Commit
•
a1352e0
1
Parent(s):
b1d3eb3
End of training
Browse files- all_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +1645 -0
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 49.95,
|
| 3 |
+
"train_loss": 1.3846193885803222,
|
| 4 |
+
"train_runtime": 2065.3687,
|
| 5 |
+
"train_samples": 164,
|
| 6 |
+
"train_samples_per_second": 3.97,
|
| 7 |
+
"train_steps_per_second": 0.121
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 49.95,
|
| 3 |
+
"train_loss": 1.3846193885803222,
|
| 4 |
+
"train_runtime": 2065.3687,
|
| 5 |
+
"train_samples": 164,
|
| 6 |
+
"train_samples_per_second": 3.97,
|
| 7 |
+
"train_steps_per_second": 0.121
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1645 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.0851063829787234,
|
| 3 |
+
"best_model_checkpoint": "xtreme_s_xlsr_300m_mt5-small_minds14.en-US/checkpoint-20",
|
| 4 |
+
"epoch": 49.95238095238095,
|
| 5 |
+
"global_step": 250,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.19,
|
| 12 |
+
"learning_rate": 2.9999999999999997e-06,
|
| 13 |
+
"loss": 2.6237,
|
| 14 |
+
"step": 1
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.38,
|
| 18 |
+
"learning_rate": 5.999999999999999e-06,
|
| 19 |
+
"loss": 2.6431,
|
| 20 |
+
"step": 2
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.57,
|
| 24 |
+
"learning_rate": 8.999999999999999e-06,
|
| 25 |
+
"loss": 2.6487,
|
| 26 |
+
"step": 3
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.76,
|
| 30 |
+
"learning_rate": 1.1999999999999999e-05,
|
| 31 |
+
"loss": 2.6581,
|
| 32 |
+
"step": 4
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.95,
|
| 36 |
+
"learning_rate": 1.4999999999999999e-05,
|
| 37 |
+
"loss": 2.6703,
|
| 38 |
+
"step": 5
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 1.19,
|
| 42 |
+
"learning_rate": 1.7999999999999997e-05,
|
| 43 |
+
"loss": 3.3039,
|
| 44 |
+
"step": 6
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 1.38,
|
| 48 |
+
"learning_rate": 2.1e-05,
|
| 49 |
+
"loss": 2.6422,
|
| 50 |
+
"step": 7
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.57,
|
| 54 |
+
"learning_rate": 2.3999999999999997e-05,
|
| 55 |
+
"loss": 2.6526,
|
| 56 |
+
"step": 8
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 1.76,
|
| 60 |
+
"learning_rate": 2.6999999999999996e-05,
|
| 61 |
+
"loss": 2.6371,
|
| 62 |
+
"step": 9
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 1.95,
|
| 66 |
+
"learning_rate": 2.9999999999999997e-05,
|
| 67 |
+
"loss": 2.6353,
|
| 68 |
+
"step": 10
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 2.19,
|
| 72 |
+
"learning_rate": 3.2999999999999996e-05,
|
| 73 |
+
"loss": 3.3038,
|
| 74 |
+
"step": 11
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 2.38,
|
| 78 |
+
"learning_rate": 3.5999999999999994e-05,
|
| 79 |
+
"loss": 2.6328,
|
| 80 |
+
"step": 12
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 2.57,
|
| 84 |
+
"learning_rate": 3.9e-05,
|
| 85 |
+
"loss": 2.6081,
|
| 86 |
+
"step": 13
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 2.76,
|
| 90 |
+
"learning_rate": 4.2e-05,
|
| 91 |
+
"loss": 2.6003,
|
| 92 |
+
"step": 14
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 2.95,
|
| 96 |
+
"learning_rate": 4.4999999999999996e-05,
|
| 97 |
+
"loss": 2.6218,
|
| 98 |
+
"step": 15
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 3.19,
|
| 102 |
+
"learning_rate": 4.7999999999999994e-05,
|
| 103 |
+
"loss": 3.2488,
|
| 104 |
+
"step": 16
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 3.38,
|
| 108 |
+
"learning_rate": 5.1e-05,
|
| 109 |
+
"loss": 2.6038,
|
| 110 |
+
"step": 17
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 3.57,
|
| 114 |
+
"learning_rate": 5.399999999999999e-05,
|
| 115 |
+
"loss": 2.6556,
|
| 116 |
+
"step": 18
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 3.76,
|
| 120 |
+
"learning_rate": 5.6999999999999996e-05,
|
| 121 |
+
"loss": 2.5489,
|
| 122 |
+
"step": 19
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 3.95,
|
| 126 |
+
"learning_rate": 5.9999999999999995e-05,
|
| 127 |
+
"loss": 2.6067,
|
| 128 |
+
"step": 20
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 3.95,
|
| 132 |
+
"eval_accuracy": 0.0851063829787234,
|
| 133 |
+
"eval_f1": 0.011204481792717087,
|
| 134 |
+
"eval_loss": 2.6500837802886963,
|
| 135 |
+
"eval_runtime": 22.888,
|
| 136 |
+
"eval_samples_per_second": 12.321,
|
| 137 |
+
"eval_steps_per_second": 0.218,
|
| 138 |
+
"step": 20
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 4.19,
|
| 142 |
+
"learning_rate": 6.299999999999999e-05,
|
| 143 |
+
"loss": 3.3092,
|
| 144 |
+
"step": 21
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 4.38,
|
| 148 |
+
"learning_rate": 6.599999999999999e-05,
|
| 149 |
+
"loss": 2.5702,
|
| 150 |
+
"step": 22
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 4.57,
|
| 154 |
+
"learning_rate": 6.9e-05,
|
| 155 |
+
"loss": 2.6135,
|
| 156 |
+
"step": 23
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 4.76,
|
| 160 |
+
"learning_rate": 7.199999999999999e-05,
|
| 161 |
+
"loss": 2.5806,
|
| 162 |
+
"step": 24
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 4.95,
|
| 166 |
+
"learning_rate": 7.5e-05,
|
| 167 |
+
"loss": 2.54,
|
| 168 |
+
"step": 25
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 5.19,
|
| 172 |
+
"learning_rate": 7.8e-05,
|
| 173 |
+
"loss": 3.2317,
|
| 174 |
+
"step": 26
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 5.38,
|
| 178 |
+
"learning_rate": 8.1e-05,
|
| 179 |
+
"loss": 2.5439,
|
| 180 |
+
"step": 27
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 5.57,
|
| 184 |
+
"learning_rate": 8.4e-05,
|
| 185 |
+
"loss": 2.5509,
|
| 186 |
+
"step": 28
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 5.76,
|
| 190 |
+
"learning_rate": 8.699999999999999e-05,
|
| 191 |
+
"loss": 2.5222,
|
| 192 |
+
"step": 29
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 5.95,
|
| 196 |
+
"learning_rate": 8.999999999999999e-05,
|
| 197 |
+
"loss": 2.5358,
|
| 198 |
+
"step": 30
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 6.19,
|
| 202 |
+
"learning_rate": 9.3e-05,
|
| 203 |
+
"loss": 3.1367,
|
| 204 |
+
"step": 31
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 6.38,
|
| 208 |
+
"learning_rate": 9.599999999999999e-05,
|
| 209 |
+
"loss": 2.4961,
|
| 210 |
+
"step": 32
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 6.57,
|
| 214 |
+
"learning_rate": 9.9e-05,
|
| 215 |
+
"loss": 2.386,
|
| 216 |
+
"step": 33
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 6.76,
|
| 220 |
+
"learning_rate": 0.000102,
|
| 221 |
+
"loss": 2.5143,
|
| 222 |
+
"step": 34
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 6.95,
|
| 226 |
+
"learning_rate": 0.00010499999999999999,
|
| 227 |
+
"loss": 2.5108,
|
| 228 |
+
"step": 35
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 7.19,
|
| 232 |
+
"learning_rate": 0.00010799999999999998,
|
| 233 |
+
"loss": 3.0678,
|
| 234 |
+
"step": 36
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 7.38,
|
| 238 |
+
"learning_rate": 0.00011099999999999999,
|
| 239 |
+
"loss": 2.3575,
|
| 240 |
+
"step": 37
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 7.57,
|
| 244 |
+
"learning_rate": 0.00011399999999999999,
|
| 245 |
+
"loss": 2.5258,
|
| 246 |
+
"step": 38
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 7.76,
|
| 250 |
+
"learning_rate": 0.000117,
|
| 251 |
+
"loss": 2.4514,
|
| 252 |
+
"step": 39
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 7.95,
|
| 256 |
+
"learning_rate": 0.00011999999999999999,
|
| 257 |
+
"loss": 2.5614,
|
| 258 |
+
"step": 40
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 7.95,
|
| 262 |
+
"eval_accuracy": 0.06028368794326241,
|
| 263 |
+
"eval_f1": 0.013335536643807322,
|
| 264 |
+
"eval_loss": 2.801831007003784,
|
| 265 |
+
"eval_runtime": 22.1671,
|
| 266 |
+
"eval_samples_per_second": 12.722,
|
| 267 |
+
"eval_steps_per_second": 0.226,
|
| 268 |
+
"step": 40
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 8.19,
|
| 272 |
+
"learning_rate": 0.00012299999999999998,
|
| 273 |
+
"loss": 2.9978,
|
| 274 |
+
"step": 41
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 8.38,
|
| 278 |
+
"learning_rate": 0.00012599999999999997,
|
| 279 |
+
"loss": 2.378,
|
| 280 |
+
"step": 42
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 8.57,
|
| 284 |
+
"learning_rate": 0.000129,
|
| 285 |
+
"loss": 2.4547,
|
| 286 |
+
"step": 43
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 8.76,
|
| 290 |
+
"learning_rate": 0.00013199999999999998,
|
| 291 |
+
"loss": 2.3469,
|
| 292 |
+
"step": 44
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 8.95,
|
| 296 |
+
"learning_rate": 0.000135,
|
| 297 |
+
"loss": 2.3036,
|
| 298 |
+
"step": 45
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 9.19,
|
| 302 |
+
"learning_rate": 0.000138,
|
| 303 |
+
"loss": 3.0764,
|
| 304 |
+
"step": 46
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 9.38,
|
| 308 |
+
"learning_rate": 0.00014099999999999998,
|
| 309 |
+
"loss": 2.2139,
|
| 310 |
+
"step": 47
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 9.57,
|
| 314 |
+
"learning_rate": 0.00014399999999999998,
|
| 315 |
+
"loss": 2.1505,
|
| 316 |
+
"step": 48
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 9.76,
|
| 320 |
+
"learning_rate": 0.000147,
|
| 321 |
+
"loss": 2.2935,
|
| 322 |
+
"step": 49
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 9.95,
|
| 326 |
+
"learning_rate": 0.00015,
|
| 327 |
+
"loss": 2.3409,
|
| 328 |
+
"step": 50
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 10.19,
|
| 332 |
+
"learning_rate": 0.00015299999999999998,
|
| 333 |
+
"loss": 2.9912,
|
| 334 |
+
"step": 51
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 10.38,
|
| 338 |
+
"learning_rate": 0.000156,
|
| 339 |
+
"loss": 2.2506,
|
| 340 |
+
"step": 52
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 10.57,
|
| 344 |
+
"learning_rate": 0.000159,
|
| 345 |
+
"loss": 2.1813,
|
| 346 |
+
"step": 53
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 10.76,
|
| 350 |
+
"learning_rate": 0.000162,
|
| 351 |
+
"loss": 2.2503,
|
| 352 |
+
"step": 54
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 10.95,
|
| 356 |
+
"learning_rate": 0.000165,
|
| 357 |
+
"loss": 2.2379,
|
| 358 |
+
"step": 55
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 11.19,
|
| 362 |
+
"learning_rate": 0.000168,
|
| 363 |
+
"loss": 2.9891,
|
| 364 |
+
"step": 56
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"epoch": 11.38,
|
| 368 |
+
"learning_rate": 0.00017099999999999998,
|
| 369 |
+
"loss": 2.1612,
|
| 370 |
+
"step": 57
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 11.57,
|
| 374 |
+
"learning_rate": 0.00017399999999999997,
|
| 375 |
+
"loss": 2.1393,
|
| 376 |
+
"step": 58
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 11.76,
|
| 380 |
+
"learning_rate": 0.00017699999999999997,
|
| 381 |
+
"loss": 2.0967,
|
| 382 |
+
"step": 59
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 11.95,
|
| 386 |
+
"learning_rate": 0.00017999999999999998,
|
| 387 |
+
"loss": 2.2836,
|
| 388 |
+
"step": 60
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 11.95,
|
| 392 |
+
"eval_accuracy": 0.06028368794326241,
|
| 393 |
+
"eval_f1": 0.008374384236453201,
|
| 394 |
+
"eval_loss": 3.078601360321045,
|
| 395 |
+
"eval_runtime": 21.8372,
|
| 396 |
+
"eval_samples_per_second": 12.914,
|
| 397 |
+
"eval_steps_per_second": 0.229,
|
| 398 |
+
"step": 60
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 12.19,
|
| 402 |
+
"learning_rate": 0.00018299999999999998,
|
| 403 |
+
"loss": 2.5805,
|
| 404 |
+
"step": 61
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 12.38,
|
| 408 |
+
"learning_rate": 0.000186,
|
| 409 |
+
"loss": 2.1918,
|
| 410 |
+
"step": 62
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 12.57,
|
| 414 |
+
"learning_rate": 0.00018899999999999999,
|
| 415 |
+
"loss": 2.0755,
|
| 416 |
+
"step": 63
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 12.76,
|
| 420 |
+
"learning_rate": 0.00019199999999999998,
|
| 421 |
+
"loss": 2.1025,
|
| 422 |
+
"step": 64
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 12.95,
|
| 426 |
+
"learning_rate": 0.000195,
|
| 427 |
+
"loss": 2.0926,
|
| 428 |
+
"step": 65
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 13.19,
|
| 432 |
+
"learning_rate": 0.000198,
|
| 433 |
+
"loss": 2.6881,
|
| 434 |
+
"step": 66
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 13.38,
|
| 438 |
+
"learning_rate": 0.000201,
|
| 439 |
+
"loss": 2.0654,
|
| 440 |
+
"step": 67
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 13.57,
|
| 444 |
+
"learning_rate": 0.000204,
|
| 445 |
+
"loss": 1.8003,
|
| 446 |
+
"step": 68
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 13.76,
|
| 450 |
+
"learning_rate": 0.00020699999999999996,
|
| 451 |
+
"loss": 1.7717,
|
| 452 |
+
"step": 69
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 13.95,
|
| 456 |
+
"learning_rate": 0.00020999999999999998,
|
| 457 |
+
"loss": 2.1953,
|
| 458 |
+
"step": 70
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 14.19,
|
| 462 |
+
"learning_rate": 0.00021299999999999997,
|
| 463 |
+
"loss": 2.4275,
|
| 464 |
+
"step": 71
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 14.38,
|
| 468 |
+
"learning_rate": 0.00021599999999999996,
|
| 469 |
+
"loss": 1.9666,
|
| 470 |
+
"step": 72
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 14.57,
|
| 474 |
+
"learning_rate": 0.00021899999999999998,
|
| 475 |
+
"loss": 1.8635,
|
| 476 |
+
"step": 73
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 14.76,
|
| 480 |
+
"learning_rate": 0.00022199999999999998,
|
| 481 |
+
"loss": 1.9375,
|
| 482 |
+
"step": 74
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 14.95,
|
| 486 |
+
"learning_rate": 0.000225,
|
| 487 |
+
"loss": 1.7644,
|
| 488 |
+
"step": 75
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 15.19,
|
| 492 |
+
"learning_rate": 0.00022799999999999999,
|
| 493 |
+
"loss": 2.3216,
|
| 494 |
+
"step": 76
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 15.38,
|
| 498 |
+
"learning_rate": 0.00023099999999999998,
|
| 499 |
+
"loss": 1.7671,
|
| 500 |
+
"step": 77
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 15.57,
|
| 504 |
+
"learning_rate": 0.000234,
|
| 505 |
+
"loss": 2.0365,
|
| 506 |
+
"step": 78
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 15.76,
|
| 510 |
+
"learning_rate": 0.000237,
|
| 511 |
+
"loss": 2.0866,
|
| 512 |
+
"step": 79
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 15.95,
|
| 516 |
+
"learning_rate": 0.00023999999999999998,
|
| 517 |
+
"loss": 1.9597,
|
| 518 |
+
"step": 80
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 15.95,
|
| 522 |
+
"eval_accuracy": 0.06382978723404255,
|
| 523 |
+
"eval_f1": 0.012634220354808589,
|
| 524 |
+
"eval_loss": 3.2287864685058594,
|
| 525 |
+
"eval_runtime": 22.1138,
|
| 526 |
+
"eval_samples_per_second": 12.752,
|
| 527 |
+
"eval_steps_per_second": 0.226,
|
| 528 |
+
"step": 80
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 16.19,
|
| 532 |
+
"learning_rate": 0.000243,
|
| 533 |
+
"loss": 2.2081,
|
| 534 |
+
"step": 81
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 16.38,
|
| 538 |
+
"learning_rate": 0.00024599999999999996,
|
| 539 |
+
"loss": 1.5558,
|
| 540 |
+
"step": 82
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 16.57,
|
| 544 |
+
"learning_rate": 0.000249,
|
| 545 |
+
"loss": 1.7294,
|
| 546 |
+
"step": 83
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 16.76,
|
| 550 |
+
"learning_rate": 0.00025199999999999995,
|
| 551 |
+
"loss": 1.4959,
|
| 552 |
+
"step": 84
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 16.95,
|
| 556 |
+
"learning_rate": 0.00025499999999999996,
|
| 557 |
+
"loss": 1.9221,
|
| 558 |
+
"step": 85
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 17.19,
|
| 562 |
+
"learning_rate": 0.000258,
|
| 563 |
+
"loss": 1.939,
|
| 564 |
+
"step": 86
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 17.38,
|
| 568 |
+
"learning_rate": 0.000261,
|
| 569 |
+
"loss": 1.6908,
|
| 570 |
+
"step": 87
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 17.57,
|
| 574 |
+
"learning_rate": 0.00026399999999999997,
|
| 575 |
+
"loss": 1.7953,
|
| 576 |
+
"step": 88
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 17.76,
|
| 580 |
+
"learning_rate": 0.000267,
|
| 581 |
+
"loss": 1.557,
|
| 582 |
+
"step": 89
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 17.95,
|
| 586 |
+
"learning_rate": 0.00027,
|
| 587 |
+
"loss": 1.8179,
|
| 588 |
+
"step": 90
|
| 589 |
+
},
|
| 590 |
+
{
|
| 591 |
+
"epoch": 18.19,
|
| 592 |
+
"learning_rate": 0.00027299999999999997,
|
| 593 |
+
"loss": 1.8706,
|
| 594 |
+
"step": 91
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 18.38,
|
| 598 |
+
"learning_rate": 0.000276,
|
| 599 |
+
"loss": 1.4066,
|
| 600 |
+
"step": 92
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 18.57,
|
| 604 |
+
"learning_rate": 0.000279,
|
| 605 |
+
"loss": 1.7786,
|
| 606 |
+
"step": 93
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 18.76,
|
| 610 |
+
"learning_rate": 0.00028199999999999997,
|
| 611 |
+
"loss": 1.5466,
|
| 612 |
+
"step": 94
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 18.95,
|
| 616 |
+
"learning_rate": 0.000285,
|
| 617 |
+
"loss": 1.7946,
|
| 618 |
+
"step": 95
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 19.19,
|
| 622 |
+
"learning_rate": 0.00028799999999999995,
|
| 623 |
+
"loss": 1.6017,
|
| 624 |
+
"step": 96
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 19.38,
|
| 628 |
+
"learning_rate": 0.00029099999999999997,
|
| 629 |
+
"loss": 1.4677,
|
| 630 |
+
"step": 97
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 19.57,
|
| 634 |
+
"learning_rate": 0.000294,
|
| 635 |
+
"loss": 1.3462,
|
| 636 |
+
"step": 98
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 19.76,
|
| 640 |
+
"learning_rate": 0.00029699999999999996,
|
| 641 |
+
"loss": 1.6485,
|
| 642 |
+
"step": 99
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 19.95,
|
| 646 |
+
"learning_rate": 0.0003,
|
| 647 |
+
"loss": 1.5566,
|
| 648 |
+
"step": 100
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 19.95,
|
| 652 |
+
"eval_accuracy": 0.05673758865248227,
|
| 653 |
+
"eval_f1": 0.017791123605077095,
|
| 654 |
+
"eval_loss": 3.6933910846710205,
|
| 655 |
+
"eval_runtime": 22.0004,
|
| 656 |
+
"eval_samples_per_second": 12.818,
|
| 657 |
+
"eval_steps_per_second": 0.227,
|
| 658 |
+
"step": 100
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 20.19,
|
| 662 |
+
"learning_rate": 0.000298,
|
| 663 |
+
"loss": 1.8716,
|
| 664 |
+
"step": 101
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 20.38,
|
| 668 |
+
"learning_rate": 0.000296,
|
| 669 |
+
"loss": 1.2853,
|
| 670 |
+
"step": 102
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 20.57,
|
| 674 |
+
"learning_rate": 0.000294,
|
| 675 |
+
"loss": 1.6498,
|
| 676 |
+
"step": 103
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 20.76,
|
| 680 |
+
"learning_rate": 0.000292,
|
| 681 |
+
"loss": 1.3141,
|
| 682 |
+
"step": 104
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 20.95,
|
| 686 |
+
"learning_rate": 0.00029,
|
| 687 |
+
"loss": 1.3928,
|
| 688 |
+
"step": 105
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 21.19,
|
| 692 |
+
"learning_rate": 0.00028799999999999995,
|
| 693 |
+
"loss": 1.719,
|
| 694 |
+
"step": 106
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 21.38,
|
| 698 |
+
"learning_rate": 0.00028599999999999996,
|
| 699 |
+
"loss": 1.6084,
|
| 700 |
+
"step": 107
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 21.57,
|
| 704 |
+
"learning_rate": 0.00028399999999999996,
|
| 705 |
+
"loss": 1.4534,
|
| 706 |
+
"step": 108
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 21.76,
|
| 710 |
+
"learning_rate": 0.00028199999999999997,
|
| 711 |
+
"loss": 1.34,
|
| 712 |
+
"step": 109
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 21.95,
|
| 716 |
+
"learning_rate": 0.00028,
|
| 717 |
+
"loss": 1.2225,
|
| 718 |
+
"step": 110
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 22.19,
|
| 722 |
+
"learning_rate": 0.000278,
|
| 723 |
+
"loss": 1.8928,
|
| 724 |
+
"step": 111
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 22.38,
|
| 728 |
+
"learning_rate": 0.000276,
|
| 729 |
+
"loss": 1.1772,
|
| 730 |
+
"step": 112
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 22.57,
|
| 734 |
+
"learning_rate": 0.000274,
|
| 735 |
+
"loss": 1.2046,
|
| 736 |
+
"step": 113
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 22.76,
|
| 740 |
+
"learning_rate": 0.00027199999999999994,
|
| 741 |
+
"loss": 1.2977,
|
| 742 |
+
"step": 114
|
| 743 |
+
},
|
| 744 |
+
{
|
| 745 |
+
"epoch": 22.95,
|
| 746 |
+
"learning_rate": 0.00027,
|
| 747 |
+
"loss": 1.3624,
|
| 748 |
+
"step": 115
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 23.19,
|
| 752 |
+
"learning_rate": 0.00026799999999999995,
|
| 753 |
+
"loss": 1.3893,
|
| 754 |
+
"step": 116
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 23.38,
|
| 758 |
+
"learning_rate": 0.000266,
|
| 759 |
+
"loss": 1.0624,
|
| 760 |
+
"step": 117
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 23.57,
|
| 764 |
+
"learning_rate": 0.00026399999999999997,
|
| 765 |
+
"loss": 1.069,
|
| 766 |
+
"step": 118
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 23.76,
|
| 770 |
+
"learning_rate": 0.00026199999999999997,
|
| 771 |
+
"loss": 1.4328,
|
| 772 |
+
"step": 119
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 23.95,
|
| 776 |
+
"learning_rate": 0.00026,
|
| 777 |
+
"loss": 1.3168,
|
| 778 |
+
"step": 120
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 23.95,
|
| 782 |
+
"eval_accuracy": 0.06382978723404255,
|
| 783 |
+
"eval_f1": 0.015014809751651858,
|
| 784 |
+
"eval_loss": 3.913482666015625,
|
| 785 |
+
"eval_runtime": 21.9303,
|
| 786 |
+
"eval_samples_per_second": 12.859,
|
| 787 |
+
"eval_steps_per_second": 0.228,
|
| 788 |
+
"step": 120
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 24.19,
|
| 792 |
+
"learning_rate": 0.000258,
|
| 793 |
+
"loss": 1.3535,
|
| 794 |
+
"step": 121
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 24.38,
|
| 798 |
+
"learning_rate": 0.000256,
|
| 799 |
+
"loss": 1.225,
|
| 800 |
+
"step": 122
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 24.57,
|
| 804 |
+
"learning_rate": 0.000254,
|
| 805 |
+
"loss": 1.382,
|
| 806 |
+
"step": 123
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 24.76,
|
| 810 |
+
"learning_rate": 0.00025199999999999995,
|
| 811 |
+
"loss": 0.9501,
|
| 812 |
+
"step": 124
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 24.95,
|
| 816 |
+
"learning_rate": 0.00025,
|
| 817 |
+
"loss": 1.2305,
|
| 818 |
+
"step": 125
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 25.19,
|
| 822 |
+
"learning_rate": 0.00024799999999999996,
|
| 823 |
+
"loss": 1.3317,
|
| 824 |
+
"step": 126
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 25.38,
|
| 828 |
+
"learning_rate": 0.00024599999999999996,
|
| 829 |
+
"loss": 1.0909,
|
| 830 |
+
"step": 127
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 25.57,
|
| 834 |
+
"learning_rate": 0.000244,
|
| 835 |
+
"loss": 1.3626,
|
| 836 |
+
"step": 128
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 25.76,
|
| 840 |
+
"learning_rate": 0.00024199999999999997,
|
| 841 |
+
"loss": 1.4076,
|
| 842 |
+
"step": 129
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 25.95,
|
| 846 |
+
"learning_rate": 0.00023999999999999998,
|
| 847 |
+
"loss": 1.178,
|
| 848 |
+
"step": 130
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 26.19,
|
| 852 |
+
"learning_rate": 0.00023799999999999998,
|
| 853 |
+
"loss": 1.4365,
|
| 854 |
+
"step": 131
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 26.38,
|
| 858 |
+
"learning_rate": 0.00023599999999999996,
|
| 859 |
+
"loss": 1.2742,
|
| 860 |
+
"step": 132
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 26.57,
|
| 864 |
+
"learning_rate": 0.000234,
|
| 865 |
+
"loss": 0.9605,
|
| 866 |
+
"step": 133
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 26.76,
|
| 870 |
+
"learning_rate": 0.00023199999999999997,
|
| 871 |
+
"loss": 1.3498,
|
| 872 |
+
"step": 134
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 26.95,
|
| 876 |
+
"learning_rate": 0.00023,
|
| 877 |
+
"loss": 1.2505,
|
| 878 |
+
"step": 135
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 27.19,
|
| 882 |
+
"learning_rate": 0.00022799999999999999,
|
| 883 |
+
"loss": 1.5257,
|
| 884 |
+
"step": 136
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 27.38,
|
| 888 |
+
"learning_rate": 0.00022599999999999996,
|
| 889 |
+
"loss": 0.9837,
|
| 890 |
+
"step": 137
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 27.57,
|
| 894 |
+
"learning_rate": 0.000224,
|
| 895 |
+
"loss": 0.9088,
|
| 896 |
+
"step": 138
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 27.76,
|
| 900 |
+
"learning_rate": 0.00022199999999999998,
|
| 901 |
+
"loss": 1.0888,
|
| 902 |
+
"step": 139
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 27.95,
|
| 906 |
+
"learning_rate": 0.00021999999999999995,
|
| 907 |
+
"loss": 1.0598,
|
| 908 |
+
"step": 140
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 27.95,
|
| 912 |
+
"eval_accuracy": 0.06028368794326241,
|
| 913 |
+
"eval_f1": 0.008403361344537815,
|
| 914 |
+
"eval_loss": 4.261812210083008,
|
| 915 |
+
"eval_runtime": 22.0694,
|
| 916 |
+
"eval_samples_per_second": 12.778,
|
| 917 |
+
"eval_steps_per_second": 0.227,
|
| 918 |
+
"step": 140
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 28.19,
|
| 922 |
+
"learning_rate": 0.00021799999999999999,
|
| 923 |
+
"loss": 1.2825,
|
| 924 |
+
"step": 141
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 28.38,
|
| 928 |
+
"learning_rate": 0.00021599999999999996,
|
| 929 |
+
"loss": 0.9134,
|
| 930 |
+
"step": 142
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 28.57,
|
| 934 |
+
"learning_rate": 0.000214,
|
| 935 |
+
"loss": 0.8605,
|
| 936 |
+
"step": 143
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 28.76,
|
| 940 |
+
"learning_rate": 0.00021199999999999998,
|
| 941 |
+
"loss": 0.977,
|
| 942 |
+
"step": 144
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 28.95,
|
| 946 |
+
"learning_rate": 0.00020999999999999998,
|
| 947 |
+
"loss": 0.9525,
|
| 948 |
+
"step": 145
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 29.19,
|
| 952 |
+
"learning_rate": 0.000208,
|
| 953 |
+
"loss": 0.9748,
|
| 954 |
+
"step": 146
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 29.38,
|
| 958 |
+
"learning_rate": 0.00020599999999999997,
|
| 959 |
+
"loss": 0.9609,
|
| 960 |
+
"step": 147
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 29.57,
|
| 964 |
+
"learning_rate": 0.000204,
|
| 965 |
+
"loss": 0.8758,
|
| 966 |
+
"step": 148
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 29.76,
|
| 970 |
+
"learning_rate": 0.00020199999999999998,
|
| 971 |
+
"loss": 0.9648,
|
| 972 |
+
"step": 149
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 29.95,
|
| 976 |
+
"learning_rate": 0.00019999999999999998,
|
| 977 |
+
"loss": 0.9435,
|
| 978 |
+
"step": 150
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 30.19,
|
| 982 |
+
"learning_rate": 0.000198,
|
| 983 |
+
"loss": 0.8495,
|
| 984 |
+
"step": 151
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 30.38,
|
| 988 |
+
"learning_rate": 0.00019599999999999997,
|
| 989 |
+
"loss": 0.7653,
|
| 990 |
+
"step": 152
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 30.57,
|
| 994 |
+
"learning_rate": 0.00019399999999999997,
|
| 995 |
+
"loss": 0.5764,
|
| 996 |
+
"step": 153
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 30.76,
|
| 1000 |
+
"learning_rate": 0.00019199999999999998,
|
| 1001 |
+
"loss": 0.6918,
|
| 1002 |
+
"step": 154
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 30.95,
|
| 1006 |
+
"learning_rate": 0.00018999999999999998,
|
| 1007 |
+
"loss": 0.9413,
|
| 1008 |
+
"step": 155
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 31.19,
|
| 1012 |
+
"learning_rate": 0.000188,
|
| 1013 |
+
"loss": 1.0015,
|
| 1014 |
+
"step": 156
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 31.38,
|
| 1018 |
+
"learning_rate": 0.000186,
|
| 1019 |
+
"loss": 0.7529,
|
| 1020 |
+
"step": 157
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 31.57,
|
| 1024 |
+
"learning_rate": 0.00018399999999999997,
|
| 1025 |
+
"loss": 0.8388,
|
| 1026 |
+
"step": 158
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 31.76,
|
| 1030 |
+
"learning_rate": 0.00018199999999999998,
|
| 1031 |
+
"loss": 0.6467,
|
| 1032 |
+
"step": 159
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 31.95,
|
| 1036 |
+
"learning_rate": 0.00017999999999999998,
|
| 1037 |
+
"loss": 0.5721,
|
| 1038 |
+
"step": 160
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 31.95,
|
| 1042 |
+
"eval_accuracy": 0.07801418439716312,
|
| 1043 |
+
"eval_f1": 0.035389118125809506,
|
| 1044 |
+
"eval_loss": 3.7973175048828125,
|
| 1045 |
+
"eval_runtime": 22.0436,
|
| 1046 |
+
"eval_samples_per_second": 12.793,
|
| 1047 |
+
"eval_steps_per_second": 0.227,
|
| 1048 |
+
"step": 160
|
| 1049 |
+
},
|
| 1050 |
+
{
|
| 1051 |
+
"epoch": 32.19,
|
| 1052 |
+
"learning_rate": 0.000178,
|
| 1053 |
+
"loss": 1.0825,
|
| 1054 |
+
"step": 161
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 32.38,
|
| 1058 |
+
"learning_rate": 0.000176,
|
| 1059 |
+
"loss": 0.5634,
|
| 1060 |
+
"step": 162
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 32.57,
|
| 1064 |
+
"learning_rate": 0.00017399999999999997,
|
| 1065 |
+
"loss": 0.8141,
|
| 1066 |
+
"step": 163
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 32.76,
|
| 1070 |
+
"learning_rate": 0.000172,
|
| 1071 |
+
"loss": 1.0187,
|
| 1072 |
+
"step": 164
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 32.95,
|
| 1076 |
+
"learning_rate": 0.00016999999999999999,
|
| 1077 |
+
"loss": 0.724,
|
| 1078 |
+
"step": 165
|
| 1079 |
+
},
|
| 1080 |
+
{
|
| 1081 |
+
"epoch": 33.19,
|
| 1082 |
+
"learning_rate": 0.000168,
|
| 1083 |
+
"loss": 1.0303,
|
| 1084 |
+
"step": 166
|
| 1085 |
+
},
|
| 1086 |
+
{
|
| 1087 |
+
"epoch": 33.38,
|
| 1088 |
+
"learning_rate": 0.000166,
|
| 1089 |
+
"loss": 0.7125,
|
| 1090 |
+
"step": 167
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 33.57,
|
| 1094 |
+
"learning_rate": 0.00016399999999999997,
|
| 1095 |
+
"loss": 0.6681,
|
| 1096 |
+
"step": 168
|
| 1097 |
+
},
|
| 1098 |
+
{
|
| 1099 |
+
"epoch": 33.76,
|
| 1100 |
+
"learning_rate": 0.000162,
|
| 1101 |
+
"loss": 0.5326,
|
| 1102 |
+
"step": 169
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 33.95,
|
| 1106 |
+
"learning_rate": 0.00015999999999999999,
|
| 1107 |
+
"loss": 0.5521,
|
| 1108 |
+
"step": 170
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 34.19,
|
| 1112 |
+
"learning_rate": 0.00015799999999999996,
|
| 1113 |
+
"loss": 0.5982,
|
| 1114 |
+
"step": 171
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 34.38,
|
| 1118 |
+
"learning_rate": 0.000156,
|
| 1119 |
+
"loss": 0.5877,
|
| 1120 |
+
"step": 172
|
| 1121 |
+
},
|
| 1122 |
+
{
|
| 1123 |
+
"epoch": 34.57,
|
| 1124 |
+
"learning_rate": 0.00015399999999999998,
|
| 1125 |
+
"loss": 0.7232,
|
| 1126 |
+
"step": 173
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 34.76,
|
| 1130 |
+
"learning_rate": 0.000152,
|
| 1131 |
+
"loss": 0.7139,
|
| 1132 |
+
"step": 174
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"epoch": 34.95,
|
| 1136 |
+
"learning_rate": 0.00015,
|
| 1137 |
+
"loss": 0.4435,
|
| 1138 |
+
"step": 175
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 35.19,
|
| 1142 |
+
"learning_rate": 0.000148,
|
| 1143 |
+
"loss": 0.5186,
|
| 1144 |
+
"step": 176
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 35.38,
|
| 1148 |
+
"learning_rate": 0.000146,
|
| 1149 |
+
"loss": 0.501,
|
| 1150 |
+
"step": 177
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 35.57,
|
| 1154 |
+
"learning_rate": 0.00014399999999999998,
|
| 1155 |
+
"loss": 0.6495,
|
| 1156 |
+
"step": 178
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 35.76,
|
| 1160 |
+
"learning_rate": 0.00014199999999999998,
|
| 1161 |
+
"loss": 0.6904,
|
| 1162 |
+
"step": 179
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 35.95,
|
| 1166 |
+
"learning_rate": 0.00014,
|
| 1167 |
+
"loss": 0.4402,
|
| 1168 |
+
"step": 180
|
| 1169 |
+
},
|
| 1170 |
+
{
|
| 1171 |
+
"epoch": 35.95,
|
| 1172 |
+
"eval_accuracy": 0.06382978723404255,
|
| 1173 |
+
"eval_f1": 0.017911912648754753,
|
| 1174 |
+
"eval_loss": 4.623348712921143,
|
| 1175 |
+
"eval_runtime": 21.8741,
|
| 1176 |
+
"eval_samples_per_second": 12.892,
|
| 1177 |
+
"eval_steps_per_second": 0.229,
|
| 1178 |
+
"step": 180
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 36.19,
|
| 1182 |
+
"learning_rate": 0.000138,
|
| 1183 |
+
"loss": 0.5381,
|
| 1184 |
+
"step": 181
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 36.38,
|
| 1188 |
+
"learning_rate": 0.00013599999999999997,
|
| 1189 |
+
"loss": 0.3591,
|
| 1190 |
+
"step": 182
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 36.57,
|
| 1194 |
+
"learning_rate": 0.00013399999999999998,
|
| 1195 |
+
"loss": 0.9615,
|
| 1196 |
+
"step": 183
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 36.76,
|
| 1200 |
+
"learning_rate": 0.00013199999999999998,
|
| 1201 |
+
"loss": 0.8523,
|
| 1202 |
+
"step": 184
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 36.95,
|
| 1206 |
+
"learning_rate": 0.00013,
|
| 1207 |
+
"loss": 0.5768,
|
| 1208 |
+
"step": 185
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 37.19,
|
| 1212 |
+
"learning_rate": 0.000128,
|
| 1213 |
+
"loss": 0.8723,
|
| 1214 |
+
"step": 186
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 37.38,
|
| 1218 |
+
"learning_rate": 0.00012599999999999997,
|
| 1219 |
+
"loss": 0.5213,
|
| 1220 |
+
"step": 187
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 37.57,
|
| 1224 |
+
"learning_rate": 0.00012399999999999998,
|
| 1225 |
+
"loss": 0.4604,
|
| 1226 |
+
"step": 188
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 37.76,
|
| 1230 |
+
"learning_rate": 0.000122,
|
| 1231 |
+
"loss": 0.3154,
|
| 1232 |
+
"step": 189
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 37.95,
|
| 1236 |
+
"learning_rate": 0.00011999999999999999,
|
| 1237 |
+
"loss": 0.8434,
|
| 1238 |
+
"step": 190
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 38.19,
|
| 1242 |
+
"learning_rate": 0.00011799999999999998,
|
| 1243 |
+
"loss": 0.6779,
|
| 1244 |
+
"step": 191
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 38.38,
|
| 1248 |
+
"learning_rate": 0.00011599999999999999,
|
| 1249 |
+
"loss": 0.394,
|
| 1250 |
+
"step": 192
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 38.57,
|
| 1254 |
+
"learning_rate": 0.00011399999999999999,
|
| 1255 |
+
"loss": 0.6969,
|
| 1256 |
+
"step": 193
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 38.76,
|
| 1260 |
+
"learning_rate": 0.000112,
|
| 1261 |
+
"loss": 0.729,
|
| 1262 |
+
"step": 194
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 38.95,
|
| 1266 |
+
"learning_rate": 0.00010999999999999998,
|
| 1267 |
+
"loss": 0.5887,
|
| 1268 |
+
"step": 195
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 39.19,
|
| 1272 |
+
"learning_rate": 0.00010799999999999998,
|
| 1273 |
+
"loss": 0.5589,
|
| 1274 |
+
"step": 196
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 39.38,
|
| 1278 |
+
"learning_rate": 0.00010599999999999999,
|
| 1279 |
+
"loss": 0.4659,
|
| 1280 |
+
"step": 197
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 39.57,
|
| 1284 |
+
"learning_rate": 0.000104,
|
| 1285 |
+
"loss": 0.3563,
|
| 1286 |
+
"step": 198
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 39.76,
|
| 1290 |
+
"learning_rate": 0.000102,
|
| 1291 |
+
"loss": 0.5006,
|
| 1292 |
+
"step": 199
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 39.95,
|
| 1296 |
+
"learning_rate": 9.999999999999999e-05,
|
| 1297 |
+
"loss": 0.6113,
|
| 1298 |
+
"step": 200
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 39.95,
|
| 1302 |
+
"eval_accuracy": 0.0673758865248227,
|
| 1303 |
+
"eval_f1": 0.020788735074449362,
|
| 1304 |
+
"eval_loss": 4.614910125732422,
|
| 1305 |
+
"eval_runtime": 22.0677,
|
| 1306 |
+
"eval_samples_per_second": 12.779,
|
| 1307 |
+
"eval_steps_per_second": 0.227,
|
| 1308 |
+
"step": 200
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 40.19,
|
| 1312 |
+
"learning_rate": 9.799999999999998e-05,
|
| 1313 |
+
"loss": 0.5995,
|
| 1314 |
+
"step": 201
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 40.38,
|
| 1318 |
+
"learning_rate": 9.599999999999999e-05,
|
| 1319 |
+
"loss": 0.3788,
|
| 1320 |
+
"step": 202
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 40.57,
|
| 1324 |
+
"learning_rate": 9.4e-05,
|
| 1325 |
+
"loss": 0.4077,
|
| 1326 |
+
"step": 203
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 40.76,
|
| 1330 |
+
"learning_rate": 9.199999999999999e-05,
|
| 1331 |
+
"loss": 0.6014,
|
| 1332 |
+
"step": 204
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 40.95,
|
| 1336 |
+
"learning_rate": 8.999999999999999e-05,
|
| 1337 |
+
"loss": 0.4386,
|
| 1338 |
+
"step": 205
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 41.19,
|
| 1342 |
+
"learning_rate": 8.8e-05,
|
| 1343 |
+
"loss": 0.6344,
|
| 1344 |
+
"step": 206
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 41.38,
|
| 1348 |
+
"learning_rate": 8.6e-05,
|
| 1349 |
+
"loss": 0.4539,
|
| 1350 |
+
"step": 207
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 41.57,
|
| 1354 |
+
"learning_rate": 8.4e-05,
|
| 1355 |
+
"loss": 0.5215,
|
| 1356 |
+
"step": 208
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 41.76,
|
| 1360 |
+
"learning_rate": 8.199999999999999e-05,
|
| 1361 |
+
"loss": 0.3119,
|
| 1362 |
+
"step": 209
|
| 1363 |
+
},
|
| 1364 |
+
{
|
| 1365 |
+
"epoch": 41.95,
|
| 1366 |
+
"learning_rate": 7.999999999999999e-05,
|
| 1367 |
+
"loss": 0.605,
|
| 1368 |
+
"step": 210
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 42.19,
|
| 1372 |
+
"learning_rate": 7.8e-05,
|
| 1373 |
+
"loss": 0.6575,
|
| 1374 |
+
"step": 211
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 42.38,
|
| 1378 |
+
"learning_rate": 7.6e-05,
|
| 1379 |
+
"loss": 0.3662,
|
| 1380 |
+
"step": 212
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 42.57,
|
| 1384 |
+
"learning_rate": 7.4e-05,
|
| 1385 |
+
"loss": 0.2629,
|
| 1386 |
+
"step": 213
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 42.76,
|
| 1390 |
+
"learning_rate": 7.199999999999999e-05,
|
| 1391 |
+
"loss": 0.4347,
|
| 1392 |
+
"step": 214
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"epoch": 42.95,
|
| 1396 |
+
"learning_rate": 7e-05,
|
| 1397 |
+
"loss": 0.5917,
|
| 1398 |
+
"step": 215
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"epoch": 43.19,
|
| 1402 |
+
"learning_rate": 6.799999999999999e-05,
|
| 1403 |
+
"loss": 0.7379,
|
| 1404 |
+
"step": 216
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 43.38,
|
| 1408 |
+
"learning_rate": 6.599999999999999e-05,
|
| 1409 |
+
"loss": 0.5227,
|
| 1410 |
+
"step": 217
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 43.57,
|
| 1414 |
+
"learning_rate": 6.4e-05,
|
| 1415 |
+
"loss": 0.2489,
|
| 1416 |
+
"step": 218
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 43.76,
|
| 1420 |
+
"learning_rate": 6.199999999999999e-05,
|
| 1421 |
+
"loss": 0.4983,
|
| 1422 |
+
"step": 219
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 43.95,
|
| 1426 |
+
"learning_rate": 5.9999999999999995e-05,
|
| 1427 |
+
"loss": 0.3938,
|
| 1428 |
+
"step": 220
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 43.95,
|
| 1432 |
+
"eval_accuracy": 0.06382978723404255,
|
| 1433 |
+
"eval_f1": 0.015878725590955805,
|
| 1434 |
+
"eval_loss": 4.788589954376221,
|
| 1435 |
+
"eval_runtime": 21.8918,
|
| 1436 |
+
"eval_samples_per_second": 12.882,
|
| 1437 |
+
"eval_steps_per_second": 0.228,
|
| 1438 |
+
"step": 220
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 44.19,
|
| 1442 |
+
"learning_rate": 5.7999999999999994e-05,
|
| 1443 |
+
"loss": 0.3637,
|
| 1444 |
+
"step": 221
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 44.38,
|
| 1448 |
+
"learning_rate": 5.6e-05,
|
| 1449 |
+
"loss": 0.4839,
|
| 1450 |
+
"step": 222
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 44.57,
|
| 1454 |
+
"learning_rate": 5.399999999999999e-05,
|
| 1455 |
+
"loss": 0.3104,
|
| 1456 |
+
"step": 223
|
| 1457 |
+
},
|
| 1458 |
+
{
|
| 1459 |
+
"epoch": 44.76,
|
| 1460 |
+
"learning_rate": 5.2e-05,
|
| 1461 |
+
"loss": 0.2964,
|
| 1462 |
+
"step": 224
|
| 1463 |
+
},
|
| 1464 |
+
{
|
| 1465 |
+
"epoch": 44.95,
|
| 1466 |
+
"learning_rate": 4.9999999999999996e-05,
|
| 1467 |
+
"loss": 0.3151,
|
| 1468 |
+
"step": 225
|
| 1469 |
+
},
|
| 1470 |
+
{
|
| 1471 |
+
"epoch": 45.19,
|
| 1472 |
+
"learning_rate": 4.7999999999999994e-05,
|
| 1473 |
+
"loss": 0.3447,
|
| 1474 |
+
"step": 226
|
| 1475 |
+
},
|
| 1476 |
+
{
|
| 1477 |
+
"epoch": 45.38,
|
| 1478 |
+
"learning_rate": 4.599999999999999e-05,
|
| 1479 |
+
"loss": 0.276,
|
| 1480 |
+
"step": 227
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 45.57,
|
| 1484 |
+
"learning_rate": 4.4e-05,
|
| 1485 |
+
"loss": 0.4807,
|
| 1486 |
+
"step": 228
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 45.76,
|
| 1490 |
+
"learning_rate": 4.2e-05,
|
| 1491 |
+
"loss": 0.4292,
|
| 1492 |
+
"step": 229
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 45.95,
|
| 1496 |
+
"learning_rate": 3.9999999999999996e-05,
|
| 1497 |
+
"loss": 0.438,
|
| 1498 |
+
"step": 230
|
| 1499 |
+
},
|
| 1500 |
+
{
|
| 1501 |
+
"epoch": 46.19,
|
| 1502 |
+
"learning_rate": 3.8e-05,
|
| 1503 |
+
"loss": 0.5133,
|
| 1504 |
+
"step": 231
|
| 1505 |
+
},
|
| 1506 |
+
{
|
| 1507 |
+
"epoch": 46.38,
|
| 1508 |
+
"learning_rate": 3.5999999999999994e-05,
|
| 1509 |
+
"loss": 0.4525,
|
| 1510 |
+
"step": 232
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"epoch": 46.57,
|
| 1514 |
+
"learning_rate": 3.399999999999999e-05,
|
| 1515 |
+
"loss": 0.3549,
|
| 1516 |
+
"step": 233
|
| 1517 |
+
},
|
| 1518 |
+
{
|
| 1519 |
+
"epoch": 46.76,
|
| 1520 |
+
"learning_rate": 3.2e-05,
|
| 1521 |
+
"loss": 0.2801,
|
| 1522 |
+
"step": 234
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 46.95,
|
| 1526 |
+
"learning_rate": 2.9999999999999997e-05,
|
| 1527 |
+
"loss": 0.2622,
|
| 1528 |
+
"step": 235
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 47.19,
|
| 1532 |
+
"learning_rate": 2.8e-05,
|
| 1533 |
+
"loss": 0.5435,
|
| 1534 |
+
"step": 236
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 47.38,
|
| 1538 |
+
"learning_rate": 2.6e-05,
|
| 1539 |
+
"loss": 0.221,
|
| 1540 |
+
"step": 237
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"epoch": 47.57,
|
| 1544 |
+
"learning_rate": 2.3999999999999997e-05,
|
| 1545 |
+
"loss": 0.2311,
|
| 1546 |
+
"step": 238
|
| 1547 |
+
},
|
| 1548 |
+
{
|
| 1549 |
+
"epoch": 47.76,
|
| 1550 |
+
"learning_rate": 2.2e-05,
|
| 1551 |
+
"loss": 0.2953,
|
| 1552 |
+
"step": 239
|
| 1553 |
+
},
|
| 1554 |
+
{
|
| 1555 |
+
"epoch": 47.95,
|
| 1556 |
+
"learning_rate": 1.9999999999999998e-05,
|
| 1557 |
+
"loss": 0.2473,
|
| 1558 |
+
"step": 240
|
| 1559 |
+
},
|
| 1560 |
+
{
|
| 1561 |
+
"epoch": 47.95,
|
| 1562 |
+
"eval_accuracy": 0.06382978723404255,
|
| 1563 |
+
"eval_f1": 0.01542207792207792,
|
| 1564 |
+
"eval_loss": 4.732133388519287,
|
| 1565 |
+
"eval_runtime": 21.9819,
|
| 1566 |
+
"eval_samples_per_second": 12.829,
|
| 1567 |
+
"eval_steps_per_second": 0.227,
|
| 1568 |
+
"step": 240
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 48.19,
|
| 1572 |
+
"learning_rate": 1.7999999999999997e-05,
|
| 1573 |
+
"loss": 0.3288,
|
| 1574 |
+
"step": 241
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 48.38,
|
| 1578 |
+
"learning_rate": 1.6e-05,
|
| 1579 |
+
"loss": 0.4118,
|
| 1580 |
+
"step": 242
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 48.57,
|
| 1584 |
+
"learning_rate": 1.4e-05,
|
| 1585 |
+
"loss": 0.2015,
|
| 1586 |
+
"step": 243
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 48.76,
|
| 1590 |
+
"learning_rate": 1.1999999999999999e-05,
|
| 1591 |
+
"loss": 0.1637,
|
| 1592 |
+
"step": 244
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 48.95,
|
| 1596 |
+
"learning_rate": 9.999999999999999e-06,
|
| 1597 |
+
"loss": 0.533,
|
| 1598 |
+
"step": 245
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 49.19,
|
| 1602 |
+
"learning_rate": 8e-06,
|
| 1603 |
+
"loss": 0.4164,
|
| 1604 |
+
"step": 246
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 49.38,
|
| 1608 |
+
"learning_rate": 5.999999999999999e-06,
|
| 1609 |
+
"loss": 0.2681,
|
| 1610 |
+
"step": 247
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 49.57,
|
| 1614 |
+
"learning_rate": 4e-06,
|
| 1615 |
+
"loss": 0.2715,
|
| 1616 |
+
"step": 248
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 49.76,
|
| 1620 |
+
"learning_rate": 2e-06,
|
| 1621 |
+
"loss": 0.2028,
|
| 1622 |
+
"step": 249
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 49.95,
|
| 1626 |
+
"learning_rate": 0.0,
|
| 1627 |
+
"loss": 0.2395,
|
| 1628 |
+
"step": 250
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 49.95,
|
| 1632 |
+
"step": 250,
|
| 1633 |
+
"total_flos": 2.497996718910248e+18,
|
| 1634 |
+
"train_loss": 1.3846193885803222,
|
| 1635 |
+
"train_runtime": 2065.3687,
|
| 1636 |
+
"train_samples_per_second": 3.97,
|
| 1637 |
+
"train_steps_per_second": 0.121
|
| 1638 |
+
}
|
| 1639 |
+
],
|
| 1640 |
+
"max_steps": 250,
|
| 1641 |
+
"num_train_epochs": 50,
|
| 1642 |
+
"total_flos": 2.497996718910248e+18,
|
| 1643 |
+
"trial_name": null,
|
| 1644 |
+
"trial_params": null
|
| 1645 |
+
}
|