End of training
Browse files- all_results.json +13 -0
- eval_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +3745 -0
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 15.0,
|
| 3 |
+
"eval_accuracy": 0.9934024505183789,
|
| 4 |
+
"eval_loss": 0.029780708253383636,
|
| 5 |
+
"eval_runtime": 94.5115,
|
| 6 |
+
"eval_samples_per_second": 22.452,
|
| 7 |
+
"eval_steps_per_second": 1.407,
|
| 8 |
+
"total_flos": 2.615871779941884e+19,
|
| 9 |
+
"train_loss": 0.02590897159305381,
|
| 10 |
+
"train_runtime": 7346.925,
|
| 11 |
+
"train_samples_per_second": 12.993,
|
| 12 |
+
"train_steps_per_second": 0.406
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 15.0,
|
| 3 |
+
"eval_accuracy": 0.9934024505183789,
|
| 4 |
+
"eval_loss": 0.029780708253383636,
|
| 5 |
+
"eval_runtime": 94.5115,
|
| 6 |
+
"eval_samples_per_second": 22.452,
|
| 7 |
+
"eval_steps_per_second": 1.407
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 15.0,
|
| 3 |
+
"total_flos": 2.615871779941884e+19,
|
| 4 |
+
"train_loss": 0.02590897159305381,
|
| 5 |
+
"train_runtime": 7346.925,
|
| 6 |
+
"train_samples_per_second": 12.993,
|
| 7 |
+
"train_steps_per_second": 0.406
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3745 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9934024505183789,
|
| 3 |
+
"best_model_checkpoint": "beit-large-patch16-224-finetuned-BreastCancer-Classification-BreakHis-AH-60-20-20-Shuffled-3rd/checkpoint-2985",
|
| 4 |
+
"epoch": 15.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 2985,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.03,
|
| 13 |
+
"learning_rate": 9.304056568663938e-09,
|
| 14 |
+
"loss": 0.798,
|
| 15 |
+
"step": 5
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.05,
|
| 19 |
+
"learning_rate": 1.8608113137327877e-08,
|
| 20 |
+
"loss": 0.7081,
|
| 21 |
+
"step": 10
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.08,
|
| 25 |
+
"learning_rate": 2.7912169705991817e-08,
|
| 26 |
+
"loss": 0.7517,
|
| 27 |
+
"step": 15
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.1,
|
| 31 |
+
"learning_rate": 3.7216226274655753e-08,
|
| 32 |
+
"loss": 0.7047,
|
| 33 |
+
"step": 20
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.13,
|
| 37 |
+
"learning_rate": 4.652028284331969e-08,
|
| 38 |
+
"loss": 0.7814,
|
| 39 |
+
"step": 25
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.15,
|
| 43 |
+
"learning_rate": 5.5824339411983633e-08,
|
| 44 |
+
"loss": 0.8109,
|
| 45 |
+
"step": 30
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.18,
|
| 49 |
+
"learning_rate": 6.512839598064757e-08,
|
| 50 |
+
"loss": 0.7576,
|
| 51 |
+
"step": 35
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.2,
|
| 55 |
+
"learning_rate": 7.443245254931151e-08,
|
| 56 |
+
"loss": 0.7828,
|
| 57 |
+
"step": 40
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.23,
|
| 61 |
+
"learning_rate": 8.373650911797545e-08,
|
| 62 |
+
"loss": 0.7467,
|
| 63 |
+
"step": 45
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.25,
|
| 67 |
+
"learning_rate": 9.304056568663937e-08,
|
| 68 |
+
"loss": 0.8058,
|
| 69 |
+
"step": 50
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.28,
|
| 73 |
+
"learning_rate": 1.0234462225530333e-07,
|
| 74 |
+
"loss": 0.7589,
|
| 75 |
+
"step": 55
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.3,
|
| 79 |
+
"learning_rate": 1.1164867882396727e-07,
|
| 80 |
+
"loss": 0.7574,
|
| 81 |
+
"step": 60
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.33,
|
| 85 |
+
"learning_rate": 1.209527353926312e-07,
|
| 86 |
+
"loss": 0.7558,
|
| 87 |
+
"step": 65
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.35,
|
| 91 |
+
"learning_rate": 1.3025679196129513e-07,
|
| 92 |
+
"loss": 0.7326,
|
| 93 |
+
"step": 70
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.38,
|
| 97 |
+
"learning_rate": 1.3956084852995907e-07,
|
| 98 |
+
"loss": 0.6907,
|
| 99 |
+
"step": 75
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.4,
|
| 103 |
+
"learning_rate": 1.4886490509862301e-07,
|
| 104 |
+
"loss": 0.7293,
|
| 105 |
+
"step": 80
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.43,
|
| 109 |
+
"learning_rate": 1.5816896166728695e-07,
|
| 110 |
+
"loss": 0.6708,
|
| 111 |
+
"step": 85
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.45,
|
| 115 |
+
"learning_rate": 1.674730182359509e-07,
|
| 116 |
+
"loss": 0.725,
|
| 117 |
+
"step": 90
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.48,
|
| 121 |
+
"learning_rate": 1.767770748046148e-07,
|
| 122 |
+
"loss": 0.7069,
|
| 123 |
+
"step": 95
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.5,
|
| 127 |
+
"learning_rate": 1.8608113137327875e-07,
|
| 128 |
+
"loss": 0.6782,
|
| 129 |
+
"step": 100
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.53,
|
| 133 |
+
"learning_rate": 1.953851879419427e-07,
|
| 134 |
+
"loss": 0.6794,
|
| 135 |
+
"step": 105
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.55,
|
| 139 |
+
"learning_rate": 2.0468924451060665e-07,
|
| 140 |
+
"loss": 0.6767,
|
| 141 |
+
"step": 110
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.58,
|
| 145 |
+
"learning_rate": 2.139933010792706e-07,
|
| 146 |
+
"loss": 0.6379,
|
| 147 |
+
"step": 115
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.6,
|
| 151 |
+
"learning_rate": 2.2329735764793453e-07,
|
| 152 |
+
"loss": 0.6491,
|
| 153 |
+
"step": 120
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.63,
|
| 157 |
+
"learning_rate": 2.3260141421659845e-07,
|
| 158 |
+
"loss": 0.6331,
|
| 159 |
+
"step": 125
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.65,
|
| 163 |
+
"learning_rate": 2.419054707852624e-07,
|
| 164 |
+
"loss": 0.6542,
|
| 165 |
+
"step": 130
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.68,
|
| 169 |
+
"learning_rate": 2.5120952735392633e-07,
|
| 170 |
+
"loss": 0.6373,
|
| 171 |
+
"step": 135
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.7,
|
| 175 |
+
"learning_rate": 2.6051358392259027e-07,
|
| 176 |
+
"loss": 0.6505,
|
| 177 |
+
"step": 140
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.73,
|
| 181 |
+
"learning_rate": 2.698176404912542e-07,
|
| 182 |
+
"loss": 0.6339,
|
| 183 |
+
"step": 145
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 0.75,
|
| 187 |
+
"learning_rate": 2.7912169705991815e-07,
|
| 188 |
+
"loss": 0.6768,
|
| 189 |
+
"step": 150
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.78,
|
| 193 |
+
"learning_rate": 2.884257536285821e-07,
|
| 194 |
+
"loss": 0.6174,
|
| 195 |
+
"step": 155
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 0.8,
|
| 199 |
+
"learning_rate": 2.9772981019724603e-07,
|
| 200 |
+
"loss": 0.6361,
|
| 201 |
+
"step": 160
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.83,
|
| 205 |
+
"learning_rate": 3.0703386676590997e-07,
|
| 206 |
+
"loss": 0.569,
|
| 207 |
+
"step": 165
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.85,
|
| 211 |
+
"learning_rate": 3.163379233345739e-07,
|
| 212 |
+
"loss": 0.5613,
|
| 213 |
+
"step": 170
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.88,
|
| 217 |
+
"learning_rate": 3.2564197990323785e-07,
|
| 218 |
+
"loss": 0.5695,
|
| 219 |
+
"step": 175
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.9,
|
| 223 |
+
"learning_rate": 3.349460364719018e-07,
|
| 224 |
+
"loss": 0.6037,
|
| 225 |
+
"step": 180
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.93,
|
| 229 |
+
"learning_rate": 3.4425009304056573e-07,
|
| 230 |
+
"loss": 0.5345,
|
| 231 |
+
"step": 185
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 0.95,
|
| 235 |
+
"learning_rate": 3.535541496092296e-07,
|
| 236 |
+
"loss": 0.5354,
|
| 237 |
+
"step": 190
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.98,
|
| 241 |
+
"learning_rate": 3.628582061778936e-07,
|
| 242 |
+
"loss": 0.5462,
|
| 243 |
+
"step": 195
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 1.0,
|
| 247 |
+
"eval_accuracy": 0.793119698397738,
|
| 248 |
+
"eval_loss": 0.4728166460990906,
|
| 249 |
+
"eval_runtime": 1950.9789,
|
| 250 |
+
"eval_samples_per_second": 1.088,
|
| 251 |
+
"eval_steps_per_second": 0.068,
|
| 252 |
+
"step": 199
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 1.01,
|
| 256 |
+
"learning_rate": 3.721622627465575e-07,
|
| 257 |
+
"loss": 0.5034,
|
| 258 |
+
"step": 200
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 1.03,
|
| 262 |
+
"learning_rate": 3.814663193152215e-07,
|
| 263 |
+
"loss": 0.548,
|
| 264 |
+
"step": 205
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 1.06,
|
| 268 |
+
"learning_rate": 3.907703758838854e-07,
|
| 269 |
+
"loss": 0.5115,
|
| 270 |
+
"step": 210
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 1.08,
|
| 274 |
+
"learning_rate": 4.0007443245254937e-07,
|
| 275 |
+
"loss": 0.5267,
|
| 276 |
+
"step": 215
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 1.11,
|
| 280 |
+
"learning_rate": 4.093784890212133e-07,
|
| 281 |
+
"loss": 0.4888,
|
| 282 |
+
"step": 220
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 1.13,
|
| 286 |
+
"learning_rate": 4.186825455898772e-07,
|
| 287 |
+
"loss": 0.4961,
|
| 288 |
+
"step": 225
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 1.16,
|
| 292 |
+
"learning_rate": 4.279866021585412e-07,
|
| 293 |
+
"loss": 0.4923,
|
| 294 |
+
"step": 230
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 1.18,
|
| 298 |
+
"learning_rate": 4.372906587272051e-07,
|
| 299 |
+
"loss": 0.4252,
|
| 300 |
+
"step": 235
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 1.21,
|
| 304 |
+
"learning_rate": 4.4659471529586907e-07,
|
| 305 |
+
"loss": 0.4606,
|
| 306 |
+
"step": 240
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 1.23,
|
| 310 |
+
"learning_rate": 4.5589877186453295e-07,
|
| 311 |
+
"loss": 0.4365,
|
| 312 |
+
"step": 245
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 1.26,
|
| 316 |
+
"learning_rate": 4.652028284331969e-07,
|
| 317 |
+
"loss": 0.4343,
|
| 318 |
+
"step": 250
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 1.28,
|
| 322 |
+
"learning_rate": 4.7450688500186083e-07,
|
| 323 |
+
"loss": 0.4589,
|
| 324 |
+
"step": 255
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 1.31,
|
| 328 |
+
"learning_rate": 4.838109415705248e-07,
|
| 329 |
+
"loss": 0.4263,
|
| 330 |
+
"step": 260
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 1.33,
|
| 334 |
+
"learning_rate": 4.931149981391888e-07,
|
| 335 |
+
"loss": 0.4504,
|
| 336 |
+
"step": 265
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 1.36,
|
| 340 |
+
"learning_rate": 5.024190547078527e-07,
|
| 341 |
+
"loss": 0.3655,
|
| 342 |
+
"step": 270
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 1.38,
|
| 346 |
+
"learning_rate": 5.117231112765166e-07,
|
| 347 |
+
"loss": 0.3757,
|
| 348 |
+
"step": 275
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 1.41,
|
| 352 |
+
"learning_rate": 5.210271678451805e-07,
|
| 353 |
+
"loss": 0.4056,
|
| 354 |
+
"step": 280
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 1.43,
|
| 358 |
+
"learning_rate": 5.303312244138445e-07,
|
| 359 |
+
"loss": 0.4005,
|
| 360 |
+
"step": 285
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 1.46,
|
| 364 |
+
"learning_rate": 5.396352809825084e-07,
|
| 365 |
+
"loss": 0.3598,
|
| 366 |
+
"step": 290
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 1.48,
|
| 370 |
+
"learning_rate": 5.489393375511724e-07,
|
| 371 |
+
"loss": 0.3617,
|
| 372 |
+
"step": 295
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 1.51,
|
| 376 |
+
"learning_rate": 5.582433941198363e-07,
|
| 377 |
+
"loss": 0.4281,
|
| 378 |
+
"step": 300
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 1.53,
|
| 382 |
+
"learning_rate": 5.675474506885002e-07,
|
| 383 |
+
"loss": 0.3632,
|
| 384 |
+
"step": 305
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 1.56,
|
| 388 |
+
"learning_rate": 5.768515072571642e-07,
|
| 389 |
+
"loss": 0.3617,
|
| 390 |
+
"step": 310
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 1.58,
|
| 394 |
+
"learning_rate": 5.861555638258281e-07,
|
| 395 |
+
"loss": 0.4292,
|
| 396 |
+
"step": 315
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 1.61,
|
| 400 |
+
"learning_rate": 5.954596203944921e-07,
|
| 401 |
+
"loss": 0.3833,
|
| 402 |
+
"step": 320
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 1.63,
|
| 406 |
+
"learning_rate": 6.047636769631559e-07,
|
| 407 |
+
"loss": 0.3883,
|
| 408 |
+
"step": 325
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 1.66,
|
| 412 |
+
"learning_rate": 6.140677335318199e-07,
|
| 413 |
+
"loss": 0.3659,
|
| 414 |
+
"step": 330
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 1.68,
|
| 418 |
+
"learning_rate": 6.233717901004838e-07,
|
| 419 |
+
"loss": 0.3971,
|
| 420 |
+
"step": 335
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 1.71,
|
| 424 |
+
"learning_rate": 6.326758466691478e-07,
|
| 425 |
+
"loss": 0.3381,
|
| 426 |
+
"step": 340
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 1.73,
|
| 430 |
+
"learning_rate": 6.419799032378117e-07,
|
| 431 |
+
"loss": 0.2912,
|
| 432 |
+
"step": 345
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 1.76,
|
| 436 |
+
"learning_rate": 6.512839598064757e-07,
|
| 437 |
+
"loss": 0.3316,
|
| 438 |
+
"step": 350
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 1.78,
|
| 442 |
+
"learning_rate": 6.605880163751397e-07,
|
| 443 |
+
"loss": 0.3128,
|
| 444 |
+
"step": 355
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 1.81,
|
| 448 |
+
"learning_rate": 6.698920729438036e-07,
|
| 449 |
+
"loss": 0.3445,
|
| 450 |
+
"step": 360
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 1.83,
|
| 454 |
+
"learning_rate": 6.791961295124675e-07,
|
| 455 |
+
"loss": 0.2811,
|
| 456 |
+
"step": 365
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 1.86,
|
| 460 |
+
"learning_rate": 6.885001860811315e-07,
|
| 461 |
+
"loss": 0.309,
|
| 462 |
+
"step": 370
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 1.88,
|
| 466 |
+
"learning_rate": 6.978042426497953e-07,
|
| 467 |
+
"loss": 0.2952,
|
| 468 |
+
"step": 375
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 1.91,
|
| 472 |
+
"learning_rate": 7.071082992184592e-07,
|
| 473 |
+
"loss": 0.2879,
|
| 474 |
+
"step": 380
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 1.93,
|
| 478 |
+
"learning_rate": 7.164123557871232e-07,
|
| 479 |
+
"loss": 0.2786,
|
| 480 |
+
"step": 385
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 1.96,
|
| 484 |
+
"learning_rate": 7.257164123557872e-07,
|
| 485 |
+
"loss": 0.2906,
|
| 486 |
+
"step": 390
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 1.98,
|
| 490 |
+
"learning_rate": 7.350204689244511e-07,
|
| 491 |
+
"loss": 0.2496,
|
| 492 |
+
"step": 395
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 2.0,
|
| 496 |
+
"eval_accuracy": 0.9184731385485391,
|
| 497 |
+
"eval_loss": 0.20740385353565216,
|
| 498 |
+
"eval_runtime": 94.0317,
|
| 499 |
+
"eval_samples_per_second": 22.567,
|
| 500 |
+
"eval_steps_per_second": 1.414,
|
| 501 |
+
"step": 398
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 2.01,
|
| 505 |
+
"learning_rate": 7.44324525493115e-07,
|
| 506 |
+
"loss": 0.2091,
|
| 507 |
+
"step": 400
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 2.04,
|
| 511 |
+
"learning_rate": 7.536285820617791e-07,
|
| 512 |
+
"loss": 0.2973,
|
| 513 |
+
"step": 405
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 2.06,
|
| 517 |
+
"learning_rate": 7.62932638630443e-07,
|
| 518 |
+
"loss": 0.2198,
|
| 519 |
+
"step": 410
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 2.09,
|
| 523 |
+
"learning_rate": 7.722366951991069e-07,
|
| 524 |
+
"loss": 0.2569,
|
| 525 |
+
"step": 415
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 2.11,
|
| 529 |
+
"learning_rate": 7.815407517677707e-07,
|
| 530 |
+
"loss": 0.2925,
|
| 531 |
+
"step": 420
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 2.14,
|
| 535 |
+
"learning_rate": 7.908448083364347e-07,
|
| 536 |
+
"loss": 0.2781,
|
| 537 |
+
"step": 425
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 2.16,
|
| 541 |
+
"learning_rate": 8.001488649050987e-07,
|
| 542 |
+
"loss": 0.1877,
|
| 543 |
+
"step": 430
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 2.19,
|
| 547 |
+
"learning_rate": 8.094529214737626e-07,
|
| 548 |
+
"loss": 0.2391,
|
| 549 |
+
"step": 435
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 2.21,
|
| 553 |
+
"learning_rate": 8.187569780424266e-07,
|
| 554 |
+
"loss": 0.2658,
|
| 555 |
+
"step": 440
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 2.24,
|
| 559 |
+
"learning_rate": 8.280610346110905e-07,
|
| 560 |
+
"loss": 0.169,
|
| 561 |
+
"step": 445
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 2.26,
|
| 565 |
+
"learning_rate": 8.373650911797544e-07,
|
| 566 |
+
"loss": 0.3144,
|
| 567 |
+
"step": 450
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 2.29,
|
| 571 |
+
"learning_rate": 8.466691477484183e-07,
|
| 572 |
+
"loss": 0.2124,
|
| 573 |
+
"step": 455
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 2.31,
|
| 577 |
+
"learning_rate": 8.559732043170824e-07,
|
| 578 |
+
"loss": 0.2338,
|
| 579 |
+
"step": 460
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 2.34,
|
| 583 |
+
"learning_rate": 8.652772608857463e-07,
|
| 584 |
+
"loss": 0.2109,
|
| 585 |
+
"step": 465
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 2.36,
|
| 589 |
+
"learning_rate": 8.745813174544101e-07,
|
| 590 |
+
"loss": 0.2428,
|
| 591 |
+
"step": 470
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 2.39,
|
| 595 |
+
"learning_rate": 8.838853740230742e-07,
|
| 596 |
+
"loss": 0.2306,
|
| 597 |
+
"step": 475
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 2.41,
|
| 601 |
+
"learning_rate": 8.931894305917381e-07,
|
| 602 |
+
"loss": 0.1823,
|
| 603 |
+
"step": 480
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 2.44,
|
| 607 |
+
"learning_rate": 9.02493487160402e-07,
|
| 608 |
+
"loss": 0.1852,
|
| 609 |
+
"step": 485
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 2.46,
|
| 613 |
+
"learning_rate": 9.117975437290659e-07,
|
| 614 |
+
"loss": 0.2175,
|
| 615 |
+
"step": 490
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 2.49,
|
| 619 |
+
"learning_rate": 9.211016002977299e-07,
|
| 620 |
+
"loss": 0.1552,
|
| 621 |
+
"step": 495
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 2.51,
|
| 625 |
+
"learning_rate": 9.304056568663938e-07,
|
| 626 |
+
"loss": 0.2086,
|
| 627 |
+
"step": 500
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 2.54,
|
| 631 |
+
"learning_rate": 9.397097134350578e-07,
|
| 632 |
+
"loss": 0.2033,
|
| 633 |
+
"step": 505
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 2.56,
|
| 637 |
+
"learning_rate": 9.490137700037217e-07,
|
| 638 |
+
"loss": 0.2066,
|
| 639 |
+
"step": 510
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 2.59,
|
| 643 |
+
"learning_rate": 9.583178265723858e-07,
|
| 644 |
+
"loss": 0.1464,
|
| 645 |
+
"step": 515
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 2.61,
|
| 649 |
+
"learning_rate": 9.676218831410495e-07,
|
| 650 |
+
"loss": 0.2294,
|
| 651 |
+
"step": 520
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 2.64,
|
| 655 |
+
"learning_rate": 9.769259397097135e-07,
|
| 656 |
+
"loss": 0.2091,
|
| 657 |
+
"step": 525
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 2.66,
|
| 661 |
+
"learning_rate": 9.862299962783775e-07,
|
| 662 |
+
"loss": 0.1767,
|
| 663 |
+
"step": 530
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 2.69,
|
| 667 |
+
"learning_rate": 9.955340528470413e-07,
|
| 668 |
+
"loss": 0.191,
|
| 669 |
+
"step": 535
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 2.71,
|
| 673 |
+
"learning_rate": 1.0048381094157053e-06,
|
| 674 |
+
"loss": 0.1785,
|
| 675 |
+
"step": 540
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 2.74,
|
| 679 |
+
"learning_rate": 1.0141421659843693e-06,
|
| 680 |
+
"loss": 0.2185,
|
| 681 |
+
"step": 545
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 2.76,
|
| 685 |
+
"learning_rate": 1.0234462225530333e-06,
|
| 686 |
+
"loss": 0.2335,
|
| 687 |
+
"step": 550
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 2.79,
|
| 691 |
+
"learning_rate": 1.032750279121697e-06,
|
| 692 |
+
"loss": 0.3026,
|
| 693 |
+
"step": 555
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 2.81,
|
| 697 |
+
"learning_rate": 1.042054335690361e-06,
|
| 698 |
+
"loss": 0.1544,
|
| 699 |
+
"step": 560
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 2.84,
|
| 703 |
+
"learning_rate": 1.051358392259025e-06,
|
| 704 |
+
"loss": 0.2014,
|
| 705 |
+
"step": 565
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 2.86,
|
| 709 |
+
"learning_rate": 1.060662448827689e-06,
|
| 710 |
+
"loss": 0.1765,
|
| 711 |
+
"step": 570
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 2.89,
|
| 715 |
+
"learning_rate": 1.0699665053963528e-06,
|
| 716 |
+
"loss": 0.2036,
|
| 717 |
+
"step": 575
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 2.91,
|
| 721 |
+
"learning_rate": 1.0792705619650168e-06,
|
| 722 |
+
"loss": 0.2085,
|
| 723 |
+
"step": 580
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 2.94,
|
| 727 |
+
"learning_rate": 1.0885746185336808e-06,
|
| 728 |
+
"loss": 0.1771,
|
| 729 |
+
"step": 585
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 2.96,
|
| 733 |
+
"learning_rate": 1.0978786751023448e-06,
|
| 734 |
+
"loss": 0.1893,
|
| 735 |
+
"step": 590
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 2.99,
|
| 739 |
+
"learning_rate": 1.1071827316710086e-06,
|
| 740 |
+
"loss": 0.2173,
|
| 741 |
+
"step": 595
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 3.0,
|
| 745 |
+
"eval_accuracy": 0.9486333647502356,
|
| 746 |
+
"eval_loss": 0.13052009046077728,
|
| 747 |
+
"eval_runtime": 93.8967,
|
| 748 |
+
"eval_samples_per_second": 22.599,
|
| 749 |
+
"eval_steps_per_second": 1.416,
|
| 750 |
+
"step": 597
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 3.02,
|
| 754 |
+
"learning_rate": 1.1164867882396726e-06,
|
| 755 |
+
"loss": 0.1243,
|
| 756 |
+
"step": 600
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 3.04,
|
| 760 |
+
"learning_rate": 1.1257908448083366e-06,
|
| 761 |
+
"loss": 0.1945,
|
| 762 |
+
"step": 605
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 3.07,
|
| 766 |
+
"learning_rate": 1.1350949013770004e-06,
|
| 767 |
+
"loss": 0.1919,
|
| 768 |
+
"step": 610
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 3.09,
|
| 772 |
+
"learning_rate": 1.1443989579456644e-06,
|
| 773 |
+
"loss": 0.1484,
|
| 774 |
+
"step": 615
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 3.12,
|
| 778 |
+
"learning_rate": 1.1537030145143283e-06,
|
| 779 |
+
"loss": 0.178,
|
| 780 |
+
"step": 620
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 3.14,
|
| 784 |
+
"learning_rate": 1.1630070710829923e-06,
|
| 785 |
+
"loss": 0.1744,
|
| 786 |
+
"step": 625
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 3.17,
|
| 790 |
+
"learning_rate": 1.1723111276516561e-06,
|
| 791 |
+
"loss": 0.1512,
|
| 792 |
+
"step": 630
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 3.19,
|
| 796 |
+
"learning_rate": 1.1816151842203201e-06,
|
| 797 |
+
"loss": 0.142,
|
| 798 |
+
"step": 635
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 3.22,
|
| 802 |
+
"learning_rate": 1.1909192407889841e-06,
|
| 803 |
+
"loss": 0.1608,
|
| 804 |
+
"step": 640
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 3.24,
|
| 808 |
+
"learning_rate": 1.200223297357648e-06,
|
| 809 |
+
"loss": 0.1279,
|
| 810 |
+
"step": 645
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 3.27,
|
| 814 |
+
"learning_rate": 1.2095273539263119e-06,
|
| 815 |
+
"loss": 0.1536,
|
| 816 |
+
"step": 650
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 3.29,
|
| 820 |
+
"learning_rate": 1.2188314104949759e-06,
|
| 821 |
+
"loss": 0.1416,
|
| 822 |
+
"step": 655
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 3.32,
|
| 826 |
+
"learning_rate": 1.2281354670636399e-06,
|
| 827 |
+
"loss": 0.1938,
|
| 828 |
+
"step": 660
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 3.34,
|
| 832 |
+
"learning_rate": 1.2374395236323039e-06,
|
| 833 |
+
"loss": 0.0906,
|
| 834 |
+
"step": 665
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 3.37,
|
| 838 |
+
"learning_rate": 1.2467435802009676e-06,
|
| 839 |
+
"loss": 0.1756,
|
| 840 |
+
"step": 670
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 3.39,
|
| 844 |
+
"learning_rate": 1.2560476367696316e-06,
|
| 845 |
+
"loss": 0.1227,
|
| 846 |
+
"step": 675
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 3.42,
|
| 850 |
+
"learning_rate": 1.2653516933382956e-06,
|
| 851 |
+
"loss": 0.1396,
|
| 852 |
+
"step": 680
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 3.44,
|
| 856 |
+
"learning_rate": 1.2746557499069594e-06,
|
| 857 |
+
"loss": 0.1443,
|
| 858 |
+
"step": 685
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 3.47,
|
| 862 |
+
"learning_rate": 1.2839598064756234e-06,
|
| 863 |
+
"loss": 0.1194,
|
| 864 |
+
"step": 690
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 3.49,
|
| 868 |
+
"learning_rate": 1.2932638630442874e-06,
|
| 869 |
+
"loss": 0.1327,
|
| 870 |
+
"step": 695
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 3.52,
|
| 874 |
+
"learning_rate": 1.3025679196129514e-06,
|
| 875 |
+
"loss": 0.1057,
|
| 876 |
+
"step": 700
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 3.54,
|
| 880 |
+
"learning_rate": 1.3118719761816154e-06,
|
| 881 |
+
"loss": 0.2053,
|
| 882 |
+
"step": 705
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 3.57,
|
| 886 |
+
"learning_rate": 1.3211760327502794e-06,
|
| 887 |
+
"loss": 0.1075,
|
| 888 |
+
"step": 710
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 3.59,
|
| 892 |
+
"learning_rate": 1.3304800893189432e-06,
|
| 893 |
+
"loss": 0.1538,
|
| 894 |
+
"step": 715
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 3.62,
|
| 898 |
+
"learning_rate": 1.3397841458876071e-06,
|
| 899 |
+
"loss": 0.1185,
|
| 900 |
+
"step": 720
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 3.64,
|
| 904 |
+
"learning_rate": 1.349088202456271e-06,
|
| 905 |
+
"loss": 0.1038,
|
| 906 |
+
"step": 725
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 3.67,
|
| 910 |
+
"learning_rate": 1.358392259024935e-06,
|
| 911 |
+
"loss": 0.1652,
|
| 912 |
+
"step": 730
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 3.69,
|
| 916 |
+
"learning_rate": 1.3676963155935991e-06,
|
| 917 |
+
"loss": 0.1213,
|
| 918 |
+
"step": 735
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 3.72,
|
| 922 |
+
"learning_rate": 1.377000372162263e-06,
|
| 923 |
+
"loss": 0.1293,
|
| 924 |
+
"step": 740
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 3.74,
|
| 928 |
+
"learning_rate": 1.386304428730927e-06,
|
| 929 |
+
"loss": 0.1298,
|
| 930 |
+
"step": 745
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 3.77,
|
| 934 |
+
"learning_rate": 1.3956084852995907e-06,
|
| 935 |
+
"loss": 0.0757,
|
| 936 |
+
"step": 750
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 3.79,
|
| 940 |
+
"learning_rate": 1.4049125418682547e-06,
|
| 941 |
+
"loss": 0.1535,
|
| 942 |
+
"step": 755
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 3.82,
|
| 946 |
+
"learning_rate": 1.4142165984369185e-06,
|
| 947 |
+
"loss": 0.138,
|
| 948 |
+
"step": 760
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 3.84,
|
| 952 |
+
"learning_rate": 1.4235206550055824e-06,
|
| 953 |
+
"loss": 0.1866,
|
| 954 |
+
"step": 765
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 3.87,
|
| 958 |
+
"learning_rate": 1.4328247115742464e-06,
|
| 959 |
+
"loss": 0.1401,
|
| 960 |
+
"step": 770
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 3.89,
|
| 964 |
+
"learning_rate": 1.4421287681429104e-06,
|
| 965 |
+
"loss": 0.1712,
|
| 966 |
+
"step": 775
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 3.92,
|
| 970 |
+
"learning_rate": 1.4514328247115744e-06,
|
| 971 |
+
"loss": 0.1425,
|
| 972 |
+
"step": 780
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 3.94,
|
| 976 |
+
"learning_rate": 1.4607368812802384e-06,
|
| 977 |
+
"loss": 0.1277,
|
| 978 |
+
"step": 785
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 3.97,
|
| 982 |
+
"learning_rate": 1.4700409378489022e-06,
|
| 983 |
+
"loss": 0.1691,
|
| 984 |
+
"step": 790
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 3.99,
|
| 988 |
+
"learning_rate": 1.4793449944175662e-06,
|
| 989 |
+
"loss": 0.1009,
|
| 990 |
+
"step": 795
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 4.0,
|
| 994 |
+
"eval_accuracy": 0.9622997172478793,
|
| 995 |
+
"eval_loss": 0.09635111689567566,
|
| 996 |
+
"eval_runtime": 94.1975,
|
| 997 |
+
"eval_samples_per_second": 22.527,
|
| 998 |
+
"eval_steps_per_second": 1.412,
|
| 999 |
+
"step": 796
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 4.02,
|
| 1003 |
+
"learning_rate": 1.48864905098623e-06,
|
| 1004 |
+
"loss": 0.1423,
|
| 1005 |
+
"step": 800
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 4.05,
|
| 1009 |
+
"learning_rate": 1.497953107554894e-06,
|
| 1010 |
+
"loss": 0.1029,
|
| 1011 |
+
"step": 805
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 4.07,
|
| 1015 |
+
"learning_rate": 1.5072571641235582e-06,
|
| 1016 |
+
"loss": 0.1342,
|
| 1017 |
+
"step": 810
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 4.1,
|
| 1021 |
+
"learning_rate": 1.516561220692222e-06,
|
| 1022 |
+
"loss": 0.0727,
|
| 1023 |
+
"step": 815
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 4.12,
|
| 1027 |
+
"learning_rate": 1.525865277260886e-06,
|
| 1028 |
+
"loss": 0.0831,
|
| 1029 |
+
"step": 820
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 4.15,
|
| 1033 |
+
"learning_rate": 1.5351693338295497e-06,
|
| 1034 |
+
"loss": 0.1447,
|
| 1035 |
+
"step": 825
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 4.17,
|
| 1039 |
+
"learning_rate": 1.5444733903982137e-06,
|
| 1040 |
+
"loss": 0.0854,
|
| 1041 |
+
"step": 830
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 4.2,
|
| 1045 |
+
"learning_rate": 1.5537774469668775e-06,
|
| 1046 |
+
"loss": 0.1395,
|
| 1047 |
+
"step": 835
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 4.22,
|
| 1051 |
+
"learning_rate": 1.5630815035355415e-06,
|
| 1052 |
+
"loss": 0.1358,
|
| 1053 |
+
"step": 840
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 4.25,
|
| 1057 |
+
"learning_rate": 1.5723855601042057e-06,
|
| 1058 |
+
"loss": 0.1026,
|
| 1059 |
+
"step": 845
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 4.27,
|
| 1063 |
+
"learning_rate": 1.5816896166728695e-06,
|
| 1064 |
+
"loss": 0.1458,
|
| 1065 |
+
"step": 850
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 4.3,
|
| 1069 |
+
"learning_rate": 1.5909936732415335e-06,
|
| 1070 |
+
"loss": 0.1487,
|
| 1071 |
+
"step": 855
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 4.32,
|
| 1075 |
+
"learning_rate": 1.6002977298101975e-06,
|
| 1076 |
+
"loss": 0.1469,
|
| 1077 |
+
"step": 860
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 4.35,
|
| 1081 |
+
"learning_rate": 1.6096017863788612e-06,
|
| 1082 |
+
"loss": 0.0948,
|
| 1083 |
+
"step": 865
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 4.37,
|
| 1087 |
+
"learning_rate": 1.6189058429475252e-06,
|
| 1088 |
+
"loss": 0.0872,
|
| 1089 |
+
"step": 870
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 4.4,
|
| 1093 |
+
"learning_rate": 1.628209899516189e-06,
|
| 1094 |
+
"loss": 0.1577,
|
| 1095 |
+
"step": 875
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 4.42,
|
| 1099 |
+
"learning_rate": 1.6375139560848532e-06,
|
| 1100 |
+
"loss": 0.1525,
|
| 1101 |
+
"step": 880
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 4.45,
|
| 1105 |
+
"learning_rate": 1.6468180126535172e-06,
|
| 1106 |
+
"loss": 0.1445,
|
| 1107 |
+
"step": 885
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 4.47,
|
| 1111 |
+
"learning_rate": 1.656122069222181e-06,
|
| 1112 |
+
"loss": 0.1282,
|
| 1113 |
+
"step": 890
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 4.5,
|
| 1117 |
+
"learning_rate": 1.665426125790845e-06,
|
| 1118 |
+
"loss": 0.063,
|
| 1119 |
+
"step": 895
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 4.52,
|
| 1123 |
+
"learning_rate": 1.6747301823595088e-06,
|
| 1124 |
+
"loss": 0.1581,
|
| 1125 |
+
"step": 900
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 4.55,
|
| 1129 |
+
"learning_rate": 1.6840342389281728e-06,
|
| 1130 |
+
"loss": 0.0998,
|
| 1131 |
+
"step": 905
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 4.57,
|
| 1135 |
+
"learning_rate": 1.6933382954968365e-06,
|
| 1136 |
+
"loss": 0.0715,
|
| 1137 |
+
"step": 910
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 4.6,
|
| 1141 |
+
"learning_rate": 1.7026423520655008e-06,
|
| 1142 |
+
"loss": 0.0756,
|
| 1143 |
+
"step": 915
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 4.62,
|
| 1147 |
+
"learning_rate": 1.7119464086341647e-06,
|
| 1148 |
+
"loss": 0.0994,
|
| 1149 |
+
"step": 920
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 4.65,
|
| 1153 |
+
"learning_rate": 1.7212504652028285e-06,
|
| 1154 |
+
"loss": 0.1515,
|
| 1155 |
+
"step": 925
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 4.67,
|
| 1159 |
+
"learning_rate": 1.7305545217714925e-06,
|
| 1160 |
+
"loss": 0.1057,
|
| 1161 |
+
"step": 930
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 4.7,
|
| 1165 |
+
"learning_rate": 1.7398585783401565e-06,
|
| 1166 |
+
"loss": 0.0981,
|
| 1167 |
+
"step": 935
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 4.72,
|
| 1171 |
+
"learning_rate": 1.7491626349088203e-06,
|
| 1172 |
+
"loss": 0.1084,
|
| 1173 |
+
"step": 940
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 4.75,
|
| 1177 |
+
"learning_rate": 1.7584666914774843e-06,
|
| 1178 |
+
"loss": 0.1648,
|
| 1179 |
+
"step": 945
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 4.77,
|
| 1183 |
+
"learning_rate": 1.7677707480461485e-06,
|
| 1184 |
+
"loss": 0.0703,
|
| 1185 |
+
"step": 950
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 4.8,
|
| 1189 |
+
"learning_rate": 1.7770748046148123e-06,
|
| 1190 |
+
"loss": 0.117,
|
| 1191 |
+
"step": 955
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 4.82,
|
| 1195 |
+
"learning_rate": 1.7863788611834763e-06,
|
| 1196 |
+
"loss": 0.0646,
|
| 1197 |
+
"step": 960
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 4.85,
|
| 1201 |
+
"learning_rate": 1.79568291775214e-06,
|
| 1202 |
+
"loss": 0.1411,
|
| 1203 |
+
"step": 965
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 4.87,
|
| 1207 |
+
"learning_rate": 1.804986974320804e-06,
|
| 1208 |
+
"loss": 0.1514,
|
| 1209 |
+
"step": 970
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 4.9,
|
| 1213 |
+
"learning_rate": 1.8142910308894678e-06,
|
| 1214 |
+
"loss": 0.1054,
|
| 1215 |
+
"step": 975
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 4.92,
|
| 1219 |
+
"learning_rate": 1.8235950874581318e-06,
|
| 1220 |
+
"loss": 0.0774,
|
| 1221 |
+
"step": 980
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 4.95,
|
| 1225 |
+
"learning_rate": 1.8328991440267956e-06,
|
| 1226 |
+
"loss": 0.0596,
|
| 1227 |
+
"step": 985
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 4.97,
|
| 1231 |
+
"learning_rate": 1.8422032005954598e-06,
|
| 1232 |
+
"loss": 0.1058,
|
| 1233 |
+
"step": 990
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 5.0,
|
| 1237 |
+
"learning_rate": 1.8515072571641238e-06,
|
| 1238 |
+
"loss": 0.1528,
|
| 1239 |
+
"step": 995
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 5.0,
|
| 1243 |
+
"eval_accuracy": 0.9717247879359096,
|
| 1244 |
+
"eval_loss": 0.06622624397277832,
|
| 1245 |
+
"eval_runtime": 93.9382,
|
| 1246 |
+
"eval_samples_per_second": 22.589,
|
| 1247 |
+
"eval_steps_per_second": 1.416,
|
| 1248 |
+
"step": 995
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 5.03,
|
| 1252 |
+
"learning_rate": 1.8608113137327876e-06,
|
| 1253 |
+
"loss": 0.1109,
|
| 1254 |
+
"step": 1000
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 5.05,
|
| 1258 |
+
"learning_rate": 1.8701153703014516e-06,
|
| 1259 |
+
"loss": 0.0903,
|
| 1260 |
+
"step": 1005
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"epoch": 5.08,
|
| 1264 |
+
"learning_rate": 1.8794194268701156e-06,
|
| 1265 |
+
"loss": 0.0937,
|
| 1266 |
+
"step": 1010
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 5.1,
|
| 1270 |
+
"learning_rate": 1.8887234834387793e-06,
|
| 1271 |
+
"loss": 0.0894,
|
| 1272 |
+
"step": 1015
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 5.13,
|
| 1276 |
+
"learning_rate": 1.8980275400074433e-06,
|
| 1277 |
+
"loss": 0.1434,
|
| 1278 |
+
"step": 1020
|
| 1279 |
+
},
|
| 1280 |
+
{
|
| 1281 |
+
"epoch": 5.15,
|
| 1282 |
+
"learning_rate": 1.9073315965761075e-06,
|
| 1283 |
+
"loss": 0.0937,
|
| 1284 |
+
"step": 1025
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 5.18,
|
| 1288 |
+
"learning_rate": 1.9166356531447715e-06,
|
| 1289 |
+
"loss": 0.1089,
|
| 1290 |
+
"step": 1030
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 5.2,
|
| 1294 |
+
"learning_rate": 1.925939709713435e-06,
|
| 1295 |
+
"loss": 0.0807,
|
| 1296 |
+
"step": 1035
|
| 1297 |
+
},
|
| 1298 |
+
{
|
| 1299 |
+
"epoch": 5.23,
|
| 1300 |
+
"learning_rate": 1.935243766282099e-06,
|
| 1301 |
+
"loss": 0.1392,
|
| 1302 |
+
"step": 1040
|
| 1303 |
+
},
|
| 1304 |
+
{
|
| 1305 |
+
"epoch": 5.25,
|
| 1306 |
+
"learning_rate": 1.944547822850763e-06,
|
| 1307 |
+
"loss": 0.1147,
|
| 1308 |
+
"step": 1045
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 5.28,
|
| 1312 |
+
"learning_rate": 1.953851879419427e-06,
|
| 1313 |
+
"loss": 0.0684,
|
| 1314 |
+
"step": 1050
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 5.3,
|
| 1318 |
+
"learning_rate": 1.9631559359880906e-06,
|
| 1319 |
+
"loss": 0.1166,
|
| 1320 |
+
"step": 1055
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 5.33,
|
| 1324 |
+
"learning_rate": 1.972459992556755e-06,
|
| 1325 |
+
"loss": 0.0921,
|
| 1326 |
+
"step": 1060
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 5.35,
|
| 1330 |
+
"learning_rate": 1.981764049125419e-06,
|
| 1331 |
+
"loss": 0.0925,
|
| 1332 |
+
"step": 1065
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 5.38,
|
| 1336 |
+
"learning_rate": 1.9910681056940826e-06,
|
| 1337 |
+
"loss": 0.0839,
|
| 1338 |
+
"step": 1070
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 5.4,
|
| 1342 |
+
"learning_rate": 2.0003721622627466e-06,
|
| 1343 |
+
"loss": 0.1073,
|
| 1344 |
+
"step": 1075
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 5.43,
|
| 1348 |
+
"learning_rate": 2.0096762188314106e-06,
|
| 1349 |
+
"loss": 0.0688,
|
| 1350 |
+
"step": 1080
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 5.45,
|
| 1354 |
+
"learning_rate": 2.0189802754000746e-06,
|
| 1355 |
+
"loss": 0.0752,
|
| 1356 |
+
"step": 1085
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 5.48,
|
| 1360 |
+
"learning_rate": 2.0282843319687386e-06,
|
| 1361 |
+
"loss": 0.068,
|
| 1362 |
+
"step": 1090
|
| 1363 |
+
},
|
| 1364 |
+
{
|
| 1365 |
+
"epoch": 5.5,
|
| 1366 |
+
"learning_rate": 2.0375883885374026e-06,
|
| 1367 |
+
"loss": 0.1445,
|
| 1368 |
+
"step": 1095
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 5.53,
|
| 1372 |
+
"learning_rate": 2.0468924451060666e-06,
|
| 1373 |
+
"loss": 0.1196,
|
| 1374 |
+
"step": 1100
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 5.55,
|
| 1378 |
+
"learning_rate": 2.0561965016747306e-06,
|
| 1379 |
+
"loss": 0.081,
|
| 1380 |
+
"step": 1105
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 5.58,
|
| 1384 |
+
"learning_rate": 2.065500558243394e-06,
|
| 1385 |
+
"loss": 0.072,
|
| 1386 |
+
"step": 1110
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 5.6,
|
| 1390 |
+
"learning_rate": 2.074804614812058e-06,
|
| 1391 |
+
"loss": 0.1298,
|
| 1392 |
+
"step": 1115
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"epoch": 5.63,
|
| 1396 |
+
"learning_rate": 2.084108671380722e-06,
|
| 1397 |
+
"loss": 0.067,
|
| 1398 |
+
"step": 1120
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"epoch": 5.65,
|
| 1402 |
+
"learning_rate": 2.093412727949386e-06,
|
| 1403 |
+
"loss": 0.1199,
|
| 1404 |
+
"step": 1125
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 5.68,
|
| 1408 |
+
"learning_rate": 2.10271678451805e-06,
|
| 1409 |
+
"loss": 0.0914,
|
| 1410 |
+
"step": 1130
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 5.7,
|
| 1414 |
+
"learning_rate": 2.112020841086714e-06,
|
| 1415 |
+
"loss": 0.0641,
|
| 1416 |
+
"step": 1135
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 5.73,
|
| 1420 |
+
"learning_rate": 2.121324897655378e-06,
|
| 1421 |
+
"loss": 0.1018,
|
| 1422 |
+
"step": 1140
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 5.75,
|
| 1426 |
+
"learning_rate": 2.1306289542240417e-06,
|
| 1427 |
+
"loss": 0.0519,
|
| 1428 |
+
"step": 1145
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 5.78,
|
| 1432 |
+
"learning_rate": 2.1399330107927057e-06,
|
| 1433 |
+
"loss": 0.1556,
|
| 1434 |
+
"step": 1150
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 5.8,
|
| 1438 |
+
"learning_rate": 2.1492370673613697e-06,
|
| 1439 |
+
"loss": 0.1224,
|
| 1440 |
+
"step": 1155
|
| 1441 |
+
},
|
| 1442 |
+
{
|
| 1443 |
+
"epoch": 5.83,
|
| 1444 |
+
"learning_rate": 2.1585411239300337e-06,
|
| 1445 |
+
"loss": 0.1218,
|
| 1446 |
+
"step": 1160
|
| 1447 |
+
},
|
| 1448 |
+
{
|
| 1449 |
+
"epoch": 5.85,
|
| 1450 |
+
"learning_rate": 2.1678451804986976e-06,
|
| 1451 |
+
"loss": 0.0527,
|
| 1452 |
+
"step": 1165
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 5.88,
|
| 1456 |
+
"learning_rate": 2.1771492370673616e-06,
|
| 1457 |
+
"loss": 0.0636,
|
| 1458 |
+
"step": 1170
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 5.9,
|
| 1462 |
+
"learning_rate": 2.1864532936360256e-06,
|
| 1463 |
+
"loss": 0.0822,
|
| 1464 |
+
"step": 1175
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 5.93,
|
| 1468 |
+
"learning_rate": 2.1957573502046896e-06,
|
| 1469 |
+
"loss": 0.0875,
|
| 1470 |
+
"step": 1180
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"epoch": 5.95,
|
| 1474 |
+
"learning_rate": 2.205061406773353e-06,
|
| 1475 |
+
"loss": 0.1289,
|
| 1476 |
+
"step": 1185
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 5.98,
|
| 1480 |
+
"learning_rate": 2.214365463342017e-06,
|
| 1481 |
+
"loss": 0.0893,
|
| 1482 |
+
"step": 1190
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 6.0,
|
| 1486 |
+
"eval_accuracy": 0.9773798303487277,
|
| 1487 |
+
"eval_loss": 0.05487333610653877,
|
| 1488 |
+
"eval_runtime": 93.9127,
|
| 1489 |
+
"eval_samples_per_second": 22.595,
|
| 1490 |
+
"eval_steps_per_second": 1.416,
|
| 1491 |
+
"step": 1194
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"epoch": 6.01,
|
| 1495 |
+
"learning_rate": 2.223669519910681e-06,
|
| 1496 |
+
"loss": 0.046,
|
| 1497 |
+
"step": 1195
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 6.03,
|
| 1501 |
+
"learning_rate": 2.232973576479345e-06,
|
| 1502 |
+
"loss": 0.1525,
|
| 1503 |
+
"step": 1200
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 6.06,
|
| 1507 |
+
"learning_rate": 2.242277633048009e-06,
|
| 1508 |
+
"loss": 0.0619,
|
| 1509 |
+
"step": 1205
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 6.08,
|
| 1513 |
+
"learning_rate": 2.251581689616673e-06,
|
| 1514 |
+
"loss": 0.0898,
|
| 1515 |
+
"step": 1210
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 6.11,
|
| 1519 |
+
"learning_rate": 2.260885746185337e-06,
|
| 1520 |
+
"loss": 0.0434,
|
| 1521 |
+
"step": 1215
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 6.13,
|
| 1525 |
+
"learning_rate": 2.2701898027540007e-06,
|
| 1526 |
+
"loss": 0.0585,
|
| 1527 |
+
"step": 1220
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 6.16,
|
| 1531 |
+
"learning_rate": 2.2794938593226647e-06,
|
| 1532 |
+
"loss": 0.0391,
|
| 1533 |
+
"step": 1225
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 6.18,
|
| 1537 |
+
"learning_rate": 2.2887979158913287e-06,
|
| 1538 |
+
"loss": 0.0917,
|
| 1539 |
+
"step": 1230
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 6.21,
|
| 1543 |
+
"learning_rate": 2.2981019724599927e-06,
|
| 1544 |
+
"loss": 0.0539,
|
| 1545 |
+
"step": 1235
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 6.23,
|
| 1549 |
+
"learning_rate": 2.3074060290286567e-06,
|
| 1550 |
+
"loss": 0.0874,
|
| 1551 |
+
"step": 1240
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 6.26,
|
| 1555 |
+
"learning_rate": 2.3167100855973207e-06,
|
| 1556 |
+
"loss": 0.1121,
|
| 1557 |
+
"step": 1245
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 6.28,
|
| 1561 |
+
"learning_rate": 2.3260141421659847e-06,
|
| 1562 |
+
"loss": 0.0649,
|
| 1563 |
+
"step": 1250
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 6.31,
|
| 1567 |
+
"learning_rate": 2.3353181987346487e-06,
|
| 1568 |
+
"loss": 0.0662,
|
| 1569 |
+
"step": 1255
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 6.33,
|
| 1573 |
+
"learning_rate": 2.3446222553033122e-06,
|
| 1574 |
+
"loss": 0.084,
|
| 1575 |
+
"step": 1260
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 6.36,
|
| 1579 |
+
"learning_rate": 2.3539263118719762e-06,
|
| 1580 |
+
"loss": 0.0776,
|
| 1581 |
+
"step": 1265
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 6.38,
|
| 1585 |
+
"learning_rate": 2.3632303684406402e-06,
|
| 1586 |
+
"loss": 0.0488,
|
| 1587 |
+
"step": 1270
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 6.41,
|
| 1591 |
+
"learning_rate": 2.3725344250093042e-06,
|
| 1592 |
+
"loss": 0.1111,
|
| 1593 |
+
"step": 1275
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 6.43,
|
| 1597 |
+
"learning_rate": 2.3818384815779682e-06,
|
| 1598 |
+
"loss": 0.0717,
|
| 1599 |
+
"step": 1280
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 6.46,
|
| 1603 |
+
"learning_rate": 2.391142538146632e-06,
|
| 1604 |
+
"loss": 0.0593,
|
| 1605 |
+
"step": 1285
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 6.48,
|
| 1609 |
+
"learning_rate": 2.400446594715296e-06,
|
| 1610 |
+
"loss": 0.0893,
|
| 1611 |
+
"step": 1290
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 6.51,
|
| 1615 |
+
"learning_rate": 2.4097506512839598e-06,
|
| 1616 |
+
"loss": 0.0503,
|
| 1617 |
+
"step": 1295
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 6.53,
|
| 1621 |
+
"learning_rate": 2.4190547078526238e-06,
|
| 1622 |
+
"loss": 0.0465,
|
| 1623 |
+
"step": 1300
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 6.56,
|
| 1627 |
+
"learning_rate": 2.4283587644212878e-06,
|
| 1628 |
+
"loss": 0.103,
|
| 1629 |
+
"step": 1305
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 6.58,
|
| 1633 |
+
"learning_rate": 2.4376628209899517e-06,
|
| 1634 |
+
"loss": 0.0893,
|
| 1635 |
+
"step": 1310
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 6.61,
|
| 1639 |
+
"learning_rate": 2.4469668775586157e-06,
|
| 1640 |
+
"loss": 0.1053,
|
| 1641 |
+
"step": 1315
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 6.63,
|
| 1645 |
+
"learning_rate": 2.4562709341272797e-06,
|
| 1646 |
+
"loss": 0.024,
|
| 1647 |
+
"step": 1320
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 6.66,
|
| 1651 |
+
"learning_rate": 2.4655749906959437e-06,
|
| 1652 |
+
"loss": 0.0774,
|
| 1653 |
+
"step": 1325
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 6.68,
|
| 1657 |
+
"learning_rate": 2.4748790472646077e-06,
|
| 1658 |
+
"loss": 0.1098,
|
| 1659 |
+
"step": 1330
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 6.71,
|
| 1663 |
+
"learning_rate": 2.4841831038332713e-06,
|
| 1664 |
+
"loss": 0.1233,
|
| 1665 |
+
"step": 1335
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 6.73,
|
| 1669 |
+
"learning_rate": 2.4934871604019353e-06,
|
| 1670 |
+
"loss": 0.0623,
|
| 1671 |
+
"step": 1340
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 6.76,
|
| 1675 |
+
"learning_rate": 2.5027912169705993e-06,
|
| 1676 |
+
"loss": 0.0948,
|
| 1677 |
+
"step": 1345
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 6.78,
|
| 1681 |
+
"learning_rate": 2.5120952735392633e-06,
|
| 1682 |
+
"loss": 0.0662,
|
| 1683 |
+
"step": 1350
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 6.81,
|
| 1687 |
+
"learning_rate": 2.521399330107927e-06,
|
| 1688 |
+
"loss": 0.0629,
|
| 1689 |
+
"step": 1355
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 6.83,
|
| 1693 |
+
"learning_rate": 2.5307033866765913e-06,
|
| 1694 |
+
"loss": 0.0507,
|
| 1695 |
+
"step": 1360
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 6.86,
|
| 1699 |
+
"learning_rate": 2.5400074432452552e-06,
|
| 1700 |
+
"loss": 0.0901,
|
| 1701 |
+
"step": 1365
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 6.88,
|
| 1705 |
+
"learning_rate": 2.549311499813919e-06,
|
| 1706 |
+
"loss": 0.0797,
|
| 1707 |
+
"step": 1370
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 6.91,
|
| 1711 |
+
"learning_rate": 2.5586155563825832e-06,
|
| 1712 |
+
"loss": 0.0698,
|
| 1713 |
+
"step": 1375
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 6.93,
|
| 1717 |
+
"learning_rate": 2.567919612951247e-06,
|
| 1718 |
+
"loss": 0.0835,
|
| 1719 |
+
"step": 1380
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 6.96,
|
| 1723 |
+
"learning_rate": 2.577223669519911e-06,
|
| 1724 |
+
"loss": 0.0561,
|
| 1725 |
+
"step": 1385
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 6.98,
|
| 1729 |
+
"learning_rate": 2.5865277260885748e-06,
|
| 1730 |
+
"loss": 0.0786,
|
| 1731 |
+
"step": 1390
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 7.0,
|
| 1735 |
+
"eval_accuracy": 0.9802073515551367,
|
| 1736 |
+
"eval_loss": 0.05052812770009041,
|
| 1737 |
+
"eval_runtime": 93.9877,
|
| 1738 |
+
"eval_samples_per_second": 22.577,
|
| 1739 |
+
"eval_steps_per_second": 1.415,
|
| 1740 |
+
"step": 1393
|
| 1741 |
+
},
|
| 1742 |
+
{
|
| 1743 |
+
"epoch": 7.01,
|
| 1744 |
+
"learning_rate": 2.5958317826572388e-06,
|
| 1745 |
+
"loss": 0.0874,
|
| 1746 |
+
"step": 1395
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 7.04,
|
| 1750 |
+
"learning_rate": 2.6051358392259028e-06,
|
| 1751 |
+
"loss": 0.0329,
|
| 1752 |
+
"step": 1400
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 7.06,
|
| 1756 |
+
"learning_rate": 2.6144398957945668e-06,
|
| 1757 |
+
"loss": 0.0419,
|
| 1758 |
+
"step": 1405
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 7.09,
|
| 1762 |
+
"learning_rate": 2.6237439523632308e-06,
|
| 1763 |
+
"loss": 0.0339,
|
| 1764 |
+
"step": 1410
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 7.11,
|
| 1768 |
+
"learning_rate": 2.6330480089318943e-06,
|
| 1769 |
+
"loss": 0.1545,
|
| 1770 |
+
"step": 1415
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 7.14,
|
| 1774 |
+
"learning_rate": 2.6423520655005587e-06,
|
| 1775 |
+
"loss": 0.0706,
|
| 1776 |
+
"step": 1420
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 7.16,
|
| 1780 |
+
"learning_rate": 2.6516561220692223e-06,
|
| 1781 |
+
"loss": 0.0726,
|
| 1782 |
+
"step": 1425
|
| 1783 |
+
},
|
| 1784 |
+
{
|
| 1785 |
+
"epoch": 7.19,
|
| 1786 |
+
"learning_rate": 2.6609601786378863e-06,
|
| 1787 |
+
"loss": 0.051,
|
| 1788 |
+
"step": 1430
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 7.21,
|
| 1792 |
+
"learning_rate": 2.6702642352065507e-06,
|
| 1793 |
+
"loss": 0.0587,
|
| 1794 |
+
"step": 1435
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 7.24,
|
| 1798 |
+
"learning_rate": 2.6795682917752143e-06,
|
| 1799 |
+
"loss": 0.0709,
|
| 1800 |
+
"step": 1440
|
| 1801 |
+
},
|
| 1802 |
+
{
|
| 1803 |
+
"epoch": 7.26,
|
| 1804 |
+
"learning_rate": 2.6888723483438783e-06,
|
| 1805 |
+
"loss": 0.0829,
|
| 1806 |
+
"step": 1445
|
| 1807 |
+
},
|
| 1808 |
+
{
|
| 1809 |
+
"epoch": 7.29,
|
| 1810 |
+
"learning_rate": 2.698176404912542e-06,
|
| 1811 |
+
"loss": 0.1267,
|
| 1812 |
+
"step": 1450
|
| 1813 |
+
},
|
| 1814 |
+
{
|
| 1815 |
+
"epoch": 7.31,
|
| 1816 |
+
"learning_rate": 2.7074804614812063e-06,
|
| 1817 |
+
"loss": 0.0488,
|
| 1818 |
+
"step": 1455
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 7.34,
|
| 1822 |
+
"learning_rate": 2.71678451804987e-06,
|
| 1823 |
+
"loss": 0.0248,
|
| 1824 |
+
"step": 1460
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 7.36,
|
| 1828 |
+
"learning_rate": 2.726088574618534e-06,
|
| 1829 |
+
"loss": 0.0757,
|
| 1830 |
+
"step": 1465
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 7.39,
|
| 1834 |
+
"learning_rate": 2.7353926311871983e-06,
|
| 1835 |
+
"loss": 0.0402,
|
| 1836 |
+
"step": 1470
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 7.41,
|
| 1840 |
+
"learning_rate": 2.744696687755862e-06,
|
| 1841 |
+
"loss": 0.0737,
|
| 1842 |
+
"step": 1475
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 7.44,
|
| 1846 |
+
"learning_rate": 2.754000744324526e-06,
|
| 1847 |
+
"loss": 0.0098,
|
| 1848 |
+
"step": 1480
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 7.46,
|
| 1852 |
+
"learning_rate": 2.7633048008931894e-06,
|
| 1853 |
+
"loss": 0.0897,
|
| 1854 |
+
"step": 1485
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 7.49,
|
| 1858 |
+
"learning_rate": 2.772608857461854e-06,
|
| 1859 |
+
"loss": 0.0949,
|
| 1860 |
+
"step": 1490
|
| 1861 |
+
},
|
| 1862 |
+
{
|
| 1863 |
+
"epoch": 7.51,
|
| 1864 |
+
"learning_rate": 2.7819129140305174e-06,
|
| 1865 |
+
"loss": 0.0791,
|
| 1866 |
+
"step": 1495
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 7.54,
|
| 1870 |
+
"learning_rate": 2.7912169705991814e-06,
|
| 1871 |
+
"loss": 0.0675,
|
| 1872 |
+
"step": 1500
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 7.56,
|
| 1876 |
+
"learning_rate": 2.8005210271678458e-06,
|
| 1877 |
+
"loss": 0.0999,
|
| 1878 |
+
"step": 1505
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 7.59,
|
| 1882 |
+
"learning_rate": 2.8098250837365093e-06,
|
| 1883 |
+
"loss": 0.0569,
|
| 1884 |
+
"step": 1510
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 7.61,
|
| 1888 |
+
"learning_rate": 2.8191291403051733e-06,
|
| 1889 |
+
"loss": 0.0698,
|
| 1890 |
+
"step": 1515
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 7.64,
|
| 1894 |
+
"learning_rate": 2.828433196873837e-06,
|
| 1895 |
+
"loss": 0.0208,
|
| 1896 |
+
"step": 1520
|
| 1897 |
+
},
|
| 1898 |
+
{
|
| 1899 |
+
"epoch": 7.66,
|
| 1900 |
+
"learning_rate": 2.8377372534425013e-06,
|
| 1901 |
+
"loss": 0.1189,
|
| 1902 |
+
"step": 1525
|
| 1903 |
+
},
|
| 1904 |
+
{
|
| 1905 |
+
"epoch": 7.69,
|
| 1906 |
+
"learning_rate": 2.847041310011165e-06,
|
| 1907 |
+
"loss": 0.066,
|
| 1908 |
+
"step": 1530
|
| 1909 |
+
},
|
| 1910 |
+
{
|
| 1911 |
+
"epoch": 7.71,
|
| 1912 |
+
"learning_rate": 2.856345366579829e-06,
|
| 1913 |
+
"loss": 0.0205,
|
| 1914 |
+
"step": 1535
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 7.74,
|
| 1918 |
+
"learning_rate": 2.865649423148493e-06,
|
| 1919 |
+
"loss": 0.0351,
|
| 1920 |
+
"step": 1540
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 7.76,
|
| 1924 |
+
"learning_rate": 2.874953479717157e-06,
|
| 1925 |
+
"loss": 0.0607,
|
| 1926 |
+
"step": 1545
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 7.79,
|
| 1930 |
+
"learning_rate": 2.884257536285821e-06,
|
| 1931 |
+
"loss": 0.0528,
|
| 1932 |
+
"step": 1550
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 7.81,
|
| 1936 |
+
"learning_rate": 2.893561592854485e-06,
|
| 1937 |
+
"loss": 0.0348,
|
| 1938 |
+
"step": 1555
|
| 1939 |
+
},
|
| 1940 |
+
{
|
| 1941 |
+
"epoch": 7.84,
|
| 1942 |
+
"learning_rate": 2.902865649423149e-06,
|
| 1943 |
+
"loss": 0.0365,
|
| 1944 |
+
"step": 1560
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 7.86,
|
| 1948 |
+
"learning_rate": 2.9121697059918124e-06,
|
| 1949 |
+
"loss": 0.0526,
|
| 1950 |
+
"step": 1565
|
| 1951 |
+
},
|
| 1952 |
+
{
|
| 1953 |
+
"epoch": 7.89,
|
| 1954 |
+
"learning_rate": 2.921473762560477e-06,
|
| 1955 |
+
"loss": 0.0519,
|
| 1956 |
+
"step": 1570
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 7.91,
|
| 1960 |
+
"learning_rate": 2.9307778191291404e-06,
|
| 1961 |
+
"loss": 0.1605,
|
| 1962 |
+
"step": 1575
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 7.94,
|
| 1966 |
+
"learning_rate": 2.9400818756978044e-06,
|
| 1967 |
+
"loss": 0.0344,
|
| 1968 |
+
"step": 1580
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 7.96,
|
| 1972 |
+
"learning_rate": 2.949385932266469e-06,
|
| 1973 |
+
"loss": 0.1139,
|
| 1974 |
+
"step": 1585
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 7.99,
|
| 1978 |
+
"learning_rate": 2.9586899888351324e-06,
|
| 1979 |
+
"loss": 0.0254,
|
| 1980 |
+
"step": 1590
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 8.0,
|
| 1984 |
+
"eval_accuracy": 0.9783223374175306,
|
| 1985 |
+
"eval_loss": 0.06433209031820297,
|
| 1986 |
+
"eval_runtime": 94.0428,
|
| 1987 |
+
"eval_samples_per_second": 22.564,
|
| 1988 |
+
"eval_steps_per_second": 1.414,
|
| 1989 |
+
"step": 1592
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 8.02,
|
| 1993 |
+
"learning_rate": 2.9679940454037964e-06,
|
| 1994 |
+
"loss": 0.0949,
|
| 1995 |
+
"step": 1595
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 8.04,
|
| 1999 |
+
"learning_rate": 2.97729810197246e-06,
|
| 2000 |
+
"loss": 0.0284,
|
| 2001 |
+
"step": 1600
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 8.07,
|
| 2005 |
+
"learning_rate": 2.9866021585411244e-06,
|
| 2006 |
+
"loss": 0.1013,
|
| 2007 |
+
"step": 1605
|
| 2008 |
+
},
|
| 2009 |
+
{
|
| 2010 |
+
"epoch": 8.09,
|
| 2011 |
+
"learning_rate": 2.995906215109788e-06,
|
| 2012 |
+
"loss": 0.0362,
|
| 2013 |
+
"step": 1610
|
| 2014 |
+
},
|
| 2015 |
+
{
|
| 2016 |
+
"epoch": 8.12,
|
| 2017 |
+
"learning_rate": 3.005210271678452e-06,
|
| 2018 |
+
"loss": 0.0551,
|
| 2019 |
+
"step": 1615
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 8.14,
|
| 2023 |
+
"learning_rate": 3.0145143282471163e-06,
|
| 2024 |
+
"loss": 0.0318,
|
| 2025 |
+
"step": 1620
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 8.17,
|
| 2029 |
+
"learning_rate": 3.02381838481578e-06,
|
| 2030 |
+
"loss": 0.0117,
|
| 2031 |
+
"step": 1625
|
| 2032 |
+
},
|
| 2033 |
+
{
|
| 2034 |
+
"epoch": 8.19,
|
| 2035 |
+
"learning_rate": 3.033122441384444e-06,
|
| 2036 |
+
"loss": 0.1075,
|
| 2037 |
+
"step": 1630
|
| 2038 |
+
},
|
| 2039 |
+
{
|
| 2040 |
+
"epoch": 8.22,
|
| 2041 |
+
"learning_rate": 3.0424264979531075e-06,
|
| 2042 |
+
"loss": 0.0389,
|
| 2043 |
+
"step": 1635
|
| 2044 |
+
},
|
| 2045 |
+
{
|
| 2046 |
+
"epoch": 8.24,
|
| 2047 |
+
"learning_rate": 3.051730554521772e-06,
|
| 2048 |
+
"loss": 0.108,
|
| 2049 |
+
"step": 1640
|
| 2050 |
+
},
|
| 2051 |
+
{
|
| 2052 |
+
"epoch": 8.27,
|
| 2053 |
+
"learning_rate": 3.0610346110904355e-06,
|
| 2054 |
+
"loss": 0.083,
|
| 2055 |
+
"step": 1645
|
| 2056 |
+
},
|
| 2057 |
+
{
|
| 2058 |
+
"epoch": 8.29,
|
| 2059 |
+
"learning_rate": 3.0703386676590995e-06,
|
| 2060 |
+
"loss": 0.0526,
|
| 2061 |
+
"step": 1650
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 8.32,
|
| 2065 |
+
"learning_rate": 3.079642724227764e-06,
|
| 2066 |
+
"loss": 0.0895,
|
| 2067 |
+
"step": 1655
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 8.34,
|
| 2071 |
+
"learning_rate": 3.0889467807964274e-06,
|
| 2072 |
+
"loss": 0.0362,
|
| 2073 |
+
"step": 1660
|
| 2074 |
+
},
|
| 2075 |
+
{
|
| 2076 |
+
"epoch": 8.37,
|
| 2077 |
+
"learning_rate": 3.0982508373650914e-06,
|
| 2078 |
+
"loss": 0.0092,
|
| 2079 |
+
"step": 1665
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 8.39,
|
| 2083 |
+
"learning_rate": 3.107554893933755e-06,
|
| 2084 |
+
"loss": 0.1012,
|
| 2085 |
+
"step": 1670
|
| 2086 |
+
},
|
| 2087 |
+
{
|
| 2088 |
+
"epoch": 8.42,
|
| 2089 |
+
"learning_rate": 3.1168589505024194e-06,
|
| 2090 |
+
"loss": 0.0868,
|
| 2091 |
+
"step": 1675
|
| 2092 |
+
},
|
| 2093 |
+
{
|
| 2094 |
+
"epoch": 8.44,
|
| 2095 |
+
"learning_rate": 3.126163007071083e-06,
|
| 2096 |
+
"loss": 0.0947,
|
| 2097 |
+
"step": 1680
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 8.47,
|
| 2101 |
+
"learning_rate": 3.135467063639747e-06,
|
| 2102 |
+
"loss": 0.0943,
|
| 2103 |
+
"step": 1685
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 8.49,
|
| 2107 |
+
"learning_rate": 3.1447711202084114e-06,
|
| 2108 |
+
"loss": 0.0277,
|
| 2109 |
+
"step": 1690
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 8.52,
|
| 2113 |
+
"learning_rate": 3.154075176777075e-06,
|
| 2114 |
+
"loss": 0.0293,
|
| 2115 |
+
"step": 1695
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 8.54,
|
| 2119 |
+
"learning_rate": 3.163379233345739e-06,
|
| 2120 |
+
"loss": 0.0897,
|
| 2121 |
+
"step": 1700
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 8.57,
|
| 2125 |
+
"learning_rate": 3.172683289914403e-06,
|
| 2126 |
+
"loss": 0.0718,
|
| 2127 |
+
"step": 1705
|
| 2128 |
+
},
|
| 2129 |
+
{
|
| 2130 |
+
"epoch": 8.59,
|
| 2131 |
+
"learning_rate": 3.181987346483067e-06,
|
| 2132 |
+
"loss": 0.0565,
|
| 2133 |
+
"step": 1710
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 8.62,
|
| 2137 |
+
"learning_rate": 3.1912914030517305e-06,
|
| 2138 |
+
"loss": 0.0482,
|
| 2139 |
+
"step": 1715
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 8.64,
|
| 2143 |
+
"learning_rate": 3.200595459620395e-06,
|
| 2144 |
+
"loss": 0.0177,
|
| 2145 |
+
"step": 1720
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 8.67,
|
| 2149 |
+
"learning_rate": 3.209899516189059e-06,
|
| 2150 |
+
"loss": 0.0588,
|
| 2151 |
+
"step": 1725
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 8.69,
|
| 2155 |
+
"learning_rate": 3.2192035727577225e-06,
|
| 2156 |
+
"loss": 0.1045,
|
| 2157 |
+
"step": 1730
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 8.72,
|
| 2161 |
+
"learning_rate": 3.228507629326387e-06,
|
| 2162 |
+
"loss": 0.0283,
|
| 2163 |
+
"step": 1735
|
| 2164 |
+
},
|
| 2165 |
+
{
|
| 2166 |
+
"epoch": 8.74,
|
| 2167 |
+
"learning_rate": 3.2378116858950505e-06,
|
| 2168 |
+
"loss": 0.1241,
|
| 2169 |
+
"step": 1740
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 8.77,
|
| 2173 |
+
"learning_rate": 3.2471157424637145e-06,
|
| 2174 |
+
"loss": 0.0139,
|
| 2175 |
+
"step": 1745
|
| 2176 |
+
},
|
| 2177 |
+
{
|
| 2178 |
+
"epoch": 8.79,
|
| 2179 |
+
"learning_rate": 3.256419799032378e-06,
|
| 2180 |
+
"loss": 0.049,
|
| 2181 |
+
"step": 1750
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 8.82,
|
| 2185 |
+
"learning_rate": 3.2657238556010425e-06,
|
| 2186 |
+
"loss": 0.0342,
|
| 2187 |
+
"step": 1755
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 8.84,
|
| 2191 |
+
"learning_rate": 3.2750279121697065e-06,
|
| 2192 |
+
"loss": 0.0535,
|
| 2193 |
+
"step": 1760
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 8.87,
|
| 2197 |
+
"learning_rate": 3.28433196873837e-06,
|
| 2198 |
+
"loss": 0.0585,
|
| 2199 |
+
"step": 1765
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 8.89,
|
| 2203 |
+
"learning_rate": 3.2936360253070344e-06,
|
| 2204 |
+
"loss": 0.16,
|
| 2205 |
+
"step": 1770
|
| 2206 |
+
},
|
| 2207 |
+
{
|
| 2208 |
+
"epoch": 8.92,
|
| 2209 |
+
"learning_rate": 3.302940081875698e-06,
|
| 2210 |
+
"loss": 0.0277,
|
| 2211 |
+
"step": 1775
|
| 2212 |
+
},
|
| 2213 |
+
{
|
| 2214 |
+
"epoch": 8.94,
|
| 2215 |
+
"learning_rate": 3.312244138444362e-06,
|
| 2216 |
+
"loss": 0.052,
|
| 2217 |
+
"step": 1780
|
| 2218 |
+
},
|
| 2219 |
+
{
|
| 2220 |
+
"epoch": 8.97,
|
| 2221 |
+
"learning_rate": 3.3215481950130256e-06,
|
| 2222 |
+
"loss": 0.0841,
|
| 2223 |
+
"step": 1785
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 8.99,
|
| 2227 |
+
"learning_rate": 3.33085225158169e-06,
|
| 2228 |
+
"loss": 0.1112,
|
| 2229 |
+
"step": 1790
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 9.0,
|
| 2233 |
+
"eval_accuracy": 0.9915174363807728,
|
| 2234 |
+
"eval_loss": 0.031130395829677582,
|
| 2235 |
+
"eval_runtime": 93.7625,
|
| 2236 |
+
"eval_samples_per_second": 22.632,
|
| 2237 |
+
"eval_steps_per_second": 1.418,
|
| 2238 |
+
"step": 1791
|
| 2239 |
+
},
|
| 2240 |
+
{
|
| 2241 |
+
"epoch": 9.02,
|
| 2242 |
+
"learning_rate": 3.340156308150354e-06,
|
| 2243 |
+
"loss": 0.066,
|
| 2244 |
+
"step": 1795
|
| 2245 |
+
},
|
| 2246 |
+
{
|
| 2247 |
+
"epoch": 9.05,
|
| 2248 |
+
"learning_rate": 3.3494603647190175e-06,
|
| 2249 |
+
"loss": 0.0461,
|
| 2250 |
+
"step": 1800
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 9.07,
|
| 2254 |
+
"learning_rate": 3.358764421287682e-06,
|
| 2255 |
+
"loss": 0.1146,
|
| 2256 |
+
"step": 1805
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 9.1,
|
| 2260 |
+
"learning_rate": 3.3680684778563455e-06,
|
| 2261 |
+
"loss": 0.0289,
|
| 2262 |
+
"step": 1810
|
| 2263 |
+
},
|
| 2264 |
+
{
|
| 2265 |
+
"epoch": 9.12,
|
| 2266 |
+
"learning_rate": 3.3773725344250095e-06,
|
| 2267 |
+
"loss": 0.018,
|
| 2268 |
+
"step": 1815
|
| 2269 |
+
},
|
| 2270 |
+
{
|
| 2271 |
+
"epoch": 9.15,
|
| 2272 |
+
"learning_rate": 3.386676590993673e-06,
|
| 2273 |
+
"loss": 0.0085,
|
| 2274 |
+
"step": 1820
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 9.17,
|
| 2278 |
+
"learning_rate": 3.3959806475623375e-06,
|
| 2279 |
+
"loss": 0.0356,
|
| 2280 |
+
"step": 1825
|
| 2281 |
+
},
|
| 2282 |
+
{
|
| 2283 |
+
"epoch": 9.2,
|
| 2284 |
+
"learning_rate": 3.4052847041310015e-06,
|
| 2285 |
+
"loss": 0.0525,
|
| 2286 |
+
"step": 1830
|
| 2287 |
+
},
|
| 2288 |
+
{
|
| 2289 |
+
"epoch": 9.22,
|
| 2290 |
+
"learning_rate": 3.414588760699665e-06,
|
| 2291 |
+
"loss": 0.0319,
|
| 2292 |
+
"step": 1835
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 9.25,
|
| 2296 |
+
"learning_rate": 3.4238928172683295e-06,
|
| 2297 |
+
"loss": 0.0463,
|
| 2298 |
+
"step": 1840
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 9.27,
|
| 2302 |
+
"learning_rate": 3.433196873836993e-06,
|
| 2303 |
+
"loss": 0.013,
|
| 2304 |
+
"step": 1845
|
| 2305 |
+
},
|
| 2306 |
+
{
|
| 2307 |
+
"epoch": 9.3,
|
| 2308 |
+
"learning_rate": 3.442500930405657e-06,
|
| 2309 |
+
"loss": 0.135,
|
| 2310 |
+
"step": 1850
|
| 2311 |
+
},
|
| 2312 |
+
{
|
| 2313 |
+
"epoch": 9.32,
|
| 2314 |
+
"learning_rate": 3.451804986974321e-06,
|
| 2315 |
+
"loss": 0.0415,
|
| 2316 |
+
"step": 1855
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 9.35,
|
| 2320 |
+
"learning_rate": 3.461109043542985e-06,
|
| 2321 |
+
"loss": 0.0242,
|
| 2322 |
+
"step": 1860
|
| 2323 |
+
},
|
| 2324 |
+
{
|
| 2325 |
+
"epoch": 9.37,
|
| 2326 |
+
"learning_rate": 3.470413100111649e-06,
|
| 2327 |
+
"loss": 0.0395,
|
| 2328 |
+
"step": 1865
|
| 2329 |
+
},
|
| 2330 |
+
{
|
| 2331 |
+
"epoch": 9.4,
|
| 2332 |
+
"learning_rate": 3.479717156680313e-06,
|
| 2333 |
+
"loss": 0.0328,
|
| 2334 |
+
"step": 1870
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 9.42,
|
| 2338 |
+
"learning_rate": 3.489021213248977e-06,
|
| 2339 |
+
"loss": 0.0525,
|
| 2340 |
+
"step": 1875
|
| 2341 |
+
},
|
| 2342 |
+
{
|
| 2343 |
+
"epoch": 9.45,
|
| 2344 |
+
"learning_rate": 3.4983252698176406e-06,
|
| 2345 |
+
"loss": 0.0676,
|
| 2346 |
+
"step": 1880
|
| 2347 |
+
},
|
| 2348 |
+
{
|
| 2349 |
+
"epoch": 9.47,
|
| 2350 |
+
"learning_rate": 3.507629326386305e-06,
|
| 2351 |
+
"loss": 0.0249,
|
| 2352 |
+
"step": 1885
|
| 2353 |
+
},
|
| 2354 |
+
{
|
| 2355 |
+
"epoch": 9.5,
|
| 2356 |
+
"learning_rate": 3.5169333829549686e-06,
|
| 2357 |
+
"loss": 0.0389,
|
| 2358 |
+
"step": 1890
|
| 2359 |
+
},
|
| 2360 |
+
{
|
| 2361 |
+
"epoch": 9.52,
|
| 2362 |
+
"learning_rate": 3.5262374395236326e-06,
|
| 2363 |
+
"loss": 0.0384,
|
| 2364 |
+
"step": 1895
|
| 2365 |
+
},
|
| 2366 |
+
{
|
| 2367 |
+
"epoch": 9.55,
|
| 2368 |
+
"learning_rate": 3.535541496092297e-06,
|
| 2369 |
+
"loss": 0.1467,
|
| 2370 |
+
"step": 1900
|
| 2371 |
+
},
|
| 2372 |
+
{
|
| 2373 |
+
"epoch": 9.57,
|
| 2374 |
+
"learning_rate": 3.5448455526609606e-06,
|
| 2375 |
+
"loss": 0.0466,
|
| 2376 |
+
"step": 1905
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 9.6,
|
| 2380 |
+
"learning_rate": 3.5541496092296245e-06,
|
| 2381 |
+
"loss": 0.0586,
|
| 2382 |
+
"step": 1910
|
| 2383 |
+
},
|
| 2384 |
+
{
|
| 2385 |
+
"epoch": 9.62,
|
| 2386 |
+
"learning_rate": 3.563453665798288e-06,
|
| 2387 |
+
"loss": 0.0593,
|
| 2388 |
+
"step": 1915
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 9.65,
|
| 2392 |
+
"learning_rate": 3.5727577223669525e-06,
|
| 2393 |
+
"loss": 0.0341,
|
| 2394 |
+
"step": 1920
|
| 2395 |
+
},
|
| 2396 |
+
{
|
| 2397 |
+
"epoch": 9.67,
|
| 2398 |
+
"learning_rate": 3.582061778935616e-06,
|
| 2399 |
+
"loss": 0.047,
|
| 2400 |
+
"step": 1925
|
| 2401 |
+
},
|
| 2402 |
+
{
|
| 2403 |
+
"epoch": 9.7,
|
| 2404 |
+
"learning_rate": 3.59136583550428e-06,
|
| 2405 |
+
"loss": 0.0496,
|
| 2406 |
+
"step": 1930
|
| 2407 |
+
},
|
| 2408 |
+
{
|
| 2409 |
+
"epoch": 9.72,
|
| 2410 |
+
"learning_rate": 3.6006698920729445e-06,
|
| 2411 |
+
"loss": 0.0828,
|
| 2412 |
+
"step": 1935
|
| 2413 |
+
},
|
| 2414 |
+
{
|
| 2415 |
+
"epoch": 9.75,
|
| 2416 |
+
"learning_rate": 3.609973948641608e-06,
|
| 2417 |
+
"loss": 0.0507,
|
| 2418 |
+
"step": 1940
|
| 2419 |
+
},
|
| 2420 |
+
{
|
| 2421 |
+
"epoch": 9.77,
|
| 2422 |
+
"learning_rate": 3.619278005210272e-06,
|
| 2423 |
+
"loss": 0.075,
|
| 2424 |
+
"step": 1945
|
| 2425 |
+
},
|
| 2426 |
+
{
|
| 2427 |
+
"epoch": 9.8,
|
| 2428 |
+
"learning_rate": 3.6285820617789356e-06,
|
| 2429 |
+
"loss": 0.0944,
|
| 2430 |
+
"step": 1950
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 9.82,
|
| 2434 |
+
"learning_rate": 3.6378861183476e-06,
|
| 2435 |
+
"loss": 0.0513,
|
| 2436 |
+
"step": 1955
|
| 2437 |
+
},
|
| 2438 |
+
{
|
| 2439 |
+
"epoch": 9.85,
|
| 2440 |
+
"learning_rate": 3.6471901749162636e-06,
|
| 2441 |
+
"loss": 0.0892,
|
| 2442 |
+
"step": 1960
|
| 2443 |
+
},
|
| 2444 |
+
{
|
| 2445 |
+
"epoch": 9.87,
|
| 2446 |
+
"learning_rate": 3.6564942314849276e-06,
|
| 2447 |
+
"loss": 0.0557,
|
| 2448 |
+
"step": 1965
|
| 2449 |
+
},
|
| 2450 |
+
{
|
| 2451 |
+
"epoch": 9.9,
|
| 2452 |
+
"learning_rate": 3.665798288053591e-06,
|
| 2453 |
+
"loss": 0.0187,
|
| 2454 |
+
"step": 1970
|
| 2455 |
+
},
|
| 2456 |
+
{
|
| 2457 |
+
"epoch": 9.92,
|
| 2458 |
+
"learning_rate": 3.6751023446222556e-06,
|
| 2459 |
+
"loss": 0.0873,
|
| 2460 |
+
"step": 1975
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"epoch": 9.95,
|
| 2464 |
+
"learning_rate": 3.6844064011909196e-06,
|
| 2465 |
+
"loss": 0.0481,
|
| 2466 |
+
"step": 1980
|
| 2467 |
+
},
|
| 2468 |
+
{
|
| 2469 |
+
"epoch": 9.97,
|
| 2470 |
+
"learning_rate": 3.693710457759583e-06,
|
| 2471 |
+
"loss": 0.0191,
|
| 2472 |
+
"step": 1985
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 10.0,
|
| 2476 |
+
"learning_rate": 3.7030145143282476e-06,
|
| 2477 |
+
"loss": 0.0428,
|
| 2478 |
+
"step": 1990
|
| 2479 |
+
},
|
| 2480 |
+
{
|
| 2481 |
+
"epoch": 10.0,
|
| 2482 |
+
"eval_accuracy": 0.9802073515551367,
|
| 2483 |
+
"eval_loss": 0.04842393100261688,
|
| 2484 |
+
"eval_runtime": 93.8775,
|
| 2485 |
+
"eval_samples_per_second": 22.604,
|
| 2486 |
+
"eval_steps_per_second": 1.417,
|
| 2487 |
+
"step": 1990
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 10.03,
|
| 2491 |
+
"learning_rate": 3.712318570896911e-06,
|
| 2492 |
+
"loss": 0.0201,
|
| 2493 |
+
"step": 1995
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 10.05,
|
| 2497 |
+
"learning_rate": 3.721622627465575e-06,
|
| 2498 |
+
"loss": 0.1922,
|
| 2499 |
+
"step": 2000
|
| 2500 |
+
},
|
| 2501 |
+
{
|
| 2502 |
+
"epoch": 10.08,
|
| 2503 |
+
"learning_rate": 3.730926684034239e-06,
|
| 2504 |
+
"loss": 0.0316,
|
| 2505 |
+
"step": 2005
|
| 2506 |
+
},
|
| 2507 |
+
{
|
| 2508 |
+
"epoch": 10.1,
|
| 2509 |
+
"learning_rate": 3.740230740602903e-06,
|
| 2510 |
+
"loss": 0.0165,
|
| 2511 |
+
"step": 2010
|
| 2512 |
+
},
|
| 2513 |
+
{
|
| 2514 |
+
"epoch": 10.13,
|
| 2515 |
+
"learning_rate": 3.749534797171567e-06,
|
| 2516 |
+
"loss": 0.0327,
|
| 2517 |
+
"step": 2015
|
| 2518 |
+
},
|
| 2519 |
+
{
|
| 2520 |
+
"epoch": 10.15,
|
| 2521 |
+
"learning_rate": 3.758838853740231e-06,
|
| 2522 |
+
"loss": 0.0651,
|
| 2523 |
+
"step": 2020
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 10.18,
|
| 2527 |
+
"learning_rate": 3.768142910308895e-06,
|
| 2528 |
+
"loss": 0.0878,
|
| 2529 |
+
"step": 2025
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 10.2,
|
| 2533 |
+
"learning_rate": 3.7774469668775587e-06,
|
| 2534 |
+
"loss": 0.0313,
|
| 2535 |
+
"step": 2030
|
| 2536 |
+
},
|
| 2537 |
+
{
|
| 2538 |
+
"epoch": 10.23,
|
| 2539 |
+
"learning_rate": 3.786751023446223e-06,
|
| 2540 |
+
"loss": 0.0416,
|
| 2541 |
+
"step": 2035
|
| 2542 |
+
},
|
| 2543 |
+
{
|
| 2544 |
+
"epoch": 10.25,
|
| 2545 |
+
"learning_rate": 3.7960550800148867e-06,
|
| 2546 |
+
"loss": 0.0398,
|
| 2547 |
+
"step": 2040
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 10.28,
|
| 2551 |
+
"learning_rate": 3.8053591365835507e-06,
|
| 2552 |
+
"loss": 0.0741,
|
| 2553 |
+
"step": 2045
|
| 2554 |
+
},
|
| 2555 |
+
{
|
| 2556 |
+
"epoch": 10.3,
|
| 2557 |
+
"learning_rate": 3.814663193152215e-06,
|
| 2558 |
+
"loss": 0.033,
|
| 2559 |
+
"step": 2050
|
| 2560 |
+
},
|
| 2561 |
+
{
|
| 2562 |
+
"epoch": 10.33,
|
| 2563 |
+
"learning_rate": 3.823967249720879e-06,
|
| 2564 |
+
"loss": 0.0131,
|
| 2565 |
+
"step": 2055
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 10.35,
|
| 2569 |
+
"learning_rate": 3.833271306289543e-06,
|
| 2570 |
+
"loss": 0.0506,
|
| 2571 |
+
"step": 2060
|
| 2572 |
+
},
|
| 2573 |
+
{
|
| 2574 |
+
"epoch": 10.38,
|
| 2575 |
+
"learning_rate": 3.842575362858207e-06,
|
| 2576 |
+
"loss": 0.0942,
|
| 2577 |
+
"step": 2065
|
| 2578 |
+
},
|
| 2579 |
+
{
|
| 2580 |
+
"epoch": 10.4,
|
| 2581 |
+
"learning_rate": 3.85187941942687e-06,
|
| 2582 |
+
"loss": 0.0292,
|
| 2583 |
+
"step": 2070
|
| 2584 |
+
},
|
| 2585 |
+
{
|
| 2586 |
+
"epoch": 10.43,
|
| 2587 |
+
"learning_rate": 3.861183475995534e-06,
|
| 2588 |
+
"loss": 0.0064,
|
| 2589 |
+
"step": 2075
|
| 2590 |
+
},
|
| 2591 |
+
{
|
| 2592 |
+
"epoch": 10.45,
|
| 2593 |
+
"learning_rate": 3.870487532564198e-06,
|
| 2594 |
+
"loss": 0.035,
|
| 2595 |
+
"step": 2080
|
| 2596 |
+
},
|
| 2597 |
+
{
|
| 2598 |
+
"epoch": 10.48,
|
| 2599 |
+
"learning_rate": 3.879791589132863e-06,
|
| 2600 |
+
"loss": 0.0083,
|
| 2601 |
+
"step": 2085
|
| 2602 |
+
},
|
| 2603 |
+
{
|
| 2604 |
+
"epoch": 10.5,
|
| 2605 |
+
"learning_rate": 3.889095645701526e-06,
|
| 2606 |
+
"loss": 0.0374,
|
| 2607 |
+
"step": 2090
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 10.53,
|
| 2611 |
+
"learning_rate": 3.898399702270191e-06,
|
| 2612 |
+
"loss": 0.0032,
|
| 2613 |
+
"step": 2095
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 10.55,
|
| 2617 |
+
"learning_rate": 3.907703758838854e-06,
|
| 2618 |
+
"loss": 0.1119,
|
| 2619 |
+
"step": 2100
|
| 2620 |
+
},
|
| 2621 |
+
{
|
| 2622 |
+
"epoch": 10.58,
|
| 2623 |
+
"learning_rate": 3.917007815407518e-06,
|
| 2624 |
+
"loss": 0.1343,
|
| 2625 |
+
"step": 2105
|
| 2626 |
+
},
|
| 2627 |
+
{
|
| 2628 |
+
"epoch": 10.6,
|
| 2629 |
+
"learning_rate": 3.926311871976181e-06,
|
| 2630 |
+
"loss": 0.1018,
|
| 2631 |
+
"step": 2110
|
| 2632 |
+
},
|
| 2633 |
+
{
|
| 2634 |
+
"epoch": 10.63,
|
| 2635 |
+
"learning_rate": 3.935615928544846e-06,
|
| 2636 |
+
"loss": 0.0402,
|
| 2637 |
+
"step": 2115
|
| 2638 |
+
},
|
| 2639 |
+
{
|
| 2640 |
+
"epoch": 10.65,
|
| 2641 |
+
"learning_rate": 3.94491998511351e-06,
|
| 2642 |
+
"loss": 0.0242,
|
| 2643 |
+
"step": 2120
|
| 2644 |
+
},
|
| 2645 |
+
{
|
| 2646 |
+
"epoch": 10.68,
|
| 2647 |
+
"learning_rate": 3.954224041682174e-06,
|
| 2648 |
+
"loss": 0.0081,
|
| 2649 |
+
"step": 2125
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 10.7,
|
| 2653 |
+
"learning_rate": 3.963528098250838e-06,
|
| 2654 |
+
"loss": 0.0405,
|
| 2655 |
+
"step": 2130
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 10.73,
|
| 2659 |
+
"learning_rate": 3.972832154819502e-06,
|
| 2660 |
+
"loss": 0.0133,
|
| 2661 |
+
"step": 2135
|
| 2662 |
+
},
|
| 2663 |
+
{
|
| 2664 |
+
"epoch": 10.75,
|
| 2665 |
+
"learning_rate": 3.982136211388165e-06,
|
| 2666 |
+
"loss": 0.0921,
|
| 2667 |
+
"step": 2140
|
| 2668 |
+
},
|
| 2669 |
+
{
|
| 2670 |
+
"epoch": 10.78,
|
| 2671 |
+
"learning_rate": 3.99144026795683e-06,
|
| 2672 |
+
"loss": 0.0668,
|
| 2673 |
+
"step": 2145
|
| 2674 |
+
},
|
| 2675 |
+
{
|
| 2676 |
+
"epoch": 10.8,
|
| 2677 |
+
"learning_rate": 4.000744324525493e-06,
|
| 2678 |
+
"loss": 0.0242,
|
| 2679 |
+
"step": 2150
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 10.83,
|
| 2683 |
+
"learning_rate": 4.010048381094158e-06,
|
| 2684 |
+
"loss": 0.0123,
|
| 2685 |
+
"step": 2155
|
| 2686 |
+
},
|
| 2687 |
+
{
|
| 2688 |
+
"epoch": 10.85,
|
| 2689 |
+
"learning_rate": 4.019352437662821e-06,
|
| 2690 |
+
"loss": 0.0287,
|
| 2691 |
+
"step": 2160
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 10.88,
|
| 2695 |
+
"learning_rate": 4.028656494231486e-06,
|
| 2696 |
+
"loss": 0.03,
|
| 2697 |
+
"step": 2165
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 10.9,
|
| 2701 |
+
"learning_rate": 4.037960550800149e-06,
|
| 2702 |
+
"loss": 0.1096,
|
| 2703 |
+
"step": 2170
|
| 2704 |
+
},
|
| 2705 |
+
{
|
| 2706 |
+
"epoch": 10.93,
|
| 2707 |
+
"learning_rate": 4.047264607368814e-06,
|
| 2708 |
+
"loss": 0.0705,
|
| 2709 |
+
"step": 2175
|
| 2710 |
+
},
|
| 2711 |
+
{
|
| 2712 |
+
"epoch": 10.95,
|
| 2713 |
+
"learning_rate": 4.056568663937477e-06,
|
| 2714 |
+
"loss": 0.0478,
|
| 2715 |
+
"step": 2180
|
| 2716 |
+
},
|
| 2717 |
+
{
|
| 2718 |
+
"epoch": 10.98,
|
| 2719 |
+
"learning_rate": 4.065872720506141e-06,
|
| 2720 |
+
"loss": 0.0842,
|
| 2721 |
+
"step": 2185
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 11.0,
|
| 2725 |
+
"eval_accuracy": 0.9863336475023563,
|
| 2726 |
+
"eval_loss": 0.04821911081671715,
|
| 2727 |
+
"eval_runtime": 93.9752,
|
| 2728 |
+
"eval_samples_per_second": 22.58,
|
| 2729 |
+
"eval_steps_per_second": 1.415,
|
| 2730 |
+
"step": 2189
|
| 2731 |
+
},
|
| 2732 |
+
{
|
| 2733 |
+
"epoch": 11.01,
|
| 2734 |
+
"learning_rate": 4.075176777074805e-06,
|
| 2735 |
+
"loss": 0.0033,
|
| 2736 |
+
"step": 2190
|
| 2737 |
+
},
|
| 2738 |
+
{
|
| 2739 |
+
"epoch": 11.03,
|
| 2740 |
+
"learning_rate": 4.084480833643469e-06,
|
| 2741 |
+
"loss": 0.077,
|
| 2742 |
+
"step": 2195
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 11.06,
|
| 2746 |
+
"learning_rate": 4.093784890212133e-06,
|
| 2747 |
+
"loss": 0.0579,
|
| 2748 |
+
"step": 2200
|
| 2749 |
+
},
|
| 2750 |
+
{
|
| 2751 |
+
"epoch": 11.08,
|
| 2752 |
+
"learning_rate": 4.103088946780797e-06,
|
| 2753 |
+
"loss": 0.0348,
|
| 2754 |
+
"step": 2205
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 11.11,
|
| 2758 |
+
"learning_rate": 4.112393003349461e-06,
|
| 2759 |
+
"loss": 0.0517,
|
| 2760 |
+
"step": 2210
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 11.13,
|
| 2764 |
+
"learning_rate": 4.121697059918125e-06,
|
| 2765 |
+
"loss": 0.0085,
|
| 2766 |
+
"step": 2215
|
| 2767 |
+
},
|
| 2768 |
+
{
|
| 2769 |
+
"epoch": 11.16,
|
| 2770 |
+
"learning_rate": 4.131001116486788e-06,
|
| 2771 |
+
"loss": 0.0242,
|
| 2772 |
+
"step": 2220
|
| 2773 |
+
},
|
| 2774 |
+
{
|
| 2775 |
+
"epoch": 11.18,
|
| 2776 |
+
"learning_rate": 4.140305173055453e-06,
|
| 2777 |
+
"loss": 0.0297,
|
| 2778 |
+
"step": 2225
|
| 2779 |
+
},
|
| 2780 |
+
{
|
| 2781 |
+
"epoch": 11.21,
|
| 2782 |
+
"learning_rate": 4.149609229624116e-06,
|
| 2783 |
+
"loss": 0.0607,
|
| 2784 |
+
"step": 2230
|
| 2785 |
+
},
|
| 2786 |
+
{
|
| 2787 |
+
"epoch": 11.23,
|
| 2788 |
+
"learning_rate": 4.158913286192781e-06,
|
| 2789 |
+
"loss": 0.0123,
|
| 2790 |
+
"step": 2235
|
| 2791 |
+
},
|
| 2792 |
+
{
|
| 2793 |
+
"epoch": 11.26,
|
| 2794 |
+
"learning_rate": 4.168217342761444e-06,
|
| 2795 |
+
"loss": 0.0944,
|
| 2796 |
+
"step": 2240
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 11.28,
|
| 2800 |
+
"learning_rate": 4.177521399330109e-06,
|
| 2801 |
+
"loss": 0.0209,
|
| 2802 |
+
"step": 2245
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 11.31,
|
| 2806 |
+
"learning_rate": 4.186825455898772e-06,
|
| 2807 |
+
"loss": 0.0577,
|
| 2808 |
+
"step": 2250
|
| 2809 |
+
},
|
| 2810 |
+
{
|
| 2811 |
+
"epoch": 11.33,
|
| 2812 |
+
"learning_rate": 4.196129512467436e-06,
|
| 2813 |
+
"loss": 0.0113,
|
| 2814 |
+
"step": 2255
|
| 2815 |
+
},
|
| 2816 |
+
{
|
| 2817 |
+
"epoch": 11.36,
|
| 2818 |
+
"learning_rate": 4.2054335690361e-06,
|
| 2819 |
+
"loss": 0.0152,
|
| 2820 |
+
"step": 2260
|
| 2821 |
+
},
|
| 2822 |
+
{
|
| 2823 |
+
"epoch": 11.38,
|
| 2824 |
+
"learning_rate": 4.214737625604764e-06,
|
| 2825 |
+
"loss": 0.1109,
|
| 2826 |
+
"step": 2265
|
| 2827 |
+
},
|
| 2828 |
+
{
|
| 2829 |
+
"epoch": 11.41,
|
| 2830 |
+
"learning_rate": 4.224041682173428e-06,
|
| 2831 |
+
"loss": 0.0049,
|
| 2832 |
+
"step": 2270
|
| 2833 |
+
},
|
| 2834 |
+
{
|
| 2835 |
+
"epoch": 11.43,
|
| 2836 |
+
"learning_rate": 4.233345738742092e-06,
|
| 2837 |
+
"loss": 0.089,
|
| 2838 |
+
"step": 2275
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 11.46,
|
| 2842 |
+
"learning_rate": 4.242649795310756e-06,
|
| 2843 |
+
"loss": 0.1244,
|
| 2844 |
+
"step": 2280
|
| 2845 |
+
},
|
| 2846 |
+
{
|
| 2847 |
+
"epoch": 11.48,
|
| 2848 |
+
"learning_rate": 4.25195385187942e-06,
|
| 2849 |
+
"loss": 0.0276,
|
| 2850 |
+
"step": 2285
|
| 2851 |
+
},
|
| 2852 |
+
{
|
| 2853 |
+
"epoch": 11.51,
|
| 2854 |
+
"learning_rate": 4.261257908448083e-06,
|
| 2855 |
+
"loss": 0.0601,
|
| 2856 |
+
"step": 2290
|
| 2857 |
+
},
|
| 2858 |
+
{
|
| 2859 |
+
"epoch": 11.53,
|
| 2860 |
+
"learning_rate": 4.270561965016748e-06,
|
| 2861 |
+
"loss": 0.0992,
|
| 2862 |
+
"step": 2295
|
| 2863 |
+
},
|
| 2864 |
+
{
|
| 2865 |
+
"epoch": 11.56,
|
| 2866 |
+
"learning_rate": 4.279866021585411e-06,
|
| 2867 |
+
"loss": 0.1146,
|
| 2868 |
+
"step": 2300
|
| 2869 |
+
},
|
| 2870 |
+
{
|
| 2871 |
+
"epoch": 11.58,
|
| 2872 |
+
"learning_rate": 4.289170078154076e-06,
|
| 2873 |
+
"loss": 0.0186,
|
| 2874 |
+
"step": 2305
|
| 2875 |
+
},
|
| 2876 |
+
{
|
| 2877 |
+
"epoch": 11.61,
|
| 2878 |
+
"learning_rate": 4.298474134722739e-06,
|
| 2879 |
+
"loss": 0.0452,
|
| 2880 |
+
"step": 2310
|
| 2881 |
+
},
|
| 2882 |
+
{
|
| 2883 |
+
"epoch": 11.63,
|
| 2884 |
+
"learning_rate": 4.307778191291404e-06,
|
| 2885 |
+
"loss": 0.09,
|
| 2886 |
+
"step": 2315
|
| 2887 |
+
},
|
| 2888 |
+
{
|
| 2889 |
+
"epoch": 11.66,
|
| 2890 |
+
"learning_rate": 4.317082247860067e-06,
|
| 2891 |
+
"loss": 0.0224,
|
| 2892 |
+
"step": 2320
|
| 2893 |
+
},
|
| 2894 |
+
{
|
| 2895 |
+
"epoch": 11.68,
|
| 2896 |
+
"learning_rate": 4.326386304428732e-06,
|
| 2897 |
+
"loss": 0.0131,
|
| 2898 |
+
"step": 2325
|
| 2899 |
+
},
|
| 2900 |
+
{
|
| 2901 |
+
"epoch": 11.71,
|
| 2902 |
+
"learning_rate": 4.335690360997395e-06,
|
| 2903 |
+
"loss": 0.0559,
|
| 2904 |
+
"step": 2330
|
| 2905 |
+
},
|
| 2906 |
+
{
|
| 2907 |
+
"epoch": 11.73,
|
| 2908 |
+
"learning_rate": 4.344994417566059e-06,
|
| 2909 |
+
"loss": 0.0779,
|
| 2910 |
+
"step": 2335
|
| 2911 |
+
},
|
| 2912 |
+
{
|
| 2913 |
+
"epoch": 11.76,
|
| 2914 |
+
"learning_rate": 4.354298474134723e-06,
|
| 2915 |
+
"loss": 0.0037,
|
| 2916 |
+
"step": 2340
|
| 2917 |
+
},
|
| 2918 |
+
{
|
| 2919 |
+
"epoch": 11.78,
|
| 2920 |
+
"learning_rate": 4.363602530703387e-06,
|
| 2921 |
+
"loss": 0.0587,
|
| 2922 |
+
"step": 2345
|
| 2923 |
+
},
|
| 2924 |
+
{
|
| 2925 |
+
"epoch": 11.81,
|
| 2926 |
+
"learning_rate": 4.372906587272051e-06,
|
| 2927 |
+
"loss": 0.0788,
|
| 2928 |
+
"step": 2350
|
| 2929 |
+
},
|
| 2930 |
+
{
|
| 2931 |
+
"epoch": 11.83,
|
| 2932 |
+
"learning_rate": 4.382210643840715e-06,
|
| 2933 |
+
"loss": 0.0082,
|
| 2934 |
+
"step": 2355
|
| 2935 |
+
},
|
| 2936 |
+
{
|
| 2937 |
+
"epoch": 11.86,
|
| 2938 |
+
"learning_rate": 4.391514700409379e-06,
|
| 2939 |
+
"loss": 0.0237,
|
| 2940 |
+
"step": 2360
|
| 2941 |
+
},
|
| 2942 |
+
{
|
| 2943 |
+
"epoch": 11.88,
|
| 2944 |
+
"learning_rate": 4.400818756978043e-06,
|
| 2945 |
+
"loss": 0.099,
|
| 2946 |
+
"step": 2365
|
| 2947 |
+
},
|
| 2948 |
+
{
|
| 2949 |
+
"epoch": 11.91,
|
| 2950 |
+
"learning_rate": 4.410122813546706e-06,
|
| 2951 |
+
"loss": 0.053,
|
| 2952 |
+
"step": 2370
|
| 2953 |
+
},
|
| 2954 |
+
{
|
| 2955 |
+
"epoch": 11.93,
|
| 2956 |
+
"learning_rate": 4.419426870115371e-06,
|
| 2957 |
+
"loss": 0.0503,
|
| 2958 |
+
"step": 2375
|
| 2959 |
+
},
|
| 2960 |
+
{
|
| 2961 |
+
"epoch": 11.96,
|
| 2962 |
+
"learning_rate": 4.428730926684034e-06,
|
| 2963 |
+
"loss": 0.0849,
|
| 2964 |
+
"step": 2380
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 11.98,
|
| 2968 |
+
"learning_rate": 4.438034983252699e-06,
|
| 2969 |
+
"loss": 0.0129,
|
| 2970 |
+
"step": 2385
|
| 2971 |
+
},
|
| 2972 |
+
{
|
| 2973 |
+
"epoch": 12.0,
|
| 2974 |
+
"eval_accuracy": 0.9896324222431668,
|
| 2975 |
+
"eval_loss": 0.0371868796646595,
|
| 2976 |
+
"eval_runtime": 93.989,
|
| 2977 |
+
"eval_samples_per_second": 22.577,
|
| 2978 |
+
"eval_steps_per_second": 1.415,
|
| 2979 |
+
"step": 2388
|
| 2980 |
+
},
|
| 2981 |
+
{
|
| 2982 |
+
"epoch": 12.01,
|
| 2983 |
+
"learning_rate": 4.447339039821362e-06,
|
| 2984 |
+
"loss": 0.045,
|
| 2985 |
+
"step": 2390
|
| 2986 |
+
},
|
| 2987 |
+
{
|
| 2988 |
+
"epoch": 12.04,
|
| 2989 |
+
"learning_rate": 4.456643096390027e-06,
|
| 2990 |
+
"loss": 0.0826,
|
| 2991 |
+
"step": 2395
|
| 2992 |
+
},
|
| 2993 |
+
{
|
| 2994 |
+
"epoch": 12.06,
|
| 2995 |
+
"learning_rate": 4.46594715295869e-06,
|
| 2996 |
+
"loss": 0.0441,
|
| 2997 |
+
"step": 2400
|
| 2998 |
+
},
|
| 2999 |
+
{
|
| 3000 |
+
"epoch": 12.09,
|
| 3001 |
+
"learning_rate": 4.475251209527354e-06,
|
| 3002 |
+
"loss": 0.1254,
|
| 3003 |
+
"step": 2405
|
| 3004 |
+
},
|
| 3005 |
+
{
|
| 3006 |
+
"epoch": 12.11,
|
| 3007 |
+
"learning_rate": 4.484555266096018e-06,
|
| 3008 |
+
"loss": 0.0109,
|
| 3009 |
+
"step": 2410
|
| 3010 |
+
},
|
| 3011 |
+
{
|
| 3012 |
+
"epoch": 12.14,
|
| 3013 |
+
"learning_rate": 4.493859322664682e-06,
|
| 3014 |
+
"loss": 0.0493,
|
| 3015 |
+
"step": 2415
|
| 3016 |
+
},
|
| 3017 |
+
{
|
| 3018 |
+
"epoch": 12.16,
|
| 3019 |
+
"learning_rate": 4.503163379233346e-06,
|
| 3020 |
+
"loss": 0.0056,
|
| 3021 |
+
"step": 2420
|
| 3022 |
+
},
|
| 3023 |
+
{
|
| 3024 |
+
"epoch": 12.19,
|
| 3025 |
+
"learning_rate": 4.51246743580201e-06,
|
| 3026 |
+
"loss": 0.0338,
|
| 3027 |
+
"step": 2425
|
| 3028 |
+
},
|
| 3029 |
+
{
|
| 3030 |
+
"epoch": 12.21,
|
| 3031 |
+
"learning_rate": 4.521771492370674e-06,
|
| 3032 |
+
"loss": 0.0402,
|
| 3033 |
+
"step": 2430
|
| 3034 |
+
},
|
| 3035 |
+
{
|
| 3036 |
+
"epoch": 12.24,
|
| 3037 |
+
"learning_rate": 4.531075548939338e-06,
|
| 3038 |
+
"loss": 0.048,
|
| 3039 |
+
"step": 2435
|
| 3040 |
+
},
|
| 3041 |
+
{
|
| 3042 |
+
"epoch": 12.26,
|
| 3043 |
+
"learning_rate": 4.5403796055080014e-06,
|
| 3044 |
+
"loss": 0.0381,
|
| 3045 |
+
"step": 2440
|
| 3046 |
+
},
|
| 3047 |
+
{
|
| 3048 |
+
"epoch": 12.29,
|
| 3049 |
+
"learning_rate": 4.549683662076666e-06,
|
| 3050 |
+
"loss": 0.0159,
|
| 3051 |
+
"step": 2445
|
| 3052 |
+
},
|
| 3053 |
+
{
|
| 3054 |
+
"epoch": 12.31,
|
| 3055 |
+
"learning_rate": 4.5589877186453294e-06,
|
| 3056 |
+
"loss": 0.0296,
|
| 3057 |
+
"step": 2450
|
| 3058 |
+
},
|
| 3059 |
+
{
|
| 3060 |
+
"epoch": 12.34,
|
| 3061 |
+
"learning_rate": 4.568291775213994e-06,
|
| 3062 |
+
"loss": 0.0202,
|
| 3063 |
+
"step": 2455
|
| 3064 |
+
},
|
| 3065 |
+
{
|
| 3066 |
+
"epoch": 12.36,
|
| 3067 |
+
"learning_rate": 4.577595831782657e-06,
|
| 3068 |
+
"loss": 0.0597,
|
| 3069 |
+
"step": 2460
|
| 3070 |
+
},
|
| 3071 |
+
{
|
| 3072 |
+
"epoch": 12.39,
|
| 3073 |
+
"learning_rate": 4.586899888351322e-06,
|
| 3074 |
+
"loss": 0.0564,
|
| 3075 |
+
"step": 2465
|
| 3076 |
+
},
|
| 3077 |
+
{
|
| 3078 |
+
"epoch": 12.41,
|
| 3079 |
+
"learning_rate": 4.596203944919985e-06,
|
| 3080 |
+
"loss": 0.0801,
|
| 3081 |
+
"step": 2470
|
| 3082 |
+
},
|
| 3083 |
+
{
|
| 3084 |
+
"epoch": 12.44,
|
| 3085 |
+
"learning_rate": 4.60550800148865e-06,
|
| 3086 |
+
"loss": 0.0081,
|
| 3087 |
+
"step": 2475
|
| 3088 |
+
},
|
| 3089 |
+
{
|
| 3090 |
+
"epoch": 12.46,
|
| 3091 |
+
"learning_rate": 4.614812058057313e-06,
|
| 3092 |
+
"loss": 0.0091,
|
| 3093 |
+
"step": 2480
|
| 3094 |
+
},
|
| 3095 |
+
{
|
| 3096 |
+
"epoch": 12.49,
|
| 3097 |
+
"learning_rate": 4.624116114625977e-06,
|
| 3098 |
+
"loss": 0.0193,
|
| 3099 |
+
"step": 2485
|
| 3100 |
+
},
|
| 3101 |
+
{
|
| 3102 |
+
"epoch": 12.51,
|
| 3103 |
+
"learning_rate": 4.633420171194641e-06,
|
| 3104 |
+
"loss": 0.0206,
|
| 3105 |
+
"step": 2490
|
| 3106 |
+
},
|
| 3107 |
+
{
|
| 3108 |
+
"epoch": 12.54,
|
| 3109 |
+
"learning_rate": 4.642724227763305e-06,
|
| 3110 |
+
"loss": 0.01,
|
| 3111 |
+
"step": 2495
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 12.56,
|
| 3115 |
+
"learning_rate": 4.652028284331969e-06,
|
| 3116 |
+
"loss": 0.0109,
|
| 3117 |
+
"step": 2500
|
| 3118 |
+
},
|
| 3119 |
+
{
|
| 3120 |
+
"epoch": 12.59,
|
| 3121 |
+
"learning_rate": 4.661332340900633e-06,
|
| 3122 |
+
"loss": 0.0122,
|
| 3123 |
+
"step": 2505
|
| 3124 |
+
},
|
| 3125 |
+
{
|
| 3126 |
+
"epoch": 12.61,
|
| 3127 |
+
"learning_rate": 4.670636397469297e-06,
|
| 3128 |
+
"loss": 0.0592,
|
| 3129 |
+
"step": 2510
|
| 3130 |
+
},
|
| 3131 |
+
{
|
| 3132 |
+
"epoch": 12.64,
|
| 3133 |
+
"learning_rate": 4.679940454037961e-06,
|
| 3134 |
+
"loss": 0.0029,
|
| 3135 |
+
"step": 2515
|
| 3136 |
+
},
|
| 3137 |
+
{
|
| 3138 |
+
"epoch": 12.66,
|
| 3139 |
+
"learning_rate": 4.6892445106066245e-06,
|
| 3140 |
+
"loss": 0.0531,
|
| 3141 |
+
"step": 2520
|
| 3142 |
+
},
|
| 3143 |
+
{
|
| 3144 |
+
"epoch": 12.69,
|
| 3145 |
+
"learning_rate": 4.698548567175289e-06,
|
| 3146 |
+
"loss": 0.0746,
|
| 3147 |
+
"step": 2525
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 12.71,
|
| 3151 |
+
"learning_rate": 4.7078526237439525e-06,
|
| 3152 |
+
"loss": 0.0235,
|
| 3153 |
+
"step": 2530
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 12.74,
|
| 3157 |
+
"learning_rate": 4.717156680312617e-06,
|
| 3158 |
+
"loss": 0.0447,
|
| 3159 |
+
"step": 2535
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 12.76,
|
| 3163 |
+
"learning_rate": 4.7264607368812805e-06,
|
| 3164 |
+
"loss": 0.0527,
|
| 3165 |
+
"step": 2540
|
| 3166 |
+
},
|
| 3167 |
+
{
|
| 3168 |
+
"epoch": 12.79,
|
| 3169 |
+
"learning_rate": 4.735764793449945e-06,
|
| 3170 |
+
"loss": 0.0347,
|
| 3171 |
+
"step": 2545
|
| 3172 |
+
},
|
| 3173 |
+
{
|
| 3174 |
+
"epoch": 12.81,
|
| 3175 |
+
"learning_rate": 4.7450688500186084e-06,
|
| 3176 |
+
"loss": 0.0774,
|
| 3177 |
+
"step": 2550
|
| 3178 |
+
},
|
| 3179 |
+
{
|
| 3180 |
+
"epoch": 12.84,
|
| 3181 |
+
"learning_rate": 4.754372906587272e-06,
|
| 3182 |
+
"loss": 0.1487,
|
| 3183 |
+
"step": 2555
|
| 3184 |
+
},
|
| 3185 |
+
{
|
| 3186 |
+
"epoch": 12.86,
|
| 3187 |
+
"learning_rate": 4.7636769631559364e-06,
|
| 3188 |
+
"loss": 0.0062,
|
| 3189 |
+
"step": 2560
|
| 3190 |
+
},
|
| 3191 |
+
{
|
| 3192 |
+
"epoch": 12.89,
|
| 3193 |
+
"learning_rate": 4.7729810197246e-06,
|
| 3194 |
+
"loss": 0.0128,
|
| 3195 |
+
"step": 2565
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 12.91,
|
| 3199 |
+
"learning_rate": 4.782285076293264e-06,
|
| 3200 |
+
"loss": 0.0236,
|
| 3201 |
+
"step": 2570
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 12.94,
|
| 3205 |
+
"learning_rate": 4.791589132861928e-06,
|
| 3206 |
+
"loss": 0.0084,
|
| 3207 |
+
"step": 2575
|
| 3208 |
+
},
|
| 3209 |
+
{
|
| 3210 |
+
"epoch": 12.96,
|
| 3211 |
+
"learning_rate": 4.800893189430592e-06,
|
| 3212 |
+
"loss": 0.0265,
|
| 3213 |
+
"step": 2580
|
| 3214 |
+
},
|
| 3215 |
+
{
|
| 3216 |
+
"epoch": 12.99,
|
| 3217 |
+
"learning_rate": 4.810197245999256e-06,
|
| 3218 |
+
"loss": 0.1463,
|
| 3219 |
+
"step": 2585
|
| 3220 |
+
},
|
| 3221 |
+
{
|
| 3222 |
+
"epoch": 13.0,
|
| 3223 |
+
"eval_accuracy": 0.9882186616399623,
|
| 3224 |
+
"eval_loss": 0.040391404181718826,
|
| 3225 |
+
"eval_runtime": 93.9387,
|
| 3226 |
+
"eval_samples_per_second": 22.589,
|
| 3227 |
+
"eval_steps_per_second": 1.416,
|
| 3228 |
+
"step": 2587
|
| 3229 |
+
},
|
| 3230 |
+
{
|
| 3231 |
+
"epoch": 13.02,
|
| 3232 |
+
"learning_rate": 4.8195013025679195e-06,
|
| 3233 |
+
"loss": 0.1094,
|
| 3234 |
+
"step": 2590
|
| 3235 |
+
},
|
| 3236 |
+
{
|
| 3237 |
+
"epoch": 13.04,
|
| 3238 |
+
"learning_rate": 4.828805359136584e-06,
|
| 3239 |
+
"loss": 0.0116,
|
| 3240 |
+
"step": 2595
|
| 3241 |
+
},
|
| 3242 |
+
{
|
| 3243 |
+
"epoch": 13.07,
|
| 3244 |
+
"learning_rate": 4.8381094157052475e-06,
|
| 3245 |
+
"loss": 0.0787,
|
| 3246 |
+
"step": 2600
|
| 3247 |
+
},
|
| 3248 |
+
{
|
| 3249 |
+
"epoch": 13.09,
|
| 3250 |
+
"learning_rate": 4.847413472273912e-06,
|
| 3251 |
+
"loss": 0.0086,
|
| 3252 |
+
"step": 2605
|
| 3253 |
+
},
|
| 3254 |
+
{
|
| 3255 |
+
"epoch": 13.12,
|
| 3256 |
+
"learning_rate": 4.8567175288425755e-06,
|
| 3257 |
+
"loss": 0.0209,
|
| 3258 |
+
"step": 2610
|
| 3259 |
+
},
|
| 3260 |
+
{
|
| 3261 |
+
"epoch": 13.14,
|
| 3262 |
+
"learning_rate": 4.86602158541124e-06,
|
| 3263 |
+
"loss": 0.089,
|
| 3264 |
+
"step": 2615
|
| 3265 |
+
},
|
| 3266 |
+
{
|
| 3267 |
+
"epoch": 13.17,
|
| 3268 |
+
"learning_rate": 4.8753256419799035e-06,
|
| 3269 |
+
"loss": 0.0516,
|
| 3270 |
+
"step": 2620
|
| 3271 |
+
},
|
| 3272 |
+
{
|
| 3273 |
+
"epoch": 13.19,
|
| 3274 |
+
"learning_rate": 4.884629698548568e-06,
|
| 3275 |
+
"loss": 0.0053,
|
| 3276 |
+
"step": 2625
|
| 3277 |
+
},
|
| 3278 |
+
{
|
| 3279 |
+
"epoch": 13.22,
|
| 3280 |
+
"learning_rate": 4.8939337551172315e-06,
|
| 3281 |
+
"loss": 0.0299,
|
| 3282 |
+
"step": 2630
|
| 3283 |
+
},
|
| 3284 |
+
{
|
| 3285 |
+
"epoch": 13.24,
|
| 3286 |
+
"learning_rate": 4.903237811685895e-06,
|
| 3287 |
+
"loss": 0.0086,
|
| 3288 |
+
"step": 2635
|
| 3289 |
+
},
|
| 3290 |
+
{
|
| 3291 |
+
"epoch": 13.27,
|
| 3292 |
+
"learning_rate": 4.9125418682545595e-06,
|
| 3293 |
+
"loss": 0.1193,
|
| 3294 |
+
"step": 2640
|
| 3295 |
+
},
|
| 3296 |
+
{
|
| 3297 |
+
"epoch": 13.29,
|
| 3298 |
+
"learning_rate": 4.921845924823223e-06,
|
| 3299 |
+
"loss": 0.0333,
|
| 3300 |
+
"step": 2645
|
| 3301 |
+
},
|
| 3302 |
+
{
|
| 3303 |
+
"epoch": 13.32,
|
| 3304 |
+
"learning_rate": 4.9311499813918875e-06,
|
| 3305 |
+
"loss": 0.0792,
|
| 3306 |
+
"step": 2650
|
| 3307 |
+
},
|
| 3308 |
+
{
|
| 3309 |
+
"epoch": 13.34,
|
| 3310 |
+
"learning_rate": 4.940454037960552e-06,
|
| 3311 |
+
"loss": 0.0455,
|
| 3312 |
+
"step": 2655
|
| 3313 |
+
},
|
| 3314 |
+
{
|
| 3315 |
+
"epoch": 13.37,
|
| 3316 |
+
"learning_rate": 4.9497580945292154e-06,
|
| 3317 |
+
"loss": 0.0256,
|
| 3318 |
+
"step": 2660
|
| 3319 |
+
},
|
| 3320 |
+
{
|
| 3321 |
+
"epoch": 13.39,
|
| 3322 |
+
"learning_rate": 4.959062151097879e-06,
|
| 3323 |
+
"loss": 0.0697,
|
| 3324 |
+
"step": 2665
|
| 3325 |
+
},
|
| 3326 |
+
{
|
| 3327 |
+
"epoch": 13.42,
|
| 3328 |
+
"learning_rate": 4.968366207666543e-06,
|
| 3329 |
+
"loss": 0.0754,
|
| 3330 |
+
"step": 2670
|
| 3331 |
+
},
|
| 3332 |
+
{
|
| 3333 |
+
"epoch": 13.44,
|
| 3334 |
+
"learning_rate": 4.977670264235207e-06,
|
| 3335 |
+
"loss": 0.0113,
|
| 3336 |
+
"step": 2675
|
| 3337 |
+
},
|
| 3338 |
+
{
|
| 3339 |
+
"epoch": 13.47,
|
| 3340 |
+
"learning_rate": 4.9869743208038706e-06,
|
| 3341 |
+
"loss": 0.0283,
|
| 3342 |
+
"step": 2680
|
| 3343 |
+
},
|
| 3344 |
+
{
|
| 3345 |
+
"epoch": 13.49,
|
| 3346 |
+
"learning_rate": 4.996278377372535e-06,
|
| 3347 |
+
"loss": 0.0223,
|
| 3348 |
+
"step": 2685
|
| 3349 |
+
},
|
| 3350 |
+
{
|
| 3351 |
+
"epoch": 13.52,
|
| 3352 |
+
"learning_rate": 4.949664429530202e-06,
|
| 3353 |
+
"loss": 0.0362,
|
| 3354 |
+
"step": 2690
|
| 3355 |
+
},
|
| 3356 |
+
{
|
| 3357 |
+
"epoch": 13.54,
|
| 3358 |
+
"learning_rate": 4.8657718120805375e-06,
|
| 3359 |
+
"loss": 0.0238,
|
| 3360 |
+
"step": 2695
|
| 3361 |
+
},
|
| 3362 |
+
{
|
| 3363 |
+
"epoch": 13.57,
|
| 3364 |
+
"learning_rate": 4.781879194630873e-06,
|
| 3365 |
+
"loss": 0.0084,
|
| 3366 |
+
"step": 2700
|
| 3367 |
+
},
|
| 3368 |
+
{
|
| 3369 |
+
"epoch": 13.59,
|
| 3370 |
+
"learning_rate": 4.697986577181208e-06,
|
| 3371 |
+
"loss": 0.0447,
|
| 3372 |
+
"step": 2705
|
| 3373 |
+
},
|
| 3374 |
+
{
|
| 3375 |
+
"epoch": 13.62,
|
| 3376 |
+
"learning_rate": 4.614093959731544e-06,
|
| 3377 |
+
"loss": 0.0665,
|
| 3378 |
+
"step": 2710
|
| 3379 |
+
},
|
| 3380 |
+
{
|
| 3381 |
+
"epoch": 13.64,
|
| 3382 |
+
"learning_rate": 4.530201342281879e-06,
|
| 3383 |
+
"loss": 0.0598,
|
| 3384 |
+
"step": 2715
|
| 3385 |
+
},
|
| 3386 |
+
{
|
| 3387 |
+
"epoch": 13.67,
|
| 3388 |
+
"learning_rate": 4.446308724832215e-06,
|
| 3389 |
+
"loss": 0.0289,
|
| 3390 |
+
"step": 2720
|
| 3391 |
+
},
|
| 3392 |
+
{
|
| 3393 |
+
"epoch": 13.69,
|
| 3394 |
+
"learning_rate": 4.362416107382551e-06,
|
| 3395 |
+
"loss": 0.0484,
|
| 3396 |
+
"step": 2725
|
| 3397 |
+
},
|
| 3398 |
+
{
|
| 3399 |
+
"epoch": 13.72,
|
| 3400 |
+
"learning_rate": 4.278523489932886e-06,
|
| 3401 |
+
"loss": 0.0302,
|
| 3402 |
+
"step": 2730
|
| 3403 |
+
},
|
| 3404 |
+
{
|
| 3405 |
+
"epoch": 13.74,
|
| 3406 |
+
"learning_rate": 4.194630872483222e-06,
|
| 3407 |
+
"loss": 0.0026,
|
| 3408 |
+
"step": 2735
|
| 3409 |
+
},
|
| 3410 |
+
{
|
| 3411 |
+
"epoch": 13.77,
|
| 3412 |
+
"learning_rate": 4.110738255033558e-06,
|
| 3413 |
+
"loss": 0.0103,
|
| 3414 |
+
"step": 2740
|
| 3415 |
+
},
|
| 3416 |
+
{
|
| 3417 |
+
"epoch": 13.79,
|
| 3418 |
+
"learning_rate": 4.026845637583892e-06,
|
| 3419 |
+
"loss": 0.0029,
|
| 3420 |
+
"step": 2745
|
| 3421 |
+
},
|
| 3422 |
+
{
|
| 3423 |
+
"epoch": 13.82,
|
| 3424 |
+
"learning_rate": 3.942953020134229e-06,
|
| 3425 |
+
"loss": 0.0118,
|
| 3426 |
+
"step": 2750
|
| 3427 |
+
},
|
| 3428 |
+
{
|
| 3429 |
+
"epoch": 13.84,
|
| 3430 |
+
"learning_rate": 3.859060402684564e-06,
|
| 3431 |
+
"loss": 0.0391,
|
| 3432 |
+
"step": 2755
|
| 3433 |
+
},
|
| 3434 |
+
{
|
| 3435 |
+
"epoch": 13.87,
|
| 3436 |
+
"learning_rate": 3.7751677852348994e-06,
|
| 3437 |
+
"loss": 0.0632,
|
| 3438 |
+
"step": 2760
|
| 3439 |
+
},
|
| 3440 |
+
{
|
| 3441 |
+
"epoch": 13.89,
|
| 3442 |
+
"learning_rate": 3.6912751677852355e-06,
|
| 3443 |
+
"loss": 0.0446,
|
| 3444 |
+
"step": 2765
|
| 3445 |
+
},
|
| 3446 |
+
{
|
| 3447 |
+
"epoch": 13.92,
|
| 3448 |
+
"learning_rate": 3.6073825503355707e-06,
|
| 3449 |
+
"loss": 0.072,
|
| 3450 |
+
"step": 2770
|
| 3451 |
+
},
|
| 3452 |
+
{
|
| 3453 |
+
"epoch": 13.94,
|
| 3454 |
+
"learning_rate": 3.5234899328859064e-06,
|
| 3455 |
+
"loss": 0.047,
|
| 3456 |
+
"step": 2775
|
| 3457 |
+
},
|
| 3458 |
+
{
|
| 3459 |
+
"epoch": 13.97,
|
| 3460 |
+
"learning_rate": 3.4395973154362416e-06,
|
| 3461 |
+
"loss": 0.0462,
|
| 3462 |
+
"step": 2780
|
| 3463 |
+
},
|
| 3464 |
+
{
|
| 3465 |
+
"epoch": 13.99,
|
| 3466 |
+
"learning_rate": 3.3557046979865777e-06,
|
| 3467 |
+
"loss": 0.0582,
|
| 3468 |
+
"step": 2785
|
| 3469 |
+
},
|
| 3470 |
+
{
|
| 3471 |
+
"epoch": 14.0,
|
| 3472 |
+
"eval_accuracy": 0.9820923656927427,
|
| 3473 |
+
"eval_loss": 0.05892978981137276,
|
| 3474 |
+
"eval_runtime": 93.6799,
|
| 3475 |
+
"eval_samples_per_second": 22.652,
|
| 3476 |
+
"eval_steps_per_second": 1.42,
|
| 3477 |
+
"step": 2786
|
| 3478 |
+
},
|
| 3479 |
+
{
|
| 3480 |
+
"epoch": 14.02,
|
| 3481 |
+
"learning_rate": 3.271812080536913e-06,
|
| 3482 |
+
"loss": 0.0142,
|
| 3483 |
+
"step": 2790
|
| 3484 |
+
},
|
| 3485 |
+
{
|
| 3486 |
+
"epoch": 14.05,
|
| 3487 |
+
"learning_rate": 3.1879194630872486e-06,
|
| 3488 |
+
"loss": 0.0553,
|
| 3489 |
+
"step": 2795
|
| 3490 |
+
},
|
| 3491 |
+
{
|
| 3492 |
+
"epoch": 14.07,
|
| 3493 |
+
"learning_rate": 3.1040268456375843e-06,
|
| 3494 |
+
"loss": 0.0318,
|
| 3495 |
+
"step": 2800
|
| 3496 |
+
},
|
| 3497 |
+
{
|
| 3498 |
+
"epoch": 14.1,
|
| 3499 |
+
"learning_rate": 3.02013422818792e-06,
|
| 3500 |
+
"loss": 0.0095,
|
| 3501 |
+
"step": 2805
|
| 3502 |
+
},
|
| 3503 |
+
{
|
| 3504 |
+
"epoch": 14.12,
|
| 3505 |
+
"learning_rate": 2.936241610738255e-06,
|
| 3506 |
+
"loss": 0.0033,
|
| 3507 |
+
"step": 2810
|
| 3508 |
+
},
|
| 3509 |
+
{
|
| 3510 |
+
"epoch": 14.15,
|
| 3511 |
+
"learning_rate": 2.8523489932885904e-06,
|
| 3512 |
+
"loss": 0.0163,
|
| 3513 |
+
"step": 2815
|
| 3514 |
+
},
|
| 3515 |
+
{
|
| 3516 |
+
"epoch": 14.17,
|
| 3517 |
+
"learning_rate": 2.7684563758389265e-06,
|
| 3518 |
+
"loss": 0.0185,
|
| 3519 |
+
"step": 2820
|
| 3520 |
+
},
|
| 3521 |
+
{
|
| 3522 |
+
"epoch": 14.2,
|
| 3523 |
+
"learning_rate": 2.684563758389262e-06,
|
| 3524 |
+
"loss": 0.0254,
|
| 3525 |
+
"step": 2825
|
| 3526 |
+
},
|
| 3527 |
+
{
|
| 3528 |
+
"epoch": 14.22,
|
| 3529 |
+
"learning_rate": 2.6006711409395974e-06,
|
| 3530 |
+
"loss": 0.002,
|
| 3531 |
+
"step": 2830
|
| 3532 |
+
},
|
| 3533 |
+
{
|
| 3534 |
+
"epoch": 14.25,
|
| 3535 |
+
"learning_rate": 2.5167785234899335e-06,
|
| 3536 |
+
"loss": 0.023,
|
| 3537 |
+
"step": 2835
|
| 3538 |
+
},
|
| 3539 |
+
{
|
| 3540 |
+
"epoch": 14.27,
|
| 3541 |
+
"learning_rate": 2.4328859060402687e-06,
|
| 3542 |
+
"loss": 0.0216,
|
| 3543 |
+
"step": 2840
|
| 3544 |
+
},
|
| 3545 |
+
{
|
| 3546 |
+
"epoch": 14.3,
|
| 3547 |
+
"learning_rate": 2.348993288590604e-06,
|
| 3548 |
+
"loss": 0.0366,
|
| 3549 |
+
"step": 2845
|
| 3550 |
+
},
|
| 3551 |
+
{
|
| 3552 |
+
"epoch": 14.32,
|
| 3553 |
+
"learning_rate": 2.2651006711409396e-06,
|
| 3554 |
+
"loss": 0.0068,
|
| 3555 |
+
"step": 2850
|
| 3556 |
+
},
|
| 3557 |
+
{
|
| 3558 |
+
"epoch": 14.35,
|
| 3559 |
+
"learning_rate": 2.1812080536912753e-06,
|
| 3560 |
+
"loss": 0.0009,
|
| 3561 |
+
"step": 2855
|
| 3562 |
+
},
|
| 3563 |
+
{
|
| 3564 |
+
"epoch": 14.37,
|
| 3565 |
+
"learning_rate": 2.097315436241611e-06,
|
| 3566 |
+
"loss": 0.0276,
|
| 3567 |
+
"step": 2860
|
| 3568 |
+
},
|
| 3569 |
+
{
|
| 3570 |
+
"epoch": 14.4,
|
| 3571 |
+
"learning_rate": 2.013422818791946e-06,
|
| 3572 |
+
"loss": 0.006,
|
| 3573 |
+
"step": 2865
|
| 3574 |
+
},
|
| 3575 |
+
{
|
| 3576 |
+
"epoch": 14.42,
|
| 3577 |
+
"learning_rate": 1.929530201342282e-06,
|
| 3578 |
+
"loss": 0.0171,
|
| 3579 |
+
"step": 2870
|
| 3580 |
+
},
|
| 3581 |
+
{
|
| 3582 |
+
"epoch": 14.45,
|
| 3583 |
+
"learning_rate": 1.8456375838926177e-06,
|
| 3584 |
+
"loss": 0.0434,
|
| 3585 |
+
"step": 2875
|
| 3586 |
+
},
|
| 3587 |
+
{
|
| 3588 |
+
"epoch": 14.47,
|
| 3589 |
+
"learning_rate": 1.7617449664429532e-06,
|
| 3590 |
+
"loss": 0.0382,
|
| 3591 |
+
"step": 2880
|
| 3592 |
+
},
|
| 3593 |
+
{
|
| 3594 |
+
"epoch": 14.5,
|
| 3595 |
+
"learning_rate": 1.6778523489932889e-06,
|
| 3596 |
+
"loss": 0.0023,
|
| 3597 |
+
"step": 2885
|
| 3598 |
+
},
|
| 3599 |
+
{
|
| 3600 |
+
"epoch": 14.52,
|
| 3601 |
+
"learning_rate": 1.5939597315436243e-06,
|
| 3602 |
+
"loss": 0.0239,
|
| 3603 |
+
"step": 2890
|
| 3604 |
+
},
|
| 3605 |
+
{
|
| 3606 |
+
"epoch": 14.55,
|
| 3607 |
+
"learning_rate": 1.51006711409396e-06,
|
| 3608 |
+
"loss": 0.0215,
|
| 3609 |
+
"step": 2895
|
| 3610 |
+
},
|
| 3611 |
+
{
|
| 3612 |
+
"epoch": 14.57,
|
| 3613 |
+
"learning_rate": 1.4261744966442952e-06,
|
| 3614 |
+
"loss": 0.0037,
|
| 3615 |
+
"step": 2900
|
| 3616 |
+
},
|
| 3617 |
+
{
|
| 3618 |
+
"epoch": 14.6,
|
| 3619 |
+
"learning_rate": 1.342281879194631e-06,
|
| 3620 |
+
"loss": 0.0601,
|
| 3621 |
+
"step": 2905
|
| 3622 |
+
},
|
| 3623 |
+
{
|
| 3624 |
+
"epoch": 14.62,
|
| 3625 |
+
"learning_rate": 1.2583892617449667e-06,
|
| 3626 |
+
"loss": 0.0113,
|
| 3627 |
+
"step": 2910
|
| 3628 |
+
},
|
| 3629 |
+
{
|
| 3630 |
+
"epoch": 14.65,
|
| 3631 |
+
"learning_rate": 1.174496644295302e-06,
|
| 3632 |
+
"loss": 0.0086,
|
| 3633 |
+
"step": 2915
|
| 3634 |
+
},
|
| 3635 |
+
{
|
| 3636 |
+
"epoch": 14.67,
|
| 3637 |
+
"learning_rate": 1.0906040268456377e-06,
|
| 3638 |
+
"loss": 0.003,
|
| 3639 |
+
"step": 2920
|
| 3640 |
+
},
|
| 3641 |
+
{
|
| 3642 |
+
"epoch": 14.7,
|
| 3643 |
+
"learning_rate": 1.006711409395973e-06,
|
| 3644 |
+
"loss": 0.0153,
|
| 3645 |
+
"step": 2925
|
| 3646 |
+
},
|
| 3647 |
+
{
|
| 3648 |
+
"epoch": 14.72,
|
| 3649 |
+
"learning_rate": 9.228187919463089e-07,
|
| 3650 |
+
"loss": 0.0414,
|
| 3651 |
+
"step": 2930
|
| 3652 |
+
},
|
| 3653 |
+
{
|
| 3654 |
+
"epoch": 14.75,
|
| 3655 |
+
"learning_rate": 8.389261744966444e-07,
|
| 3656 |
+
"loss": 0.0003,
|
| 3657 |
+
"step": 2935
|
| 3658 |
+
},
|
| 3659 |
+
{
|
| 3660 |
+
"epoch": 14.77,
|
| 3661 |
+
"learning_rate": 7.5503355704698e-07,
|
| 3662 |
+
"loss": 0.0538,
|
| 3663 |
+
"step": 2940
|
| 3664 |
+
},
|
| 3665 |
+
{
|
| 3666 |
+
"epoch": 14.8,
|
| 3667 |
+
"learning_rate": 6.711409395973155e-07,
|
| 3668 |
+
"loss": 0.1051,
|
| 3669 |
+
"step": 2945
|
| 3670 |
+
},
|
| 3671 |
+
{
|
| 3672 |
+
"epoch": 14.82,
|
| 3673 |
+
"learning_rate": 5.87248322147651e-07,
|
| 3674 |
+
"loss": 0.03,
|
| 3675 |
+
"step": 2950
|
| 3676 |
+
},
|
| 3677 |
+
{
|
| 3678 |
+
"epoch": 14.85,
|
| 3679 |
+
"learning_rate": 5.033557046979866e-07,
|
| 3680 |
+
"loss": 0.0447,
|
| 3681 |
+
"step": 2955
|
| 3682 |
+
},
|
| 3683 |
+
{
|
| 3684 |
+
"epoch": 14.87,
|
| 3685 |
+
"learning_rate": 4.194630872483222e-07,
|
| 3686 |
+
"loss": 0.0379,
|
| 3687 |
+
"step": 2960
|
| 3688 |
+
},
|
| 3689 |
+
{
|
| 3690 |
+
"epoch": 14.9,
|
| 3691 |
+
"learning_rate": 3.3557046979865777e-07,
|
| 3692 |
+
"loss": 0.0337,
|
| 3693 |
+
"step": 2965
|
| 3694 |
+
},
|
| 3695 |
+
{
|
| 3696 |
+
"epoch": 14.92,
|
| 3697 |
+
"learning_rate": 2.516778523489933e-07,
|
| 3698 |
+
"loss": 0.0071,
|
| 3699 |
+
"step": 2970
|
| 3700 |
+
},
|
| 3701 |
+
{
|
| 3702 |
+
"epoch": 14.95,
|
| 3703 |
+
"learning_rate": 1.6778523489932889e-07,
|
| 3704 |
+
"loss": 0.0266,
|
| 3705 |
+
"step": 2975
|
| 3706 |
+
},
|
| 3707 |
+
{
|
| 3708 |
+
"epoch": 14.97,
|
| 3709 |
+
"learning_rate": 8.389261744966444e-08,
|
| 3710 |
+
"loss": 0.0018,
|
| 3711 |
+
"step": 2980
|
| 3712 |
+
},
|
| 3713 |
+
{
|
| 3714 |
+
"epoch": 15.0,
|
| 3715 |
+
"learning_rate": 0.0,
|
| 3716 |
+
"loss": 0.0063,
|
| 3717 |
+
"step": 2985
|
| 3718 |
+
},
|
| 3719 |
+
{
|
| 3720 |
+
"epoch": 15.0,
|
| 3721 |
+
"eval_accuracy": 0.9934024505183789,
|
| 3722 |
+
"eval_loss": 0.029780708253383636,
|
| 3723 |
+
"eval_runtime": 93.6809,
|
| 3724 |
+
"eval_samples_per_second": 22.651,
|
| 3725 |
+
"eval_steps_per_second": 1.42,
|
| 3726 |
+
"step": 2985
|
| 3727 |
+
},
|
| 3728 |
+
{
|
| 3729 |
+
"epoch": 15.0,
|
| 3730 |
+
"step": 2985,
|
| 3731 |
+
"total_flos": 2.615871779941884e+19,
|
| 3732 |
+
"train_loss": 0.02590897159305381,
|
| 3733 |
+
"train_runtime": 7346.925,
|
| 3734 |
+
"train_samples_per_second": 12.993,
|
| 3735 |
+
"train_steps_per_second": 0.406
|
| 3736 |
+
}
|
| 3737 |
+
],
|
| 3738 |
+
"logging_steps": 5,
|
| 3739 |
+
"max_steps": 2985,
|
| 3740 |
+
"num_train_epochs": 15,
|
| 3741 |
+
"save_steps": 500,
|
| 3742 |
+
"total_flos": 2.615871779941884e+19,
|
| 3743 |
+
"trial_name": null,
|
| 3744 |
+
"trial_params": null
|
| 3745 |
+
}
|