End of training
Browse files- all_results.json +5 -5
- train_results.json +5 -5
- trainer_state.json +3014 -5
all_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 100,
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second":
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 1000.0,
|
| 3 |
+
"train_loss": 0.655489089525663,
|
| 4 |
+
"train_runtime": 67792.7328,
|
| 5 |
"train_samples": 100,
|
| 6 |
+
"train_samples_per_second": 1.475,
|
| 7 |
+
"train_steps_per_second": 0.192
|
| 8 |
}
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 100,
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second":
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 1000.0,
|
| 3 |
+
"train_loss": 0.655489089525663,
|
| 4 |
+
"train_runtime": 67792.7328,
|
| 5 |
"train_samples": 100,
|
| 6 |
+
"train_samples_per_second": 1.475,
|
| 7 |
+
"train_steps_per_second": 0.192
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -3025,13 +3025,3022 @@
|
|
| 3025 |
"train_runtime": 1.3574,
|
| 3026 |
"train_samples_per_second": 36834.222,
|
| 3027 |
"train_steps_per_second": 4788.449
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3028 |
}
|
| 3029 |
],
|
| 3030 |
"logging_steps": 500,
|
| 3031 |
-
"max_steps":
|
| 3032 |
-
"num_train_epochs":
|
| 3033 |
"save_steps": 500,
|
| 3034 |
-
"total_flos":
|
| 3035 |
"trial_name": null,
|
| 3036 |
"trial_params": null
|
| 3037 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1000.0,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 13000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 3025 |
"train_runtime": 1.3574,
|
| 3026 |
"train_samples_per_second": 36834.222,
|
| 3027 |
"train_steps_per_second": 4788.449
|
| 3028 |
+
},
|
| 3029 |
+
{
|
| 3030 |
+
"epoch": 501.0,
|
| 3031 |
+
"learning_rate": 0.003996,
|
| 3032 |
+
"loss": 1.4863,
|
| 3033 |
+
"step": 6513
|
| 3034 |
+
},
|
| 3035 |
+
{
|
| 3036 |
+
"epoch": 502.0,
|
| 3037 |
+
"learning_rate": 0.003992,
|
| 3038 |
+
"loss": 1.6645,
|
| 3039 |
+
"step": 6526
|
| 3040 |
+
},
|
| 3041 |
+
{
|
| 3042 |
+
"epoch": 503.0,
|
| 3043 |
+
"learning_rate": 0.003988,
|
| 3044 |
+
"loss": 1.7012,
|
| 3045 |
+
"step": 6539
|
| 3046 |
+
},
|
| 3047 |
+
{
|
| 3048 |
+
"epoch": 504.0,
|
| 3049 |
+
"learning_rate": 0.003984,
|
| 3050 |
+
"loss": 1.6185,
|
| 3051 |
+
"step": 6552
|
| 3052 |
+
},
|
| 3053 |
+
{
|
| 3054 |
+
"epoch": 505.0,
|
| 3055 |
+
"learning_rate": 0.00398,
|
| 3056 |
+
"loss": 1.5629,
|
| 3057 |
+
"step": 6565
|
| 3058 |
+
},
|
| 3059 |
+
{
|
| 3060 |
+
"epoch": 506.0,
|
| 3061 |
+
"learning_rate": 0.003976,
|
| 3062 |
+
"loss": 1.5867,
|
| 3063 |
+
"step": 6578
|
| 3064 |
+
},
|
| 3065 |
+
{
|
| 3066 |
+
"epoch": 507.0,
|
| 3067 |
+
"learning_rate": 0.003972,
|
| 3068 |
+
"loss": 1.6144,
|
| 3069 |
+
"step": 6591
|
| 3070 |
+
},
|
| 3071 |
+
{
|
| 3072 |
+
"epoch": 508.0,
|
| 3073 |
+
"learning_rate": 0.003968,
|
| 3074 |
+
"loss": 1.7844,
|
| 3075 |
+
"step": 6604
|
| 3076 |
+
},
|
| 3077 |
+
{
|
| 3078 |
+
"epoch": 509.0,
|
| 3079 |
+
"learning_rate": 0.003964,
|
| 3080 |
+
"loss": 1.7508,
|
| 3081 |
+
"step": 6617
|
| 3082 |
+
},
|
| 3083 |
+
{
|
| 3084 |
+
"epoch": 510.0,
|
| 3085 |
+
"learning_rate": 0.00396,
|
| 3086 |
+
"loss": 1.7693,
|
| 3087 |
+
"step": 6630
|
| 3088 |
+
},
|
| 3089 |
+
{
|
| 3090 |
+
"epoch": 511.0,
|
| 3091 |
+
"learning_rate": 0.003956,
|
| 3092 |
+
"loss": 1.8884,
|
| 3093 |
+
"step": 6643
|
| 3094 |
+
},
|
| 3095 |
+
{
|
| 3096 |
+
"epoch": 512.0,
|
| 3097 |
+
"learning_rate": 0.003952,
|
| 3098 |
+
"loss": 1.8287,
|
| 3099 |
+
"step": 6656
|
| 3100 |
+
},
|
| 3101 |
+
{
|
| 3102 |
+
"epoch": 513.0,
|
| 3103 |
+
"learning_rate": 0.003948,
|
| 3104 |
+
"loss": 1.8228,
|
| 3105 |
+
"step": 6669
|
| 3106 |
+
},
|
| 3107 |
+
{
|
| 3108 |
+
"epoch": 514.0,
|
| 3109 |
+
"learning_rate": 0.0039440000000000005,
|
| 3110 |
+
"loss": 1.7632,
|
| 3111 |
+
"step": 6682
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 515.0,
|
| 3115 |
+
"learning_rate": 0.00394,
|
| 3116 |
+
"loss": 1.7943,
|
| 3117 |
+
"step": 6695
|
| 3118 |
+
},
|
| 3119 |
+
{
|
| 3120 |
+
"epoch": 516.0,
|
| 3121 |
+
"learning_rate": 0.003936,
|
| 3122 |
+
"loss": 1.7451,
|
| 3123 |
+
"step": 6708
|
| 3124 |
+
},
|
| 3125 |
+
{
|
| 3126 |
+
"epoch": 517.0,
|
| 3127 |
+
"learning_rate": 0.003932,
|
| 3128 |
+
"loss": 1.8542,
|
| 3129 |
+
"step": 6721
|
| 3130 |
+
},
|
| 3131 |
+
{
|
| 3132 |
+
"epoch": 518.0,
|
| 3133 |
+
"learning_rate": 0.003928,
|
| 3134 |
+
"loss": 2.0283,
|
| 3135 |
+
"step": 6734
|
| 3136 |
+
},
|
| 3137 |
+
{
|
| 3138 |
+
"epoch": 519.0,
|
| 3139 |
+
"learning_rate": 0.003924,
|
| 3140 |
+
"loss": 2.0074,
|
| 3141 |
+
"step": 6747
|
| 3142 |
+
},
|
| 3143 |
+
{
|
| 3144 |
+
"epoch": 520.0,
|
| 3145 |
+
"learning_rate": 0.00392,
|
| 3146 |
+
"loss": 2.1644,
|
| 3147 |
+
"step": 6760
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 521.0,
|
| 3151 |
+
"learning_rate": 0.003916,
|
| 3152 |
+
"loss": 1.9558,
|
| 3153 |
+
"step": 6773
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 522.0,
|
| 3157 |
+
"learning_rate": 0.003912,
|
| 3158 |
+
"loss": 1.9104,
|
| 3159 |
+
"step": 6786
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 523.0,
|
| 3163 |
+
"learning_rate": 0.003908,
|
| 3164 |
+
"loss": 1.9961,
|
| 3165 |
+
"step": 6799
|
| 3166 |
+
},
|
| 3167 |
+
{
|
| 3168 |
+
"epoch": 524.0,
|
| 3169 |
+
"learning_rate": 0.003904,
|
| 3170 |
+
"loss": 2.0827,
|
| 3171 |
+
"step": 6812
|
| 3172 |
+
},
|
| 3173 |
+
{
|
| 3174 |
+
"epoch": 525.0,
|
| 3175 |
+
"learning_rate": 0.0039,
|
| 3176 |
+
"loss": 2.0293,
|
| 3177 |
+
"step": 6825
|
| 3178 |
+
},
|
| 3179 |
+
{
|
| 3180 |
+
"epoch": 526.0,
|
| 3181 |
+
"learning_rate": 0.003896,
|
| 3182 |
+
"loss": 1.9904,
|
| 3183 |
+
"step": 6838
|
| 3184 |
+
},
|
| 3185 |
+
{
|
| 3186 |
+
"epoch": 527.0,
|
| 3187 |
+
"learning_rate": 0.003892,
|
| 3188 |
+
"loss": 1.9175,
|
| 3189 |
+
"step": 6851
|
| 3190 |
+
},
|
| 3191 |
+
{
|
| 3192 |
+
"epoch": 528.0,
|
| 3193 |
+
"learning_rate": 0.003888,
|
| 3194 |
+
"loss": 1.8658,
|
| 3195 |
+
"step": 6864
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 529.0,
|
| 3199 |
+
"learning_rate": 0.003884,
|
| 3200 |
+
"loss": 1.8219,
|
| 3201 |
+
"step": 6877
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 530.0,
|
| 3205 |
+
"learning_rate": 0.0038799999999999998,
|
| 3206 |
+
"loss": 1.884,
|
| 3207 |
+
"step": 6890
|
| 3208 |
+
},
|
| 3209 |
+
{
|
| 3210 |
+
"epoch": 531.0,
|
| 3211 |
+
"learning_rate": 0.003876,
|
| 3212 |
+
"loss": 1.9361,
|
| 3213 |
+
"step": 6903
|
| 3214 |
+
},
|
| 3215 |
+
{
|
| 3216 |
+
"epoch": 532.0,
|
| 3217 |
+
"learning_rate": 0.003872,
|
| 3218 |
+
"loss": 1.8961,
|
| 3219 |
+
"step": 6916
|
| 3220 |
+
},
|
| 3221 |
+
{
|
| 3222 |
+
"epoch": 533.0,
|
| 3223 |
+
"learning_rate": 0.003868,
|
| 3224 |
+
"loss": 1.9082,
|
| 3225 |
+
"step": 6929
|
| 3226 |
+
},
|
| 3227 |
+
{
|
| 3228 |
+
"epoch": 534.0,
|
| 3229 |
+
"learning_rate": 0.003864,
|
| 3230 |
+
"loss": 2.0034,
|
| 3231 |
+
"step": 6942
|
| 3232 |
+
},
|
| 3233 |
+
{
|
| 3234 |
+
"epoch": 535.0,
|
| 3235 |
+
"learning_rate": 0.00386,
|
| 3236 |
+
"loss": 2.0058,
|
| 3237 |
+
"step": 6955
|
| 3238 |
+
},
|
| 3239 |
+
{
|
| 3240 |
+
"epoch": 536.0,
|
| 3241 |
+
"learning_rate": 0.003856,
|
| 3242 |
+
"loss": 1.9934,
|
| 3243 |
+
"step": 6968
|
| 3244 |
+
},
|
| 3245 |
+
{
|
| 3246 |
+
"epoch": 537.0,
|
| 3247 |
+
"learning_rate": 0.003852,
|
| 3248 |
+
"loss": 1.9674,
|
| 3249 |
+
"step": 6981
|
| 3250 |
+
},
|
| 3251 |
+
{
|
| 3252 |
+
"epoch": 538.0,
|
| 3253 |
+
"learning_rate": 0.003848,
|
| 3254 |
+
"loss": 1.9737,
|
| 3255 |
+
"step": 6994
|
| 3256 |
+
},
|
| 3257 |
+
{
|
| 3258 |
+
"epoch": 539.0,
|
| 3259 |
+
"learning_rate": 0.0038439999999999998,
|
| 3260 |
+
"loss": 1.9184,
|
| 3261 |
+
"step": 7007
|
| 3262 |
+
},
|
| 3263 |
+
{
|
| 3264 |
+
"epoch": 540.0,
|
| 3265 |
+
"learning_rate": 0.00384,
|
| 3266 |
+
"loss": 1.9147,
|
| 3267 |
+
"step": 7020
|
| 3268 |
+
},
|
| 3269 |
+
{
|
| 3270 |
+
"epoch": 541.0,
|
| 3271 |
+
"learning_rate": 0.003836,
|
| 3272 |
+
"loss": 1.9792,
|
| 3273 |
+
"step": 7033
|
| 3274 |
+
},
|
| 3275 |
+
{
|
| 3276 |
+
"epoch": 542.0,
|
| 3277 |
+
"learning_rate": 0.003832,
|
| 3278 |
+
"loss": 1.9448,
|
| 3279 |
+
"step": 7046
|
| 3280 |
+
},
|
| 3281 |
+
{
|
| 3282 |
+
"epoch": 543.0,
|
| 3283 |
+
"learning_rate": 0.003828,
|
| 3284 |
+
"loss": 1.8897,
|
| 3285 |
+
"step": 7059
|
| 3286 |
+
},
|
| 3287 |
+
{
|
| 3288 |
+
"epoch": 544.0,
|
| 3289 |
+
"learning_rate": 0.0038239999999999997,
|
| 3290 |
+
"loss": 1.9048,
|
| 3291 |
+
"step": 7072
|
| 3292 |
+
},
|
| 3293 |
+
{
|
| 3294 |
+
"epoch": 545.0,
|
| 3295 |
+
"learning_rate": 0.00382,
|
| 3296 |
+
"loss": 1.9577,
|
| 3297 |
+
"step": 7085
|
| 3298 |
+
},
|
| 3299 |
+
{
|
| 3300 |
+
"epoch": 546.0,
|
| 3301 |
+
"learning_rate": 0.003816,
|
| 3302 |
+
"loss": 1.9996,
|
| 3303 |
+
"step": 7098
|
| 3304 |
+
},
|
| 3305 |
+
{
|
| 3306 |
+
"epoch": 547.0,
|
| 3307 |
+
"learning_rate": 0.003812,
|
| 3308 |
+
"loss": 1.9895,
|
| 3309 |
+
"step": 7111
|
| 3310 |
+
},
|
| 3311 |
+
{
|
| 3312 |
+
"epoch": 548.0,
|
| 3313 |
+
"learning_rate": 0.0038079999999999998,
|
| 3314 |
+
"loss": 2.0381,
|
| 3315 |
+
"step": 7124
|
| 3316 |
+
},
|
| 3317 |
+
{
|
| 3318 |
+
"epoch": 549.0,
|
| 3319 |
+
"learning_rate": 0.003804,
|
| 3320 |
+
"loss": 1.9362,
|
| 3321 |
+
"step": 7137
|
| 3322 |
+
},
|
| 3323 |
+
{
|
| 3324 |
+
"epoch": 550.0,
|
| 3325 |
+
"learning_rate": 0.0038,
|
| 3326 |
+
"loss": 1.9544,
|
| 3327 |
+
"step": 7150
|
| 3328 |
+
},
|
| 3329 |
+
{
|
| 3330 |
+
"epoch": 551.0,
|
| 3331 |
+
"learning_rate": 0.003796,
|
| 3332 |
+
"loss": 1.9497,
|
| 3333 |
+
"step": 7163
|
| 3334 |
+
},
|
| 3335 |
+
{
|
| 3336 |
+
"epoch": 552.0,
|
| 3337 |
+
"learning_rate": 0.003792,
|
| 3338 |
+
"loss": 1.9105,
|
| 3339 |
+
"step": 7176
|
| 3340 |
+
},
|
| 3341 |
+
{
|
| 3342 |
+
"epoch": 553.0,
|
| 3343 |
+
"learning_rate": 0.0037879999999999997,
|
| 3344 |
+
"loss": 1.896,
|
| 3345 |
+
"step": 7189
|
| 3346 |
+
},
|
| 3347 |
+
{
|
| 3348 |
+
"epoch": 554.0,
|
| 3349 |
+
"learning_rate": 0.003784,
|
| 3350 |
+
"loss": 2.0015,
|
| 3351 |
+
"step": 7202
|
| 3352 |
+
},
|
| 3353 |
+
{
|
| 3354 |
+
"epoch": 555.0,
|
| 3355 |
+
"learning_rate": 0.00378,
|
| 3356 |
+
"loss": 1.8753,
|
| 3357 |
+
"step": 7215
|
| 3358 |
+
},
|
| 3359 |
+
{
|
| 3360 |
+
"epoch": 556.0,
|
| 3361 |
+
"learning_rate": 0.003776,
|
| 3362 |
+
"loss": 1.9087,
|
| 3363 |
+
"step": 7228
|
| 3364 |
+
},
|
| 3365 |
+
{
|
| 3366 |
+
"epoch": 557.0,
|
| 3367 |
+
"learning_rate": 0.0037719999999999997,
|
| 3368 |
+
"loss": 1.959,
|
| 3369 |
+
"step": 7241
|
| 3370 |
+
},
|
| 3371 |
+
{
|
| 3372 |
+
"epoch": 558.0,
|
| 3373 |
+
"learning_rate": 0.003768,
|
| 3374 |
+
"loss": 2.0541,
|
| 3375 |
+
"step": 7254
|
| 3376 |
+
},
|
| 3377 |
+
{
|
| 3378 |
+
"epoch": 559.0,
|
| 3379 |
+
"learning_rate": 0.003764,
|
| 3380 |
+
"loss": 2.0614,
|
| 3381 |
+
"step": 7267
|
| 3382 |
+
},
|
| 3383 |
+
{
|
| 3384 |
+
"epoch": 560.0,
|
| 3385 |
+
"learning_rate": 0.00376,
|
| 3386 |
+
"loss": 2.0349,
|
| 3387 |
+
"step": 7280
|
| 3388 |
+
},
|
| 3389 |
+
{
|
| 3390 |
+
"epoch": 561.0,
|
| 3391 |
+
"learning_rate": 0.003756,
|
| 3392 |
+
"loss": 1.9517,
|
| 3393 |
+
"step": 7293
|
| 3394 |
+
},
|
| 3395 |
+
{
|
| 3396 |
+
"epoch": 562.0,
|
| 3397 |
+
"learning_rate": 0.0037519999999999997,
|
| 3398 |
+
"loss": 2.0094,
|
| 3399 |
+
"step": 7306
|
| 3400 |
+
},
|
| 3401 |
+
{
|
| 3402 |
+
"epoch": 563.0,
|
| 3403 |
+
"learning_rate": 0.0037480000000000005,
|
| 3404 |
+
"loss": 2.0459,
|
| 3405 |
+
"step": 7319
|
| 3406 |
+
},
|
| 3407 |
+
{
|
| 3408 |
+
"epoch": 564.0,
|
| 3409 |
+
"learning_rate": 0.0037440000000000004,
|
| 3410 |
+
"loss": 2.172,
|
| 3411 |
+
"step": 7332
|
| 3412 |
+
},
|
| 3413 |
+
{
|
| 3414 |
+
"epoch": 565.0,
|
| 3415 |
+
"learning_rate": 0.0037400000000000003,
|
| 3416 |
+
"loss": 2.0798,
|
| 3417 |
+
"step": 7345
|
| 3418 |
+
},
|
| 3419 |
+
{
|
| 3420 |
+
"epoch": 566.0,
|
| 3421 |
+
"learning_rate": 0.003736,
|
| 3422 |
+
"loss": 2.0222,
|
| 3423 |
+
"step": 7358
|
| 3424 |
+
},
|
| 3425 |
+
{
|
| 3426 |
+
"epoch": 567.0,
|
| 3427 |
+
"learning_rate": 0.003732,
|
| 3428 |
+
"loss": 2.0051,
|
| 3429 |
+
"step": 7371
|
| 3430 |
+
},
|
| 3431 |
+
{
|
| 3432 |
+
"epoch": 568.0,
|
| 3433 |
+
"learning_rate": 0.0037280000000000004,
|
| 3434 |
+
"loss": 2.2096,
|
| 3435 |
+
"step": 7384
|
| 3436 |
+
},
|
| 3437 |
+
{
|
| 3438 |
+
"epoch": 569.0,
|
| 3439 |
+
"learning_rate": 0.0037240000000000003,
|
| 3440 |
+
"loss": 2.2197,
|
| 3441 |
+
"step": 7397
|
| 3442 |
+
},
|
| 3443 |
+
{
|
| 3444 |
+
"epoch": 570.0,
|
| 3445 |
+
"learning_rate": 0.00372,
|
| 3446 |
+
"loss": 2.1259,
|
| 3447 |
+
"step": 7410
|
| 3448 |
+
},
|
| 3449 |
+
{
|
| 3450 |
+
"epoch": 571.0,
|
| 3451 |
+
"learning_rate": 0.003716,
|
| 3452 |
+
"loss": 2.1098,
|
| 3453 |
+
"step": 7423
|
| 3454 |
+
},
|
| 3455 |
+
{
|
| 3456 |
+
"epoch": 572.0,
|
| 3457 |
+
"learning_rate": 0.0037120000000000005,
|
| 3458 |
+
"loss": 2.0734,
|
| 3459 |
+
"step": 7436
|
| 3460 |
+
},
|
| 3461 |
+
{
|
| 3462 |
+
"epoch": 573.0,
|
| 3463 |
+
"learning_rate": 0.0037080000000000004,
|
| 3464 |
+
"loss": 2.0822,
|
| 3465 |
+
"step": 7449
|
| 3466 |
+
},
|
| 3467 |
+
{
|
| 3468 |
+
"epoch": 574.0,
|
| 3469 |
+
"learning_rate": 0.0037040000000000003,
|
| 3470 |
+
"loss": 2.0771,
|
| 3471 |
+
"step": 7462
|
| 3472 |
+
},
|
| 3473 |
+
{
|
| 3474 |
+
"epoch": 575.0,
|
| 3475 |
+
"learning_rate": 0.0037,
|
| 3476 |
+
"loss": 2.0217,
|
| 3477 |
+
"step": 7475
|
| 3478 |
+
},
|
| 3479 |
+
{
|
| 3480 |
+
"epoch": 576.0,
|
| 3481 |
+
"learning_rate": 0.003696,
|
| 3482 |
+
"loss": 1.9762,
|
| 3483 |
+
"step": 7488
|
| 3484 |
+
},
|
| 3485 |
+
{
|
| 3486 |
+
"epoch": 577.0,
|
| 3487 |
+
"learning_rate": 0.0036920000000000004,
|
| 3488 |
+
"loss": 1.9341,
|
| 3489 |
+
"step": 7501
|
| 3490 |
+
},
|
| 3491 |
+
{
|
| 3492 |
+
"epoch": 578.0,
|
| 3493 |
+
"learning_rate": 0.0036880000000000003,
|
| 3494 |
+
"loss": 1.9837,
|
| 3495 |
+
"step": 7514
|
| 3496 |
+
},
|
| 3497 |
+
{
|
| 3498 |
+
"epoch": 579.0,
|
| 3499 |
+
"learning_rate": 0.003684,
|
| 3500 |
+
"loss": 1.9337,
|
| 3501 |
+
"step": 7527
|
| 3502 |
+
},
|
| 3503 |
+
{
|
| 3504 |
+
"epoch": 580.0,
|
| 3505 |
+
"learning_rate": 0.00368,
|
| 3506 |
+
"loss": 1.8968,
|
| 3507 |
+
"step": 7540
|
| 3508 |
+
},
|
| 3509 |
+
{
|
| 3510 |
+
"epoch": 581.0,
|
| 3511 |
+
"learning_rate": 0.0036760000000000004,
|
| 3512 |
+
"loss": 1.8705,
|
| 3513 |
+
"step": 7553
|
| 3514 |
+
},
|
| 3515 |
+
{
|
| 3516 |
+
"epoch": 582.0,
|
| 3517 |
+
"learning_rate": 0.0036720000000000004,
|
| 3518 |
+
"loss": 1.8261,
|
| 3519 |
+
"step": 7566
|
| 3520 |
+
},
|
| 3521 |
+
{
|
| 3522 |
+
"epoch": 583.0,
|
| 3523 |
+
"learning_rate": 0.0036680000000000003,
|
| 3524 |
+
"loss": 1.9411,
|
| 3525 |
+
"step": 7579
|
| 3526 |
+
},
|
| 3527 |
+
{
|
| 3528 |
+
"epoch": 584.0,
|
| 3529 |
+
"learning_rate": 0.003664,
|
| 3530 |
+
"loss": 1.9961,
|
| 3531 |
+
"step": 7592
|
| 3532 |
+
},
|
| 3533 |
+
{
|
| 3534 |
+
"epoch": 585.0,
|
| 3535 |
+
"learning_rate": 0.00366,
|
| 3536 |
+
"loss": 1.8865,
|
| 3537 |
+
"step": 7605
|
| 3538 |
+
},
|
| 3539 |
+
{
|
| 3540 |
+
"epoch": 586.0,
|
| 3541 |
+
"learning_rate": 0.0036560000000000004,
|
| 3542 |
+
"loss": 1.829,
|
| 3543 |
+
"step": 7618
|
| 3544 |
+
},
|
| 3545 |
+
{
|
| 3546 |
+
"epoch": 587.0,
|
| 3547 |
+
"learning_rate": 0.0036520000000000003,
|
| 3548 |
+
"loss": 1.8424,
|
| 3549 |
+
"step": 7631
|
| 3550 |
+
},
|
| 3551 |
+
{
|
| 3552 |
+
"epoch": 588.0,
|
| 3553 |
+
"learning_rate": 0.003648,
|
| 3554 |
+
"loss": 1.8463,
|
| 3555 |
+
"step": 7644
|
| 3556 |
+
},
|
| 3557 |
+
{
|
| 3558 |
+
"epoch": 589.0,
|
| 3559 |
+
"learning_rate": 0.003644,
|
| 3560 |
+
"loss": 1.8452,
|
| 3561 |
+
"step": 7657
|
| 3562 |
+
},
|
| 3563 |
+
{
|
| 3564 |
+
"epoch": 590.0,
|
| 3565 |
+
"learning_rate": 0.00364,
|
| 3566 |
+
"loss": 1.7974,
|
| 3567 |
+
"step": 7670
|
| 3568 |
+
},
|
| 3569 |
+
{
|
| 3570 |
+
"epoch": 591.0,
|
| 3571 |
+
"learning_rate": 0.0036360000000000003,
|
| 3572 |
+
"loss": 1.7995,
|
| 3573 |
+
"step": 7683
|
| 3574 |
+
},
|
| 3575 |
+
{
|
| 3576 |
+
"epoch": 592.0,
|
| 3577 |
+
"learning_rate": 0.0036320000000000002,
|
| 3578 |
+
"loss": 1.7664,
|
| 3579 |
+
"step": 7696
|
| 3580 |
+
},
|
| 3581 |
+
{
|
| 3582 |
+
"epoch": 593.0,
|
| 3583 |
+
"learning_rate": 0.003628,
|
| 3584 |
+
"loss": 1.7451,
|
| 3585 |
+
"step": 7709
|
| 3586 |
+
},
|
| 3587 |
+
{
|
| 3588 |
+
"epoch": 594.0,
|
| 3589 |
+
"learning_rate": 0.003624,
|
| 3590 |
+
"loss": 1.7978,
|
| 3591 |
+
"step": 7722
|
| 3592 |
+
},
|
| 3593 |
+
{
|
| 3594 |
+
"epoch": 595.0,
|
| 3595 |
+
"learning_rate": 0.0036200000000000004,
|
| 3596 |
+
"loss": 1.9067,
|
| 3597 |
+
"step": 7735
|
| 3598 |
+
},
|
| 3599 |
+
{
|
| 3600 |
+
"epoch": 596.0,
|
| 3601 |
+
"learning_rate": 0.0036160000000000003,
|
| 3602 |
+
"loss": 1.8932,
|
| 3603 |
+
"step": 7748
|
| 3604 |
+
},
|
| 3605 |
+
{
|
| 3606 |
+
"epoch": 597.0,
|
| 3607 |
+
"learning_rate": 0.003612,
|
| 3608 |
+
"loss": 1.9407,
|
| 3609 |
+
"step": 7761
|
| 3610 |
+
},
|
| 3611 |
+
{
|
| 3612 |
+
"epoch": 598.0,
|
| 3613 |
+
"learning_rate": 0.003608,
|
| 3614 |
+
"loss": 1.8776,
|
| 3615 |
+
"step": 7774
|
| 3616 |
+
},
|
| 3617 |
+
{
|
| 3618 |
+
"epoch": 599.0,
|
| 3619 |
+
"learning_rate": 0.003604,
|
| 3620 |
+
"loss": 1.8223,
|
| 3621 |
+
"step": 7787
|
| 3622 |
+
},
|
| 3623 |
+
{
|
| 3624 |
+
"epoch": 600.0,
|
| 3625 |
+
"learning_rate": 0.0036000000000000003,
|
| 3626 |
+
"loss": 1.7761,
|
| 3627 |
+
"step": 7800
|
| 3628 |
+
},
|
| 3629 |
+
{
|
| 3630 |
+
"epoch": 601.0,
|
| 3631 |
+
"learning_rate": 0.0035960000000000002,
|
| 3632 |
+
"loss": 1.7768,
|
| 3633 |
+
"step": 7813
|
| 3634 |
+
},
|
| 3635 |
+
{
|
| 3636 |
+
"epoch": 602.0,
|
| 3637 |
+
"learning_rate": 0.003592,
|
| 3638 |
+
"loss": 1.8109,
|
| 3639 |
+
"step": 7826
|
| 3640 |
+
},
|
| 3641 |
+
{
|
| 3642 |
+
"epoch": 603.0,
|
| 3643 |
+
"learning_rate": 0.003588,
|
| 3644 |
+
"loss": 1.7787,
|
| 3645 |
+
"step": 7839
|
| 3646 |
+
},
|
| 3647 |
+
{
|
| 3648 |
+
"epoch": 604.0,
|
| 3649 |
+
"learning_rate": 0.003584,
|
| 3650 |
+
"loss": 1.9842,
|
| 3651 |
+
"step": 7852
|
| 3652 |
+
},
|
| 3653 |
+
{
|
| 3654 |
+
"epoch": 605.0,
|
| 3655 |
+
"learning_rate": 0.0035800000000000003,
|
| 3656 |
+
"loss": 1.9262,
|
| 3657 |
+
"step": 7865
|
| 3658 |
+
},
|
| 3659 |
+
{
|
| 3660 |
+
"epoch": 606.0,
|
| 3661 |
+
"learning_rate": 0.003576,
|
| 3662 |
+
"loss": 1.9124,
|
| 3663 |
+
"step": 7878
|
| 3664 |
+
},
|
| 3665 |
+
{
|
| 3666 |
+
"epoch": 607.0,
|
| 3667 |
+
"learning_rate": 0.003572,
|
| 3668 |
+
"loss": 1.8407,
|
| 3669 |
+
"step": 7891
|
| 3670 |
+
},
|
| 3671 |
+
{
|
| 3672 |
+
"epoch": 608.0,
|
| 3673 |
+
"learning_rate": 0.003568,
|
| 3674 |
+
"loss": 1.8722,
|
| 3675 |
+
"step": 7904
|
| 3676 |
+
},
|
| 3677 |
+
{
|
| 3678 |
+
"epoch": 609.0,
|
| 3679 |
+
"learning_rate": 0.0035640000000000003,
|
| 3680 |
+
"loss": 1.7409,
|
| 3681 |
+
"step": 7917
|
| 3682 |
+
},
|
| 3683 |
+
{
|
| 3684 |
+
"epoch": 610.0,
|
| 3685 |
+
"learning_rate": 0.0035600000000000002,
|
| 3686 |
+
"loss": 1.712,
|
| 3687 |
+
"step": 7930
|
| 3688 |
+
},
|
| 3689 |
+
{
|
| 3690 |
+
"epoch": 611.0,
|
| 3691 |
+
"learning_rate": 0.003556,
|
| 3692 |
+
"loss": 1.6115,
|
| 3693 |
+
"step": 7943
|
| 3694 |
+
},
|
| 3695 |
+
{
|
| 3696 |
+
"epoch": 612.0,
|
| 3697 |
+
"learning_rate": 0.003552,
|
| 3698 |
+
"loss": 1.6805,
|
| 3699 |
+
"step": 7956
|
| 3700 |
+
},
|
| 3701 |
+
{
|
| 3702 |
+
"epoch": 613.0,
|
| 3703 |
+
"learning_rate": 0.003548,
|
| 3704 |
+
"loss": 1.7829,
|
| 3705 |
+
"step": 7969
|
| 3706 |
+
},
|
| 3707 |
+
{
|
| 3708 |
+
"epoch": 614.0,
|
| 3709 |
+
"learning_rate": 0.0035440000000000003,
|
| 3710 |
+
"loss": 1.7498,
|
| 3711 |
+
"step": 7982
|
| 3712 |
+
},
|
| 3713 |
+
{
|
| 3714 |
+
"epoch": 615.0,
|
| 3715 |
+
"learning_rate": 0.00354,
|
| 3716 |
+
"loss": 1.7536,
|
| 3717 |
+
"step": 7995
|
| 3718 |
+
},
|
| 3719 |
+
{
|
| 3720 |
+
"epoch": 616.0,
|
| 3721 |
+
"learning_rate": 0.003536,
|
| 3722 |
+
"loss": 1.7015,
|
| 3723 |
+
"step": 8008
|
| 3724 |
+
},
|
| 3725 |
+
{
|
| 3726 |
+
"epoch": 617.0,
|
| 3727 |
+
"learning_rate": 0.003532,
|
| 3728 |
+
"loss": 1.6556,
|
| 3729 |
+
"step": 8021
|
| 3730 |
+
},
|
| 3731 |
+
{
|
| 3732 |
+
"epoch": 618.0,
|
| 3733 |
+
"learning_rate": 0.003528,
|
| 3734 |
+
"loss": 1.7314,
|
| 3735 |
+
"step": 8034
|
| 3736 |
+
},
|
| 3737 |
+
{
|
| 3738 |
+
"epoch": 619.0,
|
| 3739 |
+
"learning_rate": 0.0035240000000000002,
|
| 3740 |
+
"loss": 1.6996,
|
| 3741 |
+
"step": 8047
|
| 3742 |
+
},
|
| 3743 |
+
{
|
| 3744 |
+
"epoch": 620.0,
|
| 3745 |
+
"learning_rate": 0.00352,
|
| 3746 |
+
"loss": 1.6819,
|
| 3747 |
+
"step": 8060
|
| 3748 |
+
},
|
| 3749 |
+
{
|
| 3750 |
+
"epoch": 621.0,
|
| 3751 |
+
"learning_rate": 0.003516,
|
| 3752 |
+
"loss": 1.6994,
|
| 3753 |
+
"step": 8073
|
| 3754 |
+
},
|
| 3755 |
+
{
|
| 3756 |
+
"epoch": 622.0,
|
| 3757 |
+
"learning_rate": 0.003512,
|
| 3758 |
+
"loss": 1.6657,
|
| 3759 |
+
"step": 8086
|
| 3760 |
+
},
|
| 3761 |
+
{
|
| 3762 |
+
"epoch": 623.0,
|
| 3763 |
+
"learning_rate": 0.0035080000000000003,
|
| 3764 |
+
"loss": 1.6558,
|
| 3765 |
+
"step": 8099
|
| 3766 |
+
},
|
| 3767 |
+
{
|
| 3768 |
+
"epoch": 624.0,
|
| 3769 |
+
"learning_rate": 0.003504,
|
| 3770 |
+
"loss": 1.6822,
|
| 3771 |
+
"step": 8112
|
| 3772 |
+
},
|
| 3773 |
+
{
|
| 3774 |
+
"epoch": 625.0,
|
| 3775 |
+
"learning_rate": 0.0035,
|
| 3776 |
+
"loss": 1.7245,
|
| 3777 |
+
"step": 8125
|
| 3778 |
+
},
|
| 3779 |
+
{
|
| 3780 |
+
"epoch": 626.0,
|
| 3781 |
+
"learning_rate": 0.003496,
|
| 3782 |
+
"loss": 1.8045,
|
| 3783 |
+
"step": 8138
|
| 3784 |
+
},
|
| 3785 |
+
{
|
| 3786 |
+
"epoch": 627.0,
|
| 3787 |
+
"learning_rate": 0.003492,
|
| 3788 |
+
"loss": 1.7307,
|
| 3789 |
+
"step": 8151
|
| 3790 |
+
},
|
| 3791 |
+
{
|
| 3792 |
+
"epoch": 628.0,
|
| 3793 |
+
"learning_rate": 0.003488,
|
| 3794 |
+
"loss": 1.7469,
|
| 3795 |
+
"step": 8164
|
| 3796 |
+
},
|
| 3797 |
+
{
|
| 3798 |
+
"epoch": 629.0,
|
| 3799 |
+
"learning_rate": 0.003484,
|
| 3800 |
+
"loss": 1.7047,
|
| 3801 |
+
"step": 8177
|
| 3802 |
+
},
|
| 3803 |
+
{
|
| 3804 |
+
"epoch": 630.0,
|
| 3805 |
+
"learning_rate": 0.00348,
|
| 3806 |
+
"loss": 1.6359,
|
| 3807 |
+
"step": 8190
|
| 3808 |
+
},
|
| 3809 |
+
{
|
| 3810 |
+
"epoch": 631.0,
|
| 3811 |
+
"learning_rate": 0.003476,
|
| 3812 |
+
"loss": 1.7324,
|
| 3813 |
+
"step": 8203
|
| 3814 |
+
},
|
| 3815 |
+
{
|
| 3816 |
+
"epoch": 632.0,
|
| 3817 |
+
"learning_rate": 0.0034720000000000003,
|
| 3818 |
+
"loss": 1.6107,
|
| 3819 |
+
"step": 8216
|
| 3820 |
+
},
|
| 3821 |
+
{
|
| 3822 |
+
"epoch": 633.0,
|
| 3823 |
+
"learning_rate": 0.003468,
|
| 3824 |
+
"loss": 1.5336,
|
| 3825 |
+
"step": 8229
|
| 3826 |
+
},
|
| 3827 |
+
{
|
| 3828 |
+
"epoch": 634.0,
|
| 3829 |
+
"learning_rate": 0.003464,
|
| 3830 |
+
"loss": 1.5587,
|
| 3831 |
+
"step": 8242
|
| 3832 |
+
},
|
| 3833 |
+
{
|
| 3834 |
+
"epoch": 635.0,
|
| 3835 |
+
"learning_rate": 0.00346,
|
| 3836 |
+
"loss": 1.581,
|
| 3837 |
+
"step": 8255
|
| 3838 |
+
},
|
| 3839 |
+
{
|
| 3840 |
+
"epoch": 636.0,
|
| 3841 |
+
"learning_rate": 0.003456,
|
| 3842 |
+
"loss": 1.5281,
|
| 3843 |
+
"step": 8268
|
| 3844 |
+
},
|
| 3845 |
+
{
|
| 3846 |
+
"epoch": 637.0,
|
| 3847 |
+
"learning_rate": 0.003452,
|
| 3848 |
+
"loss": 1.5198,
|
| 3849 |
+
"step": 8281
|
| 3850 |
+
},
|
| 3851 |
+
{
|
| 3852 |
+
"epoch": 638.0,
|
| 3853 |
+
"learning_rate": 0.003448,
|
| 3854 |
+
"loss": 1.5671,
|
| 3855 |
+
"step": 8294
|
| 3856 |
+
},
|
| 3857 |
+
{
|
| 3858 |
+
"epoch": 639.0,
|
| 3859 |
+
"learning_rate": 0.003444,
|
| 3860 |
+
"loss": 1.5257,
|
| 3861 |
+
"step": 8307
|
| 3862 |
+
},
|
| 3863 |
+
{
|
| 3864 |
+
"epoch": 640.0,
|
| 3865 |
+
"learning_rate": 0.00344,
|
| 3866 |
+
"loss": 1.5525,
|
| 3867 |
+
"step": 8320
|
| 3868 |
+
},
|
| 3869 |
+
{
|
| 3870 |
+
"epoch": 641.0,
|
| 3871 |
+
"learning_rate": 0.003436,
|
| 3872 |
+
"loss": 1.5005,
|
| 3873 |
+
"step": 8333
|
| 3874 |
+
},
|
| 3875 |
+
{
|
| 3876 |
+
"epoch": 642.0,
|
| 3877 |
+
"learning_rate": 0.003432,
|
| 3878 |
+
"loss": 1.4971,
|
| 3879 |
+
"step": 8346
|
| 3880 |
+
},
|
| 3881 |
+
{
|
| 3882 |
+
"epoch": 643.0,
|
| 3883 |
+
"learning_rate": 0.003428,
|
| 3884 |
+
"loss": 1.4738,
|
| 3885 |
+
"step": 8359
|
| 3886 |
+
},
|
| 3887 |
+
{
|
| 3888 |
+
"epoch": 644.0,
|
| 3889 |
+
"learning_rate": 0.003424,
|
| 3890 |
+
"loss": 1.5397,
|
| 3891 |
+
"step": 8372
|
| 3892 |
+
},
|
| 3893 |
+
{
|
| 3894 |
+
"epoch": 645.0,
|
| 3895 |
+
"learning_rate": 0.00342,
|
| 3896 |
+
"loss": 1.5092,
|
| 3897 |
+
"step": 8385
|
| 3898 |
+
},
|
| 3899 |
+
{
|
| 3900 |
+
"epoch": 646.0,
|
| 3901 |
+
"learning_rate": 0.003416,
|
| 3902 |
+
"loss": 1.5638,
|
| 3903 |
+
"step": 8398
|
| 3904 |
+
},
|
| 3905 |
+
{
|
| 3906 |
+
"epoch": 647.0,
|
| 3907 |
+
"learning_rate": 0.003412,
|
| 3908 |
+
"loss": 1.4813,
|
| 3909 |
+
"step": 8411
|
| 3910 |
+
},
|
| 3911 |
+
{
|
| 3912 |
+
"epoch": 648.0,
|
| 3913 |
+
"learning_rate": 0.003408,
|
| 3914 |
+
"loss": 1.4827,
|
| 3915 |
+
"step": 8424
|
| 3916 |
+
},
|
| 3917 |
+
{
|
| 3918 |
+
"epoch": 649.0,
|
| 3919 |
+
"learning_rate": 0.003404,
|
| 3920 |
+
"loss": 1.5285,
|
| 3921 |
+
"step": 8437
|
| 3922 |
+
},
|
| 3923 |
+
{
|
| 3924 |
+
"epoch": 650.0,
|
| 3925 |
+
"learning_rate": 0.0034,
|
| 3926 |
+
"loss": 1.5059,
|
| 3927 |
+
"step": 8450
|
| 3928 |
+
},
|
| 3929 |
+
{
|
| 3930 |
+
"epoch": 651.0,
|
| 3931 |
+
"learning_rate": 0.003396,
|
| 3932 |
+
"loss": 1.5452,
|
| 3933 |
+
"step": 8463
|
| 3934 |
+
},
|
| 3935 |
+
{
|
| 3936 |
+
"epoch": 652.0,
|
| 3937 |
+
"learning_rate": 0.003392,
|
| 3938 |
+
"loss": 1.6823,
|
| 3939 |
+
"step": 8476
|
| 3940 |
+
},
|
| 3941 |
+
{
|
| 3942 |
+
"epoch": 653.0,
|
| 3943 |
+
"learning_rate": 0.003388,
|
| 3944 |
+
"loss": 1.6004,
|
| 3945 |
+
"step": 8489
|
| 3946 |
+
},
|
| 3947 |
+
{
|
| 3948 |
+
"epoch": 654.0,
|
| 3949 |
+
"learning_rate": 0.003384,
|
| 3950 |
+
"loss": 1.648,
|
| 3951 |
+
"step": 8502
|
| 3952 |
+
},
|
| 3953 |
+
{
|
| 3954 |
+
"epoch": 655.0,
|
| 3955 |
+
"learning_rate": 0.0033799999999999998,
|
| 3956 |
+
"loss": 1.6614,
|
| 3957 |
+
"step": 8515
|
| 3958 |
+
},
|
| 3959 |
+
{
|
| 3960 |
+
"epoch": 656.0,
|
| 3961 |
+
"learning_rate": 0.003376,
|
| 3962 |
+
"loss": 1.6969,
|
| 3963 |
+
"step": 8528
|
| 3964 |
+
},
|
| 3965 |
+
{
|
| 3966 |
+
"epoch": 657.0,
|
| 3967 |
+
"learning_rate": 0.003372,
|
| 3968 |
+
"loss": 1.5948,
|
| 3969 |
+
"step": 8541
|
| 3970 |
+
},
|
| 3971 |
+
{
|
| 3972 |
+
"epoch": 658.0,
|
| 3973 |
+
"learning_rate": 0.003368,
|
| 3974 |
+
"loss": 1.5743,
|
| 3975 |
+
"step": 8554
|
| 3976 |
+
},
|
| 3977 |
+
{
|
| 3978 |
+
"epoch": 659.0,
|
| 3979 |
+
"learning_rate": 0.003364,
|
| 3980 |
+
"loss": 1.5521,
|
| 3981 |
+
"step": 8567
|
| 3982 |
+
},
|
| 3983 |
+
{
|
| 3984 |
+
"epoch": 660.0,
|
| 3985 |
+
"learning_rate": 0.00336,
|
| 3986 |
+
"loss": 1.5639,
|
| 3987 |
+
"step": 8580
|
| 3988 |
+
},
|
| 3989 |
+
{
|
| 3990 |
+
"epoch": 661.0,
|
| 3991 |
+
"learning_rate": 0.003356,
|
| 3992 |
+
"loss": 1.5155,
|
| 3993 |
+
"step": 8593
|
| 3994 |
+
},
|
| 3995 |
+
{
|
| 3996 |
+
"epoch": 662.0,
|
| 3997 |
+
"learning_rate": 0.003352,
|
| 3998 |
+
"loss": 1.4494,
|
| 3999 |
+
"step": 8606
|
| 4000 |
+
},
|
| 4001 |
+
{
|
| 4002 |
+
"epoch": 663.0,
|
| 4003 |
+
"learning_rate": 0.003348,
|
| 4004 |
+
"loss": 1.5803,
|
| 4005 |
+
"step": 8619
|
| 4006 |
+
},
|
| 4007 |
+
{
|
| 4008 |
+
"epoch": 664.0,
|
| 4009 |
+
"learning_rate": 0.0033439999999999998,
|
| 4010 |
+
"loss": 1.5896,
|
| 4011 |
+
"step": 8632
|
| 4012 |
+
},
|
| 4013 |
+
{
|
| 4014 |
+
"epoch": 665.0,
|
| 4015 |
+
"learning_rate": 0.00334,
|
| 4016 |
+
"loss": 1.6208,
|
| 4017 |
+
"step": 8645
|
| 4018 |
+
},
|
| 4019 |
+
{
|
| 4020 |
+
"epoch": 666.0,
|
| 4021 |
+
"learning_rate": 0.003336,
|
| 4022 |
+
"loss": 1.6345,
|
| 4023 |
+
"step": 8658
|
| 4024 |
+
},
|
| 4025 |
+
{
|
| 4026 |
+
"epoch": 667.0,
|
| 4027 |
+
"learning_rate": 0.003332,
|
| 4028 |
+
"loss": 1.6087,
|
| 4029 |
+
"step": 8671
|
| 4030 |
+
},
|
| 4031 |
+
{
|
| 4032 |
+
"epoch": 668.0,
|
| 4033 |
+
"learning_rate": 0.003328,
|
| 4034 |
+
"loss": 1.5938,
|
| 4035 |
+
"step": 8684
|
| 4036 |
+
},
|
| 4037 |
+
{
|
| 4038 |
+
"epoch": 669.0,
|
| 4039 |
+
"learning_rate": 0.0033239999999999997,
|
| 4040 |
+
"loss": 1.5391,
|
| 4041 |
+
"step": 8697
|
| 4042 |
+
},
|
| 4043 |
+
{
|
| 4044 |
+
"epoch": 670.0,
|
| 4045 |
+
"learning_rate": 0.00332,
|
| 4046 |
+
"loss": 1.5938,
|
| 4047 |
+
"step": 8710
|
| 4048 |
+
},
|
| 4049 |
+
{
|
| 4050 |
+
"epoch": 671.0,
|
| 4051 |
+
"learning_rate": 0.003316,
|
| 4052 |
+
"loss": 1.6235,
|
| 4053 |
+
"step": 8723
|
| 4054 |
+
},
|
| 4055 |
+
{
|
| 4056 |
+
"epoch": 672.0,
|
| 4057 |
+
"learning_rate": 0.003312,
|
| 4058 |
+
"loss": 1.6882,
|
| 4059 |
+
"step": 8736
|
| 4060 |
+
},
|
| 4061 |
+
{
|
| 4062 |
+
"epoch": 673.0,
|
| 4063 |
+
"learning_rate": 0.0033079999999999997,
|
| 4064 |
+
"loss": 1.5822,
|
| 4065 |
+
"step": 8749
|
| 4066 |
+
},
|
| 4067 |
+
{
|
| 4068 |
+
"epoch": 674.0,
|
| 4069 |
+
"learning_rate": 0.003304,
|
| 4070 |
+
"loss": 1.622,
|
| 4071 |
+
"step": 8762
|
| 4072 |
+
},
|
| 4073 |
+
{
|
| 4074 |
+
"epoch": 675.0,
|
| 4075 |
+
"learning_rate": 0.0033,
|
| 4076 |
+
"loss": 1.587,
|
| 4077 |
+
"step": 8775
|
| 4078 |
+
},
|
| 4079 |
+
{
|
| 4080 |
+
"epoch": 676.0,
|
| 4081 |
+
"learning_rate": 0.003296,
|
| 4082 |
+
"loss": 1.5119,
|
| 4083 |
+
"step": 8788
|
| 4084 |
+
},
|
| 4085 |
+
{
|
| 4086 |
+
"epoch": 677.0,
|
| 4087 |
+
"learning_rate": 0.003292,
|
| 4088 |
+
"loss": 1.467,
|
| 4089 |
+
"step": 8801
|
| 4090 |
+
},
|
| 4091 |
+
{
|
| 4092 |
+
"epoch": 678.0,
|
| 4093 |
+
"learning_rate": 0.0032879999999999997,
|
| 4094 |
+
"loss": 1.438,
|
| 4095 |
+
"step": 8814
|
| 4096 |
+
},
|
| 4097 |
+
{
|
| 4098 |
+
"epoch": 679.0,
|
| 4099 |
+
"learning_rate": 0.003284,
|
| 4100 |
+
"loss": 1.4497,
|
| 4101 |
+
"step": 8827
|
| 4102 |
+
},
|
| 4103 |
+
{
|
| 4104 |
+
"epoch": 680.0,
|
| 4105 |
+
"learning_rate": 0.00328,
|
| 4106 |
+
"loss": 1.4349,
|
| 4107 |
+
"step": 8840
|
| 4108 |
+
},
|
| 4109 |
+
{
|
| 4110 |
+
"epoch": 681.0,
|
| 4111 |
+
"learning_rate": 0.003276,
|
| 4112 |
+
"loss": 1.378,
|
| 4113 |
+
"step": 8853
|
| 4114 |
+
},
|
| 4115 |
+
{
|
| 4116 |
+
"epoch": 682.0,
|
| 4117 |
+
"learning_rate": 0.0032719999999999997,
|
| 4118 |
+
"loss": 1.3847,
|
| 4119 |
+
"step": 8866
|
| 4120 |
+
},
|
| 4121 |
+
{
|
| 4122 |
+
"epoch": 683.0,
|
| 4123 |
+
"learning_rate": 0.003268,
|
| 4124 |
+
"loss": 1.4556,
|
| 4125 |
+
"step": 8879
|
| 4126 |
+
},
|
| 4127 |
+
{
|
| 4128 |
+
"epoch": 684.0,
|
| 4129 |
+
"learning_rate": 0.003264,
|
| 4130 |
+
"loss": 1.4435,
|
| 4131 |
+
"step": 8892
|
| 4132 |
+
},
|
| 4133 |
+
{
|
| 4134 |
+
"epoch": 685.0,
|
| 4135 |
+
"learning_rate": 0.00326,
|
| 4136 |
+
"loss": 1.4494,
|
| 4137 |
+
"step": 8905
|
| 4138 |
+
},
|
| 4139 |
+
{
|
| 4140 |
+
"epoch": 686.0,
|
| 4141 |
+
"learning_rate": 0.0032559999999999998,
|
| 4142 |
+
"loss": 1.4065,
|
| 4143 |
+
"step": 8918
|
| 4144 |
+
},
|
| 4145 |
+
{
|
| 4146 |
+
"epoch": 687.0,
|
| 4147 |
+
"learning_rate": 0.0032519999999999997,
|
| 4148 |
+
"loss": 1.4323,
|
| 4149 |
+
"step": 8931
|
| 4150 |
+
},
|
| 4151 |
+
{
|
| 4152 |
+
"epoch": 688.0,
|
| 4153 |
+
"learning_rate": 0.0032480000000000005,
|
| 4154 |
+
"loss": 1.4698,
|
| 4155 |
+
"step": 8944
|
| 4156 |
+
},
|
| 4157 |
+
{
|
| 4158 |
+
"epoch": 689.0,
|
| 4159 |
+
"learning_rate": 0.0032440000000000004,
|
| 4160 |
+
"loss": 1.4293,
|
| 4161 |
+
"step": 8957
|
| 4162 |
+
},
|
| 4163 |
+
{
|
| 4164 |
+
"epoch": 690.0,
|
| 4165 |
+
"learning_rate": 0.0032400000000000003,
|
| 4166 |
+
"loss": 1.5205,
|
| 4167 |
+
"step": 8970
|
| 4168 |
+
},
|
| 4169 |
+
{
|
| 4170 |
+
"epoch": 691.0,
|
| 4171 |
+
"learning_rate": 0.003236,
|
| 4172 |
+
"loss": 1.5039,
|
| 4173 |
+
"step": 8983
|
| 4174 |
+
},
|
| 4175 |
+
{
|
| 4176 |
+
"epoch": 692.0,
|
| 4177 |
+
"learning_rate": 0.003232,
|
| 4178 |
+
"loss": 1.4715,
|
| 4179 |
+
"step": 8996
|
| 4180 |
+
},
|
| 4181 |
+
{
|
| 4182 |
+
"epoch": 693.0,
|
| 4183 |
+
"learning_rate": 0.0032280000000000004,
|
| 4184 |
+
"loss": 1.4448,
|
| 4185 |
+
"step": 9009
|
| 4186 |
+
},
|
| 4187 |
+
{
|
| 4188 |
+
"epoch": 694.0,
|
| 4189 |
+
"learning_rate": 0.0032240000000000003,
|
| 4190 |
+
"loss": 1.5031,
|
| 4191 |
+
"step": 9022
|
| 4192 |
+
},
|
| 4193 |
+
{
|
| 4194 |
+
"epoch": 695.0,
|
| 4195 |
+
"learning_rate": 0.00322,
|
| 4196 |
+
"loss": 1.5261,
|
| 4197 |
+
"step": 9035
|
| 4198 |
+
},
|
| 4199 |
+
{
|
| 4200 |
+
"epoch": 696.0,
|
| 4201 |
+
"learning_rate": 0.003216,
|
| 4202 |
+
"loss": 1.4389,
|
| 4203 |
+
"step": 9048
|
| 4204 |
+
},
|
| 4205 |
+
{
|
| 4206 |
+
"epoch": 697.0,
|
| 4207 |
+
"learning_rate": 0.0032120000000000004,
|
| 4208 |
+
"loss": 1.4792,
|
| 4209 |
+
"step": 9061
|
| 4210 |
+
},
|
| 4211 |
+
{
|
| 4212 |
+
"epoch": 698.0,
|
| 4213 |
+
"learning_rate": 0.0032080000000000003,
|
| 4214 |
+
"loss": 1.4317,
|
| 4215 |
+
"step": 9074
|
| 4216 |
+
},
|
| 4217 |
+
{
|
| 4218 |
+
"epoch": 699.0,
|
| 4219 |
+
"learning_rate": 0.0032040000000000003,
|
| 4220 |
+
"loss": 1.5776,
|
| 4221 |
+
"step": 9087
|
| 4222 |
+
},
|
| 4223 |
+
{
|
| 4224 |
+
"epoch": 700.0,
|
| 4225 |
+
"learning_rate": 0.0032,
|
| 4226 |
+
"loss": 1.5326,
|
| 4227 |
+
"step": 9100
|
| 4228 |
+
},
|
| 4229 |
+
{
|
| 4230 |
+
"epoch": 701.0,
|
| 4231 |
+
"learning_rate": 0.003196,
|
| 4232 |
+
"loss": 1.5345,
|
| 4233 |
+
"step": 9113
|
| 4234 |
+
},
|
| 4235 |
+
{
|
| 4236 |
+
"epoch": 702.0,
|
| 4237 |
+
"learning_rate": 0.0031920000000000004,
|
| 4238 |
+
"loss": 1.5269,
|
| 4239 |
+
"step": 9126
|
| 4240 |
+
},
|
| 4241 |
+
{
|
| 4242 |
+
"epoch": 703.0,
|
| 4243 |
+
"learning_rate": 0.0031880000000000003,
|
| 4244 |
+
"loss": 1.4819,
|
| 4245 |
+
"step": 9139
|
| 4246 |
+
},
|
| 4247 |
+
{
|
| 4248 |
+
"epoch": 704.0,
|
| 4249 |
+
"learning_rate": 0.003184,
|
| 4250 |
+
"loss": 1.5326,
|
| 4251 |
+
"step": 9152
|
| 4252 |
+
},
|
| 4253 |
+
{
|
| 4254 |
+
"epoch": 705.0,
|
| 4255 |
+
"learning_rate": 0.00318,
|
| 4256 |
+
"loss": 1.4257,
|
| 4257 |
+
"step": 9165
|
| 4258 |
+
},
|
| 4259 |
+
{
|
| 4260 |
+
"epoch": 706.0,
|
| 4261 |
+
"learning_rate": 0.0031760000000000004,
|
| 4262 |
+
"loss": 1.4306,
|
| 4263 |
+
"step": 9178
|
| 4264 |
+
},
|
| 4265 |
+
{
|
| 4266 |
+
"epoch": 707.0,
|
| 4267 |
+
"learning_rate": 0.0031720000000000003,
|
| 4268 |
+
"loss": 1.3884,
|
| 4269 |
+
"step": 9191
|
| 4270 |
+
},
|
| 4271 |
+
{
|
| 4272 |
+
"epoch": 708.0,
|
| 4273 |
+
"learning_rate": 0.0031680000000000002,
|
| 4274 |
+
"loss": 1.3421,
|
| 4275 |
+
"step": 9204
|
| 4276 |
+
},
|
| 4277 |
+
{
|
| 4278 |
+
"epoch": 709.0,
|
| 4279 |
+
"learning_rate": 0.003164,
|
| 4280 |
+
"loss": 1.394,
|
| 4281 |
+
"step": 9217
|
| 4282 |
+
},
|
| 4283 |
+
{
|
| 4284 |
+
"epoch": 710.0,
|
| 4285 |
+
"learning_rate": 0.00316,
|
| 4286 |
+
"loss": 1.3892,
|
| 4287 |
+
"step": 9230
|
| 4288 |
+
},
|
| 4289 |
+
{
|
| 4290 |
+
"epoch": 711.0,
|
| 4291 |
+
"learning_rate": 0.0031560000000000004,
|
| 4292 |
+
"loss": 1.4832,
|
| 4293 |
+
"step": 9243
|
| 4294 |
+
},
|
| 4295 |
+
{
|
| 4296 |
+
"epoch": 712.0,
|
| 4297 |
+
"learning_rate": 0.0031520000000000003,
|
| 4298 |
+
"loss": 1.4088,
|
| 4299 |
+
"step": 9256
|
| 4300 |
+
},
|
| 4301 |
+
{
|
| 4302 |
+
"epoch": 713.0,
|
| 4303 |
+
"learning_rate": 0.003148,
|
| 4304 |
+
"loss": 1.386,
|
| 4305 |
+
"step": 9269
|
| 4306 |
+
},
|
| 4307 |
+
{
|
| 4308 |
+
"epoch": 714.0,
|
| 4309 |
+
"learning_rate": 0.003144,
|
| 4310 |
+
"loss": 1.3992,
|
| 4311 |
+
"step": 9282
|
| 4312 |
+
},
|
| 4313 |
+
{
|
| 4314 |
+
"epoch": 715.0,
|
| 4315 |
+
"learning_rate": 0.00314,
|
| 4316 |
+
"loss": 1.381,
|
| 4317 |
+
"step": 9295
|
| 4318 |
+
},
|
| 4319 |
+
{
|
| 4320 |
+
"epoch": 716.0,
|
| 4321 |
+
"learning_rate": 0.0031360000000000003,
|
| 4322 |
+
"loss": 1.394,
|
| 4323 |
+
"step": 9308
|
| 4324 |
+
},
|
| 4325 |
+
{
|
| 4326 |
+
"epoch": 717.0,
|
| 4327 |
+
"learning_rate": 0.0031320000000000002,
|
| 4328 |
+
"loss": 1.4024,
|
| 4329 |
+
"step": 9321
|
| 4330 |
+
},
|
| 4331 |
+
{
|
| 4332 |
+
"epoch": 718.0,
|
| 4333 |
+
"learning_rate": 0.003128,
|
| 4334 |
+
"loss": 1.3334,
|
| 4335 |
+
"step": 9334
|
| 4336 |
+
},
|
| 4337 |
+
{
|
| 4338 |
+
"epoch": 719.0,
|
| 4339 |
+
"learning_rate": 0.003124,
|
| 4340 |
+
"loss": 1.3467,
|
| 4341 |
+
"step": 9347
|
| 4342 |
+
},
|
| 4343 |
+
{
|
| 4344 |
+
"epoch": 720.0,
|
| 4345 |
+
"learning_rate": 0.0031200000000000004,
|
| 4346 |
+
"loss": 1.284,
|
| 4347 |
+
"step": 9360
|
| 4348 |
+
},
|
| 4349 |
+
{
|
| 4350 |
+
"epoch": 721.0,
|
| 4351 |
+
"learning_rate": 0.0031160000000000003,
|
| 4352 |
+
"loss": 1.2705,
|
| 4353 |
+
"step": 9373
|
| 4354 |
+
},
|
| 4355 |
+
{
|
| 4356 |
+
"epoch": 722.0,
|
| 4357 |
+
"learning_rate": 0.003112,
|
| 4358 |
+
"loss": 1.2919,
|
| 4359 |
+
"step": 9386
|
| 4360 |
+
},
|
| 4361 |
+
{
|
| 4362 |
+
"epoch": 723.0,
|
| 4363 |
+
"learning_rate": 0.003108,
|
| 4364 |
+
"loss": 1.3071,
|
| 4365 |
+
"step": 9399
|
| 4366 |
+
},
|
| 4367 |
+
{
|
| 4368 |
+
"epoch": 724.0,
|
| 4369 |
+
"learning_rate": 0.003104,
|
| 4370 |
+
"loss": 1.3584,
|
| 4371 |
+
"step": 9412
|
| 4372 |
+
},
|
| 4373 |
+
{
|
| 4374 |
+
"epoch": 725.0,
|
| 4375 |
+
"learning_rate": 0.0031000000000000003,
|
| 4376 |
+
"loss": 1.4149,
|
| 4377 |
+
"step": 9425
|
| 4378 |
+
},
|
| 4379 |
+
{
|
| 4380 |
+
"epoch": 726.0,
|
| 4381 |
+
"learning_rate": 0.0030960000000000002,
|
| 4382 |
+
"loss": 1.3721,
|
| 4383 |
+
"step": 9438
|
| 4384 |
+
},
|
| 4385 |
+
{
|
| 4386 |
+
"epoch": 727.0,
|
| 4387 |
+
"learning_rate": 0.003092,
|
| 4388 |
+
"loss": 1.3719,
|
| 4389 |
+
"step": 9451
|
| 4390 |
+
},
|
| 4391 |
+
{
|
| 4392 |
+
"epoch": 728.0,
|
| 4393 |
+
"learning_rate": 0.003088,
|
| 4394 |
+
"loss": 1.3559,
|
| 4395 |
+
"step": 9464
|
| 4396 |
+
},
|
| 4397 |
+
{
|
| 4398 |
+
"epoch": 729.0,
|
| 4399 |
+
"learning_rate": 0.003084,
|
| 4400 |
+
"loss": 1.3108,
|
| 4401 |
+
"step": 9477
|
| 4402 |
+
},
|
| 4403 |
+
{
|
| 4404 |
+
"epoch": 730.0,
|
| 4405 |
+
"learning_rate": 0.0030800000000000003,
|
| 4406 |
+
"loss": 1.316,
|
| 4407 |
+
"step": 9490
|
| 4408 |
+
},
|
| 4409 |
+
{
|
| 4410 |
+
"epoch": 731.0,
|
| 4411 |
+
"learning_rate": 0.003076,
|
| 4412 |
+
"loss": 1.3154,
|
| 4413 |
+
"step": 9503
|
| 4414 |
+
},
|
| 4415 |
+
{
|
| 4416 |
+
"epoch": 732.0,
|
| 4417 |
+
"learning_rate": 0.003072,
|
| 4418 |
+
"loss": 1.327,
|
| 4419 |
+
"step": 9516
|
| 4420 |
+
},
|
| 4421 |
+
{
|
| 4422 |
+
"epoch": 733.0,
|
| 4423 |
+
"learning_rate": 0.003068,
|
| 4424 |
+
"loss": 1.2914,
|
| 4425 |
+
"step": 9529
|
| 4426 |
+
},
|
| 4427 |
+
{
|
| 4428 |
+
"epoch": 734.0,
|
| 4429 |
+
"learning_rate": 0.0030640000000000003,
|
| 4430 |
+
"loss": 1.2891,
|
| 4431 |
+
"step": 9542
|
| 4432 |
+
},
|
| 4433 |
+
{
|
| 4434 |
+
"epoch": 735.0,
|
| 4435 |
+
"learning_rate": 0.0030600000000000002,
|
| 4436 |
+
"loss": 1.2923,
|
| 4437 |
+
"step": 9555
|
| 4438 |
+
},
|
| 4439 |
+
{
|
| 4440 |
+
"epoch": 736.0,
|
| 4441 |
+
"learning_rate": 0.003056,
|
| 4442 |
+
"loss": 1.3608,
|
| 4443 |
+
"step": 9568
|
| 4444 |
+
},
|
| 4445 |
+
{
|
| 4446 |
+
"epoch": 737.0,
|
| 4447 |
+
"learning_rate": 0.003052,
|
| 4448 |
+
"loss": 1.3126,
|
| 4449 |
+
"step": 9581
|
| 4450 |
+
},
|
| 4451 |
+
{
|
| 4452 |
+
"epoch": 738.0,
|
| 4453 |
+
"learning_rate": 0.003048,
|
| 4454 |
+
"loss": 1.3673,
|
| 4455 |
+
"step": 9594
|
| 4456 |
+
},
|
| 4457 |
+
{
|
| 4458 |
+
"epoch": 739.0,
|
| 4459 |
+
"learning_rate": 0.0030440000000000003,
|
| 4460 |
+
"loss": 1.3951,
|
| 4461 |
+
"step": 9607
|
| 4462 |
+
},
|
| 4463 |
+
{
|
| 4464 |
+
"epoch": 740.0,
|
| 4465 |
+
"learning_rate": 0.00304,
|
| 4466 |
+
"loss": 1.3128,
|
| 4467 |
+
"step": 9620
|
| 4468 |
+
},
|
| 4469 |
+
{
|
| 4470 |
+
"epoch": 741.0,
|
| 4471 |
+
"learning_rate": 0.003036,
|
| 4472 |
+
"loss": 1.3117,
|
| 4473 |
+
"step": 9633
|
| 4474 |
+
},
|
| 4475 |
+
{
|
| 4476 |
+
"epoch": 742.0,
|
| 4477 |
+
"learning_rate": 0.003032,
|
| 4478 |
+
"loss": 1.2828,
|
| 4479 |
+
"step": 9646
|
| 4480 |
+
},
|
| 4481 |
+
{
|
| 4482 |
+
"epoch": 743.0,
|
| 4483 |
+
"learning_rate": 0.003028,
|
| 4484 |
+
"loss": 1.3054,
|
| 4485 |
+
"step": 9659
|
| 4486 |
+
},
|
| 4487 |
+
{
|
| 4488 |
+
"epoch": 744.0,
|
| 4489 |
+
"learning_rate": 0.003024,
|
| 4490 |
+
"loss": 1.289,
|
| 4491 |
+
"step": 9672
|
| 4492 |
+
},
|
| 4493 |
+
{
|
| 4494 |
+
"epoch": 745.0,
|
| 4495 |
+
"learning_rate": 0.00302,
|
| 4496 |
+
"loss": 1.3023,
|
| 4497 |
+
"step": 9685
|
| 4498 |
+
},
|
| 4499 |
+
{
|
| 4500 |
+
"epoch": 746.0,
|
| 4501 |
+
"learning_rate": 0.003016,
|
| 4502 |
+
"loss": 1.2972,
|
| 4503 |
+
"step": 9698
|
| 4504 |
+
},
|
| 4505 |
+
{
|
| 4506 |
+
"epoch": 747.0,
|
| 4507 |
+
"learning_rate": 0.003012,
|
| 4508 |
+
"loss": 1.281,
|
| 4509 |
+
"step": 9711
|
| 4510 |
+
},
|
| 4511 |
+
{
|
| 4512 |
+
"epoch": 748.0,
|
| 4513 |
+
"learning_rate": 0.0030080000000000003,
|
| 4514 |
+
"loss": 1.2475,
|
| 4515 |
+
"step": 9724
|
| 4516 |
+
},
|
| 4517 |
+
{
|
| 4518 |
+
"epoch": 749.0,
|
| 4519 |
+
"learning_rate": 0.003004,
|
| 4520 |
+
"loss": 1.2721,
|
| 4521 |
+
"step": 9737
|
| 4522 |
+
},
|
| 4523 |
+
{
|
| 4524 |
+
"epoch": 750.0,
|
| 4525 |
+
"learning_rate": 0.003,
|
| 4526 |
+
"loss": 1.3066,
|
| 4527 |
+
"step": 9750
|
| 4528 |
+
},
|
| 4529 |
+
{
|
| 4530 |
+
"epoch": 751.0,
|
| 4531 |
+
"learning_rate": 0.002996,
|
| 4532 |
+
"loss": 1.3229,
|
| 4533 |
+
"step": 9763
|
| 4534 |
+
},
|
| 4535 |
+
{
|
| 4536 |
+
"epoch": 752.0,
|
| 4537 |
+
"learning_rate": 0.002992,
|
| 4538 |
+
"loss": 1.2095,
|
| 4539 |
+
"step": 9776
|
| 4540 |
+
},
|
| 4541 |
+
{
|
| 4542 |
+
"epoch": 753.0,
|
| 4543 |
+
"learning_rate": 0.002988,
|
| 4544 |
+
"loss": 1.2389,
|
| 4545 |
+
"step": 9789
|
| 4546 |
+
},
|
| 4547 |
+
{
|
| 4548 |
+
"epoch": 754.0,
|
| 4549 |
+
"learning_rate": 0.002984,
|
| 4550 |
+
"loss": 1.2046,
|
| 4551 |
+
"step": 9802
|
| 4552 |
+
},
|
| 4553 |
+
{
|
| 4554 |
+
"epoch": 755.0,
|
| 4555 |
+
"learning_rate": 0.00298,
|
| 4556 |
+
"loss": 1.1953,
|
| 4557 |
+
"step": 9815
|
| 4558 |
+
},
|
| 4559 |
+
{
|
| 4560 |
+
"epoch": 756.0,
|
| 4561 |
+
"learning_rate": 0.002976,
|
| 4562 |
+
"loss": 1.1359,
|
| 4563 |
+
"step": 9828
|
| 4564 |
+
},
|
| 4565 |
+
{
|
| 4566 |
+
"epoch": 757.0,
|
| 4567 |
+
"learning_rate": 0.0029720000000000002,
|
| 4568 |
+
"loss": 1.13,
|
| 4569 |
+
"step": 9841
|
| 4570 |
+
},
|
| 4571 |
+
{
|
| 4572 |
+
"epoch": 758.0,
|
| 4573 |
+
"learning_rate": 0.002968,
|
| 4574 |
+
"loss": 1.1946,
|
| 4575 |
+
"step": 9854
|
| 4576 |
+
},
|
| 4577 |
+
{
|
| 4578 |
+
"epoch": 759.0,
|
| 4579 |
+
"learning_rate": 0.002964,
|
| 4580 |
+
"loss": 1.2325,
|
| 4581 |
+
"step": 9867
|
| 4582 |
+
},
|
| 4583 |
+
{
|
| 4584 |
+
"epoch": 760.0,
|
| 4585 |
+
"learning_rate": 0.00296,
|
| 4586 |
+
"loss": 1.2435,
|
| 4587 |
+
"step": 9880
|
| 4588 |
+
},
|
| 4589 |
+
{
|
| 4590 |
+
"epoch": 761.0,
|
| 4591 |
+
"learning_rate": 0.002956,
|
| 4592 |
+
"loss": 1.2878,
|
| 4593 |
+
"step": 9893
|
| 4594 |
+
},
|
| 4595 |
+
{
|
| 4596 |
+
"epoch": 762.0,
|
| 4597 |
+
"learning_rate": 0.002952,
|
| 4598 |
+
"loss": 1.2123,
|
| 4599 |
+
"step": 9906
|
| 4600 |
+
},
|
| 4601 |
+
{
|
| 4602 |
+
"epoch": 763.0,
|
| 4603 |
+
"learning_rate": 0.002948,
|
| 4604 |
+
"loss": 1.1953,
|
| 4605 |
+
"step": 9919
|
| 4606 |
+
},
|
| 4607 |
+
{
|
| 4608 |
+
"epoch": 764.0,
|
| 4609 |
+
"learning_rate": 0.002944,
|
| 4610 |
+
"loss": 1.2623,
|
| 4611 |
+
"step": 9932
|
| 4612 |
+
},
|
| 4613 |
+
{
|
| 4614 |
+
"epoch": 765.0,
|
| 4615 |
+
"learning_rate": 0.00294,
|
| 4616 |
+
"loss": 1.2676,
|
| 4617 |
+
"step": 9945
|
| 4618 |
+
},
|
| 4619 |
+
{
|
| 4620 |
+
"epoch": 766.0,
|
| 4621 |
+
"learning_rate": 0.002936,
|
| 4622 |
+
"loss": 1.1999,
|
| 4623 |
+
"step": 9958
|
| 4624 |
+
},
|
| 4625 |
+
{
|
| 4626 |
+
"epoch": 767.0,
|
| 4627 |
+
"learning_rate": 0.002932,
|
| 4628 |
+
"loss": 1.2521,
|
| 4629 |
+
"step": 9971
|
| 4630 |
+
},
|
| 4631 |
+
{
|
| 4632 |
+
"epoch": 768.0,
|
| 4633 |
+
"learning_rate": 0.002928,
|
| 4634 |
+
"loss": 1.4043,
|
| 4635 |
+
"step": 9984
|
| 4636 |
+
},
|
| 4637 |
+
{
|
| 4638 |
+
"epoch": 769.0,
|
| 4639 |
+
"learning_rate": 0.002924,
|
| 4640 |
+
"loss": 1.3043,
|
| 4641 |
+
"step": 9997
|
| 4642 |
+
},
|
| 4643 |
+
{
|
| 4644 |
+
"epoch": 770.0,
|
| 4645 |
+
"learning_rate": 0.00292,
|
| 4646 |
+
"loss": 1.1831,
|
| 4647 |
+
"step": 10010
|
| 4648 |
+
},
|
| 4649 |
+
{
|
| 4650 |
+
"epoch": 771.0,
|
| 4651 |
+
"learning_rate": 0.002916,
|
| 4652 |
+
"loss": 1.1813,
|
| 4653 |
+
"step": 10023
|
| 4654 |
+
},
|
| 4655 |
+
{
|
| 4656 |
+
"epoch": 772.0,
|
| 4657 |
+
"learning_rate": 0.002912,
|
| 4658 |
+
"loss": 1.1946,
|
| 4659 |
+
"step": 10036
|
| 4660 |
+
},
|
| 4661 |
+
{
|
| 4662 |
+
"epoch": 773.0,
|
| 4663 |
+
"learning_rate": 0.002908,
|
| 4664 |
+
"loss": 1.2182,
|
| 4665 |
+
"step": 10049
|
| 4666 |
+
},
|
| 4667 |
+
{
|
| 4668 |
+
"epoch": 774.0,
|
| 4669 |
+
"learning_rate": 0.002904,
|
| 4670 |
+
"loss": 1.2491,
|
| 4671 |
+
"step": 10062
|
| 4672 |
+
},
|
| 4673 |
+
{
|
| 4674 |
+
"epoch": 775.0,
|
| 4675 |
+
"learning_rate": 0.0029,
|
| 4676 |
+
"loss": 1.2422,
|
| 4677 |
+
"step": 10075
|
| 4678 |
+
},
|
| 4679 |
+
{
|
| 4680 |
+
"epoch": 776.0,
|
| 4681 |
+
"learning_rate": 0.002896,
|
| 4682 |
+
"loss": 1.2784,
|
| 4683 |
+
"step": 10088
|
| 4684 |
+
},
|
| 4685 |
+
{
|
| 4686 |
+
"epoch": 777.0,
|
| 4687 |
+
"learning_rate": 0.002892,
|
| 4688 |
+
"loss": 1.1924,
|
| 4689 |
+
"step": 10101
|
| 4690 |
+
},
|
| 4691 |
+
{
|
| 4692 |
+
"epoch": 778.0,
|
| 4693 |
+
"learning_rate": 0.002888,
|
| 4694 |
+
"loss": 1.1739,
|
| 4695 |
+
"step": 10114
|
| 4696 |
+
},
|
| 4697 |
+
{
|
| 4698 |
+
"epoch": 779.0,
|
| 4699 |
+
"learning_rate": 0.002884,
|
| 4700 |
+
"loss": 1.2357,
|
| 4701 |
+
"step": 10127
|
| 4702 |
+
},
|
| 4703 |
+
{
|
| 4704 |
+
"epoch": 780.0,
|
| 4705 |
+
"learning_rate": 0.0028799999999999997,
|
| 4706 |
+
"loss": 1.1845,
|
| 4707 |
+
"step": 10140
|
| 4708 |
+
},
|
| 4709 |
+
{
|
| 4710 |
+
"epoch": 781.0,
|
| 4711 |
+
"learning_rate": 0.002876,
|
| 4712 |
+
"loss": 1.1947,
|
| 4713 |
+
"step": 10153
|
| 4714 |
+
},
|
| 4715 |
+
{
|
| 4716 |
+
"epoch": 782.0,
|
| 4717 |
+
"learning_rate": 0.002872,
|
| 4718 |
+
"loss": 1.2135,
|
| 4719 |
+
"step": 10166
|
| 4720 |
+
},
|
| 4721 |
+
{
|
| 4722 |
+
"epoch": 783.0,
|
| 4723 |
+
"learning_rate": 0.002868,
|
| 4724 |
+
"loss": 1.1923,
|
| 4725 |
+
"step": 10179
|
| 4726 |
+
},
|
| 4727 |
+
{
|
| 4728 |
+
"epoch": 784.0,
|
| 4729 |
+
"learning_rate": 0.002864,
|
| 4730 |
+
"loss": 1.1954,
|
| 4731 |
+
"step": 10192
|
| 4732 |
+
},
|
| 4733 |
+
{
|
| 4734 |
+
"epoch": 785.0,
|
| 4735 |
+
"learning_rate": 0.00286,
|
| 4736 |
+
"loss": 1.2599,
|
| 4737 |
+
"step": 10205
|
| 4738 |
+
},
|
| 4739 |
+
{
|
| 4740 |
+
"epoch": 786.0,
|
| 4741 |
+
"learning_rate": 0.002856,
|
| 4742 |
+
"loss": 1.1905,
|
| 4743 |
+
"step": 10218
|
| 4744 |
+
},
|
| 4745 |
+
{
|
| 4746 |
+
"epoch": 787.0,
|
| 4747 |
+
"learning_rate": 0.002852,
|
| 4748 |
+
"loss": 1.1463,
|
| 4749 |
+
"step": 10231
|
| 4750 |
+
},
|
| 4751 |
+
{
|
| 4752 |
+
"epoch": 788.0,
|
| 4753 |
+
"learning_rate": 0.002848,
|
| 4754 |
+
"loss": 1.1417,
|
| 4755 |
+
"step": 10244
|
| 4756 |
+
},
|
| 4757 |
+
{
|
| 4758 |
+
"epoch": 789.0,
|
| 4759 |
+
"learning_rate": 0.0028439999999999997,
|
| 4760 |
+
"loss": 1.1696,
|
| 4761 |
+
"step": 10257
|
| 4762 |
+
},
|
| 4763 |
+
{
|
| 4764 |
+
"epoch": 790.0,
|
| 4765 |
+
"learning_rate": 0.00284,
|
| 4766 |
+
"loss": 1.1227,
|
| 4767 |
+
"step": 10270
|
| 4768 |
+
},
|
| 4769 |
+
{
|
| 4770 |
+
"epoch": 791.0,
|
| 4771 |
+
"learning_rate": 0.002836,
|
| 4772 |
+
"loss": 1.1858,
|
| 4773 |
+
"step": 10283
|
| 4774 |
+
},
|
| 4775 |
+
{
|
| 4776 |
+
"epoch": 792.0,
|
| 4777 |
+
"learning_rate": 0.002832,
|
| 4778 |
+
"loss": 1.1615,
|
| 4779 |
+
"step": 10296
|
| 4780 |
+
},
|
| 4781 |
+
{
|
| 4782 |
+
"epoch": 793.0,
|
| 4783 |
+
"learning_rate": 0.002828,
|
| 4784 |
+
"loss": 1.2113,
|
| 4785 |
+
"step": 10309
|
| 4786 |
+
},
|
| 4787 |
+
{
|
| 4788 |
+
"epoch": 794.0,
|
| 4789 |
+
"learning_rate": 0.0028239999999999997,
|
| 4790 |
+
"loss": 1.1995,
|
| 4791 |
+
"step": 10322
|
| 4792 |
+
},
|
| 4793 |
+
{
|
| 4794 |
+
"epoch": 795.0,
|
| 4795 |
+
"learning_rate": 0.00282,
|
| 4796 |
+
"loss": 1.2497,
|
| 4797 |
+
"step": 10335
|
| 4798 |
+
},
|
| 4799 |
+
{
|
| 4800 |
+
"epoch": 796.0,
|
| 4801 |
+
"learning_rate": 0.002816,
|
| 4802 |
+
"loss": 1.2255,
|
| 4803 |
+
"step": 10348
|
| 4804 |
+
},
|
| 4805 |
+
{
|
| 4806 |
+
"epoch": 797.0,
|
| 4807 |
+
"learning_rate": 0.002812,
|
| 4808 |
+
"loss": 1.2728,
|
| 4809 |
+
"step": 10361
|
| 4810 |
+
},
|
| 4811 |
+
{
|
| 4812 |
+
"epoch": 798.0,
|
| 4813 |
+
"learning_rate": 0.0028079999999999997,
|
| 4814 |
+
"loss": 1.2053,
|
| 4815 |
+
"step": 10374
|
| 4816 |
+
},
|
| 4817 |
+
{
|
| 4818 |
+
"epoch": 799.0,
|
| 4819 |
+
"learning_rate": 0.002804,
|
| 4820 |
+
"loss": 1.1941,
|
| 4821 |
+
"step": 10387
|
| 4822 |
+
},
|
| 4823 |
+
{
|
| 4824 |
+
"epoch": 800.0,
|
| 4825 |
+
"learning_rate": 0.0028,
|
| 4826 |
+
"loss": 1.184,
|
| 4827 |
+
"step": 10400
|
| 4828 |
+
},
|
| 4829 |
+
{
|
| 4830 |
+
"epoch": 801.0,
|
| 4831 |
+
"learning_rate": 0.002796,
|
| 4832 |
+
"loss": 1.1575,
|
| 4833 |
+
"step": 10413
|
| 4834 |
+
},
|
| 4835 |
+
{
|
| 4836 |
+
"epoch": 802.0,
|
| 4837 |
+
"learning_rate": 0.0027919999999999998,
|
| 4838 |
+
"loss": 1.1571,
|
| 4839 |
+
"step": 10426
|
| 4840 |
+
},
|
| 4841 |
+
{
|
| 4842 |
+
"epoch": 803.0,
|
| 4843 |
+
"learning_rate": 0.0027879999999999997,
|
| 4844 |
+
"loss": 1.1619,
|
| 4845 |
+
"step": 10439
|
| 4846 |
+
},
|
| 4847 |
+
{
|
| 4848 |
+
"epoch": 804.0,
|
| 4849 |
+
"learning_rate": 0.002784,
|
| 4850 |
+
"loss": 1.175,
|
| 4851 |
+
"step": 10452
|
| 4852 |
+
},
|
| 4853 |
+
{
|
| 4854 |
+
"epoch": 805.0,
|
| 4855 |
+
"learning_rate": 0.00278,
|
| 4856 |
+
"loss": 1.2,
|
| 4857 |
+
"step": 10465
|
| 4858 |
+
},
|
| 4859 |
+
{
|
| 4860 |
+
"epoch": 806.0,
|
| 4861 |
+
"learning_rate": 0.002776,
|
| 4862 |
+
"loss": 1.1761,
|
| 4863 |
+
"step": 10478
|
| 4864 |
+
},
|
| 4865 |
+
{
|
| 4866 |
+
"epoch": 807.0,
|
| 4867 |
+
"learning_rate": 0.0027719999999999997,
|
| 4868 |
+
"loss": 1.0651,
|
| 4869 |
+
"step": 10491
|
| 4870 |
+
},
|
| 4871 |
+
{
|
| 4872 |
+
"epoch": 808.0,
|
| 4873 |
+
"learning_rate": 0.002768,
|
| 4874 |
+
"loss": 1.0923,
|
| 4875 |
+
"step": 10504
|
| 4876 |
+
},
|
| 4877 |
+
{
|
| 4878 |
+
"epoch": 809.0,
|
| 4879 |
+
"learning_rate": 0.002764,
|
| 4880 |
+
"loss": 1.1537,
|
| 4881 |
+
"step": 10517
|
| 4882 |
+
},
|
| 4883 |
+
{
|
| 4884 |
+
"epoch": 810.0,
|
| 4885 |
+
"learning_rate": 0.00276,
|
| 4886 |
+
"loss": 1.1618,
|
| 4887 |
+
"step": 10530
|
| 4888 |
+
},
|
| 4889 |
+
{
|
| 4890 |
+
"epoch": 811.0,
|
| 4891 |
+
"learning_rate": 0.0027559999999999998,
|
| 4892 |
+
"loss": 1.2048,
|
| 4893 |
+
"step": 10543
|
| 4894 |
+
},
|
| 4895 |
+
{
|
| 4896 |
+
"epoch": 812.0,
|
| 4897 |
+
"learning_rate": 0.0027519999999999997,
|
| 4898 |
+
"loss": 1.1584,
|
| 4899 |
+
"step": 10556
|
| 4900 |
+
},
|
| 4901 |
+
{
|
| 4902 |
+
"epoch": 813.0,
|
| 4903 |
+
"learning_rate": 0.0027480000000000004,
|
| 4904 |
+
"loss": 1.1815,
|
| 4905 |
+
"step": 10569
|
| 4906 |
+
},
|
| 4907 |
+
{
|
| 4908 |
+
"epoch": 814.0,
|
| 4909 |
+
"learning_rate": 0.0027440000000000003,
|
| 4910 |
+
"loss": 1.1204,
|
| 4911 |
+
"step": 10582
|
| 4912 |
+
},
|
| 4913 |
+
{
|
| 4914 |
+
"epoch": 815.0,
|
| 4915 |
+
"learning_rate": 0.0027400000000000002,
|
| 4916 |
+
"loss": 1.1662,
|
| 4917 |
+
"step": 10595
|
| 4918 |
+
},
|
| 4919 |
+
{
|
| 4920 |
+
"epoch": 816.0,
|
| 4921 |
+
"learning_rate": 0.002736,
|
| 4922 |
+
"loss": 1.1275,
|
| 4923 |
+
"step": 10608
|
| 4924 |
+
},
|
| 4925 |
+
{
|
| 4926 |
+
"epoch": 817.0,
|
| 4927 |
+
"learning_rate": 0.002732,
|
| 4928 |
+
"loss": 1.1124,
|
| 4929 |
+
"step": 10621
|
| 4930 |
+
},
|
| 4931 |
+
{
|
| 4932 |
+
"epoch": 818.0,
|
| 4933 |
+
"learning_rate": 0.0027280000000000004,
|
| 4934 |
+
"loss": 1.0765,
|
| 4935 |
+
"step": 10634
|
| 4936 |
+
},
|
| 4937 |
+
{
|
| 4938 |
+
"epoch": 819.0,
|
| 4939 |
+
"learning_rate": 0.0027240000000000003,
|
| 4940 |
+
"loss": 1.1159,
|
| 4941 |
+
"step": 10647
|
| 4942 |
+
},
|
| 4943 |
+
{
|
| 4944 |
+
"epoch": 820.0,
|
| 4945 |
+
"learning_rate": 0.00272,
|
| 4946 |
+
"loss": 1.1124,
|
| 4947 |
+
"step": 10660
|
| 4948 |
+
},
|
| 4949 |
+
{
|
| 4950 |
+
"epoch": 821.0,
|
| 4951 |
+
"learning_rate": 0.002716,
|
| 4952 |
+
"loss": 1.1045,
|
| 4953 |
+
"step": 10673
|
| 4954 |
+
},
|
| 4955 |
+
{
|
| 4956 |
+
"epoch": 822.0,
|
| 4957 |
+
"learning_rate": 0.0027120000000000004,
|
| 4958 |
+
"loss": 1.1152,
|
| 4959 |
+
"step": 10686
|
| 4960 |
+
},
|
| 4961 |
+
{
|
| 4962 |
+
"epoch": 823.0,
|
| 4963 |
+
"learning_rate": 0.0027080000000000003,
|
| 4964 |
+
"loss": 1.0664,
|
| 4965 |
+
"step": 10699
|
| 4966 |
+
},
|
| 4967 |
+
{
|
| 4968 |
+
"epoch": 824.0,
|
| 4969 |
+
"learning_rate": 0.0027040000000000002,
|
| 4970 |
+
"loss": 1.0165,
|
| 4971 |
+
"step": 10712
|
| 4972 |
+
},
|
| 4973 |
+
{
|
| 4974 |
+
"epoch": 825.0,
|
| 4975 |
+
"learning_rate": 0.0027,
|
| 4976 |
+
"loss": 1.004,
|
| 4977 |
+
"step": 10725
|
| 4978 |
+
},
|
| 4979 |
+
{
|
| 4980 |
+
"epoch": 826.0,
|
| 4981 |
+
"learning_rate": 0.002696,
|
| 4982 |
+
"loss": 1.0194,
|
| 4983 |
+
"step": 10738
|
| 4984 |
+
},
|
| 4985 |
+
{
|
| 4986 |
+
"epoch": 827.0,
|
| 4987 |
+
"learning_rate": 0.0026920000000000004,
|
| 4988 |
+
"loss": 1.04,
|
| 4989 |
+
"step": 10751
|
| 4990 |
+
},
|
| 4991 |
+
{
|
| 4992 |
+
"epoch": 828.0,
|
| 4993 |
+
"learning_rate": 0.0026880000000000003,
|
| 4994 |
+
"loss": 1.0393,
|
| 4995 |
+
"step": 10764
|
| 4996 |
+
},
|
| 4997 |
+
{
|
| 4998 |
+
"epoch": 829.0,
|
| 4999 |
+
"learning_rate": 0.002684,
|
| 5000 |
+
"loss": 0.9552,
|
| 5001 |
+
"step": 10777
|
| 5002 |
+
},
|
| 5003 |
+
{
|
| 5004 |
+
"epoch": 830.0,
|
| 5005 |
+
"learning_rate": 0.00268,
|
| 5006 |
+
"loss": 0.9634,
|
| 5007 |
+
"step": 10790
|
| 5008 |
+
},
|
| 5009 |
+
{
|
| 5010 |
+
"epoch": 831.0,
|
| 5011 |
+
"learning_rate": 0.0026760000000000004,
|
| 5012 |
+
"loss": 0.9967,
|
| 5013 |
+
"step": 10803
|
| 5014 |
+
},
|
| 5015 |
+
{
|
| 5016 |
+
"epoch": 832.0,
|
| 5017 |
+
"learning_rate": 0.0026720000000000003,
|
| 5018 |
+
"loss": 0.994,
|
| 5019 |
+
"step": 10816
|
| 5020 |
+
},
|
| 5021 |
+
{
|
| 5022 |
+
"epoch": 833.0,
|
| 5023 |
+
"learning_rate": 0.0026680000000000002,
|
| 5024 |
+
"loss": 0.9906,
|
| 5025 |
+
"step": 10829
|
| 5026 |
+
},
|
| 5027 |
+
{
|
| 5028 |
+
"epoch": 834.0,
|
| 5029 |
+
"learning_rate": 0.002664,
|
| 5030 |
+
"loss": 1.0132,
|
| 5031 |
+
"step": 10842
|
| 5032 |
+
},
|
| 5033 |
+
{
|
| 5034 |
+
"epoch": 835.0,
|
| 5035 |
+
"learning_rate": 0.00266,
|
| 5036 |
+
"loss": 1.0088,
|
| 5037 |
+
"step": 10855
|
| 5038 |
+
},
|
| 5039 |
+
{
|
| 5040 |
+
"epoch": 836.0,
|
| 5041 |
+
"learning_rate": 0.0026560000000000004,
|
| 5042 |
+
"loss": 0.9914,
|
| 5043 |
+
"step": 10868
|
| 5044 |
+
},
|
| 5045 |
+
{
|
| 5046 |
+
"epoch": 837.0,
|
| 5047 |
+
"learning_rate": 0.0026520000000000003,
|
| 5048 |
+
"loss": 0.9947,
|
| 5049 |
+
"step": 10881
|
| 5050 |
+
},
|
| 5051 |
+
{
|
| 5052 |
+
"epoch": 838.0,
|
| 5053 |
+
"learning_rate": 0.002648,
|
| 5054 |
+
"loss": 1.0053,
|
| 5055 |
+
"step": 10894
|
| 5056 |
+
},
|
| 5057 |
+
{
|
| 5058 |
+
"epoch": 839.0,
|
| 5059 |
+
"learning_rate": 0.002644,
|
| 5060 |
+
"loss": 1.0138,
|
| 5061 |
+
"step": 10907
|
| 5062 |
+
},
|
| 5063 |
+
{
|
| 5064 |
+
"epoch": 840.0,
|
| 5065 |
+
"learning_rate": 0.00264,
|
| 5066 |
+
"loss": 0.9785,
|
| 5067 |
+
"step": 10920
|
| 5068 |
+
},
|
| 5069 |
+
{
|
| 5070 |
+
"epoch": 841.0,
|
| 5071 |
+
"learning_rate": 0.0026360000000000003,
|
| 5072 |
+
"loss": 1.0174,
|
| 5073 |
+
"step": 10933
|
| 5074 |
+
},
|
| 5075 |
+
{
|
| 5076 |
+
"epoch": 842.0,
|
| 5077 |
+
"learning_rate": 0.0026320000000000002,
|
| 5078 |
+
"loss": 0.9855,
|
| 5079 |
+
"step": 10946
|
| 5080 |
+
},
|
| 5081 |
+
{
|
| 5082 |
+
"epoch": 843.0,
|
| 5083 |
+
"learning_rate": 0.002628,
|
| 5084 |
+
"loss": 1.0005,
|
| 5085 |
+
"step": 10959
|
| 5086 |
+
},
|
| 5087 |
+
{
|
| 5088 |
+
"epoch": 844.0,
|
| 5089 |
+
"learning_rate": 0.002624,
|
| 5090 |
+
"loss": 0.9966,
|
| 5091 |
+
"step": 10972
|
| 5092 |
+
},
|
| 5093 |
+
{
|
| 5094 |
+
"epoch": 845.0,
|
| 5095 |
+
"learning_rate": 0.0026200000000000004,
|
| 5096 |
+
"loss": 0.9952,
|
| 5097 |
+
"step": 10985
|
| 5098 |
+
},
|
| 5099 |
+
{
|
| 5100 |
+
"epoch": 846.0,
|
| 5101 |
+
"learning_rate": 0.0026160000000000003,
|
| 5102 |
+
"loss": 1.001,
|
| 5103 |
+
"step": 10998
|
| 5104 |
+
},
|
| 5105 |
+
{
|
| 5106 |
+
"epoch": 847.0,
|
| 5107 |
+
"learning_rate": 0.002612,
|
| 5108 |
+
"loss": 1.0235,
|
| 5109 |
+
"step": 11011
|
| 5110 |
+
},
|
| 5111 |
+
{
|
| 5112 |
+
"epoch": 848.0,
|
| 5113 |
+
"learning_rate": 0.002608,
|
| 5114 |
+
"loss": 0.9835,
|
| 5115 |
+
"step": 11024
|
| 5116 |
+
},
|
| 5117 |
+
{
|
| 5118 |
+
"epoch": 849.0,
|
| 5119 |
+
"learning_rate": 0.002604,
|
| 5120 |
+
"loss": 0.9951,
|
| 5121 |
+
"step": 11037
|
| 5122 |
+
},
|
| 5123 |
+
{
|
| 5124 |
+
"epoch": 850.0,
|
| 5125 |
+
"learning_rate": 0.0026000000000000003,
|
| 5126 |
+
"loss": 1.0329,
|
| 5127 |
+
"step": 11050
|
| 5128 |
+
},
|
| 5129 |
+
{
|
| 5130 |
+
"epoch": 851.0,
|
| 5131 |
+
"learning_rate": 0.0025960000000000002,
|
| 5132 |
+
"loss": 1.0021,
|
| 5133 |
+
"step": 11063
|
| 5134 |
+
},
|
| 5135 |
+
{
|
| 5136 |
+
"epoch": 852.0,
|
| 5137 |
+
"learning_rate": 0.002592,
|
| 5138 |
+
"loss": 1.0391,
|
| 5139 |
+
"step": 11076
|
| 5140 |
+
},
|
| 5141 |
+
{
|
| 5142 |
+
"epoch": 853.0,
|
| 5143 |
+
"learning_rate": 0.002588,
|
| 5144 |
+
"loss": 1.0249,
|
| 5145 |
+
"step": 11089
|
| 5146 |
+
},
|
| 5147 |
+
{
|
| 5148 |
+
"epoch": 854.0,
|
| 5149 |
+
"learning_rate": 0.002584,
|
| 5150 |
+
"loss": 0.9974,
|
| 5151 |
+
"step": 11102
|
| 5152 |
+
},
|
| 5153 |
+
{
|
| 5154 |
+
"epoch": 855.0,
|
| 5155 |
+
"learning_rate": 0.0025800000000000003,
|
| 5156 |
+
"loss": 1.0149,
|
| 5157 |
+
"step": 11115
|
| 5158 |
+
},
|
| 5159 |
+
{
|
| 5160 |
+
"epoch": 856.0,
|
| 5161 |
+
"learning_rate": 0.002576,
|
| 5162 |
+
"loss": 1.0002,
|
| 5163 |
+
"step": 11128
|
| 5164 |
+
},
|
| 5165 |
+
{
|
| 5166 |
+
"epoch": 857.0,
|
| 5167 |
+
"learning_rate": 0.002572,
|
| 5168 |
+
"loss": 1.0379,
|
| 5169 |
+
"step": 11141
|
| 5170 |
+
},
|
| 5171 |
+
{
|
| 5172 |
+
"epoch": 858.0,
|
| 5173 |
+
"learning_rate": 0.002568,
|
| 5174 |
+
"loss": 1.0381,
|
| 5175 |
+
"step": 11154
|
| 5176 |
+
},
|
| 5177 |
+
{
|
| 5178 |
+
"epoch": 859.0,
|
| 5179 |
+
"learning_rate": 0.0025640000000000003,
|
| 5180 |
+
"loss": 0.9772,
|
| 5181 |
+
"step": 11167
|
| 5182 |
+
},
|
| 5183 |
+
{
|
| 5184 |
+
"epoch": 860.0,
|
| 5185 |
+
"learning_rate": 0.00256,
|
| 5186 |
+
"loss": 1.0263,
|
| 5187 |
+
"step": 11180
|
| 5188 |
+
},
|
| 5189 |
+
{
|
| 5190 |
+
"epoch": 861.0,
|
| 5191 |
+
"learning_rate": 0.002556,
|
| 5192 |
+
"loss": 0.982,
|
| 5193 |
+
"step": 11193
|
| 5194 |
+
},
|
| 5195 |
+
{
|
| 5196 |
+
"epoch": 862.0,
|
| 5197 |
+
"learning_rate": 0.002552,
|
| 5198 |
+
"loss": 0.9892,
|
| 5199 |
+
"step": 11206
|
| 5200 |
+
},
|
| 5201 |
+
{
|
| 5202 |
+
"epoch": 863.0,
|
| 5203 |
+
"learning_rate": 0.002548,
|
| 5204 |
+
"loss": 0.9708,
|
| 5205 |
+
"step": 11219
|
| 5206 |
+
},
|
| 5207 |
+
{
|
| 5208 |
+
"epoch": 864.0,
|
| 5209 |
+
"learning_rate": 0.0025440000000000003,
|
| 5210 |
+
"loss": 0.9883,
|
| 5211 |
+
"step": 11232
|
| 5212 |
+
},
|
| 5213 |
+
{
|
| 5214 |
+
"epoch": 865.0,
|
| 5215 |
+
"learning_rate": 0.00254,
|
| 5216 |
+
"loss": 0.9446,
|
| 5217 |
+
"step": 11245
|
| 5218 |
+
},
|
| 5219 |
+
{
|
| 5220 |
+
"epoch": 866.0,
|
| 5221 |
+
"learning_rate": 0.002536,
|
| 5222 |
+
"loss": 0.9686,
|
| 5223 |
+
"step": 11258
|
| 5224 |
+
},
|
| 5225 |
+
{
|
| 5226 |
+
"epoch": 867.0,
|
| 5227 |
+
"learning_rate": 0.002532,
|
| 5228 |
+
"loss": 1.0044,
|
| 5229 |
+
"step": 11271
|
| 5230 |
+
},
|
| 5231 |
+
{
|
| 5232 |
+
"epoch": 868.0,
|
| 5233 |
+
"learning_rate": 0.002528,
|
| 5234 |
+
"loss": 1.0128,
|
| 5235 |
+
"step": 11284
|
| 5236 |
+
},
|
| 5237 |
+
{
|
| 5238 |
+
"epoch": 869.0,
|
| 5239 |
+
"learning_rate": 0.002524,
|
| 5240 |
+
"loss": 0.9876,
|
| 5241 |
+
"step": 11297
|
| 5242 |
+
},
|
| 5243 |
+
{
|
| 5244 |
+
"epoch": 870.0,
|
| 5245 |
+
"learning_rate": 0.00252,
|
| 5246 |
+
"loss": 0.9992,
|
| 5247 |
+
"step": 11310
|
| 5248 |
+
},
|
| 5249 |
+
{
|
| 5250 |
+
"epoch": 871.0,
|
| 5251 |
+
"learning_rate": 0.002516,
|
| 5252 |
+
"loss": 1.1017,
|
| 5253 |
+
"step": 11323
|
| 5254 |
+
},
|
| 5255 |
+
{
|
| 5256 |
+
"epoch": 872.0,
|
| 5257 |
+
"learning_rate": 0.002512,
|
| 5258 |
+
"loss": 0.9853,
|
| 5259 |
+
"step": 11336
|
| 5260 |
+
},
|
| 5261 |
+
{
|
| 5262 |
+
"epoch": 873.0,
|
| 5263 |
+
"learning_rate": 0.0025080000000000002,
|
| 5264 |
+
"loss": 0.9495,
|
| 5265 |
+
"step": 11349
|
| 5266 |
+
},
|
| 5267 |
+
{
|
| 5268 |
+
"epoch": 874.0,
|
| 5269 |
+
"learning_rate": 0.002504,
|
| 5270 |
+
"loss": 0.9292,
|
| 5271 |
+
"step": 11362
|
| 5272 |
+
},
|
| 5273 |
+
{
|
| 5274 |
+
"epoch": 875.0,
|
| 5275 |
+
"learning_rate": 0.0025,
|
| 5276 |
+
"loss": 0.9339,
|
| 5277 |
+
"step": 11375
|
| 5278 |
+
},
|
| 5279 |
+
{
|
| 5280 |
+
"epoch": 876.0,
|
| 5281 |
+
"learning_rate": 0.002496,
|
| 5282 |
+
"loss": 0.9309,
|
| 5283 |
+
"step": 11388
|
| 5284 |
+
},
|
| 5285 |
+
{
|
| 5286 |
+
"epoch": 877.0,
|
| 5287 |
+
"learning_rate": 0.002492,
|
| 5288 |
+
"loss": 0.9303,
|
| 5289 |
+
"step": 11401
|
| 5290 |
+
},
|
| 5291 |
+
{
|
| 5292 |
+
"epoch": 878.0,
|
| 5293 |
+
"learning_rate": 0.002488,
|
| 5294 |
+
"loss": 0.8827,
|
| 5295 |
+
"step": 11414
|
| 5296 |
+
},
|
| 5297 |
+
{
|
| 5298 |
+
"epoch": 879.0,
|
| 5299 |
+
"learning_rate": 0.002484,
|
| 5300 |
+
"loss": 0.8898,
|
| 5301 |
+
"step": 11427
|
| 5302 |
+
},
|
| 5303 |
+
{
|
| 5304 |
+
"epoch": 880.0,
|
| 5305 |
+
"learning_rate": 0.00248,
|
| 5306 |
+
"loss": 0.8748,
|
| 5307 |
+
"step": 11440
|
| 5308 |
+
},
|
| 5309 |
+
{
|
| 5310 |
+
"epoch": 881.0,
|
| 5311 |
+
"learning_rate": 0.002476,
|
| 5312 |
+
"loss": 0.921,
|
| 5313 |
+
"step": 11453
|
| 5314 |
+
},
|
| 5315 |
+
{
|
| 5316 |
+
"epoch": 882.0,
|
| 5317 |
+
"learning_rate": 0.0024720000000000002,
|
| 5318 |
+
"loss": 0.912,
|
| 5319 |
+
"step": 11466
|
| 5320 |
+
},
|
| 5321 |
+
{
|
| 5322 |
+
"epoch": 883.0,
|
| 5323 |
+
"learning_rate": 0.002468,
|
| 5324 |
+
"loss": 0.9684,
|
| 5325 |
+
"step": 11479
|
| 5326 |
+
},
|
| 5327 |
+
{
|
| 5328 |
+
"epoch": 884.0,
|
| 5329 |
+
"learning_rate": 0.002464,
|
| 5330 |
+
"loss": 1.0113,
|
| 5331 |
+
"step": 11492
|
| 5332 |
+
},
|
| 5333 |
+
{
|
| 5334 |
+
"epoch": 885.0,
|
| 5335 |
+
"learning_rate": 0.00246,
|
| 5336 |
+
"loss": 1.0043,
|
| 5337 |
+
"step": 11505
|
| 5338 |
+
},
|
| 5339 |
+
{
|
| 5340 |
+
"epoch": 886.0,
|
| 5341 |
+
"learning_rate": 0.002456,
|
| 5342 |
+
"loss": 0.94,
|
| 5343 |
+
"step": 11518
|
| 5344 |
+
},
|
| 5345 |
+
{
|
| 5346 |
+
"epoch": 887.0,
|
| 5347 |
+
"learning_rate": 0.002452,
|
| 5348 |
+
"loss": 0.9166,
|
| 5349 |
+
"step": 11531
|
| 5350 |
+
},
|
| 5351 |
+
{
|
| 5352 |
+
"epoch": 888.0,
|
| 5353 |
+
"learning_rate": 0.002448,
|
| 5354 |
+
"loss": 0.9202,
|
| 5355 |
+
"step": 11544
|
| 5356 |
+
},
|
| 5357 |
+
{
|
| 5358 |
+
"epoch": 889.0,
|
| 5359 |
+
"learning_rate": 0.002444,
|
| 5360 |
+
"loss": 0.9179,
|
| 5361 |
+
"step": 11557
|
| 5362 |
+
},
|
| 5363 |
+
{
|
| 5364 |
+
"epoch": 890.0,
|
| 5365 |
+
"learning_rate": 0.00244,
|
| 5366 |
+
"loss": 0.8928,
|
| 5367 |
+
"step": 11570
|
| 5368 |
+
},
|
| 5369 |
+
{
|
| 5370 |
+
"epoch": 891.0,
|
| 5371 |
+
"learning_rate": 0.002436,
|
| 5372 |
+
"loss": 0.9021,
|
| 5373 |
+
"step": 11583
|
| 5374 |
+
},
|
| 5375 |
+
{
|
| 5376 |
+
"epoch": 892.0,
|
| 5377 |
+
"learning_rate": 0.002432,
|
| 5378 |
+
"loss": 0.9038,
|
| 5379 |
+
"step": 11596
|
| 5380 |
+
},
|
| 5381 |
+
{
|
| 5382 |
+
"epoch": 893.0,
|
| 5383 |
+
"learning_rate": 0.002428,
|
| 5384 |
+
"loss": 0.8446,
|
| 5385 |
+
"step": 11609
|
| 5386 |
+
},
|
| 5387 |
+
{
|
| 5388 |
+
"epoch": 894.0,
|
| 5389 |
+
"learning_rate": 0.002424,
|
| 5390 |
+
"loss": 0.9167,
|
| 5391 |
+
"step": 11622
|
| 5392 |
+
},
|
| 5393 |
+
{
|
| 5394 |
+
"epoch": 895.0,
|
| 5395 |
+
"learning_rate": 0.00242,
|
| 5396 |
+
"loss": 0.8897,
|
| 5397 |
+
"step": 11635
|
| 5398 |
+
},
|
| 5399 |
+
{
|
| 5400 |
+
"epoch": 896.0,
|
| 5401 |
+
"learning_rate": 0.002416,
|
| 5402 |
+
"loss": 0.9227,
|
| 5403 |
+
"step": 11648
|
| 5404 |
+
},
|
| 5405 |
+
{
|
| 5406 |
+
"epoch": 897.0,
|
| 5407 |
+
"learning_rate": 0.002412,
|
| 5408 |
+
"loss": 0.8956,
|
| 5409 |
+
"step": 11661
|
| 5410 |
+
},
|
| 5411 |
+
{
|
| 5412 |
+
"epoch": 898.0,
|
| 5413 |
+
"learning_rate": 0.002408,
|
| 5414 |
+
"loss": 0.8768,
|
| 5415 |
+
"step": 11674
|
| 5416 |
+
},
|
| 5417 |
+
{
|
| 5418 |
+
"epoch": 899.0,
|
| 5419 |
+
"learning_rate": 0.002404,
|
| 5420 |
+
"loss": 0.9134,
|
| 5421 |
+
"step": 11687
|
| 5422 |
+
},
|
| 5423 |
+
{
|
| 5424 |
+
"epoch": 900.0,
|
| 5425 |
+
"learning_rate": 0.0024,
|
| 5426 |
+
"loss": 0.8484,
|
| 5427 |
+
"step": 11700
|
| 5428 |
+
},
|
| 5429 |
+
{
|
| 5430 |
+
"epoch": 901.0,
|
| 5431 |
+
"learning_rate": 0.002396,
|
| 5432 |
+
"loss": 0.8616,
|
| 5433 |
+
"step": 11713
|
| 5434 |
+
},
|
| 5435 |
+
{
|
| 5436 |
+
"epoch": 902.0,
|
| 5437 |
+
"learning_rate": 0.002392,
|
| 5438 |
+
"loss": 0.8669,
|
| 5439 |
+
"step": 11726
|
| 5440 |
+
},
|
| 5441 |
+
{
|
| 5442 |
+
"epoch": 903.0,
|
| 5443 |
+
"learning_rate": 0.002388,
|
| 5444 |
+
"loss": 0.8529,
|
| 5445 |
+
"step": 11739
|
| 5446 |
+
},
|
| 5447 |
+
{
|
| 5448 |
+
"epoch": 904.0,
|
| 5449 |
+
"learning_rate": 0.002384,
|
| 5450 |
+
"loss": 0.8488,
|
| 5451 |
+
"step": 11752
|
| 5452 |
+
},
|
| 5453 |
+
{
|
| 5454 |
+
"epoch": 905.0,
|
| 5455 |
+
"learning_rate": 0.0023799999999999997,
|
| 5456 |
+
"loss": 0.8505,
|
| 5457 |
+
"step": 11765
|
| 5458 |
+
},
|
| 5459 |
+
{
|
| 5460 |
+
"epoch": 906.0,
|
| 5461 |
+
"learning_rate": 0.002376,
|
| 5462 |
+
"loss": 0.8264,
|
| 5463 |
+
"step": 11778
|
| 5464 |
+
},
|
| 5465 |
+
{
|
| 5466 |
+
"epoch": 907.0,
|
| 5467 |
+
"learning_rate": 0.002372,
|
| 5468 |
+
"loss": 0.8382,
|
| 5469 |
+
"step": 11791
|
| 5470 |
+
},
|
| 5471 |
+
{
|
| 5472 |
+
"epoch": 908.0,
|
| 5473 |
+
"learning_rate": 0.002368,
|
| 5474 |
+
"loss": 0.8176,
|
| 5475 |
+
"step": 11804
|
| 5476 |
+
},
|
| 5477 |
+
{
|
| 5478 |
+
"epoch": 909.0,
|
| 5479 |
+
"learning_rate": 0.0023639999999999998,
|
| 5480 |
+
"loss": 0.8122,
|
| 5481 |
+
"step": 11817
|
| 5482 |
+
},
|
| 5483 |
+
{
|
| 5484 |
+
"epoch": 910.0,
|
| 5485 |
+
"learning_rate": 0.00236,
|
| 5486 |
+
"loss": 0.8175,
|
| 5487 |
+
"step": 11830
|
| 5488 |
+
},
|
| 5489 |
+
{
|
| 5490 |
+
"epoch": 911.0,
|
| 5491 |
+
"learning_rate": 0.002356,
|
| 5492 |
+
"loss": 0.8345,
|
| 5493 |
+
"step": 11843
|
| 5494 |
+
},
|
| 5495 |
+
{
|
| 5496 |
+
"epoch": 912.0,
|
| 5497 |
+
"learning_rate": 0.002352,
|
| 5498 |
+
"loss": 0.8102,
|
| 5499 |
+
"step": 11856
|
| 5500 |
+
},
|
| 5501 |
+
{
|
| 5502 |
+
"epoch": 913.0,
|
| 5503 |
+
"learning_rate": 0.002348,
|
| 5504 |
+
"loss": 0.7818,
|
| 5505 |
+
"step": 11869
|
| 5506 |
+
},
|
| 5507 |
+
{
|
| 5508 |
+
"epoch": 914.0,
|
| 5509 |
+
"learning_rate": 0.0023439999999999997,
|
| 5510 |
+
"loss": 0.8027,
|
| 5511 |
+
"step": 11882
|
| 5512 |
+
},
|
| 5513 |
+
{
|
| 5514 |
+
"epoch": 915.0,
|
| 5515 |
+
"learning_rate": 0.00234,
|
| 5516 |
+
"loss": 0.7765,
|
| 5517 |
+
"step": 11895
|
| 5518 |
+
},
|
| 5519 |
+
{
|
| 5520 |
+
"epoch": 916.0,
|
| 5521 |
+
"learning_rate": 0.002336,
|
| 5522 |
+
"loss": 0.8225,
|
| 5523 |
+
"step": 11908
|
| 5524 |
+
},
|
| 5525 |
+
{
|
| 5526 |
+
"epoch": 917.0,
|
| 5527 |
+
"learning_rate": 0.002332,
|
| 5528 |
+
"loss": 0.7882,
|
| 5529 |
+
"step": 11921
|
| 5530 |
+
},
|
| 5531 |
+
{
|
| 5532 |
+
"epoch": 918.0,
|
| 5533 |
+
"learning_rate": 0.0023279999999999998,
|
| 5534 |
+
"loss": 0.7784,
|
| 5535 |
+
"step": 11934
|
| 5536 |
+
},
|
| 5537 |
+
{
|
| 5538 |
+
"epoch": 919.0,
|
| 5539 |
+
"learning_rate": 0.0023239999999999997,
|
| 5540 |
+
"loss": 0.7751,
|
| 5541 |
+
"step": 11947
|
| 5542 |
+
},
|
| 5543 |
+
{
|
| 5544 |
+
"epoch": 920.0,
|
| 5545 |
+
"learning_rate": 0.00232,
|
| 5546 |
+
"loss": 0.7837,
|
| 5547 |
+
"step": 11960
|
| 5548 |
+
},
|
| 5549 |
+
{
|
| 5550 |
+
"epoch": 921.0,
|
| 5551 |
+
"learning_rate": 0.002316,
|
| 5552 |
+
"loss": 0.7588,
|
| 5553 |
+
"step": 11973
|
| 5554 |
+
},
|
| 5555 |
+
{
|
| 5556 |
+
"epoch": 922.0,
|
| 5557 |
+
"learning_rate": 0.002312,
|
| 5558 |
+
"loss": 0.8106,
|
| 5559 |
+
"step": 11986
|
| 5560 |
+
},
|
| 5561 |
+
{
|
| 5562 |
+
"epoch": 923.0,
|
| 5563 |
+
"learning_rate": 0.0023079999999999997,
|
| 5564 |
+
"loss": 0.8359,
|
| 5565 |
+
"step": 11999
|
| 5566 |
+
},
|
| 5567 |
+
{
|
| 5568 |
+
"epoch": 924.0,
|
| 5569 |
+
"learning_rate": 0.002304,
|
| 5570 |
+
"loss": 0.7899,
|
| 5571 |
+
"step": 12012
|
| 5572 |
+
},
|
| 5573 |
+
{
|
| 5574 |
+
"epoch": 925.0,
|
| 5575 |
+
"learning_rate": 0.0023,
|
| 5576 |
+
"loss": 0.7766,
|
| 5577 |
+
"step": 12025
|
| 5578 |
+
},
|
| 5579 |
+
{
|
| 5580 |
+
"epoch": 926.0,
|
| 5581 |
+
"learning_rate": 0.002296,
|
| 5582 |
+
"loss": 0.7978,
|
| 5583 |
+
"step": 12038
|
| 5584 |
+
},
|
| 5585 |
+
{
|
| 5586 |
+
"epoch": 927.0,
|
| 5587 |
+
"learning_rate": 0.0022919999999999998,
|
| 5588 |
+
"loss": 0.8012,
|
| 5589 |
+
"step": 12051
|
| 5590 |
+
},
|
| 5591 |
+
{
|
| 5592 |
+
"epoch": 928.0,
|
| 5593 |
+
"learning_rate": 0.0022879999999999997,
|
| 5594 |
+
"loss": 0.8112,
|
| 5595 |
+
"step": 12064
|
| 5596 |
+
},
|
| 5597 |
+
{
|
| 5598 |
+
"epoch": 929.0,
|
| 5599 |
+
"learning_rate": 0.002284,
|
| 5600 |
+
"loss": 0.8725,
|
| 5601 |
+
"step": 12077
|
| 5602 |
+
},
|
| 5603 |
+
{
|
| 5604 |
+
"epoch": 930.0,
|
| 5605 |
+
"learning_rate": 0.00228,
|
| 5606 |
+
"loss": 0.8415,
|
| 5607 |
+
"step": 12090
|
| 5608 |
+
},
|
| 5609 |
+
{
|
| 5610 |
+
"epoch": 931.0,
|
| 5611 |
+
"learning_rate": 0.002276,
|
| 5612 |
+
"loss": 0.8444,
|
| 5613 |
+
"step": 12103
|
| 5614 |
+
},
|
| 5615 |
+
{
|
| 5616 |
+
"epoch": 932.0,
|
| 5617 |
+
"learning_rate": 0.0022719999999999997,
|
| 5618 |
+
"loss": 0.8459,
|
| 5619 |
+
"step": 12116
|
| 5620 |
+
},
|
| 5621 |
+
{
|
| 5622 |
+
"epoch": 933.0,
|
| 5623 |
+
"learning_rate": 0.002268,
|
| 5624 |
+
"loss": 0.7739,
|
| 5625 |
+
"step": 12129
|
| 5626 |
+
},
|
| 5627 |
+
{
|
| 5628 |
+
"epoch": 934.0,
|
| 5629 |
+
"learning_rate": 0.002264,
|
| 5630 |
+
"loss": 0.8236,
|
| 5631 |
+
"step": 12142
|
| 5632 |
+
},
|
| 5633 |
+
{
|
| 5634 |
+
"epoch": 935.0,
|
| 5635 |
+
"learning_rate": 0.00226,
|
| 5636 |
+
"loss": 0.7746,
|
| 5637 |
+
"step": 12155
|
| 5638 |
+
},
|
| 5639 |
+
{
|
| 5640 |
+
"epoch": 936.0,
|
| 5641 |
+
"learning_rate": 0.0022559999999999998,
|
| 5642 |
+
"loss": 0.807,
|
| 5643 |
+
"step": 12168
|
| 5644 |
+
},
|
| 5645 |
+
{
|
| 5646 |
+
"epoch": 937.0,
|
| 5647 |
+
"learning_rate": 0.0022519999999999997,
|
| 5648 |
+
"loss": 0.8016,
|
| 5649 |
+
"step": 12181
|
| 5650 |
+
},
|
| 5651 |
+
{
|
| 5652 |
+
"epoch": 938.0,
|
| 5653 |
+
"learning_rate": 0.0022480000000000004,
|
| 5654 |
+
"loss": 0.7812,
|
| 5655 |
+
"step": 12194
|
| 5656 |
+
},
|
| 5657 |
+
{
|
| 5658 |
+
"epoch": 939.0,
|
| 5659 |
+
"learning_rate": 0.0022440000000000003,
|
| 5660 |
+
"loss": 0.7796,
|
| 5661 |
+
"step": 12207
|
| 5662 |
+
},
|
| 5663 |
+
{
|
| 5664 |
+
"epoch": 940.0,
|
| 5665 |
+
"learning_rate": 0.0022400000000000002,
|
| 5666 |
+
"loss": 0.7743,
|
| 5667 |
+
"step": 12220
|
| 5668 |
+
},
|
| 5669 |
+
{
|
| 5670 |
+
"epoch": 941.0,
|
| 5671 |
+
"learning_rate": 0.002236,
|
| 5672 |
+
"loss": 0.8141,
|
| 5673 |
+
"step": 12233
|
| 5674 |
+
},
|
| 5675 |
+
{
|
| 5676 |
+
"epoch": 942.0,
|
| 5677 |
+
"learning_rate": 0.002232,
|
| 5678 |
+
"loss": 0.7666,
|
| 5679 |
+
"step": 12246
|
| 5680 |
+
},
|
| 5681 |
+
{
|
| 5682 |
+
"epoch": 943.0,
|
| 5683 |
+
"learning_rate": 0.0022280000000000004,
|
| 5684 |
+
"loss": 0.7668,
|
| 5685 |
+
"step": 12259
|
| 5686 |
+
},
|
| 5687 |
+
{
|
| 5688 |
+
"epoch": 944.0,
|
| 5689 |
+
"learning_rate": 0.0022240000000000003,
|
| 5690 |
+
"loss": 0.7469,
|
| 5691 |
+
"step": 12272
|
| 5692 |
+
},
|
| 5693 |
+
{
|
| 5694 |
+
"epoch": 945.0,
|
| 5695 |
+
"learning_rate": 0.00222,
|
| 5696 |
+
"loss": 0.8032,
|
| 5697 |
+
"step": 12285
|
| 5698 |
+
},
|
| 5699 |
+
{
|
| 5700 |
+
"epoch": 946.0,
|
| 5701 |
+
"learning_rate": 0.002216,
|
| 5702 |
+
"loss": 0.767,
|
| 5703 |
+
"step": 12298
|
| 5704 |
+
},
|
| 5705 |
+
{
|
| 5706 |
+
"epoch": 947.0,
|
| 5707 |
+
"learning_rate": 0.0022120000000000004,
|
| 5708 |
+
"loss": 0.7862,
|
| 5709 |
+
"step": 12311
|
| 5710 |
+
},
|
| 5711 |
+
{
|
| 5712 |
+
"epoch": 948.0,
|
| 5713 |
+
"learning_rate": 0.0022080000000000003,
|
| 5714 |
+
"loss": 0.762,
|
| 5715 |
+
"step": 12324
|
| 5716 |
+
},
|
| 5717 |
+
{
|
| 5718 |
+
"epoch": 949.0,
|
| 5719 |
+
"learning_rate": 0.0022040000000000002,
|
| 5720 |
+
"loss": 0.762,
|
| 5721 |
+
"step": 12337
|
| 5722 |
+
},
|
| 5723 |
+
{
|
| 5724 |
+
"epoch": 950.0,
|
| 5725 |
+
"learning_rate": 0.0022,
|
| 5726 |
+
"loss": 0.7546,
|
| 5727 |
+
"step": 12350
|
| 5728 |
+
},
|
| 5729 |
+
{
|
| 5730 |
+
"epoch": 951.0,
|
| 5731 |
+
"learning_rate": 0.002196,
|
| 5732 |
+
"loss": 0.721,
|
| 5733 |
+
"step": 12363
|
| 5734 |
+
},
|
| 5735 |
+
{
|
| 5736 |
+
"epoch": 952.0,
|
| 5737 |
+
"learning_rate": 0.0021920000000000004,
|
| 5738 |
+
"loss": 0.7442,
|
| 5739 |
+
"step": 12376
|
| 5740 |
+
},
|
| 5741 |
+
{
|
| 5742 |
+
"epoch": 953.0,
|
| 5743 |
+
"learning_rate": 0.0021880000000000003,
|
| 5744 |
+
"loss": 0.7331,
|
| 5745 |
+
"step": 12389
|
| 5746 |
+
},
|
| 5747 |
+
{
|
| 5748 |
+
"epoch": 954.0,
|
| 5749 |
+
"learning_rate": 0.002184,
|
| 5750 |
+
"loss": 0.7299,
|
| 5751 |
+
"step": 12402
|
| 5752 |
+
},
|
| 5753 |
+
{
|
| 5754 |
+
"epoch": 955.0,
|
| 5755 |
+
"learning_rate": 0.00218,
|
| 5756 |
+
"loss": 0.7114,
|
| 5757 |
+
"step": 12415
|
| 5758 |
+
},
|
| 5759 |
+
{
|
| 5760 |
+
"epoch": 956.0,
|
| 5761 |
+
"learning_rate": 0.0021760000000000004,
|
| 5762 |
+
"loss": 0.7443,
|
| 5763 |
+
"step": 12428
|
| 5764 |
+
},
|
| 5765 |
+
{
|
| 5766 |
+
"epoch": 957.0,
|
| 5767 |
+
"learning_rate": 0.0021720000000000003,
|
| 5768 |
+
"loss": 0.7247,
|
| 5769 |
+
"step": 12441
|
| 5770 |
+
},
|
| 5771 |
+
{
|
| 5772 |
+
"epoch": 958.0,
|
| 5773 |
+
"learning_rate": 0.0021680000000000002,
|
| 5774 |
+
"loss": 0.6941,
|
| 5775 |
+
"step": 12454
|
| 5776 |
+
},
|
| 5777 |
+
{
|
| 5778 |
+
"epoch": 959.0,
|
| 5779 |
+
"learning_rate": 0.002164,
|
| 5780 |
+
"loss": 0.6838,
|
| 5781 |
+
"step": 12467
|
| 5782 |
+
},
|
| 5783 |
+
{
|
| 5784 |
+
"epoch": 960.0,
|
| 5785 |
+
"learning_rate": 0.00216,
|
| 5786 |
+
"loss": 0.6838,
|
| 5787 |
+
"step": 12480
|
| 5788 |
+
},
|
| 5789 |
+
{
|
| 5790 |
+
"epoch": 961.0,
|
| 5791 |
+
"learning_rate": 0.0021560000000000004,
|
| 5792 |
+
"loss": 0.7048,
|
| 5793 |
+
"step": 12493
|
| 5794 |
+
},
|
| 5795 |
+
{
|
| 5796 |
+
"epoch": 962.0,
|
| 5797 |
+
"learning_rate": 0.0021520000000000003,
|
| 5798 |
+
"loss": 0.7083,
|
| 5799 |
+
"step": 12506
|
| 5800 |
+
},
|
| 5801 |
+
{
|
| 5802 |
+
"epoch": 963.0,
|
| 5803 |
+
"learning_rate": 0.002148,
|
| 5804 |
+
"loss": 0.7166,
|
| 5805 |
+
"step": 12519
|
| 5806 |
+
},
|
| 5807 |
+
{
|
| 5808 |
+
"epoch": 964.0,
|
| 5809 |
+
"learning_rate": 0.002144,
|
| 5810 |
+
"loss": 0.7128,
|
| 5811 |
+
"step": 12532
|
| 5812 |
+
},
|
| 5813 |
+
{
|
| 5814 |
+
"epoch": 965.0,
|
| 5815 |
+
"learning_rate": 0.00214,
|
| 5816 |
+
"loss": 0.7257,
|
| 5817 |
+
"step": 12545
|
| 5818 |
+
},
|
| 5819 |
+
{
|
| 5820 |
+
"epoch": 966.0,
|
| 5821 |
+
"learning_rate": 0.0021360000000000003,
|
| 5822 |
+
"loss": 0.7145,
|
| 5823 |
+
"step": 12558
|
| 5824 |
+
},
|
| 5825 |
+
{
|
| 5826 |
+
"epoch": 967.0,
|
| 5827 |
+
"learning_rate": 0.002132,
|
| 5828 |
+
"loss": 0.7173,
|
| 5829 |
+
"step": 12571
|
| 5830 |
+
},
|
| 5831 |
+
{
|
| 5832 |
+
"epoch": 968.0,
|
| 5833 |
+
"learning_rate": 0.002128,
|
| 5834 |
+
"loss": 0.7162,
|
| 5835 |
+
"step": 12584
|
| 5836 |
+
},
|
| 5837 |
+
{
|
| 5838 |
+
"epoch": 969.0,
|
| 5839 |
+
"learning_rate": 0.002124,
|
| 5840 |
+
"loss": 0.6849,
|
| 5841 |
+
"step": 12597
|
| 5842 |
+
},
|
| 5843 |
+
{
|
| 5844 |
+
"epoch": 970.0,
|
| 5845 |
+
"learning_rate": 0.0021200000000000004,
|
| 5846 |
+
"loss": 0.6859,
|
| 5847 |
+
"step": 12610
|
| 5848 |
+
},
|
| 5849 |
+
{
|
| 5850 |
+
"epoch": 971.0,
|
| 5851 |
+
"learning_rate": 0.0021160000000000003,
|
| 5852 |
+
"loss": 0.6802,
|
| 5853 |
+
"step": 12623
|
| 5854 |
+
},
|
| 5855 |
+
{
|
| 5856 |
+
"epoch": 972.0,
|
| 5857 |
+
"learning_rate": 0.002112,
|
| 5858 |
+
"loss": 0.6965,
|
| 5859 |
+
"step": 12636
|
| 5860 |
+
},
|
| 5861 |
+
{
|
| 5862 |
+
"epoch": 973.0,
|
| 5863 |
+
"learning_rate": 0.002108,
|
| 5864 |
+
"loss": 0.6941,
|
| 5865 |
+
"step": 12649
|
| 5866 |
+
},
|
| 5867 |
+
{
|
| 5868 |
+
"epoch": 974.0,
|
| 5869 |
+
"learning_rate": 0.002104,
|
| 5870 |
+
"loss": 0.6928,
|
| 5871 |
+
"step": 12662
|
| 5872 |
+
},
|
| 5873 |
+
{
|
| 5874 |
+
"epoch": 975.0,
|
| 5875 |
+
"learning_rate": 0.0021000000000000003,
|
| 5876 |
+
"loss": 0.6764,
|
| 5877 |
+
"step": 12675
|
| 5878 |
+
},
|
| 5879 |
+
{
|
| 5880 |
+
"epoch": 976.0,
|
| 5881 |
+
"learning_rate": 0.002096,
|
| 5882 |
+
"loss": 0.6569,
|
| 5883 |
+
"step": 12688
|
| 5884 |
+
},
|
| 5885 |
+
{
|
| 5886 |
+
"epoch": 977.0,
|
| 5887 |
+
"learning_rate": 0.002092,
|
| 5888 |
+
"loss": 0.6618,
|
| 5889 |
+
"step": 12701
|
| 5890 |
+
},
|
| 5891 |
+
{
|
| 5892 |
+
"epoch": 978.0,
|
| 5893 |
+
"learning_rate": 0.002088,
|
| 5894 |
+
"loss": 0.6719,
|
| 5895 |
+
"step": 12714
|
| 5896 |
+
},
|
| 5897 |
+
{
|
| 5898 |
+
"epoch": 979.0,
|
| 5899 |
+
"learning_rate": 0.002084,
|
| 5900 |
+
"loss": 0.6584,
|
| 5901 |
+
"step": 12727
|
| 5902 |
+
},
|
| 5903 |
+
{
|
| 5904 |
+
"epoch": 980.0,
|
| 5905 |
+
"learning_rate": 0.0020800000000000003,
|
| 5906 |
+
"loss": 0.6911,
|
| 5907 |
+
"step": 12740
|
| 5908 |
+
},
|
| 5909 |
+
{
|
| 5910 |
+
"epoch": 981.0,
|
| 5911 |
+
"learning_rate": 0.002076,
|
| 5912 |
+
"loss": 0.688,
|
| 5913 |
+
"step": 12753
|
| 5914 |
+
},
|
| 5915 |
+
{
|
| 5916 |
+
"epoch": 982.0,
|
| 5917 |
+
"learning_rate": 0.002072,
|
| 5918 |
+
"loss": 0.6741,
|
| 5919 |
+
"step": 12766
|
| 5920 |
+
},
|
| 5921 |
+
{
|
| 5922 |
+
"epoch": 983.0,
|
| 5923 |
+
"learning_rate": 0.002068,
|
| 5924 |
+
"loss": 0.6962,
|
| 5925 |
+
"step": 12779
|
| 5926 |
+
},
|
| 5927 |
+
{
|
| 5928 |
+
"epoch": 984.0,
|
| 5929 |
+
"learning_rate": 0.0020640000000000003,
|
| 5930 |
+
"loss": 0.6811,
|
| 5931 |
+
"step": 12792
|
| 5932 |
+
},
|
| 5933 |
+
{
|
| 5934 |
+
"epoch": 985.0,
|
| 5935 |
+
"learning_rate": 0.00206,
|
| 5936 |
+
"loss": 0.6717,
|
| 5937 |
+
"step": 12805
|
| 5938 |
+
},
|
| 5939 |
+
{
|
| 5940 |
+
"epoch": 986.0,
|
| 5941 |
+
"learning_rate": 0.002056,
|
| 5942 |
+
"loss": 0.6733,
|
| 5943 |
+
"step": 12818
|
| 5944 |
+
},
|
| 5945 |
+
{
|
| 5946 |
+
"epoch": 987.0,
|
| 5947 |
+
"learning_rate": 0.002052,
|
| 5948 |
+
"loss": 0.6813,
|
| 5949 |
+
"step": 12831
|
| 5950 |
+
},
|
| 5951 |
+
{
|
| 5952 |
+
"epoch": 988.0,
|
| 5953 |
+
"learning_rate": 0.002048,
|
| 5954 |
+
"loss": 0.6472,
|
| 5955 |
+
"step": 12844
|
| 5956 |
+
},
|
| 5957 |
+
{
|
| 5958 |
+
"epoch": 989.0,
|
| 5959 |
+
"learning_rate": 0.0020440000000000002,
|
| 5960 |
+
"loss": 0.6508,
|
| 5961 |
+
"step": 12857
|
| 5962 |
+
},
|
| 5963 |
+
{
|
| 5964 |
+
"epoch": 990.0,
|
| 5965 |
+
"learning_rate": 0.00204,
|
| 5966 |
+
"loss": 0.6576,
|
| 5967 |
+
"step": 12870
|
| 5968 |
+
},
|
| 5969 |
+
{
|
| 5970 |
+
"epoch": 991.0,
|
| 5971 |
+
"learning_rate": 0.002036,
|
| 5972 |
+
"loss": 0.6428,
|
| 5973 |
+
"step": 12883
|
| 5974 |
+
},
|
| 5975 |
+
{
|
| 5976 |
+
"epoch": 992.0,
|
| 5977 |
+
"learning_rate": 0.002032,
|
| 5978 |
+
"loss": 0.6505,
|
| 5979 |
+
"step": 12896
|
| 5980 |
+
},
|
| 5981 |
+
{
|
| 5982 |
+
"epoch": 993.0,
|
| 5983 |
+
"learning_rate": 0.002028,
|
| 5984 |
+
"loss": 0.6578,
|
| 5985 |
+
"step": 12909
|
| 5986 |
+
},
|
| 5987 |
+
{
|
| 5988 |
+
"epoch": 994.0,
|
| 5989 |
+
"learning_rate": 0.002024,
|
| 5990 |
+
"loss": 0.6689,
|
| 5991 |
+
"step": 12922
|
| 5992 |
+
},
|
| 5993 |
+
{
|
| 5994 |
+
"epoch": 995.0,
|
| 5995 |
+
"learning_rate": 0.00202,
|
| 5996 |
+
"loss": 0.6625,
|
| 5997 |
+
"step": 12935
|
| 5998 |
+
},
|
| 5999 |
+
{
|
| 6000 |
+
"epoch": 996.0,
|
| 6001 |
+
"learning_rate": 0.002016,
|
| 6002 |
+
"loss": 0.6894,
|
| 6003 |
+
"step": 12948
|
| 6004 |
+
},
|
| 6005 |
+
{
|
| 6006 |
+
"epoch": 997.0,
|
| 6007 |
+
"learning_rate": 0.002012,
|
| 6008 |
+
"loss": 0.6669,
|
| 6009 |
+
"step": 12961
|
| 6010 |
+
},
|
| 6011 |
+
{
|
| 6012 |
+
"epoch": 998.0,
|
| 6013 |
+
"learning_rate": 0.0020080000000000002,
|
| 6014 |
+
"loss": 0.6698,
|
| 6015 |
+
"step": 12974
|
| 6016 |
+
},
|
| 6017 |
+
{
|
| 6018 |
+
"epoch": 999.0,
|
| 6019 |
+
"learning_rate": 0.002004,
|
| 6020 |
+
"loss": 0.6861,
|
| 6021 |
+
"step": 12987
|
| 6022 |
+
},
|
| 6023 |
+
{
|
| 6024 |
+
"epoch": 1000.0,
|
| 6025 |
+
"learning_rate": 0.002,
|
| 6026 |
+
"loss": 0.7089,
|
| 6027 |
+
"step": 13000
|
| 6028 |
+
},
|
| 6029 |
+
{
|
| 6030 |
+
"epoch": 1000.0,
|
| 6031 |
+
"step": 13000,
|
| 6032 |
+
"total_flos": 569573692465152.0,
|
| 6033 |
+
"train_loss": 0.655489089525663,
|
| 6034 |
+
"train_runtime": 67792.7328,
|
| 6035 |
+
"train_samples_per_second": 1.475,
|
| 6036 |
+
"train_steps_per_second": 0.192
|
| 6037 |
}
|
| 6038 |
],
|
| 6039 |
"logging_steps": 500,
|
| 6040 |
+
"max_steps": 13000,
|
| 6041 |
+
"num_train_epochs": 1000,
|
| 6042 |
"save_steps": 500,
|
| 6043 |
+
"total_flos": 569573692465152.0,
|
| 6044 |
"trial_name": null,
|
| 6045 |
"trial_params": null
|
| 6046 |
}
|