+
{
|
| 3179 |
+
"epoch": 42.36,
|
| 3180 |
+
"learning_rate": 8.484848484848486e-06,
|
| 3181 |
+
"loss": 0.0872,
|
| 3182 |
+
"step": 466
|
| 3183 |
+
},
|
| 3184 |
+
{
|
| 3185 |
+
"epoch": 42.45,
|
| 3186 |
+
"learning_rate": 8.383838383838384e-06,
|
| 3187 |
+
"loss": 0.0571,
|
| 3188 |
+
"step": 467
|
| 3189 |
+
},
|
| 3190 |
+
{
|
| 3191 |
+
"epoch": 42.55,
|
| 3192 |
+
"learning_rate": 8.282828282828283e-06,
|
| 3193 |
+
"loss": 0.0991,
|
| 3194 |
+
"step": 468
|
| 3195 |
+
},
|
| 3196 |
+
{
|
| 3197 |
+
"epoch": 42.64,
|
| 3198 |
+
"learning_rate": 8.181818181818183e-06,
|
| 3199 |
+
"loss": 0.0781,
|
| 3200 |
+
"step": 469
|
| 3201 |
+
},
|
| 3202 |
+
{
|
| 3203 |
+
"epoch": 42.73,
|
| 3204 |
+
"learning_rate": 8.080808080808082e-06,
|
| 3205 |
+
"loss": 0.0814,
|
| 3206 |
+
"step": 470
|
| 3207 |
+
},
|
| 3208 |
+
{
|
| 3209 |
+
"epoch": 42.82,
|
| 3210 |
+
"learning_rate": 7.97979797979798e-06,
|
| 3211 |
+
"loss": 0.071,
|
| 3212 |
+
"step": 471
|
| 3213 |
+
},
|
| 3214 |
+
{
|
| 3215 |
+
"epoch": 42.91,
|
| 3216 |
+
"learning_rate": 7.878787878787878e-06,
|
| 3217 |
+
"loss": 0.08,
|
| 3218 |
+
"step": 472
|
| 3219 |
+
},
|
| 3220 |
+
{
|
| 3221 |
+
"epoch": 43.0,
|
| 3222 |
+
"learning_rate": 7.777777777777777e-06,
|
| 3223 |
+
"loss": 0.058,
|
| 3224 |
+
"step": 473
|
| 3225 |
+
},
|
| 3226 |
+
{
|
| 3227 |
+
"epoch": 43.0,
|
| 3228 |
+
"eval_accuracy": 0.9305555555555556,
|
| 3229 |
+
"eval_loss": 0.2058059573173523,
|
| 3230 |
+
"eval_runtime": 1.5649,
|
| 3231 |
+
"eval_samples_per_second": 184.034,
|
| 3232 |
+
"eval_steps_per_second": 3.195,
|
| 3233 |
+
"step": 473
|
| 3234 |
+
},
|
| 3235 |
+
{
|
| 3236 |
+
"epoch": 43.09,
|
| 3237 |
+
"learning_rate": 7.676767676767677e-06,
|
| 3238 |
+
"loss": 0.0719,
|
| 3239 |
+
"step": 474
|
| 3240 |
+
},
|
| 3241 |
+
{
|
| 3242 |
+
"epoch": 43.18,
|
| 3243 |
+
"learning_rate": 7.5757575757575764e-06,
|
| 3244 |
+
"loss": 0.0663,
|
| 3245 |
+
"step": 475
|
| 3246 |
+
},
|
| 3247 |
+
{
|
| 3248 |
+
"epoch": 43.27,
|
| 3249 |
+
"learning_rate": 7.474747474747475e-06,
|
| 3250 |
+
"loss": 0.0397,
|
| 3251 |
+
"step": 476
|
| 3252 |
+
},
|
| 3253 |
+
{
|
| 3254 |
+
"epoch": 43.36,
|
| 3255 |
+
"learning_rate": 7.3737373737373745e-06,
|
| 3256 |
+
"loss": 0.0452,
|
| 3257 |
+
"step": 477
|
| 3258 |
+
},
|
| 3259 |
+
{
|
| 3260 |
+
"epoch": 43.45,
|
| 3261 |
+
"learning_rate": 7.272727272727272e-06,
|
| 3262 |
+
"loss": 0.0515,
|
| 3263 |
+
"step": 478
|
| 3264 |
+
},
|
| 3265 |
+
{
|
| 3266 |
+
"epoch": 43.55,
|
| 3267 |
+
"learning_rate": 7.171717171717173e-06,
|
| 3268 |
+
"loss": 0.0958,
|
| 3269 |
+
"step": 479
|
| 3270 |
+
},
|
| 3271 |
+
{
|
| 3272 |
+
"epoch": 43.64,
|
| 3273 |
+
"learning_rate": 7.0707070707070704e-06,
|
| 3274 |
+
"loss": 0.104,
|
| 3275 |
+
"step": 480
|
| 3276 |
+
},
|
| 3277 |
+
{
|
| 3278 |
+
"epoch": 43.73,
|
| 3279 |
+
"learning_rate": 6.969696969696971e-06,
|
| 3280 |
+
"loss": 0.0882,
|
| 3281 |
+
"step": 481
|
| 3282 |
+
},
|
| 3283 |
+
{
|
| 3284 |
+
"epoch": 43.82,
|
| 3285 |
+
"learning_rate": 6.8686868686868685e-06,
|
| 3286 |
+
"loss": 0.0734,
|
| 3287 |
+
"step": 482
|
| 3288 |
+
},
|
| 3289 |
+
{
|
| 3290 |
+
"epoch": 43.91,
|
| 3291 |
+
"learning_rate": 6.767676767676769e-06,
|
| 3292 |
+
"loss": 0.0448,
|
| 3293 |
+
"step": 483
|
| 3294 |
+
},
|
| 3295 |
+
{
|
| 3296 |
+
"epoch": 44.0,
|
| 3297 |
+
"learning_rate": 6.666666666666667e-06,
|
| 3298 |
+
"loss": 0.0853,
|
| 3299 |
+
"step": 484
|
| 3300 |
+
},
|
| 3301 |
+
{
|
| 3302 |
+
"epoch": 44.0,
|
| 3303 |
+
"eval_accuracy": 0.9305555555555556,
|
| 3304 |
+
"eval_loss": 0.2266363501548767,
|
| 3305 |
+
"eval_runtime": 1.5654,
|
| 3306 |
+
"eval_samples_per_second": 183.973,
|
| 3307 |
+
"eval_steps_per_second": 3.194,
|
| 3308 |
+
"step": 484
|
| 3309 |
+
},
|
| 3310 |
+
{
|
| 3311 |
+
"epoch": 44.09,
|
| 3312 |
+
"learning_rate": 6.565656565656567e-06,
|
| 3313 |
+
"loss": 0.0558,
|
| 3314 |
+
"step": 485
|
| 3315 |
+
},
|
| 3316 |
+
{
|
| 3317 |
+
"epoch": 44.18,
|
| 3318 |
+
"learning_rate": 6.464646464646465e-06,
|
| 3319 |
+
"loss": 0.0655,
|
| 3320 |
+
"step": 486
|
| 3321 |
+
},
|
| 3322 |
+
{
|
| 3323 |
+
"epoch": 44.27,
|
| 3324 |
+
"learning_rate": 6.363636363636363e-06,
|
| 3325 |
+
"loss": 0.0781,
|
| 3326 |
+
"step": 487
|
| 3327 |
+
},
|
| 3328 |
+
{
|
| 3329 |
+
"epoch": 44.36,
|
| 3330 |
+
"learning_rate": 6.262626262626263e-06,
|
| 3331 |
+
"loss": 0.0626,
|
| 3332 |
+
"step": 488
|
| 3333 |
+
},
|
| 3334 |
+
{
|
| 3335 |
+
"epoch": 44.45,
|
| 3336 |
+
"learning_rate": 6.161616161616162e-06,
|
| 3337 |
+
"loss": 0.0671,
|
| 3338 |
+
"step": 489
|
| 3339 |
+
},
|
| 3340 |
+
{
|
| 3341 |
+
"epoch": 44.55,
|
| 3342 |
+
"learning_rate": 6.060606060606061e-06,
|
| 3343 |
+
"loss": 0.0565,
|
| 3344 |
+
"step": 490
|
| 3345 |
+
},
|
| 3346 |
+
{
|
| 3347 |
+
"epoch": 44.64,
|
| 3348 |
+
"learning_rate": 5.9595959595959605e-06,
|
| 3349 |
+
"loss": 0.0517,
|
| 3350 |
+
"step": 491
|
| 3351 |
+
},
|
| 3352 |
+
{
|
| 3353 |
+
"epoch": 44.73,
|
| 3354 |
+
"learning_rate": 5.858585858585859e-06,
|
| 3355 |
+
"loss": 0.0785,
|
| 3356 |
+
"step": 492
|
| 3357 |
+
},
|
| 3358 |
+
{
|
| 3359 |
+
"epoch": 44.82,
|
| 3360 |
+
"learning_rate": 5.7575757575757586e-06,
|
| 3361 |
+
"loss": 0.0558,
|
| 3362 |
+
"step": 493
|
| 3363 |
+
},
|
| 3364 |
+
{
|
| 3365 |
+
"epoch": 44.91,
|
| 3366 |
+
"learning_rate": 5.656565656565657e-06,
|
| 3367 |
+
"loss": 0.0655,
|
| 3368 |
+
"step": 494
|
| 3369 |
+
},
|
| 3370 |
+
{
|
| 3371 |
+
"epoch": 45.0,
|
| 3372 |
+
"learning_rate": 5.555555555555556e-06,
|
| 3373 |
+
"loss": 0.0868,
|
| 3374 |
+
"step": 495
|
| 3375 |
+
},
|
| 3376 |
+
{
|
| 3377 |
+
"epoch": 45.0,
|
| 3378 |
+
"eval_accuracy": 0.9236111111111112,
|
| 3379 |
+
"eval_loss": 0.22659315168857574,
|
| 3380 |
+
"eval_runtime": 1.5757,
|
| 3381 |
+
"eval_samples_per_second": 182.772,
|
| 3382 |
+
"eval_steps_per_second": 3.173,
|
| 3383 |
+
"step": 495
|
| 3384 |
+
},
|
| 3385 |
+
{
|
| 3386 |
+
"epoch": 45.09,
|
| 3387 |
+
"learning_rate": 5.4545454545454545e-06,
|
| 3388 |
+
"loss": 0.0681,
|
| 3389 |
+
"step": 496
|
| 3390 |
+
},
|
| 3391 |
+
{
|
| 3392 |
+
"epoch": 45.18,
|
| 3393 |
+
"learning_rate": 5.353535353535354e-06,
|
| 3394 |
+
"loss": 0.0541,
|
| 3395 |
+
"step": 497
|
| 3396 |
+
},
|
| 3397 |
+
{
|
| 3398 |
+
"epoch": 45.27,
|
| 3399 |
+
"learning_rate": 5.2525252525252526e-06,
|
| 3400 |
+
"loss": 0.0695,
|
| 3401 |
+
"step": 498
|
| 3402 |
+
},
|
| 3403 |
+
{
|
| 3404 |
+
"epoch": 45.36,
|
| 3405 |
+
"learning_rate": 5.151515151515152e-06,
|
| 3406 |
+
"loss": 0.0344,
|
| 3407 |
+
"step": 499
|
| 3408 |
+
},
|
| 3409 |
+
{
|
| 3410 |
+
"epoch": 45.45,
|
| 3411 |
+
"learning_rate": 5.050505050505051e-06,
|
| 3412 |
+
"loss": 0.0554,
|
| 3413 |
+
"step": 500
|
| 3414 |
+
},
|
| 3415 |
+
{
|
| 3416 |
+
"epoch": 45.55,
|
| 3417 |
+
"learning_rate": 4.949494949494949e-06,
|
| 3418 |
+
"loss": 0.0691,
|
| 3419 |
+
"step": 501
|
| 3420 |
+
},
|
| 3421 |
+
{
|
| 3422 |
+
"epoch": 45.64,
|
| 3423 |
+
"learning_rate": 4.848484848484849e-06,
|
| 3424 |
+
"loss": 0.0715,
|
| 3425 |
+
"step": 502
|
| 3426 |
+
},
|
| 3427 |
+
{
|
| 3428 |
+
"epoch": 45.73,
|
| 3429 |
+
"learning_rate": 4.747474747474747e-06,
|
| 3430 |
+
"loss": 0.0519,
|
| 3431 |
+
"step": 503
|
| 3432 |
+
},
|
| 3433 |
+
{
|
| 3434 |
+
"epoch": 45.82,
|
| 3435 |
+
"learning_rate": 4.646464646464647e-06,
|
| 3436 |
+
"loss": 0.0658,
|
| 3437 |
+
"step": 504
|
| 3438 |
+
},
|
| 3439 |
+
{
|
| 3440 |
+
"epoch": 45.91,
|
| 3441 |
+
"learning_rate": 4.5454545454545455e-06,
|
| 3442 |
+
"loss": 0.088,
|
| 3443 |
+
"step": 505
|
| 3444 |
+
},
|
| 3445 |
+
{
|
| 3446 |
+
"epoch": 46.0,
|
| 3447 |
+
"learning_rate": 4.444444444444445e-06,
|
| 3448 |
+
"loss": 0.0554,
|
| 3449 |
+
"step": 506
|
| 3450 |
+
},
|
| 3451 |
+
{
|
| 3452 |
+
"epoch": 46.0,
|
| 3453 |
+
"eval_accuracy": 0.9270833333333334,
|
| 3454 |
+
"eval_loss": 0.2163047045469284,
|
| 3455 |
+
"eval_runtime": 1.5704,
|
| 3456 |
+
"eval_samples_per_second": 183.395,
|
| 3457 |
+
"eval_steps_per_second": 3.184,
|
| 3458 |
+
"step": 506
|
| 3459 |
+
},
|
| 3460 |
+
{
|
| 3461 |
+
"epoch": 46.09,
|
| 3462 |
+
"learning_rate": 4.343434343434344e-06,
|
| 3463 |
+
"loss": 0.0632,
|
| 3464 |
+
"step": 507
|
| 3465 |
+
},
|
| 3466 |
+
{
|
| 3467 |
+
"epoch": 46.18,
|
| 3468 |
+
"learning_rate": 4.242424242424243e-06,
|
| 3469 |
+
"loss": 0.0838,
|
| 3470 |
+
"step": 508
|
| 3471 |
+
},
|
| 3472 |
+
{
|
| 3473 |
+
"epoch": 46.27,
|
| 3474 |
+
"learning_rate": 4.141414141414142e-06,
|
| 3475 |
+
"loss": 0.0752,
|
| 3476 |
+
"step": 509
|
| 3477 |
+
},
|
| 3478 |
+
{
|
| 3479 |
+
"epoch": 46.36,
|
| 3480 |
+
"learning_rate": 4.040404040404041e-06,
|
| 3481 |
+
"loss": 0.0767,
|
| 3482 |
+
"step": 510
|
| 3483 |
+
},
|
| 3484 |
+
{
|
| 3485 |
+
"epoch": 46.45,
|
| 3486 |
+
"learning_rate": 3.939393939393939e-06,
|
| 3487 |
+
"loss": 0.0855,
|
| 3488 |
+
"step": 511
|
| 3489 |
+
},
|
| 3490 |
+
{
|
| 3491 |
+
"epoch": 46.55,
|
| 3492 |
+
"learning_rate": 3.8383838383838385e-06,
|
| 3493 |
+
"loss": 0.0392,
|
| 3494 |
+
"step": 512
|
| 3495 |
+
},
|
| 3496 |
+
{
|
| 3497 |
+
"epoch": 46.64,
|
| 3498 |
+
"learning_rate": 3.7373737373737375e-06,
|
| 3499 |
+
"loss": 0.0681,
|
| 3500 |
+
"step": 513
|
| 3501 |
+
},
|
| 3502 |
+
{
|
| 3503 |
+
"epoch": 46.73,
|
| 3504 |
+
"learning_rate": 3.636363636363636e-06,
|
| 3505 |
+
"loss": 0.0551,
|
| 3506 |
+
"step": 514
|
| 3507 |
+
},
|
| 3508 |
+
{
|
| 3509 |
+
"epoch": 46.82,
|
| 3510 |
+
"learning_rate": 3.5353535353535352e-06,
|
| 3511 |
+
"loss": 0.0516,
|
| 3512 |
+
"step": 515
|
| 3513 |
+
},
|
| 3514 |
+
{
|
| 3515 |
+
"epoch": 46.91,
|
| 3516 |
+
"learning_rate": 3.4343434343434343e-06,
|
| 3517 |
+
"loss": 0.0778,
|
| 3518 |
+
"step": 516
|
| 3519 |
+
},
|
| 3520 |
+
{
|
| 3521 |
+
"epoch": 47.0,
|
| 3522 |
+
"learning_rate": 3.3333333333333333e-06,
|
| 3523 |
+
"loss": 0.0508,
|
| 3524 |
+
"step": 517
|
| 3525 |
+
},
|
| 3526 |
+
{
|
| 3527 |
+
"epoch": 47.0,
|
| 3528 |
+
"eval_accuracy": 0.9305555555555556,
|
| 3529 |
+
"eval_loss": 0.21035218238830566,
|
| 3530 |
+
"eval_runtime": 1.5655,
|
| 3531 |
+
"eval_samples_per_second": 183.961,
|
| 3532 |
+
"eval_steps_per_second": 3.194,
|
| 3533 |
+
"step": 517
|
| 3534 |
+
},
|
| 3535 |
+
{
|
| 3536 |
+
"epoch": 47.09,
|
| 3537 |
+
"learning_rate": 3.2323232323232324e-06,
|
| 3538 |
+
"loss": 0.041,
|
| 3539 |
+
"step": 518
|
| 3540 |
+
},
|
| 3541 |
+
{
|
| 3542 |
+
"epoch": 47.18,
|
| 3543 |
+
"learning_rate": 3.1313131313131314e-06,
|
| 3544 |
+
"loss": 0.041,
|
| 3545 |
+
"step": 519
|
| 3546 |
+
},
|
| 3547 |
+
{
|
| 3548 |
+
"epoch": 47.27,
|
| 3549 |
+
"learning_rate": 3.0303030303030305e-06,
|
| 3550 |
+
"loss": 0.0813,
|
| 3551 |
+
"step": 520
|
| 3552 |
+
},
|
| 3553 |
+
{
|
| 3554 |
+
"epoch": 47.36,
|
| 3555 |
+
"learning_rate": 2.9292929292929295e-06,
|
| 3556 |
+
"loss": 0.0567,
|
| 3557 |
+
"step": 521
|
| 3558 |
+
},
|
| 3559 |
+
{
|
| 3560 |
+
"epoch": 47.45,
|
| 3561 |
+
"learning_rate": 2.8282828282828286e-06,
|
| 3562 |
+
"loss": 0.055,
|
| 3563 |
+
"step": 522
|
| 3564 |
+
},
|
| 3565 |
+
{
|
| 3566 |
+
"epoch": 47.55,
|
| 3567 |
+
"learning_rate": 2.7272727272727272e-06,
|
| 3568 |
+
"loss": 0.0505,
|
| 3569 |
+
"step": 523
|
| 3570 |
+
},
|
| 3571 |
+
{
|
| 3572 |
+
"epoch": 47.64,
|
| 3573 |
+
"learning_rate": 2.6262626262626263e-06,
|
| 3574 |
+
"loss": 0.0239,
|
| 3575 |
+
"step": 524
|
| 3576 |
+
},
|
| 3577 |
+
{
|
| 3578 |
+
"epoch": 47.73,
|
| 3579 |
+
"learning_rate": 2.5252525252525253e-06,
|
| 3580 |
+
"loss": 0.0441,
|
| 3581 |
+
"step": 525
|
| 3582 |
+
},
|
| 3583 |
+
{
|
| 3584 |
+
"epoch": 47.82,
|
| 3585 |
+
"learning_rate": 2.4242424242424244e-06,
|
| 3586 |
+
"loss": 0.0567,
|
| 3587 |
+
"step": 526
|
| 3588 |
+
},
|
| 3589 |
+
{
|
| 3590 |
+
"epoch": 47.91,
|
| 3591 |
+
"learning_rate": 2.3232323232323234e-06,
|
| 3592 |
+
"loss": 0.0711,
|
| 3593 |
+
"step": 527
|
| 3594 |
+
},
|
| 3595 |
+
{
|
| 3596 |
+
"epoch": 48.0,
|
| 3597 |
+
"learning_rate": 2.2222222222222225e-06,
|
| 3598 |
+
"loss": 0.0589,
|
| 3599 |
+
"step": 528
|
| 3600 |
+
},
|
| 3601 |
+
{
|
| 3602 |
+
"epoch": 48.0,
|
| 3603 |
+
"eval_accuracy": 0.9270833333333334,
|
| 3604 |
+
"eval_loss": 0.21717707812786102,
|
| 3605 |
+
"eval_runtime": 1.5625,
|
| 3606 |
+
"eval_samples_per_second": 184.318,
|
| 3607 |
+
"eval_steps_per_second": 3.2,
|
| 3608 |
+
"step": 528
|
| 3609 |
+
},
|
| 3610 |
+
{
|
| 3611 |
+
"epoch": 48.09,
|
| 3612 |
+
"learning_rate": 2.1212121212121216e-06,
|
| 3613 |
+
"loss": 0.0363,
|
| 3614 |
+
"step": 529
|
| 3615 |
+
},
|
| 3616 |
+
{
|
| 3617 |
+
"epoch": 48.18,
|
| 3618 |
+
"learning_rate": 2.0202020202020206e-06,
|
| 3619 |
+
"loss": 0.0767,
|
| 3620 |
+
"step": 530
|
| 3621 |
+
},
|
| 3622 |
+
{
|
| 3623 |
+
"epoch": 48.27,
|
| 3624 |
+
"learning_rate": 1.9191919191919192e-06,
|
| 3625 |
+
"loss": 0.0835,
|
| 3626 |
+
"step": 531
|
| 3627 |
+
},
|
| 3628 |
+
{
|
| 3629 |
+
"epoch": 48.36,
|
| 3630 |
+
"learning_rate": 1.818181818181818e-06,
|
| 3631 |
+
"loss": 0.0769,
|
| 3632 |
+
"step": 532
|
| 3633 |
+
},
|
| 3634 |
+
{
|
| 3635 |
+
"epoch": 48.45,
|
| 3636 |
+
"learning_rate": 1.7171717171717171e-06,
|
| 3637 |
+
"loss": 0.0373,
|
| 3638 |
+
"step": 533
|
| 3639 |
+
},
|
| 3640 |
+
{
|
| 3641 |
+
"epoch": 48.55,
|
| 3642 |
+
"learning_rate": 1.6161616161616162e-06,
|
| 3643 |
+
"loss": 0.0473,
|
| 3644 |
+
"step": 534
|
| 3645 |
+
},
|
| 3646 |
+
{
|
| 3647 |
+
"epoch": 48.64,
|
| 3648 |
+
"learning_rate": 1.5151515151515152e-06,
|
| 3649 |
+
"loss": 0.0439,
|
| 3650 |
+
"step": 535
|
| 3651 |
+
},
|
| 3652 |
+
{
|
| 3653 |
+
"epoch": 48.73,
|
| 3654 |
+
"learning_rate": 1.4141414141414143e-06,
|
| 3655 |
+
"loss": 0.0615,
|
| 3656 |
+
"step": 536
|
| 3657 |
+
},
|
| 3658 |
+
{
|
| 3659 |
+
"epoch": 48.82,
|
| 3660 |
+
"learning_rate": 1.3131313131313131e-06,
|
| 3661 |
+
"loss": 0.0554,
|
| 3662 |
+
"step": 537
|
| 3663 |
+
},
|
| 3664 |
+
{
|
| 3665 |
+
"epoch": 48.91,
|
| 3666 |
+
"learning_rate": 1.2121212121212122e-06,
|
| 3667 |
+
"loss": 0.0669,
|
| 3668 |
+
"step": 538
|
| 3669 |
+
},
|
| 3670 |
+
{
|
| 3671 |
+
"epoch": 49.0,
|
| 3672 |
+
"learning_rate": 1.1111111111111112e-06,
|
| 3673 |
+
"loss": 0.0369,
|
| 3674 |
+
"step": 539
|
| 3675 |
+
},
|
| 3676 |
+
{
|
| 3677 |
+
"epoch": 49.0,
|
| 3678 |
+
"eval_accuracy": 0.9270833333333334,
|
| 3679 |
+
"eval_loss": 0.22139286994934082,
|
| 3680 |
+
"eval_runtime": 1.5665,
|
| 3681 |
+
"eval_samples_per_second": 183.854,
|
| 3682 |
+
"eval_steps_per_second": 3.192,
|
| 3683 |
+
"step": 539
|
| 3684 |
+
},
|
| 3685 |
+
{
|
| 3686 |
+
"epoch": 49.09,
|
| 3687 |
+
"learning_rate": 1.0101010101010103e-06,
|
| 3688 |
+
"loss": 0.043,
|
| 3689 |
+
"step": 540
|
| 3690 |
+
},
|
| 3691 |
+
{
|
| 3692 |
+
"epoch": 49.18,
|
| 3693 |
+
"learning_rate": 9.09090909090909e-07,
|
| 3694 |
+
"loss": 0.0441,
|
| 3695 |
+
"step": 541
|
| 3696 |
+
},
|
| 3697 |
+
{
|
| 3698 |
+
"epoch": 49.27,
|
| 3699 |
+
"learning_rate": 8.080808080808081e-07,
|
| 3700 |
+
"loss": 0.0507,
|
| 3701 |
+
"step": 542
|
| 3702 |
+
},
|
| 3703 |
+
{
|
| 3704 |
+
"epoch": 49.36,
|
| 3705 |
+
"learning_rate": 7.070707070707071e-07,
|
| 3706 |
+
"loss": 0.0612,
|
| 3707 |
+
"step": 543
|
| 3708 |
+
},
|
| 3709 |
+
{
|
| 3710 |
+
"epoch": 49.45,
|
| 3711 |
+
"learning_rate": 6.060606060606061e-07,
|
| 3712 |
+
"loss": 0.0719,
|
| 3713 |
+
"step": 544
|
| 3714 |
+
},
|
| 3715 |
+
{
|
| 3716 |
+
"epoch": 49.55,
|
| 3717 |
+
"learning_rate": 5.050505050505052e-07,
|
| 3718 |
+
"loss": 0.0757,
|
| 3719 |
+
"step": 545
|
| 3720 |
+
},
|
| 3721 |
+
{
|
| 3722 |
+
"epoch": 49.64,
|
| 3723 |
+
"learning_rate": 4.0404040404040405e-07,
|
| 3724 |
+
"loss": 0.0473,
|
| 3725 |
+
"step": 546
|
| 3726 |
+
},
|
| 3727 |
+
{
|
| 3728 |
+
"epoch": 49.73,
|
| 3729 |
+
"learning_rate": 3.0303030303030305e-07,
|
| 3730 |
+
"loss": 0.0859,
|
| 3731 |
+
"step": 547
|
| 3732 |
+
},
|
| 3733 |
+
{
|
| 3734 |
+
"epoch": 49.82,
|
| 3735 |
+
"learning_rate": 2.0202020202020202e-07,
|
| 3736 |
+
"loss": 0.0462,
|
| 3737 |
+
"step": 548
|
| 3738 |
+
},
|
| 3739 |
+
{
|
| 3740 |
+
"epoch": 49.91,
|
| 3741 |
+
"learning_rate": 1.0101010101010101e-07,
|
| 3742 |
+
"loss": 0.0517,
|
| 3743 |
+
"step": 549
|
| 3744 |
+
},
|
| 3745 |
+
{
|
| 3746 |
+
"epoch": 50.0,
|
| 3747 |
+
"learning_rate": 0.0,
|
| 3748 |
+
"loss": 0.0852,
|
| 3749 |
+
"step": 550
|
| 3750 |
+
},
|
| 3751 |
+
{
|
| 3752 |
+
"epoch": 50.0,
|
| 3753 |
+
"eval_accuracy": 0.9270833333333334,
|
| 3754 |
+
"eval_loss": 0.22411441802978516,
|
| 3755 |
+
"eval_runtime": 1.5744,
|
| 3756 |
+
"eval_samples_per_second": 182.932,
|
| 3757 |
+
"eval_steps_per_second": 3.176,
|
| 3758 |
+
"step": 550
|
| 3759 |
+
},
|
| 3760 |
+
{
|
| 3761 |
+
"epoch": 50.0,
|
| 3762 |
+
"step": 550,
|
| 3763 |
+
"total_flos": 1.0035482482400256e+19,
|
| 3764 |
+
"train_loss": 0.24394690781493078,
|
| 3765 |
+
"train_runtime": 2413.5397,
|
| 3766 |
+
"train_samples_per_second": 53.656,
|
| 3767 |
+
"train_steps_per_second": 0.228
|
| 3768 |
+
}
|
| 3769 |
+
],
|
| 3770 |
+
"max_steps": 550,
|
| 3771 |
+
"num_train_epochs": 50,
|
| 3772 |
+
"total_flos": 1.0035482482400256e+19,
|
| 3773 |
+
"trial_name": null,
|
| 3774 |
+
"trial_params": null
|
| 3775 |
+
}
|