Commit
·
e230b5f
1
Parent(s):
0fec038
End of training
Browse files- all_results.json +10 -5
- eval_results.json +6 -5
- train_results.json +7 -0
- trainer_state.json +1207 -0
all_results.json
CHANGED
|
@@ -1,7 +1,12 @@
|
|
| 1 |
{
|
| 2 |
-
"
|
| 3 |
-
"
|
| 4 |
-
"
|
| 5 |
-
"
|
| 6 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 30.0,
|
| 3 |
+
"eval_f1": 0.003679096298311292,
|
| 4 |
+
"eval_loss": 5.585614204406738,
|
| 5 |
+
"eval_runtime": 105.002,
|
| 6 |
+
"eval_samples_per_second": 219.158,
|
| 7 |
+
"eval_steps_per_second": 3.429,
|
| 8 |
+
"train_loss": 2.558345020757144,
|
| 9 |
+
"train_runtime": 41575.9952,
|
| 10 |
+
"train_samples_per_second": 94.093,
|
| 11 |
+
"train_steps_per_second": 1.471
|
| 12 |
}
|
eval_results.json
CHANGED
|
@@ -1,7 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"
|
| 3 |
-
"
|
| 4 |
-
"
|
| 5 |
-
"
|
| 6 |
-
"
|
|
|
|
| 7 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 30.0,
|
| 3 |
+
"eval_f1": 0.003679096298311292,
|
| 4 |
+
"eval_loss": 5.585614204406738,
|
| 5 |
+
"eval_runtime": 105.002,
|
| 6 |
+
"eval_samples_per_second": 219.158,
|
| 7 |
+
"eval_steps_per_second": 3.429
|
| 8 |
}
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 30.0,
|
| 3 |
+
"train_loss": 2.558345020757144,
|
| 4 |
+
"train_runtime": 41575.9952,
|
| 5 |
+
"train_samples_per_second": 94.093,
|
| 6 |
+
"train_steps_per_second": 1.471
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1207 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 5.585614204406738,
|
| 3 |
+
"best_model_checkpoint": "./iiif_manuscripts_label_ge_50_convnext/checkpoint-12228",
|
| 4 |
+
"epoch": 30.0,
|
| 5 |
+
"global_step": 61140,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.2,
|
| 12 |
+
"learning_rate": 0.00019869152764147858,
|
| 13 |
+
"loss": 7.9686,
|
| 14 |
+
"step": 400
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.39,
|
| 18 |
+
"learning_rate": 0.00019738632646385348,
|
| 19 |
+
"loss": 7.2475,
|
| 20 |
+
"step": 800
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.59,
|
| 24 |
+
"learning_rate": 0.00019607785410533205,
|
| 25 |
+
"loss": 6.9462,
|
| 26 |
+
"step": 1200
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.79,
|
| 30 |
+
"learning_rate": 0.0001947693817468106,
|
| 31 |
+
"loss": 6.7615,
|
| 32 |
+
"step": 1600
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.98,
|
| 36 |
+
"learning_rate": 0.0001934641805691855,
|
| 37 |
+
"loss": 6.5753,
|
| 38 |
+
"step": 2000
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 1.0,
|
| 42 |
+
"eval_f1": 0.0015526657039264577,
|
| 43 |
+
"eval_loss": 6.412076950073242,
|
| 44 |
+
"eval_runtime": 102.5706,
|
| 45 |
+
"eval_samples_per_second": 224.353,
|
| 46 |
+
"eval_steps_per_second": 3.51,
|
| 47 |
+
"step": 2038
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 1.18,
|
| 51 |
+
"learning_rate": 0.00019215570821066406,
|
| 52 |
+
"loss": 6.324,
|
| 53 |
+
"step": 2400
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 1.37,
|
| 57 |
+
"learning_rate": 0.00019084723585214262,
|
| 58 |
+
"loss": 6.2462,
|
| 59 |
+
"step": 2800
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 1.57,
|
| 63 |
+
"learning_rate": 0.00018953876349362122,
|
| 64 |
+
"loss": 6.1389,
|
| 65 |
+
"step": 3200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 1.77,
|
| 69 |
+
"learning_rate": 0.00018823029113509978,
|
| 70 |
+
"loss": 6.0919,
|
| 71 |
+
"step": 3600
|
| 72 |
+
},
|
| 73 |
+
{
|
| 74 |
+
"epoch": 1.96,
|
| 75 |
+
"learning_rate": 0.00018692508995747466,
|
| 76 |
+
"loss": 5.9865,
|
| 77 |
+
"step": 4000
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 2.0,
|
| 81 |
+
"eval_f1": 0.002120415247137495,
|
| 82 |
+
"eval_loss": 5.946579456329346,
|
| 83 |
+
"eval_runtime": 101.947,
|
| 84 |
+
"eval_samples_per_second": 225.725,
|
| 85 |
+
"eval_steps_per_second": 3.531,
|
| 86 |
+
"step": 4076
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 2.16,
|
| 90 |
+
"learning_rate": 0.00018561661759895323,
|
| 91 |
+
"loss": 5.7381,
|
| 92 |
+
"step": 4400
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 2.36,
|
| 96 |
+
"learning_rate": 0.0001843114164213281,
|
| 97 |
+
"loss": 5.6812,
|
| 98 |
+
"step": 4800
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 2.55,
|
| 102 |
+
"learning_rate": 0.0001830029440628067,
|
| 103 |
+
"loss": 5.6661,
|
| 104 |
+
"step": 5200
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 2.75,
|
| 108 |
+
"learning_rate": 0.00018169447170428526,
|
| 109 |
+
"loss": 5.6508,
|
| 110 |
+
"step": 5600
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 2.94,
|
| 114 |
+
"learning_rate": 0.00018038599934576383,
|
| 115 |
+
"loss": 5.6521,
|
| 116 |
+
"step": 6000
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 3.0,
|
| 120 |
+
"eval_f1": 0.002850390938715168,
|
| 121 |
+
"eval_loss": 5.764512062072754,
|
| 122 |
+
"eval_runtime": 99.5338,
|
| 123 |
+
"eval_samples_per_second": 231.198,
|
| 124 |
+
"eval_steps_per_second": 3.617,
|
| 125 |
+
"step": 6114
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 3.14,
|
| 129 |
+
"learning_rate": 0.0001790775269872424,
|
| 130 |
+
"loss": 5.3744,
|
| 131 |
+
"step": 6400
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 3.34,
|
| 135 |
+
"learning_rate": 0.00017776905462872097,
|
| 136 |
+
"loss": 5.3159,
|
| 137 |
+
"step": 6800
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 3.53,
|
| 141 |
+
"learning_rate": 0.00017646058227019956,
|
| 142 |
+
"loss": 5.3162,
|
| 143 |
+
"step": 7200
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 3.73,
|
| 147 |
+
"learning_rate": 0.00017515210991167813,
|
| 148 |
+
"loss": 5.3139,
|
| 149 |
+
"step": 7600
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 3.93,
|
| 153 |
+
"learning_rate": 0.0001738436375531567,
|
| 154 |
+
"loss": 5.3123,
|
| 155 |
+
"step": 8000
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 4.0,
|
| 159 |
+
"eval_f1": 0.0032535394402472113,
|
| 160 |
+
"eval_loss": 5.688981056213379,
|
| 161 |
+
"eval_runtime": 98.5752,
|
| 162 |
+
"eval_samples_per_second": 233.446,
|
| 163 |
+
"eval_steps_per_second": 3.652,
|
| 164 |
+
"step": 8152
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 4.12,
|
| 168 |
+
"learning_rate": 0.00017253516519463526,
|
| 169 |
+
"loss": 5.0622,
|
| 170 |
+
"step": 8400
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 4.32,
|
| 174 |
+
"learning_rate": 0.00017122996401701014,
|
| 175 |
+
"loss": 4.9616,
|
| 176 |
+
"step": 8800
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 4.51,
|
| 180 |
+
"learning_rate": 0.00016992476283938504,
|
| 181 |
+
"loss": 5.0117,
|
| 182 |
+
"step": 9200
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 4.71,
|
| 186 |
+
"learning_rate": 0.0001686162904808636,
|
| 187 |
+
"loss": 5.0168,
|
| 188 |
+
"step": 9600
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 4.91,
|
| 192 |
+
"learning_rate": 0.00016730781812234217,
|
| 193 |
+
"loss": 5.0337,
|
| 194 |
+
"step": 10000
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 5.0,
|
| 198 |
+
"eval_f1": 0.003394306441476293,
|
| 199 |
+
"eval_loss": 5.6691813468933105,
|
| 200 |
+
"eval_runtime": 106.5631,
|
| 201 |
+
"eval_samples_per_second": 215.947,
|
| 202 |
+
"eval_steps_per_second": 3.378,
|
| 203 |
+
"step": 10190
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 5.1,
|
| 207 |
+
"learning_rate": 0.00016599934576382074,
|
| 208 |
+
"loss": 4.8182,
|
| 209 |
+
"step": 10400
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 5.3,
|
| 213 |
+
"learning_rate": 0.0001646908734052993,
|
| 214 |
+
"loss": 4.6457,
|
| 215 |
+
"step": 10800
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 5.5,
|
| 219 |
+
"learning_rate": 0.0001633824010467779,
|
| 220 |
+
"loss": 4.6918,
|
| 221 |
+
"step": 11200
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 5.69,
|
| 225 |
+
"learning_rate": 0.00016207719986915278,
|
| 226 |
+
"loss": 4.7441,
|
| 227 |
+
"step": 11600
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 5.89,
|
| 231 |
+
"learning_rate": 0.00016076872751063134,
|
| 232 |
+
"loss": 4.743,
|
| 233 |
+
"step": 12000
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 6.0,
|
| 237 |
+
"eval_f1": 0.003679096298311292,
|
| 238 |
+
"eval_loss": 5.585614204406738,
|
| 239 |
+
"eval_runtime": 103.5743,
|
| 240 |
+
"eval_samples_per_second": 222.179,
|
| 241 |
+
"eval_steps_per_second": 3.476,
|
| 242 |
+
"step": 12228
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 6.08,
|
| 246 |
+
"learning_rate": 0.0001594602551521099,
|
| 247 |
+
"loss": 4.5617,
|
| 248 |
+
"step": 12400
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 6.28,
|
| 252 |
+
"learning_rate": 0.00015815178279358848,
|
| 253 |
+
"loss": 4.3335,
|
| 254 |
+
"step": 12800
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 6.48,
|
| 258 |
+
"learning_rate": 0.00015684331043506705,
|
| 259 |
+
"loss": 4.3856,
|
| 260 |
+
"step": 13200
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 6.67,
|
| 264 |
+
"learning_rate": 0.00015553810925744195,
|
| 265 |
+
"loss": 4.3955,
|
| 266 |
+
"step": 13600
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 6.87,
|
| 270 |
+
"learning_rate": 0.00015422963689892052,
|
| 271 |
+
"loss": 4.4387,
|
| 272 |
+
"step": 14000
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 7.0,
|
| 276 |
+
"eval_f1": 0.004160028924400371,
|
| 277 |
+
"eval_loss": 5.596897125244141,
|
| 278 |
+
"eval_runtime": 104.1264,
|
| 279 |
+
"eval_samples_per_second": 221.001,
|
| 280 |
+
"eval_steps_per_second": 3.457,
|
| 281 |
+
"step": 14266
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 7.07,
|
| 285 |
+
"learning_rate": 0.00015292116454039908,
|
| 286 |
+
"loss": 4.3069,
|
| 287 |
+
"step": 14400
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 7.26,
|
| 291 |
+
"learning_rate": 0.00015161269218187765,
|
| 292 |
+
"loss": 3.9931,
|
| 293 |
+
"step": 14800
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 7.46,
|
| 297 |
+
"learning_rate": 0.00015030421982335622,
|
| 298 |
+
"loss": 4.0845,
|
| 299 |
+
"step": 15200
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 7.65,
|
| 303 |
+
"learning_rate": 0.00014899901864573112,
|
| 304 |
+
"loss": 4.1085,
|
| 305 |
+
"step": 15600
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 7.85,
|
| 309 |
+
"learning_rate": 0.00014769054628720969,
|
| 310 |
+
"loss": 4.1422,
|
| 311 |
+
"step": 16000
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 8.0,
|
| 315 |
+
"eval_f1": 0.004266402669461609,
|
| 316 |
+
"eval_loss": 5.671103477478027,
|
| 317 |
+
"eval_runtime": 107.1495,
|
| 318 |
+
"eval_samples_per_second": 214.765,
|
| 319 |
+
"eval_steps_per_second": 3.36,
|
| 320 |
+
"step": 16304
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 8.05,
|
| 324 |
+
"learning_rate": 0.00014638207392868825,
|
| 325 |
+
"loss": 4.0586,
|
| 326 |
+
"step": 16400
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 8.24,
|
| 330 |
+
"learning_rate": 0.00014507360157016682,
|
| 331 |
+
"loss": 3.6644,
|
| 332 |
+
"step": 16800
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 8.44,
|
| 336 |
+
"learning_rate": 0.00014376512921164541,
|
| 337 |
+
"loss": 3.7197,
|
| 338 |
+
"step": 17200
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 8.64,
|
| 342 |
+
"learning_rate": 0.0001424599280340203,
|
| 343 |
+
"loss": 3.796,
|
| 344 |
+
"step": 17600
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 8.83,
|
| 348 |
+
"learning_rate": 0.00014115145567549886,
|
| 349 |
+
"loss": 3.8372,
|
| 350 |
+
"step": 18000
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 9.0,
|
| 354 |
+
"eval_f1": 0.004420450860447925,
|
| 355 |
+
"eval_loss": 5.676065921783447,
|
| 356 |
+
"eval_runtime": 104.689,
|
| 357 |
+
"eval_samples_per_second": 219.813,
|
| 358 |
+
"eval_steps_per_second": 3.439,
|
| 359 |
+
"step": 18342
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 9.03,
|
| 363 |
+
"learning_rate": 0.00013984298331697742,
|
| 364 |
+
"loss": 3.7798,
|
| 365 |
+
"step": 18400
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 9.22,
|
| 369 |
+
"learning_rate": 0.000138534510958456,
|
| 370 |
+
"loss": 3.3063,
|
| 371 |
+
"step": 18800
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 9.42,
|
| 375 |
+
"learning_rate": 0.00013722603859993459,
|
| 376 |
+
"loss": 3.389,
|
| 377 |
+
"step": 19200
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 9.62,
|
| 381 |
+
"learning_rate": 0.00013592083742230946,
|
| 382 |
+
"loss": 3.4438,
|
| 383 |
+
"step": 19600
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 9.81,
|
| 387 |
+
"learning_rate": 0.00013461236506378803,
|
| 388 |
+
"loss": 3.5244,
|
| 389 |
+
"step": 20000
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 10.0,
|
| 393 |
+
"eval_f1": 0.004167666888342888,
|
| 394 |
+
"eval_loss": 5.8468732833862305,
|
| 395 |
+
"eval_runtime": 105.8207,
|
| 396 |
+
"eval_samples_per_second": 217.462,
|
| 397 |
+
"eval_steps_per_second": 3.402,
|
| 398 |
+
"step": 20380
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 10.01,
|
| 402 |
+
"learning_rate": 0.0001333038927052666,
|
| 403 |
+
"loss": 3.542,
|
| 404 |
+
"step": 20400
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 10.21,
|
| 408 |
+
"learning_rate": 0.0001319954203467452,
|
| 409 |
+
"loss": 3.023,
|
| 410 |
+
"step": 20800
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 10.4,
|
| 414 |
+
"learning_rate": 0.00013068694798822376,
|
| 415 |
+
"loss": 3.0507,
|
| 416 |
+
"step": 21200
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 10.6,
|
| 420 |
+
"learning_rate": 0.00012938174681059863,
|
| 421 |
+
"loss": 3.132,
|
| 422 |
+
"step": 21600
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 10.79,
|
| 426 |
+
"learning_rate": 0.0001280732744520772,
|
| 427 |
+
"loss": 3.1857,
|
| 428 |
+
"step": 22000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 10.99,
|
| 432 |
+
"learning_rate": 0.0001267648020935558,
|
| 433 |
+
"loss": 3.2321,
|
| 434 |
+
"step": 22400
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 11.0,
|
| 438 |
+
"eval_f1": 0.004533528112732692,
|
| 439 |
+
"eval_loss": 5.87736177444458,
|
| 440 |
+
"eval_runtime": 105.6895,
|
| 441 |
+
"eval_samples_per_second": 217.732,
|
| 442 |
+
"eval_steps_per_second": 3.406,
|
| 443 |
+
"step": 22418
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 11.19,
|
| 447 |
+
"learning_rate": 0.00012545632973503436,
|
| 448 |
+
"loss": 2.705,
|
| 449 |
+
"step": 22800
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 11.38,
|
| 453 |
+
"learning_rate": 0.00012414785737651293,
|
| 454 |
+
"loss": 2.7688,
|
| 455 |
+
"step": 23200
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 11.58,
|
| 459 |
+
"learning_rate": 0.0001228393850179915,
|
| 460 |
+
"loss": 2.8223,
|
| 461 |
+
"step": 23600
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 11.78,
|
| 465 |
+
"learning_rate": 0.00012153745502126268,
|
| 466 |
+
"loss": 2.8647,
|
| 467 |
+
"step": 24000
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 11.97,
|
| 471 |
+
"learning_rate": 0.00012022898266274125,
|
| 472 |
+
"loss": 2.9004,
|
| 473 |
+
"step": 24400
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 12.0,
|
| 477 |
+
"eval_f1": 0.004656155608992903,
|
| 478 |
+
"eval_loss": 6.118570804595947,
|
| 479 |
+
"eval_runtime": 103.8763,
|
| 480 |
+
"eval_samples_per_second": 221.533,
|
| 481 |
+
"eval_steps_per_second": 3.466,
|
| 482 |
+
"step": 24456
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 12.17,
|
| 486 |
+
"learning_rate": 0.00011892051030421983,
|
| 487 |
+
"loss": 2.4632,
|
| 488 |
+
"step": 24800
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 12.37,
|
| 492 |
+
"learning_rate": 0.0001176120379456984,
|
| 493 |
+
"loss": 2.4821,
|
| 494 |
+
"step": 25200
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 12.56,
|
| 498 |
+
"learning_rate": 0.00011630356558717696,
|
| 499 |
+
"loss": 2.5336,
|
| 500 |
+
"step": 25600
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 12.76,
|
| 504 |
+
"learning_rate": 0.00011499509322865556,
|
| 505 |
+
"loss": 2.5646,
|
| 506 |
+
"step": 26000
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 12.95,
|
| 510 |
+
"learning_rate": 0.00011368662087013414,
|
| 511 |
+
"loss": 2.5937,
|
| 512 |
+
"step": 26400
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 13.0,
|
| 516 |
+
"eval_f1": 0.004601475826989957,
|
| 517 |
+
"eval_loss": 6.239825248718262,
|
| 518 |
+
"eval_runtime": 107.5857,
|
| 519 |
+
"eval_samples_per_second": 213.895,
|
| 520 |
+
"eval_steps_per_second": 3.346,
|
| 521 |
+
"step": 26494
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 13.15,
|
| 525 |
+
"learning_rate": 0.0001123781485116127,
|
| 526 |
+
"loss": 2.231,
|
| 527 |
+
"step": 26800
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 13.35,
|
| 531 |
+
"learning_rate": 0.00011106967615309127,
|
| 532 |
+
"loss": 2.149,
|
| 533 |
+
"step": 27200
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 13.54,
|
| 537 |
+
"learning_rate": 0.00010976120379456985,
|
| 538 |
+
"loss": 2.2258,
|
| 539 |
+
"step": 27600
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 13.74,
|
| 543 |
+
"learning_rate": 0.00010845600261694473,
|
| 544 |
+
"loss": 2.2778,
|
| 545 |
+
"step": 28000
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 13.94,
|
| 549 |
+
"learning_rate": 0.00010715080143931959,
|
| 550 |
+
"loss": 2.2983,
|
| 551 |
+
"step": 28400
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 14.0,
|
| 555 |
+
"eval_f1": 0.004861690958274386,
|
| 556 |
+
"eval_loss": 6.3732194900512695,
|
| 557 |
+
"eval_runtime": 106.2123,
|
| 558 |
+
"eval_samples_per_second": 216.66,
|
| 559 |
+
"eval_steps_per_second": 3.389,
|
| 560 |
+
"step": 28532
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 14.13,
|
| 564 |
+
"learning_rate": 0.00010584232908079817,
|
| 565 |
+
"loss": 2.0303,
|
| 566 |
+
"step": 28800
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 14.33,
|
| 570 |
+
"learning_rate": 0.00010453385672227676,
|
| 571 |
+
"loss": 1.9106,
|
| 572 |
+
"step": 29200
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 14.52,
|
| 576 |
+
"learning_rate": 0.00010322538436375533,
|
| 577 |
+
"loss": 1.9854,
|
| 578 |
+
"step": 29600
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 14.72,
|
| 582 |
+
"learning_rate": 0.0001019169120052339,
|
| 583 |
+
"loss": 2.0189,
|
| 584 |
+
"step": 30000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 14.92,
|
| 588 |
+
"learning_rate": 0.00010060843964671246,
|
| 589 |
+
"loss": 2.0611,
|
| 590 |
+
"step": 30400
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 15.0,
|
| 594 |
+
"eval_f1": 0.004546370201449235,
|
| 595 |
+
"eval_loss": 6.502394199371338,
|
| 596 |
+
"eval_runtime": 106.7026,
|
| 597 |
+
"eval_samples_per_second": 215.665,
|
| 598 |
+
"eval_steps_per_second": 3.374,
|
| 599 |
+
"step": 30570
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 15.11,
|
| 603 |
+
"learning_rate": 9.929996728819105e-05,
|
| 604 |
+
"loss": 1.8034,
|
| 605 |
+
"step": 30800
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 15.31,
|
| 609 |
+
"learning_rate": 9.799476611056592e-05,
|
| 610 |
+
"loss": 1.7043,
|
| 611 |
+
"step": 31200
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 15.51,
|
| 615 |
+
"learning_rate": 9.668629375204449e-05,
|
| 616 |
+
"loss": 1.731,
|
| 617 |
+
"step": 31600
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 15.7,
|
| 621 |
+
"learning_rate": 9.537782139352307e-05,
|
| 622 |
+
"loss": 1.7611,
|
| 623 |
+
"step": 32000
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 15.9,
|
| 627 |
+
"learning_rate": 9.406934903500164e-05,
|
| 628 |
+
"loss": 1.8153,
|
| 629 |
+
"step": 32400
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 16.0,
|
| 633 |
+
"eval_f1": 0.004668311047328734,
|
| 634 |
+
"eval_loss": 6.658481121063232,
|
| 635 |
+
"eval_runtime": 106.8749,
|
| 636 |
+
"eval_samples_per_second": 215.317,
|
| 637 |
+
"eval_steps_per_second": 3.368,
|
| 638 |
+
"step": 32608
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 16.09,
|
| 642 |
+
"learning_rate": 9.276087667648022e-05,
|
| 643 |
+
"loss": 1.6449,
|
| 644 |
+
"step": 32800
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 16.29,
|
| 648 |
+
"learning_rate": 9.145567549885509e-05,
|
| 649 |
+
"loss": 1.4815,
|
| 650 |
+
"step": 33200
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 16.49,
|
| 654 |
+
"learning_rate": 9.014720314033367e-05,
|
| 655 |
+
"loss": 1.5311,
|
| 656 |
+
"step": 33600
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 16.68,
|
| 660 |
+
"learning_rate": 8.883873078181224e-05,
|
| 661 |
+
"loss": 1.5722,
|
| 662 |
+
"step": 34000
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 16.88,
|
| 666 |
+
"learning_rate": 8.75302584232908e-05,
|
| 667 |
+
"loss": 1.6075,
|
| 668 |
+
"step": 34400
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 17.0,
|
| 672 |
+
"eval_f1": 0.004309847995944622,
|
| 673 |
+
"eval_loss": 6.833281993865967,
|
| 674 |
+
"eval_runtime": 104.6363,
|
| 675 |
+
"eval_samples_per_second": 219.924,
|
| 676 |
+
"eval_steps_per_second": 3.44,
|
| 677 |
+
"step": 34646
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 17.08,
|
| 681 |
+
"learning_rate": 8.622178606476939e-05,
|
| 682 |
+
"loss": 1.4882,
|
| 683 |
+
"step": 34800
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 17.27,
|
| 687 |
+
"learning_rate": 8.491331370624795e-05,
|
| 688 |
+
"loss": 1.3091,
|
| 689 |
+
"step": 35200
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 17.47,
|
| 693 |
+
"learning_rate": 8.360811252862284e-05,
|
| 694 |
+
"loss": 1.3476,
|
| 695 |
+
"step": 35600
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 17.66,
|
| 699 |
+
"learning_rate": 8.229964017010141e-05,
|
| 700 |
+
"loss": 1.3937,
|
| 701 |
+
"step": 36000
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 17.86,
|
| 705 |
+
"learning_rate": 8.099116781157998e-05,
|
| 706 |
+
"loss": 1.4342,
|
| 707 |
+
"step": 36400
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 18.0,
|
| 711 |
+
"eval_f1": 0.004381425028314271,
|
| 712 |
+
"eval_loss": 6.952894687652588,
|
| 713 |
+
"eval_runtime": 106.0924,
|
| 714 |
+
"eval_samples_per_second": 216.905,
|
| 715 |
+
"eval_steps_per_second": 3.393,
|
| 716 |
+
"step": 36684
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 18.06,
|
| 720 |
+
"learning_rate": 7.968269545305856e-05,
|
| 721 |
+
"loss": 1.3502,
|
| 722 |
+
"step": 36800
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 18.25,
|
| 726 |
+
"learning_rate": 7.837422309453714e-05,
|
| 727 |
+
"loss": 1.1797,
|
| 728 |
+
"step": 37200
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 18.45,
|
| 732 |
+
"learning_rate": 7.70657507360157e-05,
|
| 733 |
+
"loss": 1.234,
|
| 734 |
+
"step": 37600
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 18.65,
|
| 738 |
+
"learning_rate": 7.575727837749427e-05,
|
| 739 |
+
"loss": 1.2631,
|
| 740 |
+
"step": 38000
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 18.84,
|
| 744 |
+
"learning_rate": 7.445207719986915e-05,
|
| 745 |
+
"loss": 1.2614,
|
| 746 |
+
"step": 38400
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 19.0,
|
| 750 |
+
"eval_f1": 0.0045730417031692346,
|
| 751 |
+
"eval_loss": 7.112914085388184,
|
| 752 |
+
"eval_runtime": 106.9736,
|
| 753 |
+
"eval_samples_per_second": 215.118,
|
| 754 |
+
"eval_steps_per_second": 3.365,
|
| 755 |
+
"step": 38722
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 19.04,
|
| 759 |
+
"learning_rate": 7.314360484134773e-05,
|
| 760 |
+
"loss": 1.2353,
|
| 761 |
+
"step": 38800
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 19.23,
|
| 765 |
+
"learning_rate": 7.183513248282631e-05,
|
| 766 |
+
"loss": 1.0694,
|
| 767 |
+
"step": 39200
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 19.43,
|
| 771 |
+
"learning_rate": 7.052666012430488e-05,
|
| 772 |
+
"loss": 1.0988,
|
| 773 |
+
"step": 39600
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 19.63,
|
| 777 |
+
"learning_rate": 6.921818776578344e-05,
|
| 778 |
+
"loss": 1.125,
|
| 779 |
+
"step": 40000
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 19.82,
|
| 783 |
+
"learning_rate": 6.791298658815833e-05,
|
| 784 |
+
"loss": 1.1463,
|
| 785 |
+
"step": 40400
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 20.0,
|
| 789 |
+
"eval_f1": 0.003947589716911441,
|
| 790 |
+
"eval_loss": 7.197678089141846,
|
| 791 |
+
"eval_runtime": 105.623,
|
| 792 |
+
"eval_samples_per_second": 217.869,
|
| 793 |
+
"eval_steps_per_second": 3.408,
|
| 794 |
+
"step": 40760
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 20.02,
|
| 798 |
+
"learning_rate": 6.66045142296369e-05,
|
| 799 |
+
"loss": 1.1429,
|
| 800 |
+
"step": 40800
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 20.22,
|
| 804 |
+
"learning_rate": 6.529604187111548e-05,
|
| 805 |
+
"loss": 0.9563,
|
| 806 |
+
"step": 41200
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 20.41,
|
| 810 |
+
"learning_rate": 6.398756951259405e-05,
|
| 811 |
+
"loss": 1.0085,
|
| 812 |
+
"step": 41600
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 20.61,
|
| 816 |
+
"learning_rate": 6.267909715407262e-05,
|
| 817 |
+
"loss": 1.0361,
|
| 818 |
+
"step": 42000
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 20.8,
|
| 822 |
+
"learning_rate": 6.13738959764475e-05,
|
| 823 |
+
"loss": 1.0387,
|
| 824 |
+
"step": 42400
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 21.0,
|
| 828 |
+
"eval_f1": 0.004382334218618053,
|
| 829 |
+
"eval_loss": 7.270018100738525,
|
| 830 |
+
"eval_runtime": 104.2085,
|
| 831 |
+
"eval_samples_per_second": 220.827,
|
| 832 |
+
"eval_steps_per_second": 3.455,
|
| 833 |
+
"step": 42798
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 21.0,
|
| 837 |
+
"learning_rate": 6.006542361792608e-05,
|
| 838 |
+
"loss": 1.0315,
|
| 839 |
+
"step": 42800
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 21.2,
|
| 843 |
+
"learning_rate": 5.8756951259404646e-05,
|
| 844 |
+
"loss": 0.8991,
|
| 845 |
+
"step": 43200
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 21.39,
|
| 849 |
+
"learning_rate": 5.744847890088322e-05,
|
| 850 |
+
"loss": 0.907,
|
| 851 |
+
"step": 43600
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 21.59,
|
| 855 |
+
"learning_rate": 5.61400065423618e-05,
|
| 856 |
+
"loss": 0.9289,
|
| 857 |
+
"step": 44000
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 21.79,
|
| 861 |
+
"learning_rate": 5.4831534183840375e-05,
|
| 862 |
+
"loss": 0.9542,
|
| 863 |
+
"step": 44400
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 21.98,
|
| 867 |
+
"learning_rate": 5.352306182531894e-05,
|
| 868 |
+
"loss": 0.9635,
|
| 869 |
+
"step": 44800
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 22.0,
|
| 873 |
+
"eval_f1": 0.0039901118782942465,
|
| 874 |
+
"eval_loss": 7.337534427642822,
|
| 875 |
+
"eval_runtime": 105.9291,
|
| 876 |
+
"eval_samples_per_second": 217.24,
|
| 877 |
+
"eval_steps_per_second": 3.399,
|
| 878 |
+
"step": 44836
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 22.18,
|
| 882 |
+
"learning_rate": 5.2214589466797516e-05,
|
| 883 |
+
"loss": 0.8345,
|
| 884 |
+
"step": 45200
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 22.37,
|
| 888 |
+
"learning_rate": 5.090938828917239e-05,
|
| 889 |
+
"loss": 0.8363,
|
| 890 |
+
"step": 45600
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 22.57,
|
| 894 |
+
"learning_rate": 4.9600915930650965e-05,
|
| 895 |
+
"loss": 0.8663,
|
| 896 |
+
"step": 46000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 22.77,
|
| 900 |
+
"learning_rate": 4.829244357212954e-05,
|
| 901 |
+
"loss": 0.8618,
|
| 902 |
+
"step": 46400
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 22.96,
|
| 906 |
+
"learning_rate": 4.698397121360811e-05,
|
| 907 |
+
"loss": 0.8872,
|
| 908 |
+
"step": 46800
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 23.0,
|
| 912 |
+
"eval_f1": 0.003935013281715019,
|
| 913 |
+
"eval_loss": 7.400303363800049,
|
| 914 |
+
"eval_runtime": 115.223,
|
| 915 |
+
"eval_samples_per_second": 199.717,
|
| 916 |
+
"eval_steps_per_second": 3.124,
|
| 917 |
+
"step": 46874
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 23.16,
|
| 921 |
+
"learning_rate": 4.5675498855086694e-05,
|
| 922 |
+
"loss": 0.7942,
|
| 923 |
+
"step": 47200
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 23.36,
|
| 927 |
+
"learning_rate": 4.437029767746156e-05,
|
| 928 |
+
"loss": 0.803,
|
| 929 |
+
"step": 47600
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 23.55,
|
| 933 |
+
"learning_rate": 4.306182531894014e-05,
|
| 934 |
+
"loss": 0.7919,
|
| 935 |
+
"step": 48000
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 23.75,
|
| 939 |
+
"learning_rate": 4.175335296041871e-05,
|
| 940 |
+
"loss": 0.8018,
|
| 941 |
+
"step": 48400
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 23.95,
|
| 945 |
+
"learning_rate": 4.044488060189729e-05,
|
| 946 |
+
"loss": 0.8156,
|
| 947 |
+
"step": 48800
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 24.0,
|
| 951 |
+
"eval_f1": 0.003947052519025861,
|
| 952 |
+
"eval_loss": 7.488423824310303,
|
| 953 |
+
"eval_runtime": 106.3546,
|
| 954 |
+
"eval_samples_per_second": 216.371,
|
| 955 |
+
"eval_steps_per_second": 3.385,
|
| 956 |
+
"step": 48912
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 24.14,
|
| 960 |
+
"learning_rate": 3.913640824337586e-05,
|
| 961 |
+
"loss": 0.7462,
|
| 962 |
+
"step": 49200
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 24.34,
|
| 966 |
+
"learning_rate": 3.782793588485443e-05,
|
| 967 |
+
"loss": 0.7192,
|
| 968 |
+
"step": 49600
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 24.53,
|
| 972 |
+
"learning_rate": 3.6519463526333006e-05,
|
| 973 |
+
"loss": 0.7476,
|
| 974 |
+
"step": 50000
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 24.73,
|
| 978 |
+
"learning_rate": 3.521099116781158e-05,
|
| 979 |
+
"loss": 0.7602,
|
| 980 |
+
"step": 50400
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 24.93,
|
| 984 |
+
"learning_rate": 3.3902518809290154e-05,
|
| 985 |
+
"loss": 0.7544,
|
| 986 |
+
"step": 50800
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 25.0,
|
| 990 |
+
"eval_f1": 0.003877386201510839,
|
| 991 |
+
"eval_loss": 7.476434707641602,
|
| 992 |
+
"eval_runtime": 105.8268,
|
| 993 |
+
"eval_samples_per_second": 217.45,
|
| 994 |
+
"eval_steps_per_second": 3.402,
|
| 995 |
+
"step": 50950
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 25.12,
|
| 999 |
+
"learning_rate": 3.259404645076873e-05,
|
| 1000 |
+
"loss": 0.7114,
|
| 1001 |
+
"step": 51200
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 25.32,
|
| 1005 |
+
"learning_rate": 3.12855740922473e-05,
|
| 1006 |
+
"loss": 0.683,
|
| 1007 |
+
"step": 51600
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 25.52,
|
| 1011 |
+
"learning_rate": 2.998037291462218e-05,
|
| 1012 |
+
"loss": 0.6972,
|
| 1013 |
+
"step": 52000
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 25.71,
|
| 1017 |
+
"learning_rate": 2.8671900556100754e-05,
|
| 1018 |
+
"loss": 0.7087,
|
| 1019 |
+
"step": 52400
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 25.91,
|
| 1023 |
+
"learning_rate": 2.736669937847563e-05,
|
| 1024 |
+
"loss": 0.6893,
|
| 1025 |
+
"step": 52800
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 26.0,
|
| 1029 |
+
"eval_f1": 0.004237152566679073,
|
| 1030 |
+
"eval_loss": 7.515334606170654,
|
| 1031 |
+
"eval_runtime": 104.8855,
|
| 1032 |
+
"eval_samples_per_second": 219.401,
|
| 1033 |
+
"eval_steps_per_second": 3.432,
|
| 1034 |
+
"step": 52988
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 26.1,
|
| 1038 |
+
"learning_rate": 2.6058227019954207e-05,
|
| 1039 |
+
"loss": 0.6589,
|
| 1040 |
+
"step": 53200
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 26.3,
|
| 1044 |
+
"learning_rate": 2.4749754661432777e-05,
|
| 1045 |
+
"loss": 0.6609,
|
| 1046 |
+
"step": 53600
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 26.5,
|
| 1050 |
+
"learning_rate": 2.344128230291135e-05,
|
| 1051 |
+
"loss": 0.6485,
|
| 1052 |
+
"step": 54000
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 26.69,
|
| 1056 |
+
"learning_rate": 2.2132809944389925e-05,
|
| 1057 |
+
"loss": 0.667,
|
| 1058 |
+
"step": 54400
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 26.89,
|
| 1062 |
+
"learning_rate": 2.08243375858685e-05,
|
| 1063 |
+
"loss": 0.6767,
|
| 1064 |
+
"step": 54800
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 27.0,
|
| 1068 |
+
"eval_f1": 0.004296555657055152,
|
| 1069 |
+
"eval_loss": 7.542654514312744,
|
| 1070 |
+
"eval_runtime": 104.6361,
|
| 1071 |
+
"eval_samples_per_second": 219.924,
|
| 1072 |
+
"eval_steps_per_second": 3.44,
|
| 1073 |
+
"step": 55026
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 27.09,
|
| 1077 |
+
"learning_rate": 1.9515865227347073e-05,
|
| 1078 |
+
"loss": 0.639,
|
| 1079 |
+
"step": 55200
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 27.28,
|
| 1083 |
+
"learning_rate": 1.8207392868825644e-05,
|
| 1084 |
+
"loss": 0.6216,
|
| 1085 |
+
"step": 55600
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 27.48,
|
| 1089 |
+
"learning_rate": 1.689892051030422e-05,
|
| 1090 |
+
"loss": 0.627,
|
| 1091 |
+
"step": 56000
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 27.67,
|
| 1095 |
+
"learning_rate": 1.5590448151782795e-05,
|
| 1096 |
+
"loss": 0.613,
|
| 1097 |
+
"step": 56400
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 27.87,
|
| 1101 |
+
"learning_rate": 1.428524697415767e-05,
|
| 1102 |
+
"loss": 0.6098,
|
| 1103 |
+
"step": 56800
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 28.0,
|
| 1107 |
+
"eval_f1": 0.004164219476454643,
|
| 1108 |
+
"eval_loss": 7.554748058319092,
|
| 1109 |
+
"eval_runtime": 103.9745,
|
| 1110 |
+
"eval_samples_per_second": 221.324,
|
| 1111 |
+
"eval_steps_per_second": 3.462,
|
| 1112 |
+
"step": 57064
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 28.07,
|
| 1116 |
+
"learning_rate": 1.2976774615636244e-05,
|
| 1117 |
+
"loss": 0.6228,
|
| 1118 |
+
"step": 57200
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 28.26,
|
| 1122 |
+
"learning_rate": 1.1668302257114818e-05,
|
| 1123 |
+
"loss": 0.5921,
|
| 1124 |
+
"step": 57600
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 28.46,
|
| 1128 |
+
"learning_rate": 1.0359829898593392e-05,
|
| 1129 |
+
"loss": 0.6023,
|
| 1130 |
+
"step": 58000
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 28.66,
|
| 1134 |
+
"learning_rate": 9.051357540071966e-06,
|
| 1135 |
+
"loss": 0.5876,
|
| 1136 |
+
"step": 58400
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 28.85,
|
| 1140 |
+
"learning_rate": 7.746156362446843e-06,
|
| 1141 |
+
"loss": 0.5871,
|
| 1142 |
+
"step": 58800
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 29.0,
|
| 1146 |
+
"eval_f1": 0.00413622990961682,
|
| 1147 |
+
"eval_loss": 7.553284645080566,
|
| 1148 |
+
"eval_runtime": 104.0213,
|
| 1149 |
+
"eval_samples_per_second": 221.224,
|
| 1150 |
+
"eval_steps_per_second": 3.461,
|
| 1151 |
+
"step": 59102
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 29.05,
|
| 1155 |
+
"learning_rate": 6.437684003925418e-06,
|
| 1156 |
+
"loss": 0.5787,
|
| 1157 |
+
"step": 59200
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 29.24,
|
| 1161 |
+
"learning_rate": 5.129211645403991e-06,
|
| 1162 |
+
"loss": 0.5671,
|
| 1163 |
+
"step": 59600
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 29.44,
|
| 1167 |
+
"learning_rate": 3.820739286882564e-06,
|
| 1168 |
+
"loss": 0.5596,
|
| 1169 |
+
"step": 60000
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 29.64,
|
| 1173 |
+
"learning_rate": 2.5122669283611384e-06,
|
| 1174 |
+
"loss": 0.5595,
|
| 1175 |
+
"step": 60400
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 29.83,
|
| 1179 |
+
"learning_rate": 1.2070657507360158e-06,
|
| 1180 |
+
"loss": 0.5696,
|
| 1181 |
+
"step": 60800
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 30.0,
|
| 1185 |
+
"eval_f1": 0.004123942255601062,
|
| 1186 |
+
"eval_loss": 7.559465408325195,
|
| 1187 |
+
"eval_runtime": 98.4712,
|
| 1188 |
+
"eval_samples_per_second": 233.693,
|
| 1189 |
+
"eval_steps_per_second": 3.656,
|
| 1190 |
+
"step": 61140
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 30.0,
|
| 1194 |
+
"step": 61140,
|
| 1195 |
+
"total_flos": 3.687496208074506e+20,
|
| 1196 |
+
"train_loss": 2.558345020757144,
|
| 1197 |
+
"train_runtime": 41575.9952,
|
| 1198 |
+
"train_samples_per_second": 94.093,
|
| 1199 |
+
"train_steps_per_second": 1.471
|
| 1200 |
+
}
|
| 1201 |
+
],
|
| 1202 |
+
"max_steps": 61140,
|
| 1203 |
+
"num_train_epochs": 30,
|
| 1204 |
+
"total_flos": 3.687496208074506e+20,
|
| 1205 |
+
"trial_name": null,
|
| 1206 |
+
"trial_params": null
|
| 1207 |
+
}
|