End of training
Browse files- all_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +3424 -0
all_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 19.94,
|
3 |
+
"total_flos": 1.7046597288139905e+19,
|
4 |
+
"train_loss": 0.5607940738325689,
|
5 |
+
"train_runtime": 20096.6246,
|
6 |
+
"train_samples_per_second": 34.218,
|
7 |
+
"train_steps_per_second": 0.267
|
8 |
+
}
|
train_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 19.94,
|
3 |
+
"total_flos": 1.7046597288139905e+19,
|
4 |
+
"train_loss": 0.5607940738325689,
|
5 |
+
"train_runtime": 20096.6246,
|
6 |
+
"train_samples_per_second": 34.218,
|
7 |
+
"train_steps_per_second": 0.267
|
8 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,3424 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 0.7557471264367817,
|
3 |
+
"best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned-plant-classification/checkpoint-4837",
|
4 |
+
"epoch": 19.94418604651163,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 5360,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.04,
|
13 |
+
"learning_rate": 9.328358208955223e-07,
|
14 |
+
"loss": 1.5971,
|
15 |
+
"step": 10
|
16 |
+
},
|
17 |
+
{
|
18 |
+
"epoch": 0.07,
|
19 |
+
"learning_rate": 1.8656716417910446e-06,
|
20 |
+
"loss": 1.567,
|
21 |
+
"step": 20
|
22 |
+
},
|
23 |
+
{
|
24 |
+
"epoch": 0.11,
|
25 |
+
"learning_rate": 2.7985074626865674e-06,
|
26 |
+
"loss": 1.4699,
|
27 |
+
"step": 30
|
28 |
+
},
|
29 |
+
{
|
30 |
+
"epoch": 0.15,
|
31 |
+
"learning_rate": 3.7313432835820893e-06,
|
32 |
+
"loss": 1.3608,
|
33 |
+
"step": 40
|
34 |
+
},
|
35 |
+
{
|
36 |
+
"epoch": 0.19,
|
37 |
+
"learning_rate": 4.664179104477613e-06,
|
38 |
+
"loss": 1.2558,
|
39 |
+
"step": 50
|
40 |
+
},
|
41 |
+
{
|
42 |
+
"epoch": 0.22,
|
43 |
+
"learning_rate": 5.597014925373135e-06,
|
44 |
+
"loss": 1.1765,
|
45 |
+
"step": 60
|
46 |
+
},
|
47 |
+
{
|
48 |
+
"epoch": 0.26,
|
49 |
+
"learning_rate": 6.529850746268657e-06,
|
50 |
+
"loss": 1.0991,
|
51 |
+
"step": 70
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.3,
|
55 |
+
"learning_rate": 7.4626865671641785e-06,
|
56 |
+
"loss": 1.0631,
|
57 |
+
"step": 80
|
58 |
+
},
|
59 |
+
{
|
60 |
+
"epoch": 0.33,
|
61 |
+
"learning_rate": 8.395522388059703e-06,
|
62 |
+
"loss": 1.0134,
|
63 |
+
"step": 90
|
64 |
+
},
|
65 |
+
{
|
66 |
+
"epoch": 0.37,
|
67 |
+
"learning_rate": 9.328358208955226e-06,
|
68 |
+
"loss": 0.9662,
|
69 |
+
"step": 100
|
70 |
+
},
|
71 |
+
{
|
72 |
+
"epoch": 0.41,
|
73 |
+
"learning_rate": 1.0261194029850747e-05,
|
74 |
+
"loss": 0.9108,
|
75 |
+
"step": 110
|
76 |
+
},
|
77 |
+
{
|
78 |
+
"epoch": 0.45,
|
79 |
+
"learning_rate": 1.119402985074627e-05,
|
80 |
+
"loss": 0.915,
|
81 |
+
"step": 120
|
82 |
+
},
|
83 |
+
{
|
84 |
+
"epoch": 0.48,
|
85 |
+
"learning_rate": 1.2126865671641792e-05,
|
86 |
+
"loss": 0.842,
|
87 |
+
"step": 130
|
88 |
+
},
|
89 |
+
{
|
90 |
+
"epoch": 0.52,
|
91 |
+
"learning_rate": 1.3059701492537313e-05,
|
92 |
+
"loss": 0.8596,
|
93 |
+
"step": 140
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.56,
|
97 |
+
"learning_rate": 1.3992537313432836e-05,
|
98 |
+
"loss": 0.8443,
|
99 |
+
"step": 150
|
100 |
+
},
|
101 |
+
{
|
102 |
+
"epoch": 0.6,
|
103 |
+
"learning_rate": 1.4925373134328357e-05,
|
104 |
+
"loss": 0.8628,
|
105 |
+
"step": 160
|
106 |
+
},
|
107 |
+
{
|
108 |
+
"epoch": 0.63,
|
109 |
+
"learning_rate": 1.585820895522388e-05,
|
110 |
+
"loss": 0.8044,
|
111 |
+
"step": 170
|
112 |
+
},
|
113 |
+
{
|
114 |
+
"epoch": 0.67,
|
115 |
+
"learning_rate": 1.6791044776119406e-05,
|
116 |
+
"loss": 0.8323,
|
117 |
+
"step": 180
|
118 |
+
},
|
119 |
+
{
|
120 |
+
"epoch": 0.71,
|
121 |
+
"learning_rate": 1.7723880597014927e-05,
|
122 |
+
"loss": 0.8068,
|
123 |
+
"step": 190
|
124 |
+
},
|
125 |
+
{
|
126 |
+
"epoch": 0.74,
|
127 |
+
"learning_rate": 1.865671641791045e-05,
|
128 |
+
"loss": 0.8638,
|
129 |
+
"step": 200
|
130 |
+
},
|
131 |
+
{
|
132 |
+
"epoch": 0.78,
|
133 |
+
"learning_rate": 1.9589552238805972e-05,
|
134 |
+
"loss": 0.8378,
|
135 |
+
"step": 210
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 0.82,
|
139 |
+
"learning_rate": 2.0522388059701493e-05,
|
140 |
+
"loss": 0.7997,
|
141 |
+
"step": 220
|
142 |
+
},
|
143 |
+
{
|
144 |
+
"epoch": 0.86,
|
145 |
+
"learning_rate": 2.1455223880597018e-05,
|
146 |
+
"loss": 0.8079,
|
147 |
+
"step": 230
|
148 |
+
},
|
149 |
+
{
|
150 |
+
"epoch": 0.89,
|
151 |
+
"learning_rate": 2.238805970149254e-05,
|
152 |
+
"loss": 0.8171,
|
153 |
+
"step": 240
|
154 |
+
},
|
155 |
+
{
|
156 |
+
"epoch": 0.93,
|
157 |
+
"learning_rate": 2.332089552238806e-05,
|
158 |
+
"loss": 0.7909,
|
159 |
+
"step": 250
|
160 |
+
},
|
161 |
+
{
|
162 |
+
"epoch": 0.97,
|
163 |
+
"learning_rate": 2.4253731343283584e-05,
|
164 |
+
"loss": 0.8257,
|
165 |
+
"step": 260
|
166 |
+
},
|
167 |
+
{
|
168 |
+
"epoch": 1.0,
|
169 |
+
"eval_accuracy": 0.6695402298850575,
|
170 |
+
"eval_loss": 0.7940567135810852,
|
171 |
+
"eval_runtime": 7.9455,
|
172 |
+
"eval_samples_per_second": 43.798,
|
173 |
+
"eval_steps_per_second": 1.384,
|
174 |
+
"step": 268
|
175 |
+
},
|
176 |
+
{
|
177 |
+
"epoch": 1.0,
|
178 |
+
"learning_rate": 2.5186567164179102e-05,
|
179 |
+
"loss": 0.7715,
|
180 |
+
"step": 270
|
181 |
+
},
|
182 |
+
{
|
183 |
+
"epoch": 1.04,
|
184 |
+
"learning_rate": 2.6119402985074626e-05,
|
185 |
+
"loss": 0.7877,
|
186 |
+
"step": 280
|
187 |
+
},
|
188 |
+
{
|
189 |
+
"epoch": 1.08,
|
190 |
+
"learning_rate": 2.7052238805970147e-05,
|
191 |
+
"loss": 0.7481,
|
192 |
+
"step": 290
|
193 |
+
},
|
194 |
+
{
|
195 |
+
"epoch": 1.12,
|
196 |
+
"learning_rate": 2.7985074626865672e-05,
|
197 |
+
"loss": 0.761,
|
198 |
+
"step": 300
|
199 |
+
},
|
200 |
+
{
|
201 |
+
"epoch": 1.15,
|
202 |
+
"learning_rate": 2.8917910447761193e-05,
|
203 |
+
"loss": 0.7673,
|
204 |
+
"step": 310
|
205 |
+
},
|
206 |
+
{
|
207 |
+
"epoch": 1.19,
|
208 |
+
"learning_rate": 2.9850746268656714e-05,
|
209 |
+
"loss": 0.7708,
|
210 |
+
"step": 320
|
211 |
+
},
|
212 |
+
{
|
213 |
+
"epoch": 1.23,
|
214 |
+
"learning_rate": 3.078358208955224e-05,
|
215 |
+
"loss": 0.7464,
|
216 |
+
"step": 330
|
217 |
+
},
|
218 |
+
{
|
219 |
+
"epoch": 1.27,
|
220 |
+
"learning_rate": 3.171641791044776e-05,
|
221 |
+
"loss": 0.7877,
|
222 |
+
"step": 340
|
223 |
+
},
|
224 |
+
{
|
225 |
+
"epoch": 1.3,
|
226 |
+
"learning_rate": 3.264925373134329e-05,
|
227 |
+
"loss": 0.761,
|
228 |
+
"step": 350
|
229 |
+
},
|
230 |
+
{
|
231 |
+
"epoch": 1.34,
|
232 |
+
"learning_rate": 3.358208955223881e-05,
|
233 |
+
"loss": 0.7607,
|
234 |
+
"step": 360
|
235 |
+
},
|
236 |
+
{
|
237 |
+
"epoch": 1.38,
|
238 |
+
"learning_rate": 3.451492537313433e-05,
|
239 |
+
"loss": 0.7372,
|
240 |
+
"step": 370
|
241 |
+
},
|
242 |
+
{
|
243 |
+
"epoch": 1.41,
|
244 |
+
"learning_rate": 3.5447761194029854e-05,
|
245 |
+
"loss": 0.7314,
|
246 |
+
"step": 380
|
247 |
+
},
|
248 |
+
{
|
249 |
+
"epoch": 1.45,
|
250 |
+
"learning_rate": 3.638059701492538e-05,
|
251 |
+
"loss": 0.777,
|
252 |
+
"step": 390
|
253 |
+
},
|
254 |
+
{
|
255 |
+
"epoch": 1.49,
|
256 |
+
"learning_rate": 3.73134328358209e-05,
|
257 |
+
"loss": 0.7383,
|
258 |
+
"step": 400
|
259 |
+
},
|
260 |
+
{
|
261 |
+
"epoch": 1.53,
|
262 |
+
"learning_rate": 3.824626865671642e-05,
|
263 |
+
"loss": 0.7482,
|
264 |
+
"step": 410
|
265 |
+
},
|
266 |
+
{
|
267 |
+
"epoch": 1.56,
|
268 |
+
"learning_rate": 3.9179104477611945e-05,
|
269 |
+
"loss": 0.7893,
|
270 |
+
"step": 420
|
271 |
+
},
|
272 |
+
{
|
273 |
+
"epoch": 1.6,
|
274 |
+
"learning_rate": 4.011194029850747e-05,
|
275 |
+
"loss": 0.7754,
|
276 |
+
"step": 430
|
277 |
+
},
|
278 |
+
{
|
279 |
+
"epoch": 1.64,
|
280 |
+
"learning_rate": 4.104477611940299e-05,
|
281 |
+
"loss": 0.7645,
|
282 |
+
"step": 440
|
283 |
+
},
|
284 |
+
{
|
285 |
+
"epoch": 1.67,
|
286 |
+
"learning_rate": 4.197761194029851e-05,
|
287 |
+
"loss": 0.8007,
|
288 |
+
"step": 450
|
289 |
+
},
|
290 |
+
{
|
291 |
+
"epoch": 1.71,
|
292 |
+
"learning_rate": 4.2910447761194036e-05,
|
293 |
+
"loss": 0.7467,
|
294 |
+
"step": 460
|
295 |
+
},
|
296 |
+
{
|
297 |
+
"epoch": 1.75,
|
298 |
+
"learning_rate": 4.384328358208955e-05,
|
299 |
+
"loss": 0.7646,
|
300 |
+
"step": 470
|
301 |
+
},
|
302 |
+
{
|
303 |
+
"epoch": 1.79,
|
304 |
+
"learning_rate": 4.477611940298508e-05,
|
305 |
+
"loss": 0.7797,
|
306 |
+
"step": 480
|
307 |
+
},
|
308 |
+
{
|
309 |
+
"epoch": 1.82,
|
310 |
+
"learning_rate": 4.57089552238806e-05,
|
311 |
+
"loss": 0.7368,
|
312 |
+
"step": 490
|
313 |
+
},
|
314 |
+
{
|
315 |
+
"epoch": 1.86,
|
316 |
+
"learning_rate": 4.664179104477612e-05,
|
317 |
+
"loss": 0.7156,
|
318 |
+
"step": 500
|
319 |
+
},
|
320 |
+
{
|
321 |
+
"epoch": 1.9,
|
322 |
+
"learning_rate": 4.7574626865671644e-05,
|
323 |
+
"loss": 0.7469,
|
324 |
+
"step": 510
|
325 |
+
},
|
326 |
+
{
|
327 |
+
"epoch": 1.93,
|
328 |
+
"learning_rate": 4.850746268656717e-05,
|
329 |
+
"loss": 0.7605,
|
330 |
+
"step": 520
|
331 |
+
},
|
332 |
+
{
|
333 |
+
"epoch": 1.97,
|
334 |
+
"learning_rate": 4.944029850746269e-05,
|
335 |
+
"loss": 0.7235,
|
336 |
+
"step": 530
|
337 |
+
},
|
338 |
+
{
|
339 |
+
"epoch": 2.0,
|
340 |
+
"eval_accuracy": 0.6695402298850575,
|
341 |
+
"eval_loss": 0.7695909738540649,
|
342 |
+
"eval_runtime": 8.0334,
|
343 |
+
"eval_samples_per_second": 43.319,
|
344 |
+
"eval_steps_per_second": 1.369,
|
345 |
+
"step": 537
|
346 |
+
},
|
347 |
+
{
|
348 |
+
"epoch": 2.01,
|
349 |
+
"learning_rate": 4.995854063018242e-05,
|
350 |
+
"loss": 0.7533,
|
351 |
+
"step": 540
|
352 |
+
},
|
353 |
+
{
|
354 |
+
"epoch": 2.05,
|
355 |
+
"learning_rate": 4.985489220563848e-05,
|
356 |
+
"loss": 0.7302,
|
357 |
+
"step": 550
|
358 |
+
},
|
359 |
+
{
|
360 |
+
"epoch": 2.08,
|
361 |
+
"learning_rate": 4.975124378109453e-05,
|
362 |
+
"loss": 0.6607,
|
363 |
+
"step": 560
|
364 |
+
},
|
365 |
+
{
|
366 |
+
"epoch": 2.12,
|
367 |
+
"learning_rate": 4.964759535655058e-05,
|
368 |
+
"loss": 0.7344,
|
369 |
+
"step": 570
|
370 |
+
},
|
371 |
+
{
|
372 |
+
"epoch": 2.16,
|
373 |
+
"learning_rate": 4.954394693200663e-05,
|
374 |
+
"loss": 0.7249,
|
375 |
+
"step": 580
|
376 |
+
},
|
377 |
+
{
|
378 |
+
"epoch": 2.2,
|
379 |
+
"learning_rate": 4.944029850746269e-05,
|
380 |
+
"loss": 0.6936,
|
381 |
+
"step": 590
|
382 |
+
},
|
383 |
+
{
|
384 |
+
"epoch": 2.23,
|
385 |
+
"learning_rate": 4.933665008291874e-05,
|
386 |
+
"loss": 0.7419,
|
387 |
+
"step": 600
|
388 |
+
},
|
389 |
+
{
|
390 |
+
"epoch": 2.27,
|
391 |
+
"learning_rate": 4.9233001658374794e-05,
|
392 |
+
"loss": 0.7075,
|
393 |
+
"step": 610
|
394 |
+
},
|
395 |
+
{
|
396 |
+
"epoch": 2.31,
|
397 |
+
"learning_rate": 4.912935323383085e-05,
|
398 |
+
"loss": 0.708,
|
399 |
+
"step": 620
|
400 |
+
},
|
401 |
+
{
|
402 |
+
"epoch": 2.34,
|
403 |
+
"learning_rate": 4.90257048092869e-05,
|
404 |
+
"loss": 0.6972,
|
405 |
+
"step": 630
|
406 |
+
},
|
407 |
+
{
|
408 |
+
"epoch": 2.38,
|
409 |
+
"learning_rate": 4.8922056384742954e-05,
|
410 |
+
"loss": 0.7685,
|
411 |
+
"step": 640
|
412 |
+
},
|
413 |
+
{
|
414 |
+
"epoch": 2.42,
|
415 |
+
"learning_rate": 4.881840796019901e-05,
|
416 |
+
"loss": 0.7331,
|
417 |
+
"step": 650
|
418 |
+
},
|
419 |
+
{
|
420 |
+
"epoch": 2.46,
|
421 |
+
"learning_rate": 4.8714759535655055e-05,
|
422 |
+
"loss": 0.7438,
|
423 |
+
"step": 660
|
424 |
+
},
|
425 |
+
{
|
426 |
+
"epoch": 2.49,
|
427 |
+
"learning_rate": 4.8611111111111115e-05,
|
428 |
+
"loss": 0.7604,
|
429 |
+
"step": 670
|
430 |
+
},
|
431 |
+
{
|
432 |
+
"epoch": 2.53,
|
433 |
+
"learning_rate": 4.850746268656717e-05,
|
434 |
+
"loss": 0.7316,
|
435 |
+
"step": 680
|
436 |
+
},
|
437 |
+
{
|
438 |
+
"epoch": 2.57,
|
439 |
+
"learning_rate": 4.8403814262023216e-05,
|
440 |
+
"loss": 0.744,
|
441 |
+
"step": 690
|
442 |
+
},
|
443 |
+
{
|
444 |
+
"epoch": 2.6,
|
445 |
+
"learning_rate": 4.830016583747927e-05,
|
446 |
+
"loss": 0.689,
|
447 |
+
"step": 700
|
448 |
+
},
|
449 |
+
{
|
450 |
+
"epoch": 2.64,
|
451 |
+
"learning_rate": 4.819651741293533e-05,
|
452 |
+
"loss": 0.7152,
|
453 |
+
"step": 710
|
454 |
+
},
|
455 |
+
{
|
456 |
+
"epoch": 2.68,
|
457 |
+
"learning_rate": 4.8092868988391376e-05,
|
458 |
+
"loss": 0.7138,
|
459 |
+
"step": 720
|
460 |
+
},
|
461 |
+
{
|
462 |
+
"epoch": 2.72,
|
463 |
+
"learning_rate": 4.798922056384743e-05,
|
464 |
+
"loss": 0.7114,
|
465 |
+
"step": 730
|
466 |
+
},
|
467 |
+
{
|
468 |
+
"epoch": 2.75,
|
469 |
+
"learning_rate": 4.7885572139303484e-05,
|
470 |
+
"loss": 0.7164,
|
471 |
+
"step": 740
|
472 |
+
},
|
473 |
+
{
|
474 |
+
"epoch": 2.79,
|
475 |
+
"learning_rate": 4.778192371475954e-05,
|
476 |
+
"loss": 0.6972,
|
477 |
+
"step": 750
|
478 |
+
},
|
479 |
+
{
|
480 |
+
"epoch": 2.83,
|
481 |
+
"learning_rate": 4.767827529021559e-05,
|
482 |
+
"loss": 0.6989,
|
483 |
+
"step": 760
|
484 |
+
},
|
485 |
+
{
|
486 |
+
"epoch": 2.87,
|
487 |
+
"learning_rate": 4.7574626865671644e-05,
|
488 |
+
"loss": 0.7195,
|
489 |
+
"step": 770
|
490 |
+
},
|
491 |
+
{
|
492 |
+
"epoch": 2.9,
|
493 |
+
"learning_rate": 4.74709784411277e-05,
|
494 |
+
"loss": 0.7047,
|
495 |
+
"step": 780
|
496 |
+
},
|
497 |
+
{
|
498 |
+
"epoch": 2.94,
|
499 |
+
"learning_rate": 4.736733001658375e-05,
|
500 |
+
"loss": 0.714,
|
501 |
+
"step": 790
|
502 |
+
},
|
503 |
+
{
|
504 |
+
"epoch": 2.98,
|
505 |
+
"learning_rate": 4.7263681592039805e-05,
|
506 |
+
"loss": 0.6939,
|
507 |
+
"step": 800
|
508 |
+
},
|
509 |
+
{
|
510 |
+
"epoch": 3.0,
|
511 |
+
"eval_accuracy": 0.6724137931034483,
|
512 |
+
"eval_loss": 0.7427859902381897,
|
513 |
+
"eval_runtime": 7.9341,
|
514 |
+
"eval_samples_per_second": 43.861,
|
515 |
+
"eval_steps_per_second": 1.386,
|
516 |
+
"step": 806
|
517 |
+
},
|
518 |
+
{
|
519 |
+
"epoch": 3.01,
|
520 |
+
"learning_rate": 4.716003316749585e-05,
|
521 |
+
"loss": 0.7368,
|
522 |
+
"step": 810
|
523 |
+
},
|
524 |
+
{
|
525 |
+
"epoch": 3.05,
|
526 |
+
"learning_rate": 4.705638474295191e-05,
|
527 |
+
"loss": 0.6497,
|
528 |
+
"step": 820
|
529 |
+
},
|
530 |
+
{
|
531 |
+
"epoch": 3.09,
|
532 |
+
"learning_rate": 4.6952736318407966e-05,
|
533 |
+
"loss": 0.6358,
|
534 |
+
"step": 830
|
535 |
+
},
|
536 |
+
{
|
537 |
+
"epoch": 3.13,
|
538 |
+
"learning_rate": 4.684908789386401e-05,
|
539 |
+
"loss": 0.6734,
|
540 |
+
"step": 840
|
541 |
+
},
|
542 |
+
{
|
543 |
+
"epoch": 3.16,
|
544 |
+
"learning_rate": 4.6745439469320066e-05,
|
545 |
+
"loss": 0.698,
|
546 |
+
"step": 850
|
547 |
+
},
|
548 |
+
{
|
549 |
+
"epoch": 3.2,
|
550 |
+
"learning_rate": 4.664179104477612e-05,
|
551 |
+
"loss": 0.6064,
|
552 |
+
"step": 860
|
553 |
+
},
|
554 |
+
{
|
555 |
+
"epoch": 3.24,
|
556 |
+
"learning_rate": 4.6538142620232173e-05,
|
557 |
+
"loss": 0.6898,
|
558 |
+
"step": 870
|
559 |
+
},
|
560 |
+
{
|
561 |
+
"epoch": 3.27,
|
562 |
+
"learning_rate": 4.643449419568823e-05,
|
563 |
+
"loss": 0.6642,
|
564 |
+
"step": 880
|
565 |
+
},
|
566 |
+
{
|
567 |
+
"epoch": 3.31,
|
568 |
+
"learning_rate": 4.633084577114428e-05,
|
569 |
+
"loss": 0.6794,
|
570 |
+
"step": 890
|
571 |
+
},
|
572 |
+
{
|
573 |
+
"epoch": 3.35,
|
574 |
+
"learning_rate": 4.6227197346600334e-05,
|
575 |
+
"loss": 0.7127,
|
576 |
+
"step": 900
|
577 |
+
},
|
578 |
+
{
|
579 |
+
"epoch": 3.39,
|
580 |
+
"learning_rate": 4.612354892205639e-05,
|
581 |
+
"loss": 0.683,
|
582 |
+
"step": 910
|
583 |
+
},
|
584 |
+
{
|
585 |
+
"epoch": 3.42,
|
586 |
+
"learning_rate": 4.601990049751244e-05,
|
587 |
+
"loss": 0.6746,
|
588 |
+
"step": 920
|
589 |
+
},
|
590 |
+
{
|
591 |
+
"epoch": 3.46,
|
592 |
+
"learning_rate": 4.591625207296849e-05,
|
593 |
+
"loss": 0.7107,
|
594 |
+
"step": 930
|
595 |
+
},
|
596 |
+
{
|
597 |
+
"epoch": 3.5,
|
598 |
+
"learning_rate": 4.581260364842455e-05,
|
599 |
+
"loss": 0.6652,
|
600 |
+
"step": 940
|
601 |
+
},
|
602 |
+
{
|
603 |
+
"epoch": 3.53,
|
604 |
+
"learning_rate": 4.57089552238806e-05,
|
605 |
+
"loss": 0.6877,
|
606 |
+
"step": 950
|
607 |
+
},
|
608 |
+
{
|
609 |
+
"epoch": 3.57,
|
610 |
+
"learning_rate": 4.560530679933665e-05,
|
611 |
+
"loss": 0.7108,
|
612 |
+
"step": 960
|
613 |
+
},
|
614 |
+
{
|
615 |
+
"epoch": 3.61,
|
616 |
+
"learning_rate": 4.55016583747927e-05,
|
617 |
+
"loss": 0.696,
|
618 |
+
"step": 970
|
619 |
+
},
|
620 |
+
{
|
621 |
+
"epoch": 3.65,
|
622 |
+
"learning_rate": 4.539800995024876e-05,
|
623 |
+
"loss": 0.6727,
|
624 |
+
"step": 980
|
625 |
+
},
|
626 |
+
{
|
627 |
+
"epoch": 3.68,
|
628 |
+
"learning_rate": 4.529436152570481e-05,
|
629 |
+
"loss": 0.6806,
|
630 |
+
"step": 990
|
631 |
+
},
|
632 |
+
{
|
633 |
+
"epoch": 3.72,
|
634 |
+
"learning_rate": 4.5190713101160863e-05,
|
635 |
+
"loss": 0.6604,
|
636 |
+
"step": 1000
|
637 |
+
},
|
638 |
+
{
|
639 |
+
"epoch": 3.76,
|
640 |
+
"learning_rate": 4.508706467661692e-05,
|
641 |
+
"loss": 0.6893,
|
642 |
+
"step": 1010
|
643 |
+
},
|
644 |
+
{
|
645 |
+
"epoch": 3.8,
|
646 |
+
"learning_rate": 4.498341625207297e-05,
|
647 |
+
"loss": 0.711,
|
648 |
+
"step": 1020
|
649 |
+
},
|
650 |
+
{
|
651 |
+
"epoch": 3.83,
|
652 |
+
"learning_rate": 4.4879767827529024e-05,
|
653 |
+
"loss": 0.691,
|
654 |
+
"step": 1030
|
655 |
+
},
|
656 |
+
{
|
657 |
+
"epoch": 3.87,
|
658 |
+
"learning_rate": 4.477611940298508e-05,
|
659 |
+
"loss": 0.6595,
|
660 |
+
"step": 1040
|
661 |
+
},
|
662 |
+
{
|
663 |
+
"epoch": 3.91,
|
664 |
+
"learning_rate": 4.4672470978441125e-05,
|
665 |
+
"loss": 0.664,
|
666 |
+
"step": 1050
|
667 |
+
},
|
668 |
+
{
|
669 |
+
"epoch": 3.94,
|
670 |
+
"learning_rate": 4.4568822553897185e-05,
|
671 |
+
"loss": 0.6802,
|
672 |
+
"step": 1060
|
673 |
+
},
|
674 |
+
{
|
675 |
+
"epoch": 3.98,
|
676 |
+
"learning_rate": 4.446517412935324e-05,
|
677 |
+
"loss": 0.665,
|
678 |
+
"step": 1070
|
679 |
+
},
|
680 |
+
{
|
681 |
+
"epoch": 4.0,
|
682 |
+
"eval_accuracy": 0.7327586206896551,
|
683 |
+
"eval_loss": 0.6883933544158936,
|
684 |
+
"eval_runtime": 7.9722,
|
685 |
+
"eval_samples_per_second": 43.652,
|
686 |
+
"eval_steps_per_second": 1.38,
|
687 |
+
"step": 1075
|
688 |
+
},
|
689 |
+
{
|
690 |
+
"epoch": 4.02,
|
691 |
+
"learning_rate": 4.4361525704809285e-05,
|
692 |
+
"loss": 0.6826,
|
693 |
+
"step": 1080
|
694 |
+
},
|
695 |
+
{
|
696 |
+
"epoch": 4.06,
|
697 |
+
"learning_rate": 4.425787728026534e-05,
|
698 |
+
"loss": 0.6568,
|
699 |
+
"step": 1090
|
700 |
+
},
|
701 |
+
{
|
702 |
+
"epoch": 4.09,
|
703 |
+
"learning_rate": 4.41542288557214e-05,
|
704 |
+
"loss": 0.6656,
|
705 |
+
"step": 1100
|
706 |
+
},
|
707 |
+
{
|
708 |
+
"epoch": 4.13,
|
709 |
+
"learning_rate": 4.4050580431177446e-05,
|
710 |
+
"loss": 0.6584,
|
711 |
+
"step": 1110
|
712 |
+
},
|
713 |
+
{
|
714 |
+
"epoch": 4.17,
|
715 |
+
"learning_rate": 4.39469320066335e-05,
|
716 |
+
"loss": 0.6941,
|
717 |
+
"step": 1120
|
718 |
+
},
|
719 |
+
{
|
720 |
+
"epoch": 4.2,
|
721 |
+
"learning_rate": 4.384328358208955e-05,
|
722 |
+
"loss": 0.6603,
|
723 |
+
"step": 1130
|
724 |
+
},
|
725 |
+
{
|
726 |
+
"epoch": 4.24,
|
727 |
+
"learning_rate": 4.373963515754561e-05,
|
728 |
+
"loss": 0.671,
|
729 |
+
"step": 1140
|
730 |
+
},
|
731 |
+
{
|
732 |
+
"epoch": 4.28,
|
733 |
+
"learning_rate": 4.363598673300166e-05,
|
734 |
+
"loss": 0.6237,
|
735 |
+
"step": 1150
|
736 |
+
},
|
737 |
+
{
|
738 |
+
"epoch": 4.32,
|
739 |
+
"learning_rate": 4.3532338308457714e-05,
|
740 |
+
"loss": 0.6486,
|
741 |
+
"step": 1160
|
742 |
+
},
|
743 |
+
{
|
744 |
+
"epoch": 4.35,
|
745 |
+
"learning_rate": 4.342868988391377e-05,
|
746 |
+
"loss": 0.6779,
|
747 |
+
"step": 1170
|
748 |
+
},
|
749 |
+
{
|
750 |
+
"epoch": 4.39,
|
751 |
+
"learning_rate": 4.332504145936982e-05,
|
752 |
+
"loss": 0.6249,
|
753 |
+
"step": 1180
|
754 |
+
},
|
755 |
+
{
|
756 |
+
"epoch": 4.43,
|
757 |
+
"learning_rate": 4.3221393034825875e-05,
|
758 |
+
"loss": 0.6742,
|
759 |
+
"step": 1190
|
760 |
+
},
|
761 |
+
{
|
762 |
+
"epoch": 4.47,
|
763 |
+
"learning_rate": 4.311774461028192e-05,
|
764 |
+
"loss": 0.6204,
|
765 |
+
"step": 1200
|
766 |
+
},
|
767 |
+
{
|
768 |
+
"epoch": 4.5,
|
769 |
+
"learning_rate": 4.301409618573798e-05,
|
770 |
+
"loss": 0.6319,
|
771 |
+
"step": 1210
|
772 |
+
},
|
773 |
+
{
|
774 |
+
"epoch": 4.54,
|
775 |
+
"learning_rate": 4.2910447761194036e-05,
|
776 |
+
"loss": 0.6557,
|
777 |
+
"step": 1220
|
778 |
+
},
|
779 |
+
{
|
780 |
+
"epoch": 4.58,
|
781 |
+
"learning_rate": 4.280679933665008e-05,
|
782 |
+
"loss": 0.6654,
|
783 |
+
"step": 1230
|
784 |
+
},
|
785 |
+
{
|
786 |
+
"epoch": 4.61,
|
787 |
+
"learning_rate": 4.2703150912106136e-05,
|
788 |
+
"loss": 0.6299,
|
789 |
+
"step": 1240
|
790 |
+
},
|
791 |
+
{
|
792 |
+
"epoch": 4.65,
|
793 |
+
"learning_rate": 4.259950248756219e-05,
|
794 |
+
"loss": 0.6283,
|
795 |
+
"step": 1250
|
796 |
+
},
|
797 |
+
{
|
798 |
+
"epoch": 4.69,
|
799 |
+
"learning_rate": 4.249585406301824e-05,
|
800 |
+
"loss": 0.6497,
|
801 |
+
"step": 1260
|
802 |
+
},
|
803 |
+
{
|
804 |
+
"epoch": 4.73,
|
805 |
+
"learning_rate": 4.23922056384743e-05,
|
806 |
+
"loss": 0.6255,
|
807 |
+
"step": 1270
|
808 |
+
},
|
809 |
+
{
|
810 |
+
"epoch": 4.76,
|
811 |
+
"learning_rate": 4.228855721393035e-05,
|
812 |
+
"loss": 0.6651,
|
813 |
+
"step": 1280
|
814 |
+
},
|
815 |
+
{
|
816 |
+
"epoch": 4.8,
|
817 |
+
"learning_rate": 4.2184908789386404e-05,
|
818 |
+
"loss": 0.6903,
|
819 |
+
"step": 1290
|
820 |
+
},
|
821 |
+
{
|
822 |
+
"epoch": 4.84,
|
823 |
+
"learning_rate": 4.208126036484246e-05,
|
824 |
+
"loss": 0.6359,
|
825 |
+
"step": 1300
|
826 |
+
},
|
827 |
+
{
|
828 |
+
"epoch": 4.87,
|
829 |
+
"learning_rate": 4.197761194029851e-05,
|
830 |
+
"loss": 0.6503,
|
831 |
+
"step": 1310
|
832 |
+
},
|
833 |
+
{
|
834 |
+
"epoch": 4.91,
|
835 |
+
"learning_rate": 4.187396351575456e-05,
|
836 |
+
"loss": 0.6313,
|
837 |
+
"step": 1320
|
838 |
+
},
|
839 |
+
{
|
840 |
+
"epoch": 4.95,
|
841 |
+
"learning_rate": 4.177031509121062e-05,
|
842 |
+
"loss": 0.6241,
|
843 |
+
"step": 1330
|
844 |
+
},
|
845 |
+
{
|
846 |
+
"epoch": 4.99,
|
847 |
+
"learning_rate": 4.166666666666667e-05,
|
848 |
+
"loss": 0.6846,
|
849 |
+
"step": 1340
|
850 |
+
},
|
851 |
+
{
|
852 |
+
"epoch": 5.0,
|
853 |
+
"eval_accuracy": 0.6954022988505747,
|
854 |
+
"eval_loss": 0.7143946886062622,
|
855 |
+
"eval_runtime": 7.9668,
|
856 |
+
"eval_samples_per_second": 43.681,
|
857 |
+
"eval_steps_per_second": 1.381,
|
858 |
+
"step": 1343
|
859 |
+
},
|
860 |
+
{
|
861 |
+
"epoch": 5.02,
|
862 |
+
"learning_rate": 4.156301824212272e-05,
|
863 |
+
"loss": 0.6513,
|
864 |
+
"step": 1350
|
865 |
+
},
|
866 |
+
{
|
867 |
+
"epoch": 5.06,
|
868 |
+
"learning_rate": 4.145936981757877e-05,
|
869 |
+
"loss": 0.6148,
|
870 |
+
"step": 1360
|
871 |
+
},
|
872 |
+
{
|
873 |
+
"epoch": 5.1,
|
874 |
+
"learning_rate": 4.135572139303483e-05,
|
875 |
+
"loss": 0.5983,
|
876 |
+
"step": 1370
|
877 |
+
},
|
878 |
+
{
|
879 |
+
"epoch": 5.13,
|
880 |
+
"learning_rate": 4.125207296849088e-05,
|
881 |
+
"loss": 0.6122,
|
882 |
+
"step": 1380
|
883 |
+
},
|
884 |
+
{
|
885 |
+
"epoch": 5.17,
|
886 |
+
"learning_rate": 4.114842454394693e-05,
|
887 |
+
"loss": 0.6419,
|
888 |
+
"step": 1390
|
889 |
+
},
|
890 |
+
{
|
891 |
+
"epoch": 5.21,
|
892 |
+
"learning_rate": 4.104477611940299e-05,
|
893 |
+
"loss": 0.621,
|
894 |
+
"step": 1400
|
895 |
+
},
|
896 |
+
{
|
897 |
+
"epoch": 5.25,
|
898 |
+
"learning_rate": 4.094112769485904e-05,
|
899 |
+
"loss": 0.6155,
|
900 |
+
"step": 1410
|
901 |
+
},
|
902 |
+
{
|
903 |
+
"epoch": 5.28,
|
904 |
+
"learning_rate": 4.0837479270315094e-05,
|
905 |
+
"loss": 0.6246,
|
906 |
+
"step": 1420
|
907 |
+
},
|
908 |
+
{
|
909 |
+
"epoch": 5.32,
|
910 |
+
"learning_rate": 4.073383084577115e-05,
|
911 |
+
"loss": 0.6416,
|
912 |
+
"step": 1430
|
913 |
+
},
|
914 |
+
{
|
915 |
+
"epoch": 5.36,
|
916 |
+
"learning_rate": 4.0630182421227194e-05,
|
917 |
+
"loss": 0.5965,
|
918 |
+
"step": 1440
|
919 |
+
},
|
920 |
+
{
|
921 |
+
"epoch": 5.4,
|
922 |
+
"learning_rate": 4.0526533996683255e-05,
|
923 |
+
"loss": 0.6338,
|
924 |
+
"step": 1450
|
925 |
+
},
|
926 |
+
{
|
927 |
+
"epoch": 5.43,
|
928 |
+
"learning_rate": 4.042288557213931e-05,
|
929 |
+
"loss": 0.608,
|
930 |
+
"step": 1460
|
931 |
+
},
|
932 |
+
{
|
933 |
+
"epoch": 5.47,
|
934 |
+
"learning_rate": 4.0319237147595355e-05,
|
935 |
+
"loss": 0.6114,
|
936 |
+
"step": 1470
|
937 |
+
},
|
938 |
+
{
|
939 |
+
"epoch": 5.51,
|
940 |
+
"learning_rate": 4.021558872305141e-05,
|
941 |
+
"loss": 0.5846,
|
942 |
+
"step": 1480
|
943 |
+
},
|
944 |
+
{
|
945 |
+
"epoch": 5.54,
|
946 |
+
"learning_rate": 4.011194029850747e-05,
|
947 |
+
"loss": 0.5931,
|
948 |
+
"step": 1490
|
949 |
+
},
|
950 |
+
{
|
951 |
+
"epoch": 5.58,
|
952 |
+
"learning_rate": 4.0008291873963516e-05,
|
953 |
+
"loss": 0.6483,
|
954 |
+
"step": 1500
|
955 |
+
},
|
956 |
+
{
|
957 |
+
"epoch": 5.62,
|
958 |
+
"learning_rate": 3.990464344941957e-05,
|
959 |
+
"loss": 0.6507,
|
960 |
+
"step": 1510
|
961 |
+
},
|
962 |
+
{
|
963 |
+
"epoch": 5.66,
|
964 |
+
"learning_rate": 3.980099502487562e-05,
|
965 |
+
"loss": 0.6008,
|
966 |
+
"step": 1520
|
967 |
+
},
|
968 |
+
{
|
969 |
+
"epoch": 5.69,
|
970 |
+
"learning_rate": 3.969734660033168e-05,
|
971 |
+
"loss": 0.6521,
|
972 |
+
"step": 1530
|
973 |
+
},
|
974 |
+
{
|
975 |
+
"epoch": 5.73,
|
976 |
+
"learning_rate": 3.959369817578773e-05,
|
977 |
+
"loss": 0.5868,
|
978 |
+
"step": 1540
|
979 |
+
},
|
980 |
+
{
|
981 |
+
"epoch": 5.77,
|
982 |
+
"learning_rate": 3.9490049751243784e-05,
|
983 |
+
"loss": 0.6469,
|
984 |
+
"step": 1550
|
985 |
+
},
|
986 |
+
{
|
987 |
+
"epoch": 5.8,
|
988 |
+
"learning_rate": 3.938640132669984e-05,
|
989 |
+
"loss": 0.6324,
|
990 |
+
"step": 1560
|
991 |
+
},
|
992 |
+
{
|
993 |
+
"epoch": 5.84,
|
994 |
+
"learning_rate": 3.928275290215589e-05,
|
995 |
+
"loss": 0.677,
|
996 |
+
"step": 1570
|
997 |
+
},
|
998 |
+
{
|
999 |
+
"epoch": 5.88,
|
1000 |
+
"learning_rate": 3.9179104477611945e-05,
|
1001 |
+
"loss": 0.6405,
|
1002 |
+
"step": 1580
|
1003 |
+
},
|
1004 |
+
{
|
1005 |
+
"epoch": 5.92,
|
1006 |
+
"learning_rate": 3.907545605306799e-05,
|
1007 |
+
"loss": 0.6127,
|
1008 |
+
"step": 1590
|
1009 |
+
},
|
1010 |
+
{
|
1011 |
+
"epoch": 5.95,
|
1012 |
+
"learning_rate": 3.8971807628524045e-05,
|
1013 |
+
"loss": 0.6458,
|
1014 |
+
"step": 1600
|
1015 |
+
},
|
1016 |
+
{
|
1017 |
+
"epoch": 5.99,
|
1018 |
+
"learning_rate": 3.8868159203980106e-05,
|
1019 |
+
"loss": 0.6391,
|
1020 |
+
"step": 1610
|
1021 |
+
},
|
1022 |
+
{
|
1023 |
+
"epoch": 6.0,
|
1024 |
+
"eval_accuracy": 0.7155172413793104,
|
1025 |
+
"eval_loss": 0.6854252219200134,
|
1026 |
+
"eval_runtime": 7.8664,
|
1027 |
+
"eval_samples_per_second": 44.239,
|
1028 |
+
"eval_steps_per_second": 1.398,
|
1029 |
+
"step": 1612
|
1030 |
+
},
|
1031 |
+
{
|
1032 |
+
"epoch": 6.03,
|
1033 |
+
"learning_rate": 3.876451077943615e-05,
|
1034 |
+
"loss": 0.5935,
|
1035 |
+
"step": 1620
|
1036 |
+
},
|
1037 |
+
{
|
1038 |
+
"epoch": 6.07,
|
1039 |
+
"learning_rate": 3.8660862354892206e-05,
|
1040 |
+
"loss": 0.6403,
|
1041 |
+
"step": 1630
|
1042 |
+
},
|
1043 |
+
{
|
1044 |
+
"epoch": 6.1,
|
1045 |
+
"learning_rate": 3.855721393034826e-05,
|
1046 |
+
"loss": 0.5986,
|
1047 |
+
"step": 1640
|
1048 |
+
},
|
1049 |
+
{
|
1050 |
+
"epoch": 6.14,
|
1051 |
+
"learning_rate": 3.845356550580431e-05,
|
1052 |
+
"loss": 0.586,
|
1053 |
+
"step": 1650
|
1054 |
+
},
|
1055 |
+
{
|
1056 |
+
"epoch": 6.18,
|
1057 |
+
"learning_rate": 3.834991708126037e-05,
|
1058 |
+
"loss": 0.6035,
|
1059 |
+
"step": 1660
|
1060 |
+
},
|
1061 |
+
{
|
1062 |
+
"epoch": 6.21,
|
1063 |
+
"learning_rate": 3.824626865671642e-05,
|
1064 |
+
"loss": 0.5918,
|
1065 |
+
"step": 1670
|
1066 |
+
},
|
1067 |
+
{
|
1068 |
+
"epoch": 6.25,
|
1069 |
+
"learning_rate": 3.8142620232172474e-05,
|
1070 |
+
"loss": 0.634,
|
1071 |
+
"step": 1680
|
1072 |
+
},
|
1073 |
+
{
|
1074 |
+
"epoch": 6.29,
|
1075 |
+
"learning_rate": 3.803897180762853e-05,
|
1076 |
+
"loss": 0.5972,
|
1077 |
+
"step": 1690
|
1078 |
+
},
|
1079 |
+
{
|
1080 |
+
"epoch": 6.33,
|
1081 |
+
"learning_rate": 3.793532338308458e-05,
|
1082 |
+
"loss": 0.6018,
|
1083 |
+
"step": 1700
|
1084 |
+
},
|
1085 |
+
{
|
1086 |
+
"epoch": 6.36,
|
1087 |
+
"learning_rate": 3.783167495854063e-05,
|
1088 |
+
"loss": 0.6128,
|
1089 |
+
"step": 1710
|
1090 |
+
},
|
1091 |
+
{
|
1092 |
+
"epoch": 6.4,
|
1093 |
+
"learning_rate": 3.772802653399669e-05,
|
1094 |
+
"loss": 0.5853,
|
1095 |
+
"step": 1720
|
1096 |
+
},
|
1097 |
+
{
|
1098 |
+
"epoch": 6.44,
|
1099 |
+
"learning_rate": 3.762437810945274e-05,
|
1100 |
+
"loss": 0.621,
|
1101 |
+
"step": 1730
|
1102 |
+
},
|
1103 |
+
{
|
1104 |
+
"epoch": 6.47,
|
1105 |
+
"learning_rate": 3.752072968490879e-05,
|
1106 |
+
"loss": 0.5998,
|
1107 |
+
"step": 1740
|
1108 |
+
},
|
1109 |
+
{
|
1110 |
+
"epoch": 6.51,
|
1111 |
+
"learning_rate": 3.741708126036484e-05,
|
1112 |
+
"loss": 0.6023,
|
1113 |
+
"step": 1750
|
1114 |
+
},
|
1115 |
+
{
|
1116 |
+
"epoch": 6.55,
|
1117 |
+
"learning_rate": 3.73134328358209e-05,
|
1118 |
+
"loss": 0.5987,
|
1119 |
+
"step": 1760
|
1120 |
+
},
|
1121 |
+
{
|
1122 |
+
"epoch": 6.59,
|
1123 |
+
"learning_rate": 3.720978441127695e-05,
|
1124 |
+
"loss": 0.5815,
|
1125 |
+
"step": 1770
|
1126 |
+
},
|
1127 |
+
{
|
1128 |
+
"epoch": 6.62,
|
1129 |
+
"learning_rate": 3.7106135986733e-05,
|
1130 |
+
"loss": 0.6187,
|
1131 |
+
"step": 1780
|
1132 |
+
},
|
1133 |
+
{
|
1134 |
+
"epoch": 6.66,
|
1135 |
+
"learning_rate": 3.700248756218906e-05,
|
1136 |
+
"loss": 0.6047,
|
1137 |
+
"step": 1790
|
1138 |
+
},
|
1139 |
+
{
|
1140 |
+
"epoch": 6.7,
|
1141 |
+
"learning_rate": 3.689883913764511e-05,
|
1142 |
+
"loss": 0.657,
|
1143 |
+
"step": 1800
|
1144 |
+
},
|
1145 |
+
{
|
1146 |
+
"epoch": 6.73,
|
1147 |
+
"learning_rate": 3.6795190713101164e-05,
|
1148 |
+
"loss": 0.5752,
|
1149 |
+
"step": 1810
|
1150 |
+
},
|
1151 |
+
{
|
1152 |
+
"epoch": 6.77,
|
1153 |
+
"learning_rate": 3.669154228855722e-05,
|
1154 |
+
"loss": 0.6431,
|
1155 |
+
"step": 1820
|
1156 |
+
},
|
1157 |
+
{
|
1158 |
+
"epoch": 6.81,
|
1159 |
+
"learning_rate": 3.6587893864013264e-05,
|
1160 |
+
"loss": 0.5964,
|
1161 |
+
"step": 1830
|
1162 |
+
},
|
1163 |
+
{
|
1164 |
+
"epoch": 6.85,
|
1165 |
+
"learning_rate": 3.6484245439469325e-05,
|
1166 |
+
"loss": 0.5821,
|
1167 |
+
"step": 1840
|
1168 |
+
},
|
1169 |
+
{
|
1170 |
+
"epoch": 6.88,
|
1171 |
+
"learning_rate": 3.638059701492538e-05,
|
1172 |
+
"loss": 0.6084,
|
1173 |
+
"step": 1850
|
1174 |
+
},
|
1175 |
+
{
|
1176 |
+
"epoch": 6.92,
|
1177 |
+
"learning_rate": 3.6276948590381425e-05,
|
1178 |
+
"loss": 0.606,
|
1179 |
+
"step": 1860
|
1180 |
+
},
|
1181 |
+
{
|
1182 |
+
"epoch": 6.96,
|
1183 |
+
"learning_rate": 3.617330016583748e-05,
|
1184 |
+
"loss": 0.6085,
|
1185 |
+
"step": 1870
|
1186 |
+
},
|
1187 |
+
{
|
1188 |
+
"epoch": 7.0,
|
1189 |
+
"learning_rate": 3.606965174129354e-05,
|
1190 |
+
"loss": 0.6172,
|
1191 |
+
"step": 1880
|
1192 |
+
},
|
1193 |
+
{
|
1194 |
+
"epoch": 7.0,
|
1195 |
+
"eval_accuracy": 0.7011494252873564,
|
1196 |
+
"eval_loss": 0.6697583794593811,
|
1197 |
+
"eval_runtime": 7.9293,
|
1198 |
+
"eval_samples_per_second": 43.888,
|
1199 |
+
"eval_steps_per_second": 1.387,
|
1200 |
+
"step": 1881
|
1201 |
+
},
|
1202 |
+
{
|
1203 |
+
"epoch": 7.03,
|
1204 |
+
"learning_rate": 3.5966003316749586e-05,
|
1205 |
+
"loss": 0.5844,
|
1206 |
+
"step": 1890
|
1207 |
+
},
|
1208 |
+
{
|
1209 |
+
"epoch": 7.07,
|
1210 |
+
"learning_rate": 3.586235489220564e-05,
|
1211 |
+
"loss": 0.5703,
|
1212 |
+
"step": 1900
|
1213 |
+
},
|
1214 |
+
{
|
1215 |
+
"epoch": 7.11,
|
1216 |
+
"learning_rate": 3.575870646766169e-05,
|
1217 |
+
"loss": 0.5721,
|
1218 |
+
"step": 1910
|
1219 |
+
},
|
1220 |
+
{
|
1221 |
+
"epoch": 7.14,
|
1222 |
+
"learning_rate": 3.565505804311775e-05,
|
1223 |
+
"loss": 0.5754,
|
1224 |
+
"step": 1920
|
1225 |
+
},
|
1226 |
+
{
|
1227 |
+
"epoch": 7.18,
|
1228 |
+
"learning_rate": 3.55514096185738e-05,
|
1229 |
+
"loss": 0.5667,
|
1230 |
+
"step": 1930
|
1231 |
+
},
|
1232 |
+
{
|
1233 |
+
"epoch": 7.22,
|
1234 |
+
"learning_rate": 3.5447761194029854e-05,
|
1235 |
+
"loss": 0.587,
|
1236 |
+
"step": 1940
|
1237 |
+
},
|
1238 |
+
{
|
1239 |
+
"epoch": 7.26,
|
1240 |
+
"learning_rate": 3.534411276948591e-05,
|
1241 |
+
"loss": 0.5664,
|
1242 |
+
"step": 1950
|
1243 |
+
},
|
1244 |
+
{
|
1245 |
+
"epoch": 7.29,
|
1246 |
+
"learning_rate": 3.524046434494196e-05,
|
1247 |
+
"loss": 0.5692,
|
1248 |
+
"step": 1960
|
1249 |
+
},
|
1250 |
+
{
|
1251 |
+
"epoch": 7.33,
|
1252 |
+
"learning_rate": 3.5136815920398015e-05,
|
1253 |
+
"loss": 0.6177,
|
1254 |
+
"step": 1970
|
1255 |
+
},
|
1256 |
+
{
|
1257 |
+
"epoch": 7.37,
|
1258 |
+
"learning_rate": 3.503316749585406e-05,
|
1259 |
+
"loss": 0.5813,
|
1260 |
+
"step": 1980
|
1261 |
+
},
|
1262 |
+
{
|
1263 |
+
"epoch": 7.4,
|
1264 |
+
"learning_rate": 3.4929519071310115e-05,
|
1265 |
+
"loss": 0.5878,
|
1266 |
+
"step": 1990
|
1267 |
+
},
|
1268 |
+
{
|
1269 |
+
"epoch": 7.44,
|
1270 |
+
"learning_rate": 3.4825870646766175e-05,
|
1271 |
+
"loss": 0.5915,
|
1272 |
+
"step": 2000
|
1273 |
+
},
|
1274 |
+
{
|
1275 |
+
"epoch": 7.48,
|
1276 |
+
"learning_rate": 3.472222222222222e-05,
|
1277 |
+
"loss": 0.5825,
|
1278 |
+
"step": 2010
|
1279 |
+
},
|
1280 |
+
{
|
1281 |
+
"epoch": 7.52,
|
1282 |
+
"learning_rate": 3.4618573797678276e-05,
|
1283 |
+
"loss": 0.5945,
|
1284 |
+
"step": 2020
|
1285 |
+
},
|
1286 |
+
{
|
1287 |
+
"epoch": 7.55,
|
1288 |
+
"learning_rate": 3.451492537313433e-05,
|
1289 |
+
"loss": 0.614,
|
1290 |
+
"step": 2030
|
1291 |
+
},
|
1292 |
+
{
|
1293 |
+
"epoch": 7.59,
|
1294 |
+
"learning_rate": 3.441127694859038e-05,
|
1295 |
+
"loss": 0.585,
|
1296 |
+
"step": 2040
|
1297 |
+
},
|
1298 |
+
{
|
1299 |
+
"epoch": 7.63,
|
1300 |
+
"learning_rate": 3.4307628524046437e-05,
|
1301 |
+
"loss": 0.5715,
|
1302 |
+
"step": 2050
|
1303 |
+
},
|
1304 |
+
{
|
1305 |
+
"epoch": 7.67,
|
1306 |
+
"learning_rate": 3.420398009950249e-05,
|
1307 |
+
"loss": 0.6171,
|
1308 |
+
"step": 2060
|
1309 |
+
},
|
1310 |
+
{
|
1311 |
+
"epoch": 7.7,
|
1312 |
+
"learning_rate": 3.4100331674958544e-05,
|
1313 |
+
"loss": 0.6081,
|
1314 |
+
"step": 2070
|
1315 |
+
},
|
1316 |
+
{
|
1317 |
+
"epoch": 7.74,
|
1318 |
+
"learning_rate": 3.39966832504146e-05,
|
1319 |
+
"loss": 0.6111,
|
1320 |
+
"step": 2080
|
1321 |
+
},
|
1322 |
+
{
|
1323 |
+
"epoch": 7.78,
|
1324 |
+
"learning_rate": 3.389303482587065e-05,
|
1325 |
+
"loss": 0.5688,
|
1326 |
+
"step": 2090
|
1327 |
+
},
|
1328 |
+
{
|
1329 |
+
"epoch": 7.81,
|
1330 |
+
"learning_rate": 3.37893864013267e-05,
|
1331 |
+
"loss": 0.5773,
|
1332 |
+
"step": 2100
|
1333 |
+
},
|
1334 |
+
{
|
1335 |
+
"epoch": 7.85,
|
1336 |
+
"learning_rate": 3.368573797678276e-05,
|
1337 |
+
"loss": 0.6,
|
1338 |
+
"step": 2110
|
1339 |
+
},
|
1340 |
+
{
|
1341 |
+
"epoch": 7.89,
|
1342 |
+
"learning_rate": 3.358208955223881e-05,
|
1343 |
+
"loss": 0.564,
|
1344 |
+
"step": 2120
|
1345 |
+
},
|
1346 |
+
{
|
1347 |
+
"epoch": 7.93,
|
1348 |
+
"learning_rate": 3.347844112769486e-05,
|
1349 |
+
"loss": 0.562,
|
1350 |
+
"step": 2130
|
1351 |
+
},
|
1352 |
+
{
|
1353 |
+
"epoch": 7.96,
|
1354 |
+
"learning_rate": 3.337479270315091e-05,
|
1355 |
+
"loss": 0.5699,
|
1356 |
+
"step": 2140
|
1357 |
+
},
|
1358 |
+
{
|
1359 |
+
"epoch": 8.0,
|
1360 |
+
"learning_rate": 3.327114427860697e-05,
|
1361 |
+
"loss": 0.6332,
|
1362 |
+
"step": 2150
|
1363 |
+
},
|
1364 |
+
{
|
1365 |
+
"epoch": 8.0,
|
1366 |
+
"eval_accuracy": 0.7126436781609196,
|
1367 |
+
"eval_loss": 0.6509921550750732,
|
1368 |
+
"eval_runtime": 7.7743,
|
1369 |
+
"eval_samples_per_second": 44.763,
|
1370 |
+
"eval_steps_per_second": 1.415,
|
1371 |
+
"step": 2150
|
1372 |
+
},
|
1373 |
+
{
|
1374 |
+
"epoch": 8.04,
|
1375 |
+
"learning_rate": 3.316749585406302e-05,
|
1376 |
+
"loss": 0.515,
|
1377 |
+
"step": 2160
|
1378 |
+
},
|
1379 |
+
{
|
1380 |
+
"epoch": 8.07,
|
1381 |
+
"learning_rate": 3.306384742951907e-05,
|
1382 |
+
"loss": 0.6073,
|
1383 |
+
"step": 2170
|
1384 |
+
},
|
1385 |
+
{
|
1386 |
+
"epoch": 8.11,
|
1387 |
+
"learning_rate": 3.2960199004975127e-05,
|
1388 |
+
"loss": 0.5479,
|
1389 |
+
"step": 2180
|
1390 |
+
},
|
1391 |
+
{
|
1392 |
+
"epoch": 8.15,
|
1393 |
+
"learning_rate": 3.285655058043118e-05,
|
1394 |
+
"loss": 0.5566,
|
1395 |
+
"step": 2190
|
1396 |
+
},
|
1397 |
+
{
|
1398 |
+
"epoch": 8.19,
|
1399 |
+
"learning_rate": 3.2752902155887234e-05,
|
1400 |
+
"loss": 0.5654,
|
1401 |
+
"step": 2200
|
1402 |
+
},
|
1403 |
+
{
|
1404 |
+
"epoch": 8.22,
|
1405 |
+
"learning_rate": 3.264925373134329e-05,
|
1406 |
+
"loss": 0.5673,
|
1407 |
+
"step": 2210
|
1408 |
+
},
|
1409 |
+
{
|
1410 |
+
"epoch": 8.26,
|
1411 |
+
"learning_rate": 3.2545605306799334e-05,
|
1412 |
+
"loss": 0.5645,
|
1413 |
+
"step": 2220
|
1414 |
+
},
|
1415 |
+
{
|
1416 |
+
"epoch": 8.3,
|
1417 |
+
"learning_rate": 3.2441956882255394e-05,
|
1418 |
+
"loss": 0.5641,
|
1419 |
+
"step": 2230
|
1420 |
+
},
|
1421 |
+
{
|
1422 |
+
"epoch": 8.33,
|
1423 |
+
"learning_rate": 3.233830845771145e-05,
|
1424 |
+
"loss": 0.5799,
|
1425 |
+
"step": 2240
|
1426 |
+
},
|
1427 |
+
{
|
1428 |
+
"epoch": 8.37,
|
1429 |
+
"learning_rate": 3.2234660033167495e-05,
|
1430 |
+
"loss": 0.5581,
|
1431 |
+
"step": 2250
|
1432 |
+
},
|
1433 |
+
{
|
1434 |
+
"epoch": 8.41,
|
1435 |
+
"learning_rate": 3.213101160862355e-05,
|
1436 |
+
"loss": 0.525,
|
1437 |
+
"step": 2260
|
1438 |
+
},
|
1439 |
+
{
|
1440 |
+
"epoch": 8.45,
|
1441 |
+
"learning_rate": 3.202736318407961e-05,
|
1442 |
+
"loss": 0.561,
|
1443 |
+
"step": 2270
|
1444 |
+
},
|
1445 |
+
{
|
1446 |
+
"epoch": 8.48,
|
1447 |
+
"learning_rate": 3.1923714759535656e-05,
|
1448 |
+
"loss": 0.5366,
|
1449 |
+
"step": 2280
|
1450 |
+
},
|
1451 |
+
{
|
1452 |
+
"epoch": 8.52,
|
1453 |
+
"learning_rate": 3.182006633499171e-05,
|
1454 |
+
"loss": 0.5483,
|
1455 |
+
"step": 2290
|
1456 |
+
},
|
1457 |
+
{
|
1458 |
+
"epoch": 8.56,
|
1459 |
+
"learning_rate": 3.171641791044776e-05,
|
1460 |
+
"loss": 0.5534,
|
1461 |
+
"step": 2300
|
1462 |
+
},
|
1463 |
+
{
|
1464 |
+
"epoch": 8.6,
|
1465 |
+
"learning_rate": 3.1612769485903816e-05,
|
1466 |
+
"loss": 0.5736,
|
1467 |
+
"step": 2310
|
1468 |
+
},
|
1469 |
+
{
|
1470 |
+
"epoch": 8.63,
|
1471 |
+
"learning_rate": 3.150912106135987e-05,
|
1472 |
+
"loss": 0.5819,
|
1473 |
+
"step": 2320
|
1474 |
+
},
|
1475 |
+
{
|
1476 |
+
"epoch": 8.67,
|
1477 |
+
"learning_rate": 3.1405472636815924e-05,
|
1478 |
+
"loss": 0.542,
|
1479 |
+
"step": 2330
|
1480 |
+
},
|
1481 |
+
{
|
1482 |
+
"epoch": 8.71,
|
1483 |
+
"learning_rate": 3.130182421227198e-05,
|
1484 |
+
"loss": 0.577,
|
1485 |
+
"step": 2340
|
1486 |
+
},
|
1487 |
+
{
|
1488 |
+
"epoch": 8.74,
|
1489 |
+
"learning_rate": 3.119817578772803e-05,
|
1490 |
+
"loss": 0.5382,
|
1491 |
+
"step": 2350
|
1492 |
+
},
|
1493 |
+
{
|
1494 |
+
"epoch": 8.78,
|
1495 |
+
"learning_rate": 3.109452736318408e-05,
|
1496 |
+
"loss": 0.5709,
|
1497 |
+
"step": 2360
|
1498 |
+
},
|
1499 |
+
{
|
1500 |
+
"epoch": 8.82,
|
1501 |
+
"learning_rate": 3.099087893864013e-05,
|
1502 |
+
"loss": 0.6235,
|
1503 |
+
"step": 2370
|
1504 |
+
},
|
1505 |
+
{
|
1506 |
+
"epoch": 8.86,
|
1507 |
+
"learning_rate": 3.0887230514096185e-05,
|
1508 |
+
"loss": 0.5996,
|
1509 |
+
"step": 2380
|
1510 |
+
},
|
1511 |
+
{
|
1512 |
+
"epoch": 8.89,
|
1513 |
+
"learning_rate": 3.078358208955224e-05,
|
1514 |
+
"loss": 0.573,
|
1515 |
+
"step": 2390
|
1516 |
+
},
|
1517 |
+
{
|
1518 |
+
"epoch": 8.93,
|
1519 |
+
"learning_rate": 3.067993366500829e-05,
|
1520 |
+
"loss": 0.5648,
|
1521 |
+
"step": 2400
|
1522 |
+
},
|
1523 |
+
{
|
1524 |
+
"epoch": 8.97,
|
1525 |
+
"learning_rate": 3.0576285240464346e-05,
|
1526 |
+
"loss": 0.5679,
|
1527 |
+
"step": 2410
|
1528 |
+
},
|
1529 |
+
{
|
1530 |
+
"epoch": 9.0,
|
1531 |
+
"eval_accuracy": 0.7298850574712644,
|
1532 |
+
"eval_loss": 0.6322771310806274,
|
1533 |
+
"eval_runtime": 7.9472,
|
1534 |
+
"eval_samples_per_second": 43.789,
|
1535 |
+
"eval_steps_per_second": 1.384,
|
1536 |
+
"step": 2418
|
1537 |
+
},
|
1538 |
+
{
|
1539 |
+
"epoch": 9.0,
|
1540 |
+
"learning_rate": 3.0472636815920396e-05,
|
1541 |
+
"loss": 0.5413,
|
1542 |
+
"step": 2420
|
1543 |
+
},
|
1544 |
+
{
|
1545 |
+
"epoch": 9.04,
|
1546 |
+
"learning_rate": 3.0368988391376453e-05,
|
1547 |
+
"loss": 0.5374,
|
1548 |
+
"step": 2430
|
1549 |
+
},
|
1550 |
+
{
|
1551 |
+
"epoch": 9.08,
|
1552 |
+
"learning_rate": 3.0265339966832506e-05,
|
1553 |
+
"loss": 0.5222,
|
1554 |
+
"step": 2440
|
1555 |
+
},
|
1556 |
+
{
|
1557 |
+
"epoch": 9.12,
|
1558 |
+
"learning_rate": 3.0161691542288557e-05,
|
1559 |
+
"loss": 0.5412,
|
1560 |
+
"step": 2450
|
1561 |
+
},
|
1562 |
+
{
|
1563 |
+
"epoch": 9.15,
|
1564 |
+
"learning_rate": 3.005804311774461e-05,
|
1565 |
+
"loss": 0.5444,
|
1566 |
+
"step": 2460
|
1567 |
+
},
|
1568 |
+
{
|
1569 |
+
"epoch": 9.19,
|
1570 |
+
"learning_rate": 2.9954394693200667e-05,
|
1571 |
+
"loss": 0.5489,
|
1572 |
+
"step": 2470
|
1573 |
+
},
|
1574 |
+
{
|
1575 |
+
"epoch": 9.23,
|
1576 |
+
"learning_rate": 2.9850746268656714e-05,
|
1577 |
+
"loss": 0.5491,
|
1578 |
+
"step": 2480
|
1579 |
+
},
|
1580 |
+
{
|
1581 |
+
"epoch": 9.27,
|
1582 |
+
"learning_rate": 2.974709784411277e-05,
|
1583 |
+
"loss": 0.5785,
|
1584 |
+
"step": 2490
|
1585 |
+
},
|
1586 |
+
{
|
1587 |
+
"epoch": 9.3,
|
1588 |
+
"learning_rate": 2.9643449419568825e-05,
|
1589 |
+
"loss": 0.5328,
|
1590 |
+
"step": 2500
|
1591 |
+
},
|
1592 |
+
{
|
1593 |
+
"epoch": 9.34,
|
1594 |
+
"learning_rate": 2.9539800995024875e-05,
|
1595 |
+
"loss": 0.5176,
|
1596 |
+
"step": 2510
|
1597 |
+
},
|
1598 |
+
{
|
1599 |
+
"epoch": 9.38,
|
1600 |
+
"learning_rate": 2.943615257048093e-05,
|
1601 |
+
"loss": 0.542,
|
1602 |
+
"step": 2520
|
1603 |
+
},
|
1604 |
+
{
|
1605 |
+
"epoch": 9.41,
|
1606 |
+
"learning_rate": 2.9332504145936985e-05,
|
1607 |
+
"loss": 0.5868,
|
1608 |
+
"step": 2530
|
1609 |
+
},
|
1610 |
+
{
|
1611 |
+
"epoch": 9.45,
|
1612 |
+
"learning_rate": 2.9228855721393032e-05,
|
1613 |
+
"loss": 0.5542,
|
1614 |
+
"step": 2540
|
1615 |
+
},
|
1616 |
+
{
|
1617 |
+
"epoch": 9.49,
|
1618 |
+
"learning_rate": 2.912520729684909e-05,
|
1619 |
+
"loss": 0.5843,
|
1620 |
+
"step": 2550
|
1621 |
+
},
|
1622 |
+
{
|
1623 |
+
"epoch": 9.53,
|
1624 |
+
"learning_rate": 2.9021558872305143e-05,
|
1625 |
+
"loss": 0.5357,
|
1626 |
+
"step": 2560
|
1627 |
+
},
|
1628 |
+
{
|
1629 |
+
"epoch": 9.56,
|
1630 |
+
"learning_rate": 2.8917910447761193e-05,
|
1631 |
+
"loss": 0.5512,
|
1632 |
+
"step": 2570
|
1633 |
+
},
|
1634 |
+
{
|
1635 |
+
"epoch": 9.6,
|
1636 |
+
"learning_rate": 2.8814262023217247e-05,
|
1637 |
+
"loss": 0.5022,
|
1638 |
+
"step": 2580
|
1639 |
+
},
|
1640 |
+
{
|
1641 |
+
"epoch": 9.64,
|
1642 |
+
"learning_rate": 2.8710613598673304e-05,
|
1643 |
+
"loss": 0.5884,
|
1644 |
+
"step": 2590
|
1645 |
+
},
|
1646 |
+
{
|
1647 |
+
"epoch": 9.67,
|
1648 |
+
"learning_rate": 2.8606965174129354e-05,
|
1649 |
+
"loss": 0.5547,
|
1650 |
+
"step": 2600
|
1651 |
+
},
|
1652 |
+
{
|
1653 |
+
"epoch": 9.71,
|
1654 |
+
"learning_rate": 2.8503316749585407e-05,
|
1655 |
+
"loss": 0.5201,
|
1656 |
+
"step": 2610
|
1657 |
+
},
|
1658 |
+
{
|
1659 |
+
"epoch": 9.75,
|
1660 |
+
"learning_rate": 2.839966832504146e-05,
|
1661 |
+
"loss": 0.5794,
|
1662 |
+
"step": 2620
|
1663 |
+
},
|
1664 |
+
{
|
1665 |
+
"epoch": 9.79,
|
1666 |
+
"learning_rate": 2.829601990049751e-05,
|
1667 |
+
"loss": 0.5359,
|
1668 |
+
"step": 2630
|
1669 |
+
},
|
1670 |
+
{
|
1671 |
+
"epoch": 9.82,
|
1672 |
+
"learning_rate": 2.8192371475953565e-05,
|
1673 |
+
"loss": 0.5503,
|
1674 |
+
"step": 2640
|
1675 |
+
},
|
1676 |
+
{
|
1677 |
+
"epoch": 9.86,
|
1678 |
+
"learning_rate": 2.8088723051409622e-05,
|
1679 |
+
"loss": 0.5548,
|
1680 |
+
"step": 2650
|
1681 |
+
},
|
1682 |
+
{
|
1683 |
+
"epoch": 9.9,
|
1684 |
+
"learning_rate": 2.7985074626865672e-05,
|
1685 |
+
"loss": 0.5461,
|
1686 |
+
"step": 2660
|
1687 |
+
},
|
1688 |
+
{
|
1689 |
+
"epoch": 9.93,
|
1690 |
+
"learning_rate": 2.7881426202321726e-05,
|
1691 |
+
"loss": 0.5332,
|
1692 |
+
"step": 2670
|
1693 |
+
},
|
1694 |
+
{
|
1695 |
+
"epoch": 9.97,
|
1696 |
+
"learning_rate": 2.777777777777778e-05,
|
1697 |
+
"loss": 0.5109,
|
1698 |
+
"step": 2680
|
1699 |
+
},
|
1700 |
+
{
|
1701 |
+
"epoch": 10.0,
|
1702 |
+
"eval_accuracy": 0.7097701149425287,
|
1703 |
+
"eval_loss": 0.6628959774971008,
|
1704 |
+
"eval_runtime": 7.7553,
|
1705 |
+
"eval_samples_per_second": 44.872,
|
1706 |
+
"eval_steps_per_second": 1.418,
|
1707 |
+
"step": 2687
|
1708 |
+
},
|
1709 |
+
{
|
1710 |
+
"epoch": 10.01,
|
1711 |
+
"learning_rate": 2.767412935323383e-05,
|
1712 |
+
"loss": 0.5619,
|
1713 |
+
"step": 2690
|
1714 |
+
},
|
1715 |
+
{
|
1716 |
+
"epoch": 10.05,
|
1717 |
+
"learning_rate": 2.7570480928689886e-05,
|
1718 |
+
"loss": 0.5471,
|
1719 |
+
"step": 2700
|
1720 |
+
},
|
1721 |
+
{
|
1722 |
+
"epoch": 10.08,
|
1723 |
+
"learning_rate": 2.746683250414594e-05,
|
1724 |
+
"loss": 0.5053,
|
1725 |
+
"step": 2710
|
1726 |
+
},
|
1727 |
+
{
|
1728 |
+
"epoch": 10.12,
|
1729 |
+
"learning_rate": 2.736318407960199e-05,
|
1730 |
+
"loss": 0.5172,
|
1731 |
+
"step": 2720
|
1732 |
+
},
|
1733 |
+
{
|
1734 |
+
"epoch": 10.16,
|
1735 |
+
"learning_rate": 2.7259535655058044e-05,
|
1736 |
+
"loss": 0.5415,
|
1737 |
+
"step": 2730
|
1738 |
+
},
|
1739 |
+
{
|
1740 |
+
"epoch": 10.2,
|
1741 |
+
"learning_rate": 2.7155887230514097e-05,
|
1742 |
+
"loss": 0.5219,
|
1743 |
+
"step": 2740
|
1744 |
+
},
|
1745 |
+
{
|
1746 |
+
"epoch": 10.23,
|
1747 |
+
"learning_rate": 2.7052238805970147e-05,
|
1748 |
+
"loss": 0.4921,
|
1749 |
+
"step": 2750
|
1750 |
+
},
|
1751 |
+
{
|
1752 |
+
"epoch": 10.27,
|
1753 |
+
"learning_rate": 2.6948590381426204e-05,
|
1754 |
+
"loss": 0.5044,
|
1755 |
+
"step": 2760
|
1756 |
+
},
|
1757 |
+
{
|
1758 |
+
"epoch": 10.31,
|
1759 |
+
"learning_rate": 2.6844941956882258e-05,
|
1760 |
+
"loss": 0.5319,
|
1761 |
+
"step": 2770
|
1762 |
+
},
|
1763 |
+
{
|
1764 |
+
"epoch": 10.34,
|
1765 |
+
"learning_rate": 2.6741293532338308e-05,
|
1766 |
+
"loss": 0.5005,
|
1767 |
+
"step": 2780
|
1768 |
+
},
|
1769 |
+
{
|
1770 |
+
"epoch": 10.38,
|
1771 |
+
"learning_rate": 2.6637645107794362e-05,
|
1772 |
+
"loss": 0.5261,
|
1773 |
+
"step": 2790
|
1774 |
+
},
|
1775 |
+
{
|
1776 |
+
"epoch": 10.42,
|
1777 |
+
"learning_rate": 2.653399668325042e-05,
|
1778 |
+
"loss": 0.5689,
|
1779 |
+
"step": 2800
|
1780 |
+
},
|
1781 |
+
{
|
1782 |
+
"epoch": 10.46,
|
1783 |
+
"learning_rate": 2.6430348258706466e-05,
|
1784 |
+
"loss": 0.5411,
|
1785 |
+
"step": 2810
|
1786 |
+
},
|
1787 |
+
{
|
1788 |
+
"epoch": 10.49,
|
1789 |
+
"learning_rate": 2.6326699834162523e-05,
|
1790 |
+
"loss": 0.5128,
|
1791 |
+
"step": 2820
|
1792 |
+
},
|
1793 |
+
{
|
1794 |
+
"epoch": 10.53,
|
1795 |
+
"learning_rate": 2.6223051409618576e-05,
|
1796 |
+
"loss": 0.502,
|
1797 |
+
"step": 2830
|
1798 |
+
},
|
1799 |
+
{
|
1800 |
+
"epoch": 10.57,
|
1801 |
+
"learning_rate": 2.6119402985074626e-05,
|
1802 |
+
"loss": 0.4911,
|
1803 |
+
"step": 2840
|
1804 |
+
},
|
1805 |
+
{
|
1806 |
+
"epoch": 10.6,
|
1807 |
+
"learning_rate": 2.601575456053068e-05,
|
1808 |
+
"loss": 0.491,
|
1809 |
+
"step": 2850
|
1810 |
+
},
|
1811 |
+
{
|
1812 |
+
"epoch": 10.64,
|
1813 |
+
"learning_rate": 2.5912106135986737e-05,
|
1814 |
+
"loss": 0.5436,
|
1815 |
+
"step": 2860
|
1816 |
+
},
|
1817 |
+
{
|
1818 |
+
"epoch": 10.68,
|
1819 |
+
"learning_rate": 2.5808457711442784e-05,
|
1820 |
+
"loss": 0.5418,
|
1821 |
+
"step": 2870
|
1822 |
+
},
|
1823 |
+
{
|
1824 |
+
"epoch": 10.72,
|
1825 |
+
"learning_rate": 2.570480928689884e-05,
|
1826 |
+
"loss": 0.5358,
|
1827 |
+
"step": 2880
|
1828 |
+
},
|
1829 |
+
{
|
1830 |
+
"epoch": 10.75,
|
1831 |
+
"learning_rate": 2.5601160862354894e-05,
|
1832 |
+
"loss": 0.4977,
|
1833 |
+
"step": 2890
|
1834 |
+
},
|
1835 |
+
{
|
1836 |
+
"epoch": 10.79,
|
1837 |
+
"learning_rate": 2.5497512437810945e-05,
|
1838 |
+
"loss": 0.5357,
|
1839 |
+
"step": 2900
|
1840 |
+
},
|
1841 |
+
{
|
1842 |
+
"epoch": 10.83,
|
1843 |
+
"learning_rate": 2.5393864013266998e-05,
|
1844 |
+
"loss": 0.5652,
|
1845 |
+
"step": 2910
|
1846 |
+
},
|
1847 |
+
{
|
1848 |
+
"epoch": 10.87,
|
1849 |
+
"learning_rate": 2.5290215588723055e-05,
|
1850 |
+
"loss": 0.5394,
|
1851 |
+
"step": 2920
|
1852 |
+
},
|
1853 |
+
{
|
1854 |
+
"epoch": 10.9,
|
1855 |
+
"learning_rate": 2.5186567164179102e-05,
|
1856 |
+
"loss": 0.5331,
|
1857 |
+
"step": 2930
|
1858 |
+
},
|
1859 |
+
{
|
1860 |
+
"epoch": 10.94,
|
1861 |
+
"learning_rate": 2.508291873963516e-05,
|
1862 |
+
"loss": 0.5152,
|
1863 |
+
"step": 2940
|
1864 |
+
},
|
1865 |
+
{
|
1866 |
+
"epoch": 10.98,
|
1867 |
+
"learning_rate": 2.497927031509121e-05,
|
1868 |
+
"loss": 0.5594,
|
1869 |
+
"step": 2950
|
1870 |
+
},
|
1871 |
+
{
|
1872 |
+
"epoch": 11.0,
|
1873 |
+
"eval_accuracy": 0.7270114942528736,
|
1874 |
+
"eval_loss": 0.6555773615837097,
|
1875 |
+
"eval_runtime": 7.8406,
|
1876 |
+
"eval_samples_per_second": 44.384,
|
1877 |
+
"eval_steps_per_second": 1.403,
|
1878 |
+
"step": 2956
|
1879 |
+
},
|
1880 |
+
{
|
1881 |
+
"epoch": 11.01,
|
1882 |
+
"learning_rate": 2.4875621890547266e-05,
|
1883 |
+
"loss": 0.5397,
|
1884 |
+
"step": 2960
|
1885 |
+
},
|
1886 |
+
{
|
1887 |
+
"epoch": 11.05,
|
1888 |
+
"learning_rate": 2.4771973466003316e-05,
|
1889 |
+
"loss": 0.5241,
|
1890 |
+
"step": 2970
|
1891 |
+
},
|
1892 |
+
{
|
1893 |
+
"epoch": 11.09,
|
1894 |
+
"learning_rate": 2.466832504145937e-05,
|
1895 |
+
"loss": 0.513,
|
1896 |
+
"step": 2980
|
1897 |
+
},
|
1898 |
+
{
|
1899 |
+
"epoch": 11.13,
|
1900 |
+
"learning_rate": 2.4564676616915424e-05,
|
1901 |
+
"loss": 0.4784,
|
1902 |
+
"step": 2990
|
1903 |
+
},
|
1904 |
+
{
|
1905 |
+
"epoch": 11.16,
|
1906 |
+
"learning_rate": 2.4461028192371477e-05,
|
1907 |
+
"loss": 0.4926,
|
1908 |
+
"step": 3000
|
1909 |
+
},
|
1910 |
+
{
|
1911 |
+
"epoch": 11.2,
|
1912 |
+
"learning_rate": 2.4357379767827527e-05,
|
1913 |
+
"loss": 0.4658,
|
1914 |
+
"step": 3010
|
1915 |
+
},
|
1916 |
+
{
|
1917 |
+
"epoch": 11.24,
|
1918 |
+
"learning_rate": 2.4253731343283584e-05,
|
1919 |
+
"loss": 0.5156,
|
1920 |
+
"step": 3020
|
1921 |
+
},
|
1922 |
+
{
|
1923 |
+
"epoch": 11.27,
|
1924 |
+
"learning_rate": 2.4150082918739635e-05,
|
1925 |
+
"loss": 0.4829,
|
1926 |
+
"step": 3030
|
1927 |
+
},
|
1928 |
+
{
|
1929 |
+
"epoch": 11.31,
|
1930 |
+
"learning_rate": 2.4046434494195688e-05,
|
1931 |
+
"loss": 0.5036,
|
1932 |
+
"step": 3040
|
1933 |
+
},
|
1934 |
+
{
|
1935 |
+
"epoch": 11.35,
|
1936 |
+
"learning_rate": 2.3942786069651742e-05,
|
1937 |
+
"loss": 0.4961,
|
1938 |
+
"step": 3050
|
1939 |
+
},
|
1940 |
+
{
|
1941 |
+
"epoch": 11.39,
|
1942 |
+
"learning_rate": 2.3839137645107795e-05,
|
1943 |
+
"loss": 0.5211,
|
1944 |
+
"step": 3060
|
1945 |
+
},
|
1946 |
+
{
|
1947 |
+
"epoch": 11.42,
|
1948 |
+
"learning_rate": 2.373548922056385e-05,
|
1949 |
+
"loss": 0.4947,
|
1950 |
+
"step": 3070
|
1951 |
+
},
|
1952 |
+
{
|
1953 |
+
"epoch": 11.46,
|
1954 |
+
"learning_rate": 2.3631840796019903e-05,
|
1955 |
+
"loss": 0.5268,
|
1956 |
+
"step": 3080
|
1957 |
+
},
|
1958 |
+
{
|
1959 |
+
"epoch": 11.5,
|
1960 |
+
"learning_rate": 2.3528192371475956e-05,
|
1961 |
+
"loss": 0.5437,
|
1962 |
+
"step": 3090
|
1963 |
+
},
|
1964 |
+
{
|
1965 |
+
"epoch": 11.53,
|
1966 |
+
"learning_rate": 2.3424543946932006e-05,
|
1967 |
+
"loss": 0.4871,
|
1968 |
+
"step": 3100
|
1969 |
+
},
|
1970 |
+
{
|
1971 |
+
"epoch": 11.57,
|
1972 |
+
"learning_rate": 2.332089552238806e-05,
|
1973 |
+
"loss": 0.5179,
|
1974 |
+
"step": 3110
|
1975 |
+
},
|
1976 |
+
{
|
1977 |
+
"epoch": 11.61,
|
1978 |
+
"learning_rate": 2.3217247097844114e-05,
|
1979 |
+
"loss": 0.5482,
|
1980 |
+
"step": 3120
|
1981 |
+
},
|
1982 |
+
{
|
1983 |
+
"epoch": 11.65,
|
1984 |
+
"learning_rate": 2.3113598673300167e-05,
|
1985 |
+
"loss": 0.5322,
|
1986 |
+
"step": 3130
|
1987 |
+
},
|
1988 |
+
{
|
1989 |
+
"epoch": 11.68,
|
1990 |
+
"learning_rate": 2.300995024875622e-05,
|
1991 |
+
"loss": 0.4994,
|
1992 |
+
"step": 3140
|
1993 |
+
},
|
1994 |
+
{
|
1995 |
+
"epoch": 11.72,
|
1996 |
+
"learning_rate": 2.2906301824212274e-05,
|
1997 |
+
"loss": 0.5002,
|
1998 |
+
"step": 3150
|
1999 |
+
},
|
2000 |
+
{
|
2001 |
+
"epoch": 11.76,
|
2002 |
+
"learning_rate": 2.2802653399668325e-05,
|
2003 |
+
"loss": 0.4943,
|
2004 |
+
"step": 3160
|
2005 |
+
},
|
2006 |
+
{
|
2007 |
+
"epoch": 11.8,
|
2008 |
+
"learning_rate": 2.269900497512438e-05,
|
2009 |
+
"loss": 0.4861,
|
2010 |
+
"step": 3170
|
2011 |
+
},
|
2012 |
+
{
|
2013 |
+
"epoch": 11.83,
|
2014 |
+
"learning_rate": 2.2595356550580432e-05,
|
2015 |
+
"loss": 0.5137,
|
2016 |
+
"step": 3180
|
2017 |
+
},
|
2018 |
+
{
|
2019 |
+
"epoch": 11.87,
|
2020 |
+
"learning_rate": 2.2491708126036485e-05,
|
2021 |
+
"loss": 0.4797,
|
2022 |
+
"step": 3190
|
2023 |
+
},
|
2024 |
+
{
|
2025 |
+
"epoch": 11.91,
|
2026 |
+
"learning_rate": 2.238805970149254e-05,
|
2027 |
+
"loss": 0.4931,
|
2028 |
+
"step": 3200
|
2029 |
+
},
|
2030 |
+
{
|
2031 |
+
"epoch": 11.94,
|
2032 |
+
"learning_rate": 2.2284411276948592e-05,
|
2033 |
+
"loss": 0.4794,
|
2034 |
+
"step": 3210
|
2035 |
+
},
|
2036 |
+
{
|
2037 |
+
"epoch": 11.98,
|
2038 |
+
"learning_rate": 2.2180762852404643e-05,
|
2039 |
+
"loss": 0.4874,
|
2040 |
+
"step": 3220
|
2041 |
+
},
|
2042 |
+
{
|
2043 |
+
"epoch": 12.0,
|
2044 |
+
"eval_accuracy": 0.7155172413793104,
|
2045 |
+
"eval_loss": 0.6626778841018677,
|
2046 |
+
"eval_runtime": 7.7968,
|
2047 |
+
"eval_samples_per_second": 44.634,
|
2048 |
+
"eval_steps_per_second": 1.411,
|
2049 |
+
"step": 3225
|
2050 |
+
},
|
2051 |
+
{
|
2052 |
+
"epoch": 12.02,
|
2053 |
+
"learning_rate": 2.20771144278607e-05,
|
2054 |
+
"loss": 0.4701,
|
2055 |
+
"step": 3230
|
2056 |
+
},
|
2057 |
+
{
|
2058 |
+
"epoch": 12.06,
|
2059 |
+
"learning_rate": 2.197346600331675e-05,
|
2060 |
+
"loss": 0.5069,
|
2061 |
+
"step": 3240
|
2062 |
+
},
|
2063 |
+
{
|
2064 |
+
"epoch": 12.09,
|
2065 |
+
"learning_rate": 2.1869817578772803e-05,
|
2066 |
+
"loss": 0.4329,
|
2067 |
+
"step": 3250
|
2068 |
+
},
|
2069 |
+
{
|
2070 |
+
"epoch": 12.13,
|
2071 |
+
"learning_rate": 2.1766169154228857e-05,
|
2072 |
+
"loss": 0.4718,
|
2073 |
+
"step": 3260
|
2074 |
+
},
|
2075 |
+
{
|
2076 |
+
"epoch": 12.17,
|
2077 |
+
"learning_rate": 2.166252072968491e-05,
|
2078 |
+
"loss": 0.5176,
|
2079 |
+
"step": 3270
|
2080 |
+
},
|
2081 |
+
{
|
2082 |
+
"epoch": 12.2,
|
2083 |
+
"learning_rate": 2.155887230514096e-05,
|
2084 |
+
"loss": 0.4949,
|
2085 |
+
"step": 3280
|
2086 |
+
},
|
2087 |
+
{
|
2088 |
+
"epoch": 12.24,
|
2089 |
+
"learning_rate": 2.1455223880597018e-05,
|
2090 |
+
"loss": 0.4502,
|
2091 |
+
"step": 3290
|
2092 |
+
},
|
2093 |
+
{
|
2094 |
+
"epoch": 12.28,
|
2095 |
+
"learning_rate": 2.1351575456053068e-05,
|
2096 |
+
"loss": 0.4591,
|
2097 |
+
"step": 3300
|
2098 |
+
},
|
2099 |
+
{
|
2100 |
+
"epoch": 12.32,
|
2101 |
+
"learning_rate": 2.124792703150912e-05,
|
2102 |
+
"loss": 0.5132,
|
2103 |
+
"step": 3310
|
2104 |
+
},
|
2105 |
+
{
|
2106 |
+
"epoch": 12.35,
|
2107 |
+
"learning_rate": 2.1144278606965175e-05,
|
2108 |
+
"loss": 0.4763,
|
2109 |
+
"step": 3320
|
2110 |
+
},
|
2111 |
+
{
|
2112 |
+
"epoch": 12.39,
|
2113 |
+
"learning_rate": 2.104063018242123e-05,
|
2114 |
+
"loss": 0.4915,
|
2115 |
+
"step": 3330
|
2116 |
+
},
|
2117 |
+
{
|
2118 |
+
"epoch": 12.43,
|
2119 |
+
"learning_rate": 2.093698175787728e-05,
|
2120 |
+
"loss": 0.4536,
|
2121 |
+
"step": 3340
|
2122 |
+
},
|
2123 |
+
{
|
2124 |
+
"epoch": 12.47,
|
2125 |
+
"learning_rate": 2.0833333333333336e-05,
|
2126 |
+
"loss": 0.49,
|
2127 |
+
"step": 3350
|
2128 |
+
},
|
2129 |
+
{
|
2130 |
+
"epoch": 12.5,
|
2131 |
+
"learning_rate": 2.0729684908789386e-05,
|
2132 |
+
"loss": 0.4827,
|
2133 |
+
"step": 3360
|
2134 |
+
},
|
2135 |
+
{
|
2136 |
+
"epoch": 12.54,
|
2137 |
+
"learning_rate": 2.062603648424544e-05,
|
2138 |
+
"loss": 0.4844,
|
2139 |
+
"step": 3370
|
2140 |
+
},
|
2141 |
+
{
|
2142 |
+
"epoch": 12.58,
|
2143 |
+
"learning_rate": 2.0522388059701493e-05,
|
2144 |
+
"loss": 0.5079,
|
2145 |
+
"step": 3380
|
2146 |
+
},
|
2147 |
+
{
|
2148 |
+
"epoch": 12.61,
|
2149 |
+
"learning_rate": 2.0418739635157547e-05,
|
2150 |
+
"loss": 0.5262,
|
2151 |
+
"step": 3390
|
2152 |
+
},
|
2153 |
+
{
|
2154 |
+
"epoch": 12.65,
|
2155 |
+
"learning_rate": 2.0315091210613597e-05,
|
2156 |
+
"loss": 0.4884,
|
2157 |
+
"step": 3400
|
2158 |
+
},
|
2159 |
+
{
|
2160 |
+
"epoch": 12.69,
|
2161 |
+
"learning_rate": 2.0211442786069654e-05,
|
2162 |
+
"loss": 0.4665,
|
2163 |
+
"step": 3410
|
2164 |
+
},
|
2165 |
+
{
|
2166 |
+
"epoch": 12.73,
|
2167 |
+
"learning_rate": 2.0107794361525704e-05,
|
2168 |
+
"loss": 0.5148,
|
2169 |
+
"step": 3420
|
2170 |
+
},
|
2171 |
+
{
|
2172 |
+
"epoch": 12.76,
|
2173 |
+
"learning_rate": 2.0004145936981758e-05,
|
2174 |
+
"loss": 0.4784,
|
2175 |
+
"step": 3430
|
2176 |
+
},
|
2177 |
+
{
|
2178 |
+
"epoch": 12.8,
|
2179 |
+
"learning_rate": 1.990049751243781e-05,
|
2180 |
+
"loss": 0.4917,
|
2181 |
+
"step": 3440
|
2182 |
+
},
|
2183 |
+
{
|
2184 |
+
"epoch": 12.84,
|
2185 |
+
"learning_rate": 1.9796849087893865e-05,
|
2186 |
+
"loss": 0.4707,
|
2187 |
+
"step": 3450
|
2188 |
+
},
|
2189 |
+
{
|
2190 |
+
"epoch": 12.87,
|
2191 |
+
"learning_rate": 1.969320066334992e-05,
|
2192 |
+
"loss": 0.4375,
|
2193 |
+
"step": 3460
|
2194 |
+
},
|
2195 |
+
{
|
2196 |
+
"epoch": 12.91,
|
2197 |
+
"learning_rate": 1.9589552238805972e-05,
|
2198 |
+
"loss": 0.5316,
|
2199 |
+
"step": 3470
|
2200 |
+
},
|
2201 |
+
{
|
2202 |
+
"epoch": 12.95,
|
2203 |
+
"learning_rate": 1.9485903814262023e-05,
|
2204 |
+
"loss": 0.459,
|
2205 |
+
"step": 3480
|
2206 |
+
},
|
2207 |
+
{
|
2208 |
+
"epoch": 12.99,
|
2209 |
+
"learning_rate": 1.9382255389718076e-05,
|
2210 |
+
"loss": 0.4687,
|
2211 |
+
"step": 3490
|
2212 |
+
},
|
2213 |
+
{
|
2214 |
+
"epoch": 13.0,
|
2215 |
+
"eval_accuracy": 0.7298850574712644,
|
2216 |
+
"eval_loss": 0.6644836068153381,
|
2217 |
+
"eval_runtime": 8.0326,
|
2218 |
+
"eval_samples_per_second": 43.324,
|
2219 |
+
"eval_steps_per_second": 1.369,
|
2220 |
+
"step": 3493
|
2221 |
+
},
|
2222 |
+
{
|
2223 |
+
"epoch": 13.02,
|
2224 |
+
"learning_rate": 1.927860696517413e-05,
|
2225 |
+
"loss": 0.4732,
|
2226 |
+
"step": 3500
|
2227 |
+
},
|
2228 |
+
{
|
2229 |
+
"epoch": 13.06,
|
2230 |
+
"learning_rate": 1.9174958540630183e-05,
|
2231 |
+
"loss": 0.4884,
|
2232 |
+
"step": 3510
|
2233 |
+
},
|
2234 |
+
{
|
2235 |
+
"epoch": 13.1,
|
2236 |
+
"learning_rate": 1.9071310116086237e-05,
|
2237 |
+
"loss": 0.4424,
|
2238 |
+
"step": 3520
|
2239 |
+
},
|
2240 |
+
{
|
2241 |
+
"epoch": 13.13,
|
2242 |
+
"learning_rate": 1.896766169154229e-05,
|
2243 |
+
"loss": 0.4737,
|
2244 |
+
"step": 3530
|
2245 |
+
},
|
2246 |
+
{
|
2247 |
+
"epoch": 13.17,
|
2248 |
+
"learning_rate": 1.8864013266998344e-05,
|
2249 |
+
"loss": 0.4467,
|
2250 |
+
"step": 3540
|
2251 |
+
},
|
2252 |
+
{
|
2253 |
+
"epoch": 13.21,
|
2254 |
+
"learning_rate": 1.8760364842454394e-05,
|
2255 |
+
"loss": 0.4727,
|
2256 |
+
"step": 3550
|
2257 |
+
},
|
2258 |
+
{
|
2259 |
+
"epoch": 13.25,
|
2260 |
+
"learning_rate": 1.865671641791045e-05,
|
2261 |
+
"loss": 0.4547,
|
2262 |
+
"step": 3560
|
2263 |
+
},
|
2264 |
+
{
|
2265 |
+
"epoch": 13.28,
|
2266 |
+
"learning_rate": 1.85530679933665e-05,
|
2267 |
+
"loss": 0.4437,
|
2268 |
+
"step": 3570
|
2269 |
+
},
|
2270 |
+
{
|
2271 |
+
"epoch": 13.32,
|
2272 |
+
"learning_rate": 1.8449419568822555e-05,
|
2273 |
+
"loss": 0.5015,
|
2274 |
+
"step": 3580
|
2275 |
+
},
|
2276 |
+
{
|
2277 |
+
"epoch": 13.36,
|
2278 |
+
"learning_rate": 1.834577114427861e-05,
|
2279 |
+
"loss": 0.4562,
|
2280 |
+
"step": 3590
|
2281 |
+
},
|
2282 |
+
{
|
2283 |
+
"epoch": 13.4,
|
2284 |
+
"learning_rate": 1.8242122719734662e-05,
|
2285 |
+
"loss": 0.4313,
|
2286 |
+
"step": 3600
|
2287 |
+
},
|
2288 |
+
{
|
2289 |
+
"epoch": 13.43,
|
2290 |
+
"learning_rate": 1.8138474295190713e-05,
|
2291 |
+
"loss": 0.4655,
|
2292 |
+
"step": 3610
|
2293 |
+
},
|
2294 |
+
{
|
2295 |
+
"epoch": 13.47,
|
2296 |
+
"learning_rate": 1.803482587064677e-05,
|
2297 |
+
"loss": 0.484,
|
2298 |
+
"step": 3620
|
2299 |
+
},
|
2300 |
+
{
|
2301 |
+
"epoch": 13.51,
|
2302 |
+
"learning_rate": 1.793117744610282e-05,
|
2303 |
+
"loss": 0.4701,
|
2304 |
+
"step": 3630
|
2305 |
+
},
|
2306 |
+
{
|
2307 |
+
"epoch": 13.54,
|
2308 |
+
"learning_rate": 1.7827529021558873e-05,
|
2309 |
+
"loss": 0.472,
|
2310 |
+
"step": 3640
|
2311 |
+
},
|
2312 |
+
{
|
2313 |
+
"epoch": 13.58,
|
2314 |
+
"learning_rate": 1.7723880597014927e-05,
|
2315 |
+
"loss": 0.4556,
|
2316 |
+
"step": 3650
|
2317 |
+
},
|
2318 |
+
{
|
2319 |
+
"epoch": 13.62,
|
2320 |
+
"learning_rate": 1.762023217247098e-05,
|
2321 |
+
"loss": 0.487,
|
2322 |
+
"step": 3660
|
2323 |
+
},
|
2324 |
+
{
|
2325 |
+
"epoch": 13.66,
|
2326 |
+
"learning_rate": 1.751658374792703e-05,
|
2327 |
+
"loss": 0.4726,
|
2328 |
+
"step": 3670
|
2329 |
+
},
|
2330 |
+
{
|
2331 |
+
"epoch": 13.69,
|
2332 |
+
"learning_rate": 1.7412935323383088e-05,
|
2333 |
+
"loss": 0.4641,
|
2334 |
+
"step": 3680
|
2335 |
+
},
|
2336 |
+
{
|
2337 |
+
"epoch": 13.73,
|
2338 |
+
"learning_rate": 1.7309286898839138e-05,
|
2339 |
+
"loss": 0.4812,
|
2340 |
+
"step": 3690
|
2341 |
+
},
|
2342 |
+
{
|
2343 |
+
"epoch": 13.77,
|
2344 |
+
"learning_rate": 1.720563847429519e-05,
|
2345 |
+
"loss": 0.4553,
|
2346 |
+
"step": 3700
|
2347 |
+
},
|
2348 |
+
{
|
2349 |
+
"epoch": 13.8,
|
2350 |
+
"learning_rate": 1.7101990049751245e-05,
|
2351 |
+
"loss": 0.444,
|
2352 |
+
"step": 3710
|
2353 |
+
},
|
2354 |
+
{
|
2355 |
+
"epoch": 13.84,
|
2356 |
+
"learning_rate": 1.69983416252073e-05,
|
2357 |
+
"loss": 0.4735,
|
2358 |
+
"step": 3720
|
2359 |
+
},
|
2360 |
+
{
|
2361 |
+
"epoch": 13.88,
|
2362 |
+
"learning_rate": 1.689469320066335e-05,
|
2363 |
+
"loss": 0.4832,
|
2364 |
+
"step": 3730
|
2365 |
+
},
|
2366 |
+
{
|
2367 |
+
"epoch": 13.92,
|
2368 |
+
"learning_rate": 1.6791044776119406e-05,
|
2369 |
+
"loss": 0.4632,
|
2370 |
+
"step": 3740
|
2371 |
+
},
|
2372 |
+
{
|
2373 |
+
"epoch": 13.95,
|
2374 |
+
"learning_rate": 1.6687396351575456e-05,
|
2375 |
+
"loss": 0.4713,
|
2376 |
+
"step": 3750
|
2377 |
+
},
|
2378 |
+
{
|
2379 |
+
"epoch": 13.99,
|
2380 |
+
"learning_rate": 1.658374792703151e-05,
|
2381 |
+
"loss": 0.4686,
|
2382 |
+
"step": 3760
|
2383 |
+
},
|
2384 |
+
{
|
2385 |
+
"epoch": 14.0,
|
2386 |
+
"eval_accuracy": 0.7212643678160919,
|
2387 |
+
"eval_loss": 0.6468998789787292,
|
2388 |
+
"eval_runtime": 7.9385,
|
2389 |
+
"eval_samples_per_second": 43.837,
|
2390 |
+
"eval_steps_per_second": 1.386,
|
2391 |
+
"step": 3762
|
2392 |
+
},
|
2393 |
+
{
|
2394 |
+
"epoch": 14.03,
|
2395 |
+
"learning_rate": 1.6480099502487563e-05,
|
2396 |
+
"loss": 0.48,
|
2397 |
+
"step": 3770
|
2398 |
+
},
|
2399 |
+
{
|
2400 |
+
"epoch": 14.07,
|
2401 |
+
"learning_rate": 1.6376451077943617e-05,
|
2402 |
+
"loss": 0.4606,
|
2403 |
+
"step": 3780
|
2404 |
+
},
|
2405 |
+
{
|
2406 |
+
"epoch": 14.1,
|
2407 |
+
"learning_rate": 1.6272802653399667e-05,
|
2408 |
+
"loss": 0.4746,
|
2409 |
+
"step": 3790
|
2410 |
+
},
|
2411 |
+
{
|
2412 |
+
"epoch": 14.14,
|
2413 |
+
"learning_rate": 1.6169154228855724e-05,
|
2414 |
+
"loss": 0.4824,
|
2415 |
+
"step": 3800
|
2416 |
+
},
|
2417 |
+
{
|
2418 |
+
"epoch": 14.18,
|
2419 |
+
"learning_rate": 1.6065505804311774e-05,
|
2420 |
+
"loss": 0.4345,
|
2421 |
+
"step": 3810
|
2422 |
+
},
|
2423 |
+
{
|
2424 |
+
"epoch": 14.21,
|
2425 |
+
"learning_rate": 1.5961857379767828e-05,
|
2426 |
+
"loss": 0.4398,
|
2427 |
+
"step": 3820
|
2428 |
+
},
|
2429 |
+
{
|
2430 |
+
"epoch": 14.25,
|
2431 |
+
"learning_rate": 1.585820895522388e-05,
|
2432 |
+
"loss": 0.4238,
|
2433 |
+
"step": 3830
|
2434 |
+
},
|
2435 |
+
{
|
2436 |
+
"epoch": 14.29,
|
2437 |
+
"learning_rate": 1.5754560530679935e-05,
|
2438 |
+
"loss": 0.4516,
|
2439 |
+
"step": 3840
|
2440 |
+
},
|
2441 |
+
{
|
2442 |
+
"epoch": 14.33,
|
2443 |
+
"learning_rate": 1.565091210613599e-05,
|
2444 |
+
"loss": 0.4432,
|
2445 |
+
"step": 3850
|
2446 |
+
},
|
2447 |
+
{
|
2448 |
+
"epoch": 14.36,
|
2449 |
+
"learning_rate": 1.554726368159204e-05,
|
2450 |
+
"loss": 0.4261,
|
2451 |
+
"step": 3860
|
2452 |
+
},
|
2453 |
+
{
|
2454 |
+
"epoch": 14.4,
|
2455 |
+
"learning_rate": 1.5443615257048092e-05,
|
2456 |
+
"loss": 0.431,
|
2457 |
+
"step": 3870
|
2458 |
+
},
|
2459 |
+
{
|
2460 |
+
"epoch": 14.44,
|
2461 |
+
"learning_rate": 1.5339966832504146e-05,
|
2462 |
+
"loss": 0.4333,
|
2463 |
+
"step": 3880
|
2464 |
+
},
|
2465 |
+
{
|
2466 |
+
"epoch": 14.47,
|
2467 |
+
"learning_rate": 1.5236318407960198e-05,
|
2468 |
+
"loss": 0.4381,
|
2469 |
+
"step": 3890
|
2470 |
+
},
|
2471 |
+
{
|
2472 |
+
"epoch": 14.51,
|
2473 |
+
"learning_rate": 1.5132669983416253e-05,
|
2474 |
+
"loss": 0.4489,
|
2475 |
+
"step": 3900
|
2476 |
+
},
|
2477 |
+
{
|
2478 |
+
"epoch": 14.55,
|
2479 |
+
"learning_rate": 1.5029021558872305e-05,
|
2480 |
+
"loss": 0.4202,
|
2481 |
+
"step": 3910
|
2482 |
+
},
|
2483 |
+
{
|
2484 |
+
"epoch": 14.59,
|
2485 |
+
"learning_rate": 1.4925373134328357e-05,
|
2486 |
+
"loss": 0.4581,
|
2487 |
+
"step": 3920
|
2488 |
+
},
|
2489 |
+
{
|
2490 |
+
"epoch": 14.62,
|
2491 |
+
"learning_rate": 1.4821724709784412e-05,
|
2492 |
+
"loss": 0.483,
|
2493 |
+
"step": 3930
|
2494 |
+
},
|
2495 |
+
{
|
2496 |
+
"epoch": 14.66,
|
2497 |
+
"learning_rate": 1.4718076285240464e-05,
|
2498 |
+
"loss": 0.4667,
|
2499 |
+
"step": 3940
|
2500 |
+
},
|
2501 |
+
{
|
2502 |
+
"epoch": 14.7,
|
2503 |
+
"learning_rate": 1.4614427860696516e-05,
|
2504 |
+
"loss": 0.4774,
|
2505 |
+
"step": 3950
|
2506 |
+
},
|
2507 |
+
{
|
2508 |
+
"epoch": 14.73,
|
2509 |
+
"learning_rate": 1.4510779436152571e-05,
|
2510 |
+
"loss": 0.4491,
|
2511 |
+
"step": 3960
|
2512 |
+
},
|
2513 |
+
{
|
2514 |
+
"epoch": 14.77,
|
2515 |
+
"learning_rate": 1.4407131011608623e-05,
|
2516 |
+
"loss": 0.4417,
|
2517 |
+
"step": 3970
|
2518 |
+
},
|
2519 |
+
{
|
2520 |
+
"epoch": 14.81,
|
2521 |
+
"learning_rate": 1.4303482587064677e-05,
|
2522 |
+
"loss": 0.439,
|
2523 |
+
"step": 3980
|
2524 |
+
},
|
2525 |
+
{
|
2526 |
+
"epoch": 14.85,
|
2527 |
+
"learning_rate": 1.419983416252073e-05,
|
2528 |
+
"loss": 0.4474,
|
2529 |
+
"step": 3990
|
2530 |
+
},
|
2531 |
+
{
|
2532 |
+
"epoch": 14.88,
|
2533 |
+
"learning_rate": 1.4096185737976782e-05,
|
2534 |
+
"loss": 0.4677,
|
2535 |
+
"step": 4000
|
2536 |
+
},
|
2537 |
+
{
|
2538 |
+
"epoch": 14.92,
|
2539 |
+
"learning_rate": 1.3992537313432836e-05,
|
2540 |
+
"loss": 0.4155,
|
2541 |
+
"step": 4010
|
2542 |
+
},
|
2543 |
+
{
|
2544 |
+
"epoch": 14.96,
|
2545 |
+
"learning_rate": 1.388888888888889e-05,
|
2546 |
+
"loss": 0.4397,
|
2547 |
+
"step": 4020
|
2548 |
+
},
|
2549 |
+
{
|
2550 |
+
"epoch": 15.0,
|
2551 |
+
"learning_rate": 1.3785240464344943e-05,
|
2552 |
+
"loss": 0.4862,
|
2553 |
+
"step": 4030
|
2554 |
+
},
|
2555 |
+
{
|
2556 |
+
"epoch": 15.0,
|
2557 |
+
"eval_accuracy": 0.735632183908046,
|
2558 |
+
"eval_loss": 0.6602159738540649,
|
2559 |
+
"eval_runtime": 8.016,
|
2560 |
+
"eval_samples_per_second": 43.413,
|
2561 |
+
"eval_steps_per_second": 1.372,
|
2562 |
+
"step": 4031
|
2563 |
+
},
|
2564 |
+
{
|
2565 |
+
"epoch": 15.03,
|
2566 |
+
"learning_rate": 1.3681592039800995e-05,
|
2567 |
+
"loss": 0.4322,
|
2568 |
+
"step": 4040
|
2569 |
+
},
|
2570 |
+
{
|
2571 |
+
"epoch": 15.07,
|
2572 |
+
"learning_rate": 1.3577943615257049e-05,
|
2573 |
+
"loss": 0.4379,
|
2574 |
+
"step": 4050
|
2575 |
+
},
|
2576 |
+
{
|
2577 |
+
"epoch": 15.11,
|
2578 |
+
"learning_rate": 1.3474295190713102e-05,
|
2579 |
+
"loss": 0.4406,
|
2580 |
+
"step": 4060
|
2581 |
+
},
|
2582 |
+
{
|
2583 |
+
"epoch": 15.14,
|
2584 |
+
"learning_rate": 1.3370646766169154e-05,
|
2585 |
+
"loss": 0.4178,
|
2586 |
+
"step": 4070
|
2587 |
+
},
|
2588 |
+
{
|
2589 |
+
"epoch": 15.18,
|
2590 |
+
"learning_rate": 1.326699834162521e-05,
|
2591 |
+
"loss": 0.4156,
|
2592 |
+
"step": 4080
|
2593 |
+
},
|
2594 |
+
{
|
2595 |
+
"epoch": 15.22,
|
2596 |
+
"learning_rate": 1.3163349917081261e-05,
|
2597 |
+
"loss": 0.4421,
|
2598 |
+
"step": 4090
|
2599 |
+
},
|
2600 |
+
{
|
2601 |
+
"epoch": 15.26,
|
2602 |
+
"learning_rate": 1.3059701492537313e-05,
|
2603 |
+
"loss": 0.4278,
|
2604 |
+
"step": 4100
|
2605 |
+
},
|
2606 |
+
{
|
2607 |
+
"epoch": 15.29,
|
2608 |
+
"learning_rate": 1.2956053067993369e-05,
|
2609 |
+
"loss": 0.4174,
|
2610 |
+
"step": 4110
|
2611 |
+
},
|
2612 |
+
{
|
2613 |
+
"epoch": 15.33,
|
2614 |
+
"learning_rate": 1.285240464344942e-05,
|
2615 |
+
"loss": 0.4091,
|
2616 |
+
"step": 4120
|
2617 |
+
},
|
2618 |
+
{
|
2619 |
+
"epoch": 15.37,
|
2620 |
+
"learning_rate": 1.2748756218905472e-05,
|
2621 |
+
"loss": 0.443,
|
2622 |
+
"step": 4130
|
2623 |
+
},
|
2624 |
+
{
|
2625 |
+
"epoch": 15.4,
|
2626 |
+
"learning_rate": 1.2645107794361528e-05,
|
2627 |
+
"loss": 0.4121,
|
2628 |
+
"step": 4140
|
2629 |
+
},
|
2630 |
+
{
|
2631 |
+
"epoch": 15.44,
|
2632 |
+
"learning_rate": 1.254145936981758e-05,
|
2633 |
+
"loss": 0.4408,
|
2634 |
+
"step": 4150
|
2635 |
+
},
|
2636 |
+
{
|
2637 |
+
"epoch": 15.48,
|
2638 |
+
"learning_rate": 1.2437810945273633e-05,
|
2639 |
+
"loss": 0.4411,
|
2640 |
+
"step": 4160
|
2641 |
+
},
|
2642 |
+
{
|
2643 |
+
"epoch": 15.52,
|
2644 |
+
"learning_rate": 1.2334162520729685e-05,
|
2645 |
+
"loss": 0.4615,
|
2646 |
+
"step": 4170
|
2647 |
+
},
|
2648 |
+
{
|
2649 |
+
"epoch": 15.55,
|
2650 |
+
"learning_rate": 1.2230514096185739e-05,
|
2651 |
+
"loss": 0.4123,
|
2652 |
+
"step": 4180
|
2653 |
+
},
|
2654 |
+
{
|
2655 |
+
"epoch": 15.59,
|
2656 |
+
"learning_rate": 1.2126865671641792e-05,
|
2657 |
+
"loss": 0.4801,
|
2658 |
+
"step": 4190
|
2659 |
+
},
|
2660 |
+
{
|
2661 |
+
"epoch": 15.63,
|
2662 |
+
"learning_rate": 1.2023217247097844e-05,
|
2663 |
+
"loss": 0.4289,
|
2664 |
+
"step": 4200
|
2665 |
+
},
|
2666 |
+
{
|
2667 |
+
"epoch": 15.67,
|
2668 |
+
"learning_rate": 1.1919568822553898e-05,
|
2669 |
+
"loss": 0.4341,
|
2670 |
+
"step": 4210
|
2671 |
+
},
|
2672 |
+
{
|
2673 |
+
"epoch": 15.7,
|
2674 |
+
"learning_rate": 1.1815920398009951e-05,
|
2675 |
+
"loss": 0.4095,
|
2676 |
+
"step": 4220
|
2677 |
+
},
|
2678 |
+
{
|
2679 |
+
"epoch": 15.74,
|
2680 |
+
"learning_rate": 1.1712271973466003e-05,
|
2681 |
+
"loss": 0.4384,
|
2682 |
+
"step": 4230
|
2683 |
+
},
|
2684 |
+
{
|
2685 |
+
"epoch": 15.78,
|
2686 |
+
"learning_rate": 1.1608623548922057e-05,
|
2687 |
+
"loss": 0.4546,
|
2688 |
+
"step": 4240
|
2689 |
+
},
|
2690 |
+
{
|
2691 |
+
"epoch": 15.81,
|
2692 |
+
"learning_rate": 1.150497512437811e-05,
|
2693 |
+
"loss": 0.4643,
|
2694 |
+
"step": 4250
|
2695 |
+
},
|
2696 |
+
{
|
2697 |
+
"epoch": 15.85,
|
2698 |
+
"learning_rate": 1.1401326699834162e-05,
|
2699 |
+
"loss": 0.4105,
|
2700 |
+
"step": 4260
|
2701 |
+
},
|
2702 |
+
{
|
2703 |
+
"epoch": 15.89,
|
2704 |
+
"learning_rate": 1.1297678275290216e-05,
|
2705 |
+
"loss": 0.4155,
|
2706 |
+
"step": 4270
|
2707 |
+
},
|
2708 |
+
{
|
2709 |
+
"epoch": 15.93,
|
2710 |
+
"learning_rate": 1.119402985074627e-05,
|
2711 |
+
"loss": 0.4002,
|
2712 |
+
"step": 4280
|
2713 |
+
},
|
2714 |
+
{
|
2715 |
+
"epoch": 15.96,
|
2716 |
+
"learning_rate": 1.1090381426202321e-05,
|
2717 |
+
"loss": 0.4344,
|
2718 |
+
"step": 4290
|
2719 |
+
},
|
2720 |
+
{
|
2721 |
+
"epoch": 16.0,
|
2722 |
+
"learning_rate": 1.0986733001658375e-05,
|
2723 |
+
"loss": 0.4432,
|
2724 |
+
"step": 4300
|
2725 |
+
},
|
2726 |
+
{
|
2727 |
+
"epoch": 16.0,
|
2728 |
+
"eval_accuracy": 0.7270114942528736,
|
2729 |
+
"eval_loss": 0.655029296875,
|
2730 |
+
"eval_runtime": 7.9863,
|
2731 |
+
"eval_samples_per_second": 43.574,
|
2732 |
+
"eval_steps_per_second": 1.377,
|
2733 |
+
"step": 4300
|
2734 |
+
},
|
2735 |
+
{
|
2736 |
+
"epoch": 16.04,
|
2737 |
+
"learning_rate": 1.0883084577114429e-05,
|
2738 |
+
"loss": 0.433,
|
2739 |
+
"step": 4310
|
2740 |
+
},
|
2741 |
+
{
|
2742 |
+
"epoch": 16.07,
|
2743 |
+
"learning_rate": 1.077943615257048e-05,
|
2744 |
+
"loss": 0.434,
|
2745 |
+
"step": 4320
|
2746 |
+
},
|
2747 |
+
{
|
2748 |
+
"epoch": 16.11,
|
2749 |
+
"learning_rate": 1.0675787728026534e-05,
|
2750 |
+
"loss": 0.3986,
|
2751 |
+
"step": 4330
|
2752 |
+
},
|
2753 |
+
{
|
2754 |
+
"epoch": 16.15,
|
2755 |
+
"learning_rate": 1.0572139303482588e-05,
|
2756 |
+
"loss": 0.4175,
|
2757 |
+
"step": 4340
|
2758 |
+
},
|
2759 |
+
{
|
2760 |
+
"epoch": 16.19,
|
2761 |
+
"learning_rate": 1.046849087893864e-05,
|
2762 |
+
"loss": 0.4284,
|
2763 |
+
"step": 4350
|
2764 |
+
},
|
2765 |
+
{
|
2766 |
+
"epoch": 16.22,
|
2767 |
+
"learning_rate": 1.0364842454394693e-05,
|
2768 |
+
"loss": 0.4392,
|
2769 |
+
"step": 4360
|
2770 |
+
},
|
2771 |
+
{
|
2772 |
+
"epoch": 16.26,
|
2773 |
+
"learning_rate": 1.0261194029850747e-05,
|
2774 |
+
"loss": 0.3935,
|
2775 |
+
"step": 4370
|
2776 |
+
},
|
2777 |
+
{
|
2778 |
+
"epoch": 16.3,
|
2779 |
+
"learning_rate": 1.0157545605306799e-05,
|
2780 |
+
"loss": 0.4262,
|
2781 |
+
"step": 4380
|
2782 |
+
},
|
2783 |
+
{
|
2784 |
+
"epoch": 16.33,
|
2785 |
+
"learning_rate": 1.0053897180762852e-05,
|
2786 |
+
"loss": 0.446,
|
2787 |
+
"step": 4390
|
2788 |
+
},
|
2789 |
+
{
|
2790 |
+
"epoch": 16.37,
|
2791 |
+
"learning_rate": 9.950248756218906e-06,
|
2792 |
+
"loss": 0.4231,
|
2793 |
+
"step": 4400
|
2794 |
+
},
|
2795 |
+
{
|
2796 |
+
"epoch": 16.41,
|
2797 |
+
"learning_rate": 9.84660033167496e-06,
|
2798 |
+
"loss": 0.442,
|
2799 |
+
"step": 4410
|
2800 |
+
},
|
2801 |
+
{
|
2802 |
+
"epoch": 16.45,
|
2803 |
+
"learning_rate": 9.742951907131011e-06,
|
2804 |
+
"loss": 0.4138,
|
2805 |
+
"step": 4420
|
2806 |
+
},
|
2807 |
+
{
|
2808 |
+
"epoch": 16.48,
|
2809 |
+
"learning_rate": 9.639303482587065e-06,
|
2810 |
+
"loss": 0.3881,
|
2811 |
+
"step": 4430
|
2812 |
+
},
|
2813 |
+
{
|
2814 |
+
"epoch": 16.52,
|
2815 |
+
"learning_rate": 9.535655058043118e-06,
|
2816 |
+
"loss": 0.3992,
|
2817 |
+
"step": 4440
|
2818 |
+
},
|
2819 |
+
{
|
2820 |
+
"epoch": 16.56,
|
2821 |
+
"learning_rate": 9.432006633499172e-06,
|
2822 |
+
"loss": 0.4051,
|
2823 |
+
"step": 4450
|
2824 |
+
},
|
2825 |
+
{
|
2826 |
+
"epoch": 16.6,
|
2827 |
+
"learning_rate": 9.328358208955226e-06,
|
2828 |
+
"loss": 0.4369,
|
2829 |
+
"step": 4460
|
2830 |
+
},
|
2831 |
+
{
|
2832 |
+
"epoch": 16.63,
|
2833 |
+
"learning_rate": 9.224709784411278e-06,
|
2834 |
+
"loss": 0.3802,
|
2835 |
+
"step": 4470
|
2836 |
+
},
|
2837 |
+
{
|
2838 |
+
"epoch": 16.67,
|
2839 |
+
"learning_rate": 9.121061359867331e-06,
|
2840 |
+
"loss": 0.4153,
|
2841 |
+
"step": 4480
|
2842 |
+
},
|
2843 |
+
{
|
2844 |
+
"epoch": 16.71,
|
2845 |
+
"learning_rate": 9.017412935323385e-06,
|
2846 |
+
"loss": 0.4561,
|
2847 |
+
"step": 4490
|
2848 |
+
},
|
2849 |
+
{
|
2850 |
+
"epoch": 16.74,
|
2851 |
+
"learning_rate": 8.913764510779437e-06,
|
2852 |
+
"loss": 0.4244,
|
2853 |
+
"step": 4500
|
2854 |
+
},
|
2855 |
+
{
|
2856 |
+
"epoch": 16.78,
|
2857 |
+
"learning_rate": 8.81011608623549e-06,
|
2858 |
+
"loss": 0.3932,
|
2859 |
+
"step": 4510
|
2860 |
+
},
|
2861 |
+
{
|
2862 |
+
"epoch": 16.82,
|
2863 |
+
"learning_rate": 8.706467661691544e-06,
|
2864 |
+
"loss": 0.4317,
|
2865 |
+
"step": 4520
|
2866 |
+
},
|
2867 |
+
{
|
2868 |
+
"epoch": 16.86,
|
2869 |
+
"learning_rate": 8.602819237147596e-06,
|
2870 |
+
"loss": 0.4566,
|
2871 |
+
"step": 4530
|
2872 |
+
},
|
2873 |
+
{
|
2874 |
+
"epoch": 16.89,
|
2875 |
+
"learning_rate": 8.49917081260365e-06,
|
2876 |
+
"loss": 0.451,
|
2877 |
+
"step": 4540
|
2878 |
+
},
|
2879 |
+
{
|
2880 |
+
"epoch": 16.93,
|
2881 |
+
"learning_rate": 8.395522388059703e-06,
|
2882 |
+
"loss": 0.4418,
|
2883 |
+
"step": 4550
|
2884 |
+
},
|
2885 |
+
{
|
2886 |
+
"epoch": 16.97,
|
2887 |
+
"learning_rate": 8.291873963515755e-06,
|
2888 |
+
"loss": 0.4368,
|
2889 |
+
"step": 4560
|
2890 |
+
},
|
2891 |
+
{
|
2892 |
+
"epoch": 17.0,
|
2893 |
+
"eval_accuracy": 0.7385057471264368,
|
2894 |
+
"eval_loss": 0.6471846699714661,
|
2895 |
+
"eval_runtime": 7.5887,
|
2896 |
+
"eval_samples_per_second": 45.857,
|
2897 |
+
"eval_steps_per_second": 1.45,
|
2898 |
+
"step": 4568
|
2899 |
+
},
|
2900 |
+
{
|
2901 |
+
"epoch": 17.0,
|
2902 |
+
"learning_rate": 8.188225538971808e-06,
|
2903 |
+
"loss": 0.4044,
|
2904 |
+
"step": 4570
|
2905 |
+
},
|
2906 |
+
{
|
2907 |
+
"epoch": 17.04,
|
2908 |
+
"learning_rate": 8.084577114427862e-06,
|
2909 |
+
"loss": 0.4333,
|
2910 |
+
"step": 4580
|
2911 |
+
},
|
2912 |
+
{
|
2913 |
+
"epoch": 17.08,
|
2914 |
+
"learning_rate": 7.980928689883914e-06,
|
2915 |
+
"loss": 0.3968,
|
2916 |
+
"step": 4590
|
2917 |
+
},
|
2918 |
+
{
|
2919 |
+
"epoch": 17.12,
|
2920 |
+
"learning_rate": 7.877280265339968e-06,
|
2921 |
+
"loss": 0.3895,
|
2922 |
+
"step": 4600
|
2923 |
+
},
|
2924 |
+
{
|
2925 |
+
"epoch": 17.15,
|
2926 |
+
"learning_rate": 7.77363184079602e-06,
|
2927 |
+
"loss": 0.3924,
|
2928 |
+
"step": 4610
|
2929 |
+
},
|
2930 |
+
{
|
2931 |
+
"epoch": 17.19,
|
2932 |
+
"learning_rate": 7.669983416252073e-06,
|
2933 |
+
"loss": 0.417,
|
2934 |
+
"step": 4620
|
2935 |
+
},
|
2936 |
+
{
|
2937 |
+
"epoch": 17.23,
|
2938 |
+
"learning_rate": 7.566334991708127e-06,
|
2939 |
+
"loss": 0.3854,
|
2940 |
+
"step": 4630
|
2941 |
+
},
|
2942 |
+
{
|
2943 |
+
"epoch": 17.27,
|
2944 |
+
"learning_rate": 7.4626865671641785e-06,
|
2945 |
+
"loss": 0.4233,
|
2946 |
+
"step": 4640
|
2947 |
+
},
|
2948 |
+
{
|
2949 |
+
"epoch": 17.3,
|
2950 |
+
"learning_rate": 7.359038142620232e-06,
|
2951 |
+
"loss": 0.4202,
|
2952 |
+
"step": 4650
|
2953 |
+
},
|
2954 |
+
{
|
2955 |
+
"epoch": 17.34,
|
2956 |
+
"learning_rate": 7.255389718076286e-06,
|
2957 |
+
"loss": 0.4143,
|
2958 |
+
"step": 4660
|
2959 |
+
},
|
2960 |
+
{
|
2961 |
+
"epoch": 17.38,
|
2962 |
+
"learning_rate": 7.1517412935323384e-06,
|
2963 |
+
"loss": 0.4241,
|
2964 |
+
"step": 4670
|
2965 |
+
},
|
2966 |
+
{
|
2967 |
+
"epoch": 17.41,
|
2968 |
+
"learning_rate": 7.048092868988391e-06,
|
2969 |
+
"loss": 0.3974,
|
2970 |
+
"step": 4680
|
2971 |
+
},
|
2972 |
+
{
|
2973 |
+
"epoch": 17.45,
|
2974 |
+
"learning_rate": 6.944444444444445e-06,
|
2975 |
+
"loss": 0.4151,
|
2976 |
+
"step": 4690
|
2977 |
+
},
|
2978 |
+
{
|
2979 |
+
"epoch": 17.49,
|
2980 |
+
"learning_rate": 6.8407960199004975e-06,
|
2981 |
+
"loss": 0.4151,
|
2982 |
+
"step": 4700
|
2983 |
+
},
|
2984 |
+
{
|
2985 |
+
"epoch": 17.53,
|
2986 |
+
"learning_rate": 6.737147595356551e-06,
|
2987 |
+
"loss": 0.4355,
|
2988 |
+
"step": 4710
|
2989 |
+
},
|
2990 |
+
{
|
2991 |
+
"epoch": 17.56,
|
2992 |
+
"learning_rate": 6.633499170812605e-06,
|
2993 |
+
"loss": 0.4039,
|
2994 |
+
"step": 4720
|
2995 |
+
},
|
2996 |
+
{
|
2997 |
+
"epoch": 17.6,
|
2998 |
+
"learning_rate": 6.529850746268657e-06,
|
2999 |
+
"loss": 0.3873,
|
3000 |
+
"step": 4730
|
3001 |
+
},
|
3002 |
+
{
|
3003 |
+
"epoch": 17.64,
|
3004 |
+
"learning_rate": 6.42620232172471e-06,
|
3005 |
+
"loss": 0.4179,
|
3006 |
+
"step": 4740
|
3007 |
+
},
|
3008 |
+
{
|
3009 |
+
"epoch": 17.67,
|
3010 |
+
"learning_rate": 6.322553897180764e-06,
|
3011 |
+
"loss": 0.3985,
|
3012 |
+
"step": 4750
|
3013 |
+
},
|
3014 |
+
{
|
3015 |
+
"epoch": 17.71,
|
3016 |
+
"learning_rate": 6.2189054726368165e-06,
|
3017 |
+
"loss": 0.4186,
|
3018 |
+
"step": 4760
|
3019 |
+
},
|
3020 |
+
{
|
3021 |
+
"epoch": 17.75,
|
3022 |
+
"learning_rate": 6.115257048092869e-06,
|
3023 |
+
"loss": 0.449,
|
3024 |
+
"step": 4770
|
3025 |
+
},
|
3026 |
+
{
|
3027 |
+
"epoch": 17.79,
|
3028 |
+
"learning_rate": 6.011608623548922e-06,
|
3029 |
+
"loss": 0.3952,
|
3030 |
+
"step": 4780
|
3031 |
+
},
|
3032 |
+
{
|
3033 |
+
"epoch": 17.82,
|
3034 |
+
"learning_rate": 5.907960199004976e-06,
|
3035 |
+
"loss": 0.4167,
|
3036 |
+
"step": 4790
|
3037 |
+
},
|
3038 |
+
{
|
3039 |
+
"epoch": 17.86,
|
3040 |
+
"learning_rate": 5.804311774461028e-06,
|
3041 |
+
"loss": 0.4018,
|
3042 |
+
"step": 4800
|
3043 |
+
},
|
3044 |
+
{
|
3045 |
+
"epoch": 17.9,
|
3046 |
+
"learning_rate": 5.700663349917081e-06,
|
3047 |
+
"loss": 0.4053,
|
3048 |
+
"step": 4810
|
3049 |
+
},
|
3050 |
+
{
|
3051 |
+
"epoch": 17.93,
|
3052 |
+
"learning_rate": 5.597014925373135e-06,
|
3053 |
+
"loss": 0.4287,
|
3054 |
+
"step": 4820
|
3055 |
+
},
|
3056 |
+
{
|
3057 |
+
"epoch": 17.97,
|
3058 |
+
"learning_rate": 5.4933665008291875e-06,
|
3059 |
+
"loss": 0.3815,
|
3060 |
+
"step": 4830
|
3061 |
+
},
|
3062 |
+
{
|
3063 |
+
"epoch": 18.0,
|
3064 |
+
"eval_accuracy": 0.7557471264367817,
|
3065 |
+
"eval_loss": 0.6556899547576904,
|
3066 |
+
"eval_runtime": 7.8702,
|
3067 |
+
"eval_samples_per_second": 44.217,
|
3068 |
+
"eval_steps_per_second": 1.398,
|
3069 |
+
"step": 4837
|
3070 |
+
},
|
3071 |
+
{
|
3072 |
+
"epoch": 18.01,
|
3073 |
+
"learning_rate": 5.38971807628524e-06,
|
3074 |
+
"loss": 0.4203,
|
3075 |
+
"step": 4840
|
3076 |
+
},
|
3077 |
+
{
|
3078 |
+
"epoch": 18.05,
|
3079 |
+
"learning_rate": 5.286069651741294e-06,
|
3080 |
+
"loss": 0.4077,
|
3081 |
+
"step": 4850
|
3082 |
+
},
|
3083 |
+
{
|
3084 |
+
"epoch": 18.08,
|
3085 |
+
"learning_rate": 5.1824212271973466e-06,
|
3086 |
+
"loss": 0.3997,
|
3087 |
+
"step": 4860
|
3088 |
+
},
|
3089 |
+
{
|
3090 |
+
"epoch": 18.12,
|
3091 |
+
"learning_rate": 5.078772802653399e-06,
|
3092 |
+
"loss": 0.4031,
|
3093 |
+
"step": 4870
|
3094 |
+
},
|
3095 |
+
{
|
3096 |
+
"epoch": 18.16,
|
3097 |
+
"learning_rate": 4.975124378109453e-06,
|
3098 |
+
"loss": 0.3897,
|
3099 |
+
"step": 4880
|
3100 |
+
},
|
3101 |
+
{
|
3102 |
+
"epoch": 18.2,
|
3103 |
+
"learning_rate": 4.871475953565506e-06,
|
3104 |
+
"loss": 0.3964,
|
3105 |
+
"step": 4890
|
3106 |
+
},
|
3107 |
+
{
|
3108 |
+
"epoch": 18.23,
|
3109 |
+
"learning_rate": 4.767827529021559e-06,
|
3110 |
+
"loss": 0.392,
|
3111 |
+
"step": 4900
|
3112 |
+
},
|
3113 |
+
{
|
3114 |
+
"epoch": 18.27,
|
3115 |
+
"learning_rate": 4.664179104477613e-06,
|
3116 |
+
"loss": 0.4087,
|
3117 |
+
"step": 4910
|
3118 |
+
},
|
3119 |
+
{
|
3120 |
+
"epoch": 18.31,
|
3121 |
+
"learning_rate": 4.560530679933666e-06,
|
3122 |
+
"loss": 0.3859,
|
3123 |
+
"step": 4920
|
3124 |
+
},
|
3125 |
+
{
|
3126 |
+
"epoch": 18.34,
|
3127 |
+
"learning_rate": 4.456882255389718e-06,
|
3128 |
+
"loss": 0.4284,
|
3129 |
+
"step": 4930
|
3130 |
+
},
|
3131 |
+
{
|
3132 |
+
"epoch": 18.38,
|
3133 |
+
"learning_rate": 4.353233830845772e-06,
|
3134 |
+
"loss": 0.3872,
|
3135 |
+
"step": 4940
|
3136 |
+
},
|
3137 |
+
{
|
3138 |
+
"epoch": 18.42,
|
3139 |
+
"learning_rate": 4.249585406301825e-06,
|
3140 |
+
"loss": 0.3822,
|
3141 |
+
"step": 4950
|
3142 |
+
},
|
3143 |
+
{
|
3144 |
+
"epoch": 18.46,
|
3145 |
+
"learning_rate": 4.145936981757877e-06,
|
3146 |
+
"loss": 0.4059,
|
3147 |
+
"step": 4960
|
3148 |
+
},
|
3149 |
+
{
|
3150 |
+
"epoch": 18.49,
|
3151 |
+
"learning_rate": 4.042288557213931e-06,
|
3152 |
+
"loss": 0.4052,
|
3153 |
+
"step": 4970
|
3154 |
+
},
|
3155 |
+
{
|
3156 |
+
"epoch": 18.53,
|
3157 |
+
"learning_rate": 3.938640132669984e-06,
|
3158 |
+
"loss": 0.4147,
|
3159 |
+
"step": 4980
|
3160 |
+
},
|
3161 |
+
{
|
3162 |
+
"epoch": 18.57,
|
3163 |
+
"learning_rate": 3.8349917081260365e-06,
|
3164 |
+
"loss": 0.4213,
|
3165 |
+
"step": 4990
|
3166 |
+
},
|
3167 |
+
{
|
3168 |
+
"epoch": 18.6,
|
3169 |
+
"learning_rate": 3.7313432835820893e-06,
|
3170 |
+
"loss": 0.39,
|
3171 |
+
"step": 5000
|
3172 |
+
},
|
3173 |
+
{
|
3174 |
+
"epoch": 18.64,
|
3175 |
+
"learning_rate": 3.627694859038143e-06,
|
3176 |
+
"loss": 0.3517,
|
3177 |
+
"step": 5010
|
3178 |
+
},
|
3179 |
+
{
|
3180 |
+
"epoch": 18.68,
|
3181 |
+
"learning_rate": 3.5240464344941956e-06,
|
3182 |
+
"loss": 0.374,
|
3183 |
+
"step": 5020
|
3184 |
+
},
|
3185 |
+
{
|
3186 |
+
"epoch": 18.72,
|
3187 |
+
"learning_rate": 3.4203980099502488e-06,
|
3188 |
+
"loss": 0.3799,
|
3189 |
+
"step": 5030
|
3190 |
+
},
|
3191 |
+
{
|
3192 |
+
"epoch": 18.75,
|
3193 |
+
"learning_rate": 3.3167495854063024e-06,
|
3194 |
+
"loss": 0.3898,
|
3195 |
+
"step": 5040
|
3196 |
+
},
|
3197 |
+
{
|
3198 |
+
"epoch": 18.79,
|
3199 |
+
"learning_rate": 3.213101160862355e-06,
|
3200 |
+
"loss": 0.4238,
|
3201 |
+
"step": 5050
|
3202 |
+
},
|
3203 |
+
{
|
3204 |
+
"epoch": 18.83,
|
3205 |
+
"learning_rate": 3.1094527363184083e-06,
|
3206 |
+
"loss": 0.3948,
|
3207 |
+
"step": 5060
|
3208 |
+
},
|
3209 |
+
{
|
3210 |
+
"epoch": 18.87,
|
3211 |
+
"learning_rate": 3.005804311774461e-06,
|
3212 |
+
"loss": 0.4003,
|
3213 |
+
"step": 5070
|
3214 |
+
},
|
3215 |
+
{
|
3216 |
+
"epoch": 18.9,
|
3217 |
+
"learning_rate": 2.902155887230514e-06,
|
3218 |
+
"loss": 0.4015,
|
3219 |
+
"step": 5080
|
3220 |
+
},
|
3221 |
+
{
|
3222 |
+
"epoch": 18.94,
|
3223 |
+
"learning_rate": 2.7985074626865674e-06,
|
3224 |
+
"loss": 0.3958,
|
3225 |
+
"step": 5090
|
3226 |
+
},
|
3227 |
+
{
|
3228 |
+
"epoch": 18.98,
|
3229 |
+
"learning_rate": 2.69485903814262e-06,
|
3230 |
+
"loss": 0.3674,
|
3231 |
+
"step": 5100
|
3232 |
+
},
|
3233 |
+
{
|
3234 |
+
"epoch": 19.0,
|
3235 |
+
"eval_accuracy": 0.7528735632183908,
|
3236 |
+
"eval_loss": 0.6638280749320984,
|
3237 |
+
"eval_runtime": 7.9682,
|
3238 |
+
"eval_samples_per_second": 43.673,
|
3239 |
+
"eval_steps_per_second": 1.38,
|
3240 |
+
"step": 5106
|
3241 |
+
},
|
3242 |
+
{
|
3243 |
+
"epoch": 19.01,
|
3244 |
+
"learning_rate": 2.5912106135986733e-06,
|
3245 |
+
"loss": 0.3796,
|
3246 |
+
"step": 5110
|
3247 |
+
},
|
3248 |
+
{
|
3249 |
+
"epoch": 19.05,
|
3250 |
+
"learning_rate": 2.4875621890547264e-06,
|
3251 |
+
"loss": 0.4141,
|
3252 |
+
"step": 5120
|
3253 |
+
},
|
3254 |
+
{
|
3255 |
+
"epoch": 19.09,
|
3256 |
+
"learning_rate": 2.3839137645107796e-06,
|
3257 |
+
"loss": 0.4136,
|
3258 |
+
"step": 5130
|
3259 |
+
},
|
3260 |
+
{
|
3261 |
+
"epoch": 19.13,
|
3262 |
+
"learning_rate": 2.280265339966833e-06,
|
3263 |
+
"loss": 0.3852,
|
3264 |
+
"step": 5140
|
3265 |
+
},
|
3266 |
+
{
|
3267 |
+
"epoch": 19.16,
|
3268 |
+
"learning_rate": 2.176616915422886e-06,
|
3269 |
+
"loss": 0.3842,
|
3270 |
+
"step": 5150
|
3271 |
+
},
|
3272 |
+
{
|
3273 |
+
"epoch": 19.2,
|
3274 |
+
"learning_rate": 2.0729684908789387e-06,
|
3275 |
+
"loss": 0.3745,
|
3276 |
+
"step": 5160
|
3277 |
+
},
|
3278 |
+
{
|
3279 |
+
"epoch": 19.24,
|
3280 |
+
"learning_rate": 1.969320066334992e-06,
|
3281 |
+
"loss": 0.3694,
|
3282 |
+
"step": 5170
|
3283 |
+
},
|
3284 |
+
{
|
3285 |
+
"epoch": 19.27,
|
3286 |
+
"learning_rate": 1.8656716417910446e-06,
|
3287 |
+
"loss": 0.4001,
|
3288 |
+
"step": 5180
|
3289 |
+
},
|
3290 |
+
{
|
3291 |
+
"epoch": 19.31,
|
3292 |
+
"learning_rate": 1.7620232172470978e-06,
|
3293 |
+
"loss": 0.4035,
|
3294 |
+
"step": 5190
|
3295 |
+
},
|
3296 |
+
{
|
3297 |
+
"epoch": 19.35,
|
3298 |
+
"learning_rate": 1.6583747927031512e-06,
|
3299 |
+
"loss": 0.4029,
|
3300 |
+
"step": 5200
|
3301 |
+
},
|
3302 |
+
{
|
3303 |
+
"epoch": 19.39,
|
3304 |
+
"learning_rate": 1.5547263681592041e-06,
|
3305 |
+
"loss": 0.3915,
|
3306 |
+
"step": 5210
|
3307 |
+
},
|
3308 |
+
{
|
3309 |
+
"epoch": 19.42,
|
3310 |
+
"learning_rate": 1.451077943615257e-06,
|
3311 |
+
"loss": 0.3579,
|
3312 |
+
"step": 5220
|
3313 |
+
},
|
3314 |
+
{
|
3315 |
+
"epoch": 19.46,
|
3316 |
+
"learning_rate": 1.34742951907131e-06,
|
3317 |
+
"loss": 0.403,
|
3318 |
+
"step": 5230
|
3319 |
+
},
|
3320 |
+
{
|
3321 |
+
"epoch": 19.5,
|
3322 |
+
"learning_rate": 1.2437810945273632e-06,
|
3323 |
+
"loss": 0.3859,
|
3324 |
+
"step": 5240
|
3325 |
+
},
|
3326 |
+
{
|
3327 |
+
"epoch": 19.53,
|
3328 |
+
"learning_rate": 1.1401326699834164e-06,
|
3329 |
+
"loss": 0.3895,
|
3330 |
+
"step": 5250
|
3331 |
+
},
|
3332 |
+
{
|
3333 |
+
"epoch": 19.57,
|
3334 |
+
"learning_rate": 1.0364842454394694e-06,
|
3335 |
+
"loss": 0.3489,
|
3336 |
+
"step": 5260
|
3337 |
+
},
|
3338 |
+
{
|
3339 |
+
"epoch": 19.61,
|
3340 |
+
"learning_rate": 9.328358208955223e-07,
|
3341 |
+
"loss": 0.4045,
|
3342 |
+
"step": 5270
|
3343 |
+
},
|
3344 |
+
{
|
3345 |
+
"epoch": 19.65,
|
3346 |
+
"learning_rate": 8.291873963515756e-07,
|
3347 |
+
"loss": 0.3953,
|
3348 |
+
"step": 5280
|
3349 |
+
},
|
3350 |
+
{
|
3351 |
+
"epoch": 19.68,
|
3352 |
+
"learning_rate": 7.255389718076285e-07,
|
3353 |
+
"loss": 0.3901,
|
3354 |
+
"step": 5290
|
3355 |
+
},
|
3356 |
+
{
|
3357 |
+
"epoch": 19.72,
|
3358 |
+
"learning_rate": 6.218905472636816e-07,
|
3359 |
+
"loss": 0.3985,
|
3360 |
+
"step": 5300
|
3361 |
+
},
|
3362 |
+
{
|
3363 |
+
"epoch": 19.76,
|
3364 |
+
"learning_rate": 5.182421227197347e-07,
|
3365 |
+
"loss": 0.3804,
|
3366 |
+
"step": 5310
|
3367 |
+
},
|
3368 |
+
{
|
3369 |
+
"epoch": 19.8,
|
3370 |
+
"learning_rate": 4.145936981757878e-07,
|
3371 |
+
"loss": 0.3659,
|
3372 |
+
"step": 5320
|
3373 |
+
},
|
3374 |
+
{
|
3375 |
+
"epoch": 19.83,
|
3376 |
+
"learning_rate": 3.109452736318408e-07,
|
3377 |
+
"loss": 0.4105,
|
3378 |
+
"step": 5330
|
3379 |
+
},
|
3380 |
+
{
|
3381 |
+
"epoch": 19.87,
|
3382 |
+
"learning_rate": 2.072968490878939e-07,
|
3383 |
+
"loss": 0.3751,
|
3384 |
+
"step": 5340
|
3385 |
+
},
|
3386 |
+
{
|
3387 |
+
"epoch": 19.91,
|
3388 |
+
"learning_rate": 1.0364842454394695e-07,
|
3389 |
+
"loss": 0.3928,
|
3390 |
+
"step": 5350
|
3391 |
+
},
|
3392 |
+
{
|
3393 |
+
"epoch": 19.94,
|
3394 |
+
"learning_rate": 0.0,
|
3395 |
+
"loss": 0.4224,
|
3396 |
+
"step": 5360
|
3397 |
+
},
|
3398 |
+
{
|
3399 |
+
"epoch": 19.94,
|
3400 |
+
"eval_accuracy": 0.7557471264367817,
|
3401 |
+
"eval_loss": 0.6591867208480835,
|
3402 |
+
"eval_runtime": 7.744,
|
3403 |
+
"eval_samples_per_second": 44.938,
|
3404 |
+
"eval_steps_per_second": 1.42,
|
3405 |
+
"step": 5360
|
3406 |
+
},
|
3407 |
+
{
|
3408 |
+
"epoch": 19.94,
|
3409 |
+
"step": 5360,
|
3410 |
+
"total_flos": 1.7046597288139905e+19,
|
3411 |
+
"train_loss": 0.5607940738325689,
|
3412 |
+
"train_runtime": 20096.6246,
|
3413 |
+
"train_samples_per_second": 34.218,
|
3414 |
+
"train_steps_per_second": 0.267
|
3415 |
+
}
|
3416 |
+
],
|
3417 |
+
"logging_steps": 10,
|
3418 |
+
"max_steps": 5360,
|
3419 |
+
"num_train_epochs": 20,
|
3420 |
+
"save_steps": 500,
|
3421 |
+
"total_flos": 1.7046597288139905e+19,
|
3422 |
+
"trial_name": null,
|
3423 |
+
"trial_params": null
|
3424 |
+
}
|