Commit
•
94487ae
1
Parent(s):
aed4f77
ThyroidTumorClassification
Browse files- all_results.json +13 -0
- eval_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +1447 -0
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 149.89,
|
| 3 |
+
"eval_accuracy": 0.9142857142857143,
|
| 4 |
+
"eval_loss": 0.5618674755096436,
|
| 5 |
+
"eval_runtime": 0.4031,
|
| 6 |
+
"eval_samples_per_second": 173.666,
|
| 7 |
+
"eval_steps_per_second": 7.443,
|
| 8 |
+
"total_flos": 1.0435256966870508e+18,
|
| 9 |
+
"train_loss": 0.39242326736450195,
|
| 10 |
+
"train_runtime": 677.2441,
|
| 11 |
+
"train_samples_per_second": 61.352,
|
| 12 |
+
"train_steps_per_second": 0.443
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 149.89,
|
| 3 |
+
"eval_accuracy": 0.9142857142857143,
|
| 4 |
+
"eval_loss": 0.5618674755096436,
|
| 5 |
+
"eval_runtime": 0.4031,
|
| 6 |
+
"eval_samples_per_second": 173.666,
|
| 7 |
+
"eval_steps_per_second": 7.443
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 149.89,
|
| 3 |
+
"total_flos": 1.0435256966870508e+18,
|
| 4 |
+
"train_loss": 0.39242326736450195,
|
| 5 |
+
"train_runtime": 677.2441,
|
| 6 |
+
"train_samples_per_second": 61.352,
|
| 7 |
+
"train_steps_per_second": 0.443
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1447 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9142857142857143,
|
| 3 |
+
"best_model_checkpoint": "Cvt-finetuned-thyroid/checkpoint-18",
|
| 4 |
+
"epoch": 149.88888888888889,
|
| 5 |
+
"global_step": 300,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.89,
|
| 12 |
+
"eval_accuracy": 0.6857142857142857,
|
| 13 |
+
"eval_loss": 0.6600533127784729,
|
| 14 |
+
"eval_runtime": 0.4026,
|
| 15 |
+
"eval_samples_per_second": 173.882,
|
| 16 |
+
"eval_steps_per_second": 7.452,
|
| 17 |
+
"step": 2
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 1.89,
|
| 21 |
+
"eval_accuracy": 0.7,
|
| 22 |
+
"eval_loss": 0.6563708186149597,
|
| 23 |
+
"eval_runtime": 0.3922,
|
| 24 |
+
"eval_samples_per_second": 178.495,
|
| 25 |
+
"eval_steps_per_second": 7.65,
|
| 26 |
+
"step": 4
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 2.89,
|
| 30 |
+
"eval_accuracy": 0.7428571428571429,
|
| 31 |
+
"eval_loss": 0.6498632431030273,
|
| 32 |
+
"eval_runtime": 0.3916,
|
| 33 |
+
"eval_samples_per_second": 178.774,
|
| 34 |
+
"eval_steps_per_second": 7.662,
|
| 35 |
+
"step": 6
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"epoch": 3.89,
|
| 39 |
+
"eval_accuracy": 0.7714285714285715,
|
| 40 |
+
"eval_loss": 0.6404139399528503,
|
| 41 |
+
"eval_runtime": 0.3913,
|
| 42 |
+
"eval_samples_per_second": 178.878,
|
| 43 |
+
"eval_steps_per_second": 7.666,
|
| 44 |
+
"step": 8
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 4.89,
|
| 48 |
+
"eval_accuracy": 0.8,
|
| 49 |
+
"eval_loss": 0.62859708070755,
|
| 50 |
+
"eval_runtime": 0.3923,
|
| 51 |
+
"eval_samples_per_second": 178.448,
|
| 52 |
+
"eval_steps_per_second": 7.648,
|
| 53 |
+
"step": 10
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 5.89,
|
| 57 |
+
"eval_accuracy": 0.8142857142857143,
|
| 58 |
+
"eval_loss": 0.6146815419197083,
|
| 59 |
+
"eval_runtime": 0.3861,
|
| 60 |
+
"eval_samples_per_second": 181.308,
|
| 61 |
+
"eval_steps_per_second": 7.77,
|
| 62 |
+
"step": 12
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 6.89,
|
| 66 |
+
"eval_accuracy": 0.8571428571428571,
|
| 67 |
+
"eval_loss": 0.5986873507499695,
|
| 68 |
+
"eval_runtime": 0.3853,
|
| 69 |
+
"eval_samples_per_second": 181.687,
|
| 70 |
+
"eval_steps_per_second": 7.787,
|
| 71 |
+
"step": 14
|
| 72 |
+
},
|
| 73 |
+
{
|
| 74 |
+
"epoch": 7.89,
|
| 75 |
+
"eval_accuracy": 0.9,
|
| 76 |
+
"eval_loss": 0.5808569192886353,
|
| 77 |
+
"eval_runtime": 0.391,
|
| 78 |
+
"eval_samples_per_second": 179.013,
|
| 79 |
+
"eval_steps_per_second": 7.672,
|
| 80 |
+
"step": 16
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 8.89,
|
| 84 |
+
"eval_accuracy": 0.9142857142857143,
|
| 85 |
+
"eval_loss": 0.5618674755096436,
|
| 86 |
+
"eval_runtime": 0.3861,
|
| 87 |
+
"eval_samples_per_second": 181.318,
|
| 88 |
+
"eval_steps_per_second": 7.771,
|
| 89 |
+
"step": 18
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 9.89,
|
| 93 |
+
"eval_accuracy": 0.9142857142857143,
|
| 94 |
+
"eval_loss": 0.5425492525100708,
|
| 95 |
+
"eval_runtime": 0.3895,
|
| 96 |
+
"eval_samples_per_second": 179.718,
|
| 97 |
+
"eval_steps_per_second": 7.702,
|
| 98 |
+
"step": 20
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 10.89,
|
| 102 |
+
"eval_accuracy": 0.9142857142857143,
|
| 103 |
+
"eval_loss": 0.5230604410171509,
|
| 104 |
+
"eval_runtime": 0.3908,
|
| 105 |
+
"eval_samples_per_second": 179.109,
|
| 106 |
+
"eval_steps_per_second": 7.676,
|
| 107 |
+
"step": 22
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 11.89,
|
| 111 |
+
"eval_accuracy": 0.9142857142857143,
|
| 112 |
+
"eval_loss": 0.5035112500190735,
|
| 113 |
+
"eval_runtime": 0.3841,
|
| 114 |
+
"eval_samples_per_second": 182.262,
|
| 115 |
+
"eval_steps_per_second": 7.811,
|
| 116 |
+
"step": 24
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 12.44,
|
| 120 |
+
"learning_rate": 8.333333333333334e-06,
|
| 121 |
+
"loss": 0.6786,
|
| 122 |
+
"step": 25
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 12.89,
|
| 126 |
+
"eval_accuracy": 0.9142857142857143,
|
| 127 |
+
"eval_loss": 0.4828525483608246,
|
| 128 |
+
"eval_runtime": 0.3958,
|
| 129 |
+
"eval_samples_per_second": 176.835,
|
| 130 |
+
"eval_steps_per_second": 7.579,
|
| 131 |
+
"step": 26
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 13.89,
|
| 135 |
+
"eval_accuracy": 0.9142857142857143,
|
| 136 |
+
"eval_loss": 0.4621497690677643,
|
| 137 |
+
"eval_runtime": 0.3893,
|
| 138 |
+
"eval_samples_per_second": 179.815,
|
| 139 |
+
"eval_steps_per_second": 7.706,
|
| 140 |
+
"step": 28
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 14.89,
|
| 144 |
+
"eval_accuracy": 0.9142857142857143,
|
| 145 |
+
"eval_loss": 0.44225531816482544,
|
| 146 |
+
"eval_runtime": 0.3936,
|
| 147 |
+
"eval_samples_per_second": 177.866,
|
| 148 |
+
"eval_steps_per_second": 7.623,
|
| 149 |
+
"step": 30
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 15.89,
|
| 153 |
+
"eval_accuracy": 0.9142857142857143,
|
| 154 |
+
"eval_loss": 0.4237878918647766,
|
| 155 |
+
"eval_runtime": 0.3885,
|
| 156 |
+
"eval_samples_per_second": 180.201,
|
| 157 |
+
"eval_steps_per_second": 7.723,
|
| 158 |
+
"step": 32
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 16.89,
|
| 162 |
+
"eval_accuracy": 0.9142857142857143,
|
| 163 |
+
"eval_loss": 0.4087783396244049,
|
| 164 |
+
"eval_runtime": 0.3869,
|
| 165 |
+
"eval_samples_per_second": 180.93,
|
| 166 |
+
"eval_steps_per_second": 7.754,
|
| 167 |
+
"step": 34
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 17.89,
|
| 171 |
+
"eval_accuracy": 0.9142857142857143,
|
| 172 |
+
"eval_loss": 0.39638686180114746,
|
| 173 |
+
"eval_runtime": 0.3919,
|
| 174 |
+
"eval_samples_per_second": 178.599,
|
| 175 |
+
"eval_steps_per_second": 7.654,
|
| 176 |
+
"step": 36
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 18.89,
|
| 180 |
+
"eval_accuracy": 0.9142857142857143,
|
| 181 |
+
"eval_loss": 0.3871336281299591,
|
| 182 |
+
"eval_runtime": 0.3895,
|
| 183 |
+
"eval_samples_per_second": 179.725,
|
| 184 |
+
"eval_steps_per_second": 7.703,
|
| 185 |
+
"step": 38
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 19.89,
|
| 189 |
+
"eval_accuracy": 0.9142857142857143,
|
| 190 |
+
"eval_loss": 0.37913817167282104,
|
| 191 |
+
"eval_runtime": 0.3975,
|
| 192 |
+
"eval_samples_per_second": 176.113,
|
| 193 |
+
"eval_steps_per_second": 7.548,
|
| 194 |
+
"step": 40
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 20.89,
|
| 198 |
+
"eval_accuracy": 0.9142857142857143,
|
| 199 |
+
"eval_loss": 0.37276408076286316,
|
| 200 |
+
"eval_runtime": 0.3884,
|
| 201 |
+
"eval_samples_per_second": 180.205,
|
| 202 |
+
"eval_steps_per_second": 7.723,
|
| 203 |
+
"step": 42
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 21.89,
|
| 207 |
+
"eval_accuracy": 0.9142857142857143,
|
| 208 |
+
"eval_loss": 0.36989572644233704,
|
| 209 |
+
"eval_runtime": 0.3857,
|
| 210 |
+
"eval_samples_per_second": 181.502,
|
| 211 |
+
"eval_steps_per_second": 7.779,
|
| 212 |
+
"step": 44
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 22.89,
|
| 216 |
+
"eval_accuracy": 0.9142857142857143,
|
| 217 |
+
"eval_loss": 0.36899128556251526,
|
| 218 |
+
"eval_runtime": 0.3889,
|
| 219 |
+
"eval_samples_per_second": 180.008,
|
| 220 |
+
"eval_steps_per_second": 7.715,
|
| 221 |
+
"step": 46
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 23.89,
|
| 225 |
+
"eval_accuracy": 0.9142857142857143,
|
| 226 |
+
"eval_loss": 0.36841970682144165,
|
| 227 |
+
"eval_runtime": 0.3867,
|
| 228 |
+
"eval_samples_per_second": 181.0,
|
| 229 |
+
"eval_steps_per_second": 7.757,
|
| 230 |
+
"step": 48
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 24.89,
|
| 234 |
+
"learning_rate": 9.25925925925926e-06,
|
| 235 |
+
"loss": 0.544,
|
| 236 |
+
"step": 50
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 24.89,
|
| 240 |
+
"eval_accuracy": 0.9142857142857143,
|
| 241 |
+
"eval_loss": 0.36657989025115967,
|
| 242 |
+
"eval_runtime": 0.4003,
|
| 243 |
+
"eval_samples_per_second": 174.876,
|
| 244 |
+
"eval_steps_per_second": 7.495,
|
| 245 |
+
"step": 50
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 25.89,
|
| 249 |
+
"eval_accuracy": 0.9142857142857143,
|
| 250 |
+
"eval_loss": 0.36433330178260803,
|
| 251 |
+
"eval_runtime": 0.386,
|
| 252 |
+
"eval_samples_per_second": 181.342,
|
| 253 |
+
"eval_steps_per_second": 7.772,
|
| 254 |
+
"step": 52
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 26.89,
|
| 258 |
+
"eval_accuracy": 0.9142857142857143,
|
| 259 |
+
"eval_loss": 0.3611242175102234,
|
| 260 |
+
"eval_runtime": 0.3881,
|
| 261 |
+
"eval_samples_per_second": 180.356,
|
| 262 |
+
"eval_steps_per_second": 7.73,
|
| 263 |
+
"step": 54
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 27.89,
|
| 267 |
+
"eval_accuracy": 0.9142857142857143,
|
| 268 |
+
"eval_loss": 0.356433242559433,
|
| 269 |
+
"eval_runtime": 0.3885,
|
| 270 |
+
"eval_samples_per_second": 180.192,
|
| 271 |
+
"eval_steps_per_second": 7.723,
|
| 272 |
+
"step": 56
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 28.89,
|
| 276 |
+
"eval_accuracy": 0.9142857142857143,
|
| 277 |
+
"eval_loss": 0.3520072400569916,
|
| 278 |
+
"eval_runtime": 0.3869,
|
| 279 |
+
"eval_samples_per_second": 180.937,
|
| 280 |
+
"eval_steps_per_second": 7.754,
|
| 281 |
+
"step": 58
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 29.89,
|
| 285 |
+
"eval_accuracy": 0.9142857142857143,
|
| 286 |
+
"eval_loss": 0.3477487862110138,
|
| 287 |
+
"eval_runtime": 0.39,
|
| 288 |
+
"eval_samples_per_second": 179.492,
|
| 289 |
+
"eval_steps_per_second": 7.693,
|
| 290 |
+
"step": 60
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 30.89,
|
| 294 |
+
"eval_accuracy": 0.9142857142857143,
|
| 295 |
+
"eval_loss": 0.3445895314216614,
|
| 296 |
+
"eval_runtime": 0.4111,
|
| 297 |
+
"eval_samples_per_second": 170.276,
|
| 298 |
+
"eval_steps_per_second": 7.298,
|
| 299 |
+
"step": 62
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 31.89,
|
| 303 |
+
"eval_accuracy": 0.9142857142857143,
|
| 304 |
+
"eval_loss": 0.3390456736087799,
|
| 305 |
+
"eval_runtime": 0.3852,
|
| 306 |
+
"eval_samples_per_second": 181.703,
|
| 307 |
+
"eval_steps_per_second": 7.787,
|
| 308 |
+
"step": 64
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 32.89,
|
| 312 |
+
"eval_accuracy": 0.9142857142857143,
|
| 313 |
+
"eval_loss": 0.3347129821777344,
|
| 314 |
+
"eval_runtime": 0.4506,
|
| 315 |
+
"eval_samples_per_second": 155.353,
|
| 316 |
+
"eval_steps_per_second": 6.658,
|
| 317 |
+
"step": 66
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 33.89,
|
| 321 |
+
"eval_accuracy": 0.9142857142857143,
|
| 322 |
+
"eval_loss": 0.33145877718925476,
|
| 323 |
+
"eval_runtime": 0.394,
|
| 324 |
+
"eval_samples_per_second": 177.648,
|
| 325 |
+
"eval_steps_per_second": 7.613,
|
| 326 |
+
"step": 68
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 34.89,
|
| 330 |
+
"eval_accuracy": 0.9142857142857143,
|
| 331 |
+
"eval_loss": 0.3286542296409607,
|
| 332 |
+
"eval_runtime": 0.4325,
|
| 333 |
+
"eval_samples_per_second": 161.841,
|
| 334 |
+
"eval_steps_per_second": 6.936,
|
| 335 |
+
"step": 70
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 35.89,
|
| 339 |
+
"eval_accuracy": 0.9142857142857143,
|
| 340 |
+
"eval_loss": 0.3254566192626953,
|
| 341 |
+
"eval_runtime": 0.3901,
|
| 342 |
+
"eval_samples_per_second": 179.436,
|
| 343 |
+
"eval_steps_per_second": 7.69,
|
| 344 |
+
"step": 72
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 36.89,
|
| 348 |
+
"eval_accuracy": 0.9142857142857143,
|
| 349 |
+
"eval_loss": 0.3226519525051117,
|
| 350 |
+
"eval_runtime": 0.3957,
|
| 351 |
+
"eval_samples_per_second": 176.889,
|
| 352 |
+
"eval_steps_per_second": 7.581,
|
| 353 |
+
"step": 74
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 37.44,
|
| 357 |
+
"learning_rate": 8.333333333333334e-06,
|
| 358 |
+
"loss": 0.4904,
|
| 359 |
+
"step": 75
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 37.89,
|
| 363 |
+
"eval_accuracy": 0.9142857142857143,
|
| 364 |
+
"eval_loss": 0.32086578011512756,
|
| 365 |
+
"eval_runtime": 0.3917,
|
| 366 |
+
"eval_samples_per_second": 178.712,
|
| 367 |
+
"eval_steps_per_second": 7.659,
|
| 368 |
+
"step": 76
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 38.89,
|
| 372 |
+
"eval_accuracy": 0.9142857142857143,
|
| 373 |
+
"eval_loss": 0.32166367769241333,
|
| 374 |
+
"eval_runtime": 0.3879,
|
| 375 |
+
"eval_samples_per_second": 180.476,
|
| 376 |
+
"eval_steps_per_second": 7.735,
|
| 377 |
+
"step": 78
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 39.89,
|
| 381 |
+
"eval_accuracy": 0.9142857142857143,
|
| 382 |
+
"eval_loss": 0.3228186070919037,
|
| 383 |
+
"eval_runtime": 0.3945,
|
| 384 |
+
"eval_samples_per_second": 177.444,
|
| 385 |
+
"eval_steps_per_second": 7.605,
|
| 386 |
+
"step": 80
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 40.89,
|
| 390 |
+
"eval_accuracy": 0.9142857142857143,
|
| 391 |
+
"eval_loss": 0.32248714566230774,
|
| 392 |
+
"eval_runtime": 0.3925,
|
| 393 |
+
"eval_samples_per_second": 178.361,
|
| 394 |
+
"eval_steps_per_second": 7.644,
|
| 395 |
+
"step": 82
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 41.89,
|
| 399 |
+
"eval_accuracy": 0.9142857142857143,
|
| 400 |
+
"eval_loss": 0.32159554958343506,
|
| 401 |
+
"eval_runtime": 0.3914,
|
| 402 |
+
"eval_samples_per_second": 178.848,
|
| 403 |
+
"eval_steps_per_second": 7.665,
|
| 404 |
+
"step": 84
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 42.89,
|
| 408 |
+
"eval_accuracy": 0.9142857142857143,
|
| 409 |
+
"eval_loss": 0.31996870040893555,
|
| 410 |
+
"eval_runtime": 0.3902,
|
| 411 |
+
"eval_samples_per_second": 179.374,
|
| 412 |
+
"eval_steps_per_second": 7.687,
|
| 413 |
+
"step": 86
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 43.89,
|
| 417 |
+
"eval_accuracy": 0.9142857142857143,
|
| 418 |
+
"eval_loss": 0.31838318705558777,
|
| 419 |
+
"eval_runtime": 0.3951,
|
| 420 |
+
"eval_samples_per_second": 177.183,
|
| 421 |
+
"eval_steps_per_second": 7.594,
|
| 422 |
+
"step": 88
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 44.89,
|
| 426 |
+
"eval_accuracy": 0.9142857142857143,
|
| 427 |
+
"eval_loss": 0.3160267174243927,
|
| 428 |
+
"eval_runtime": 0.3892,
|
| 429 |
+
"eval_samples_per_second": 179.849,
|
| 430 |
+
"eval_steps_per_second": 7.708,
|
| 431 |
+
"step": 90
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 45.89,
|
| 435 |
+
"eval_accuracy": 0.9142857142857143,
|
| 436 |
+
"eval_loss": 0.31339019536972046,
|
| 437 |
+
"eval_runtime": 0.3914,
|
| 438 |
+
"eval_samples_per_second": 178.826,
|
| 439 |
+
"eval_steps_per_second": 7.664,
|
| 440 |
+
"step": 92
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 46.89,
|
| 444 |
+
"eval_accuracy": 0.9142857142857143,
|
| 445 |
+
"eval_loss": 0.3097696900367737,
|
| 446 |
+
"eval_runtime": 0.3909,
|
| 447 |
+
"eval_samples_per_second": 179.096,
|
| 448 |
+
"eval_steps_per_second": 7.676,
|
| 449 |
+
"step": 94
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 47.89,
|
| 453 |
+
"eval_accuracy": 0.9,
|
| 454 |
+
"eval_loss": 0.3083226680755615,
|
| 455 |
+
"eval_runtime": 0.3975,
|
| 456 |
+
"eval_samples_per_second": 176.094,
|
| 457 |
+
"eval_steps_per_second": 7.547,
|
| 458 |
+
"step": 96
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 48.89,
|
| 462 |
+
"eval_accuracy": 0.9,
|
| 463 |
+
"eval_loss": 0.306194543838501,
|
| 464 |
+
"eval_runtime": 0.39,
|
| 465 |
+
"eval_samples_per_second": 179.464,
|
| 466 |
+
"eval_steps_per_second": 7.691,
|
| 467 |
+
"step": 98
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 49.89,
|
| 471 |
+
"learning_rate": 7.4074074074074075e-06,
|
| 472 |
+
"loss": 0.4468,
|
| 473 |
+
"step": 100
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 49.89,
|
| 477 |
+
"eval_accuracy": 0.9,
|
| 478 |
+
"eval_loss": 0.3034059703350067,
|
| 479 |
+
"eval_runtime": 0.3988,
|
| 480 |
+
"eval_samples_per_second": 175.515,
|
| 481 |
+
"eval_steps_per_second": 7.522,
|
| 482 |
+
"step": 100
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 50.89,
|
| 486 |
+
"eval_accuracy": 0.9,
|
| 487 |
+
"eval_loss": 0.3030487298965454,
|
| 488 |
+
"eval_runtime": 0.3912,
|
| 489 |
+
"eval_samples_per_second": 178.919,
|
| 490 |
+
"eval_steps_per_second": 7.668,
|
| 491 |
+
"step": 102
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 51.89,
|
| 495 |
+
"eval_accuracy": 0.9,
|
| 496 |
+
"eval_loss": 0.3037838339805603,
|
| 497 |
+
"eval_runtime": 0.4036,
|
| 498 |
+
"eval_samples_per_second": 173.422,
|
| 499 |
+
"eval_steps_per_second": 7.432,
|
| 500 |
+
"step": 104
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 52.89,
|
| 504 |
+
"eval_accuracy": 0.9,
|
| 505 |
+
"eval_loss": 0.30441269278526306,
|
| 506 |
+
"eval_runtime": 0.3905,
|
| 507 |
+
"eval_samples_per_second": 179.272,
|
| 508 |
+
"eval_steps_per_second": 7.683,
|
| 509 |
+
"step": 106
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 53.89,
|
| 513 |
+
"eval_accuracy": 0.9,
|
| 514 |
+
"eval_loss": 0.30462566018104553,
|
| 515 |
+
"eval_runtime": 0.4004,
|
| 516 |
+
"eval_samples_per_second": 174.832,
|
| 517 |
+
"eval_steps_per_second": 7.493,
|
| 518 |
+
"step": 108
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 54.89,
|
| 522 |
+
"eval_accuracy": 0.9,
|
| 523 |
+
"eval_loss": 0.30405017733573914,
|
| 524 |
+
"eval_runtime": 0.3926,
|
| 525 |
+
"eval_samples_per_second": 178.307,
|
| 526 |
+
"eval_steps_per_second": 7.642,
|
| 527 |
+
"step": 110
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 55.89,
|
| 531 |
+
"eval_accuracy": 0.9,
|
| 532 |
+
"eval_loss": 0.3027400076389313,
|
| 533 |
+
"eval_runtime": 0.3921,
|
| 534 |
+
"eval_samples_per_second": 178.52,
|
| 535 |
+
"eval_steps_per_second": 7.651,
|
| 536 |
+
"step": 112
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 56.89,
|
| 540 |
+
"eval_accuracy": 0.9,
|
| 541 |
+
"eval_loss": 0.3015061914920807,
|
| 542 |
+
"eval_runtime": 0.3917,
|
| 543 |
+
"eval_samples_per_second": 178.717,
|
| 544 |
+
"eval_steps_per_second": 7.659,
|
| 545 |
+
"step": 114
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 57.89,
|
| 549 |
+
"eval_accuracy": 0.9,
|
| 550 |
+
"eval_loss": 0.298301637172699,
|
| 551 |
+
"eval_runtime": 0.3901,
|
| 552 |
+
"eval_samples_per_second": 179.444,
|
| 553 |
+
"eval_steps_per_second": 7.69,
|
| 554 |
+
"step": 116
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 58.89,
|
| 558 |
+
"eval_accuracy": 0.9,
|
| 559 |
+
"eval_loss": 0.2965472638607025,
|
| 560 |
+
"eval_runtime": 0.3941,
|
| 561 |
+
"eval_samples_per_second": 177.631,
|
| 562 |
+
"eval_steps_per_second": 7.613,
|
| 563 |
+
"step": 118
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 59.89,
|
| 567 |
+
"eval_accuracy": 0.9,
|
| 568 |
+
"eval_loss": 0.2955659329891205,
|
| 569 |
+
"eval_runtime": 0.3929,
|
| 570 |
+
"eval_samples_per_second": 178.161,
|
| 571 |
+
"eval_steps_per_second": 7.635,
|
| 572 |
+
"step": 120
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 60.89,
|
| 576 |
+
"eval_accuracy": 0.9,
|
| 577 |
+
"eval_loss": 0.2953043282032013,
|
| 578 |
+
"eval_runtime": 0.3961,
|
| 579 |
+
"eval_samples_per_second": 176.718,
|
| 580 |
+
"eval_steps_per_second": 7.574,
|
| 581 |
+
"step": 122
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 61.89,
|
| 585 |
+
"eval_accuracy": 0.9,
|
| 586 |
+
"eval_loss": 0.2960428297519684,
|
| 587 |
+
"eval_runtime": 0.3894,
|
| 588 |
+
"eval_samples_per_second": 179.751,
|
| 589 |
+
"eval_steps_per_second": 7.704,
|
| 590 |
+
"step": 124
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 62.44,
|
| 594 |
+
"learning_rate": 6.481481481481482e-06,
|
| 595 |
+
"loss": 0.411,
|
| 596 |
+
"step": 125
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 62.89,
|
| 600 |
+
"eval_accuracy": 0.9,
|
| 601 |
+
"eval_loss": 0.29606738686561584,
|
| 602 |
+
"eval_runtime": 0.3907,
|
| 603 |
+
"eval_samples_per_second": 179.177,
|
| 604 |
+
"eval_steps_per_second": 7.679,
|
| 605 |
+
"step": 126
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 63.89,
|
| 609 |
+
"eval_accuracy": 0.9,
|
| 610 |
+
"eval_loss": 0.29615068435668945,
|
| 611 |
+
"eval_runtime": 0.3957,
|
| 612 |
+
"eval_samples_per_second": 176.884,
|
| 613 |
+
"eval_steps_per_second": 7.581,
|
| 614 |
+
"step": 128
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 64.89,
|
| 618 |
+
"eval_accuracy": 0.9,
|
| 619 |
+
"eval_loss": 0.29546239972114563,
|
| 620 |
+
"eval_runtime": 0.4028,
|
| 621 |
+
"eval_samples_per_second": 173.791,
|
| 622 |
+
"eval_steps_per_second": 7.448,
|
| 623 |
+
"step": 130
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 65.89,
|
| 627 |
+
"eval_accuracy": 0.9,
|
| 628 |
+
"eval_loss": 0.29375123977661133,
|
| 629 |
+
"eval_runtime": 0.3984,
|
| 630 |
+
"eval_samples_per_second": 175.721,
|
| 631 |
+
"eval_steps_per_second": 7.531,
|
| 632 |
+
"step": 132
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 66.89,
|
| 636 |
+
"eval_accuracy": 0.9,
|
| 637 |
+
"eval_loss": 0.293165922164917,
|
| 638 |
+
"eval_runtime": 0.3896,
|
| 639 |
+
"eval_samples_per_second": 179.657,
|
| 640 |
+
"eval_steps_per_second": 7.7,
|
| 641 |
+
"step": 134
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 67.89,
|
| 645 |
+
"eval_accuracy": 0.9,
|
| 646 |
+
"eval_loss": 0.2927437126636505,
|
| 647 |
+
"eval_runtime": 0.3965,
|
| 648 |
+
"eval_samples_per_second": 176.563,
|
| 649 |
+
"eval_steps_per_second": 7.567,
|
| 650 |
+
"step": 136
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 68.89,
|
| 654 |
+
"eval_accuracy": 0.9,
|
| 655 |
+
"eval_loss": 0.2911137044429779,
|
| 656 |
+
"eval_runtime": 0.3982,
|
| 657 |
+
"eval_samples_per_second": 175.784,
|
| 658 |
+
"eval_steps_per_second": 7.534,
|
| 659 |
+
"step": 138
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 69.89,
|
| 663 |
+
"eval_accuracy": 0.9,
|
| 664 |
+
"eval_loss": 0.29145926237106323,
|
| 665 |
+
"eval_runtime": 0.3948,
|
| 666 |
+
"eval_samples_per_second": 177.308,
|
| 667 |
+
"eval_steps_per_second": 7.599,
|
| 668 |
+
"step": 140
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 70.89,
|
| 672 |
+
"eval_accuracy": 0.9,
|
| 673 |
+
"eval_loss": 0.291074275970459,
|
| 674 |
+
"eval_runtime": 0.3933,
|
| 675 |
+
"eval_samples_per_second": 177.964,
|
| 676 |
+
"eval_steps_per_second": 7.627,
|
| 677 |
+
"step": 142
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 71.89,
|
| 681 |
+
"eval_accuracy": 0.9,
|
| 682 |
+
"eval_loss": 0.2896248400211334,
|
| 683 |
+
"eval_runtime": 0.3962,
|
| 684 |
+
"eval_samples_per_second": 176.685,
|
| 685 |
+
"eval_steps_per_second": 7.572,
|
| 686 |
+
"step": 144
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 72.89,
|
| 690 |
+
"eval_accuracy": 0.9,
|
| 691 |
+
"eval_loss": 0.2880937457084656,
|
| 692 |
+
"eval_runtime": 0.3944,
|
| 693 |
+
"eval_samples_per_second": 177.463,
|
| 694 |
+
"eval_steps_per_second": 7.606,
|
| 695 |
+
"step": 146
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 73.89,
|
| 699 |
+
"eval_accuracy": 0.9,
|
| 700 |
+
"eval_loss": 0.28765279054641724,
|
| 701 |
+
"eval_runtime": 0.3966,
|
| 702 |
+
"eval_samples_per_second": 176.488,
|
| 703 |
+
"eval_steps_per_second": 7.564,
|
| 704 |
+
"step": 148
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 74.89,
|
| 708 |
+
"learning_rate": 5.555555555555557e-06,
|
| 709 |
+
"loss": 0.3705,
|
| 710 |
+
"step": 150
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 74.89,
|
| 714 |
+
"eval_accuracy": 0.9,
|
| 715 |
+
"eval_loss": 0.2876226603984833,
|
| 716 |
+
"eval_runtime": 0.3888,
|
| 717 |
+
"eval_samples_per_second": 180.054,
|
| 718 |
+
"eval_steps_per_second": 7.717,
|
| 719 |
+
"step": 150
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 75.89,
|
| 723 |
+
"eval_accuracy": 0.9,
|
| 724 |
+
"eval_loss": 0.28748172521591187,
|
| 725 |
+
"eval_runtime": 0.411,
|
| 726 |
+
"eval_samples_per_second": 170.337,
|
| 727 |
+
"eval_steps_per_second": 7.3,
|
| 728 |
+
"step": 152
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 76.89,
|
| 732 |
+
"eval_accuracy": 0.8857142857142857,
|
| 733 |
+
"eval_loss": 0.2876095473766327,
|
| 734 |
+
"eval_runtime": 0.4294,
|
| 735 |
+
"eval_samples_per_second": 163.0,
|
| 736 |
+
"eval_steps_per_second": 6.986,
|
| 737 |
+
"step": 154
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 77.89,
|
| 741 |
+
"eval_accuracy": 0.8857142857142857,
|
| 742 |
+
"eval_loss": 0.28670457005500793,
|
| 743 |
+
"eval_runtime": 0.3949,
|
| 744 |
+
"eval_samples_per_second": 177.274,
|
| 745 |
+
"eval_steps_per_second": 7.597,
|
| 746 |
+
"step": 156
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 78.89,
|
| 750 |
+
"eval_accuracy": 0.8857142857142857,
|
| 751 |
+
"eval_loss": 0.28570595383644104,
|
| 752 |
+
"eval_runtime": 0.3939,
|
| 753 |
+
"eval_samples_per_second": 177.69,
|
| 754 |
+
"eval_steps_per_second": 7.615,
|
| 755 |
+
"step": 158
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 79.89,
|
| 759 |
+
"eval_accuracy": 0.8857142857142857,
|
| 760 |
+
"eval_loss": 0.28557589650154114,
|
| 761 |
+
"eval_runtime": 0.3949,
|
| 762 |
+
"eval_samples_per_second": 177.268,
|
| 763 |
+
"eval_steps_per_second": 7.597,
|
| 764 |
+
"step": 160
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 80.89,
|
| 768 |
+
"eval_accuracy": 0.8857142857142857,
|
| 769 |
+
"eval_loss": 0.2853609621524811,
|
| 770 |
+
"eval_runtime": 0.3898,
|
| 771 |
+
"eval_samples_per_second": 179.59,
|
| 772 |
+
"eval_steps_per_second": 7.697,
|
| 773 |
+
"step": 162
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 81.89,
|
| 777 |
+
"eval_accuracy": 0.8714285714285714,
|
| 778 |
+
"eval_loss": 0.2859923243522644,
|
| 779 |
+
"eval_runtime": 0.3917,
|
| 780 |
+
"eval_samples_per_second": 178.729,
|
| 781 |
+
"eval_steps_per_second": 7.66,
|
| 782 |
+
"step": 164
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 82.89,
|
| 786 |
+
"eval_accuracy": 0.8714285714285714,
|
| 787 |
+
"eval_loss": 0.2853332459926605,
|
| 788 |
+
"eval_runtime": 0.3898,
|
| 789 |
+
"eval_samples_per_second": 179.566,
|
| 790 |
+
"eval_steps_per_second": 7.696,
|
| 791 |
+
"step": 166
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 83.89,
|
| 795 |
+
"eval_accuracy": 0.8714285714285714,
|
| 796 |
+
"eval_loss": 0.28681549429893494,
|
| 797 |
+
"eval_runtime": 0.3903,
|
| 798 |
+
"eval_samples_per_second": 179.364,
|
| 799 |
+
"eval_steps_per_second": 7.687,
|
| 800 |
+
"step": 168
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 84.89,
|
| 804 |
+
"eval_accuracy": 0.8714285714285714,
|
| 805 |
+
"eval_loss": 0.2884899079799652,
|
| 806 |
+
"eval_runtime": 0.4004,
|
| 807 |
+
"eval_samples_per_second": 174.808,
|
| 808 |
+
"eval_steps_per_second": 7.492,
|
| 809 |
+
"step": 170
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 85.89,
|
| 813 |
+
"eval_accuracy": 0.8714285714285714,
|
| 814 |
+
"eval_loss": 0.2893112599849701,
|
| 815 |
+
"eval_runtime": 0.3932,
|
| 816 |
+
"eval_samples_per_second": 178.006,
|
| 817 |
+
"eval_steps_per_second": 7.629,
|
| 818 |
+
"step": 172
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 86.89,
|
| 822 |
+
"eval_accuracy": 0.8714285714285714,
|
| 823 |
+
"eval_loss": 0.29076695442199707,
|
| 824 |
+
"eval_runtime": 0.4006,
|
| 825 |
+
"eval_samples_per_second": 174.727,
|
| 826 |
+
"eval_steps_per_second": 7.488,
|
| 827 |
+
"step": 174
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 87.44,
|
| 831 |
+
"learning_rate": 4.62962962962963e-06,
|
| 832 |
+
"loss": 0.3442,
|
| 833 |
+
"step": 175
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 87.89,
|
| 837 |
+
"eval_accuracy": 0.8714285714285714,
|
| 838 |
+
"eval_loss": 0.29055681824684143,
|
| 839 |
+
"eval_runtime": 0.3948,
|
| 840 |
+
"eval_samples_per_second": 177.302,
|
| 841 |
+
"eval_steps_per_second": 7.599,
|
| 842 |
+
"step": 176
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 88.89,
|
| 846 |
+
"eval_accuracy": 0.8714285714285714,
|
| 847 |
+
"eval_loss": 0.29001426696777344,
|
| 848 |
+
"eval_runtime": 0.4034,
|
| 849 |
+
"eval_samples_per_second": 173.528,
|
| 850 |
+
"eval_steps_per_second": 7.437,
|
| 851 |
+
"step": 178
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 89.89,
|
| 855 |
+
"eval_accuracy": 0.8714285714285714,
|
| 856 |
+
"eval_loss": 0.287276953458786,
|
| 857 |
+
"eval_runtime": 0.3919,
|
| 858 |
+
"eval_samples_per_second": 178.62,
|
| 859 |
+
"eval_steps_per_second": 7.655,
|
| 860 |
+
"step": 180
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 90.89,
|
| 864 |
+
"eval_accuracy": 0.8714285714285714,
|
| 865 |
+
"eval_loss": 0.28426897525787354,
|
| 866 |
+
"eval_runtime": 0.4018,
|
| 867 |
+
"eval_samples_per_second": 174.231,
|
| 868 |
+
"eval_steps_per_second": 7.467,
|
| 869 |
+
"step": 182
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 91.89,
|
| 873 |
+
"eval_accuracy": 0.8857142857142857,
|
| 874 |
+
"eval_loss": 0.28223827481269836,
|
| 875 |
+
"eval_runtime": 0.3886,
|
| 876 |
+
"eval_samples_per_second": 180.111,
|
| 877 |
+
"eval_steps_per_second": 7.719,
|
| 878 |
+
"step": 184
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 92.89,
|
| 882 |
+
"eval_accuracy": 0.8714285714285714,
|
| 883 |
+
"eval_loss": 0.28137001395225525,
|
| 884 |
+
"eval_runtime": 0.3867,
|
| 885 |
+
"eval_samples_per_second": 181.016,
|
| 886 |
+
"eval_steps_per_second": 7.758,
|
| 887 |
+
"step": 186
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 93.89,
|
| 891 |
+
"eval_accuracy": 0.8714285714285714,
|
| 892 |
+
"eval_loss": 0.2821663022041321,
|
| 893 |
+
"eval_runtime": 0.395,
|
| 894 |
+
"eval_samples_per_second": 177.23,
|
| 895 |
+
"eval_steps_per_second": 7.596,
|
| 896 |
+
"step": 188
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 94.89,
|
| 900 |
+
"eval_accuracy": 0.8714285714285714,
|
| 901 |
+
"eval_loss": 0.2825068235397339,
|
| 902 |
+
"eval_runtime": 0.3994,
|
| 903 |
+
"eval_samples_per_second": 175.263,
|
| 904 |
+
"eval_steps_per_second": 7.511,
|
| 905 |
+
"step": 190
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 95.89,
|
| 909 |
+
"eval_accuracy": 0.8857142857142857,
|
| 910 |
+
"eval_loss": 0.2831081449985504,
|
| 911 |
+
"eval_runtime": 0.3997,
|
| 912 |
+
"eval_samples_per_second": 175.112,
|
| 913 |
+
"eval_steps_per_second": 7.505,
|
| 914 |
+
"step": 192
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 96.89,
|
| 918 |
+
"eval_accuracy": 0.8857142857142857,
|
| 919 |
+
"eval_loss": 0.2835954427719116,
|
| 920 |
+
"eval_runtime": 0.3968,
|
| 921 |
+
"eval_samples_per_second": 176.408,
|
| 922 |
+
"eval_steps_per_second": 7.56,
|
| 923 |
+
"step": 194
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 97.89,
|
| 927 |
+
"eval_accuracy": 0.8857142857142857,
|
| 928 |
+
"eval_loss": 0.28326863050460815,
|
| 929 |
+
"eval_runtime": 0.3965,
|
| 930 |
+
"eval_samples_per_second": 176.561,
|
| 931 |
+
"eval_steps_per_second": 7.567,
|
| 932 |
+
"step": 196
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 98.89,
|
| 936 |
+
"eval_accuracy": 0.8857142857142857,
|
| 937 |
+
"eval_loss": 0.281973272562027,
|
| 938 |
+
"eval_runtime": 0.3973,
|
| 939 |
+
"eval_samples_per_second": 176.206,
|
| 940 |
+
"eval_steps_per_second": 7.552,
|
| 941 |
+
"step": 198
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 99.89,
|
| 945 |
+
"learning_rate": 3.7037037037037037e-06,
|
| 946 |
+
"loss": 0.3159,
|
| 947 |
+
"step": 200
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 99.89,
|
| 951 |
+
"eval_accuracy": 0.8857142857142857,
|
| 952 |
+
"eval_loss": 0.28054317831993103,
|
| 953 |
+
"eval_runtime": 0.3966,
|
| 954 |
+
"eval_samples_per_second": 176.5,
|
| 955 |
+
"eval_steps_per_second": 7.564,
|
| 956 |
+
"step": 200
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 100.89,
|
| 960 |
+
"eval_accuracy": 0.8714285714285714,
|
| 961 |
+
"eval_loss": 0.27876371145248413,
|
| 962 |
+
"eval_runtime": 0.393,
|
| 963 |
+
"eval_samples_per_second": 178.117,
|
| 964 |
+
"eval_steps_per_second": 7.634,
|
| 965 |
+
"step": 202
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 101.89,
|
| 969 |
+
"eval_accuracy": 0.8714285714285714,
|
| 970 |
+
"eval_loss": 0.27795442938804626,
|
| 971 |
+
"eval_runtime": 0.388,
|
| 972 |
+
"eval_samples_per_second": 180.426,
|
| 973 |
+
"eval_steps_per_second": 7.733,
|
| 974 |
+
"step": 204
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 102.89,
|
| 978 |
+
"eval_accuracy": 0.8714285714285714,
|
| 979 |
+
"eval_loss": 0.27797621488571167,
|
| 980 |
+
"eval_runtime": 0.3898,
|
| 981 |
+
"eval_samples_per_second": 179.574,
|
| 982 |
+
"eval_steps_per_second": 7.696,
|
| 983 |
+
"step": 206
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 103.89,
|
| 987 |
+
"eval_accuracy": 0.8714285714285714,
|
| 988 |
+
"eval_loss": 0.27778205275535583,
|
| 989 |
+
"eval_runtime": 0.3946,
|
| 990 |
+
"eval_samples_per_second": 177.402,
|
| 991 |
+
"eval_steps_per_second": 7.603,
|
| 992 |
+
"step": 208
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 104.89,
|
| 996 |
+
"eval_accuracy": 0.8714285714285714,
|
| 997 |
+
"eval_loss": 0.2783416509628296,
|
| 998 |
+
"eval_runtime": 0.3956,
|
| 999 |
+
"eval_samples_per_second": 176.929,
|
| 1000 |
+
"eval_steps_per_second": 7.583,
|
| 1001 |
+
"step": 210
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 105.89,
|
| 1005 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1006 |
+
"eval_loss": 0.2788624167442322,
|
| 1007 |
+
"eval_runtime": 0.3965,
|
| 1008 |
+
"eval_samples_per_second": 176.537,
|
| 1009 |
+
"eval_steps_per_second": 7.566,
|
| 1010 |
+
"step": 212
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 106.89,
|
| 1014 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1015 |
+
"eval_loss": 0.27885138988494873,
|
| 1016 |
+
"eval_runtime": 0.401,
|
| 1017 |
+
"eval_samples_per_second": 174.562,
|
| 1018 |
+
"eval_steps_per_second": 7.481,
|
| 1019 |
+
"step": 214
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 107.89,
|
| 1023 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1024 |
+
"eval_loss": 0.2788878381252289,
|
| 1025 |
+
"eval_runtime": 0.3934,
|
| 1026 |
+
"eval_samples_per_second": 177.953,
|
| 1027 |
+
"eval_steps_per_second": 7.627,
|
| 1028 |
+
"step": 216
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 108.89,
|
| 1032 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1033 |
+
"eval_loss": 0.27941665053367615,
|
| 1034 |
+
"eval_runtime": 0.3952,
|
| 1035 |
+
"eval_samples_per_second": 177.139,
|
| 1036 |
+
"eval_steps_per_second": 7.592,
|
| 1037 |
+
"step": 218
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 109.89,
|
| 1041 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1042 |
+
"eval_loss": 0.2792874872684479,
|
| 1043 |
+
"eval_runtime": 0.3992,
|
| 1044 |
+
"eval_samples_per_second": 175.358,
|
| 1045 |
+
"eval_steps_per_second": 7.515,
|
| 1046 |
+
"step": 220
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 110.89,
|
| 1050 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1051 |
+
"eval_loss": 0.27895817160606384,
|
| 1052 |
+
"eval_runtime": 0.393,
|
| 1053 |
+
"eval_samples_per_second": 178.121,
|
| 1054 |
+
"eval_steps_per_second": 7.634,
|
| 1055 |
+
"step": 222
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 111.89,
|
| 1059 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1060 |
+
"eval_loss": 0.2787097692489624,
|
| 1061 |
+
"eval_runtime": 0.3941,
|
| 1062 |
+
"eval_samples_per_second": 177.636,
|
| 1063 |
+
"eval_steps_per_second": 7.613,
|
| 1064 |
+
"step": 224
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 112.44,
|
| 1068 |
+
"learning_rate": 2.7777777777777783e-06,
|
| 1069 |
+
"loss": 0.2982,
|
| 1070 |
+
"step": 225
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 112.89,
|
| 1074 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1075 |
+
"eval_loss": 0.2774018943309784,
|
| 1076 |
+
"eval_runtime": 0.3953,
|
| 1077 |
+
"eval_samples_per_second": 177.084,
|
| 1078 |
+
"eval_steps_per_second": 7.589,
|
| 1079 |
+
"step": 226
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 113.89,
|
| 1083 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1084 |
+
"eval_loss": 0.2765597105026245,
|
| 1085 |
+
"eval_runtime": 0.389,
|
| 1086 |
+
"eval_samples_per_second": 179.928,
|
| 1087 |
+
"eval_steps_per_second": 7.711,
|
| 1088 |
+
"step": 228
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 114.89,
|
| 1092 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1093 |
+
"eval_loss": 0.27541905641555786,
|
| 1094 |
+
"eval_runtime": 0.3931,
|
| 1095 |
+
"eval_samples_per_second": 178.054,
|
| 1096 |
+
"eval_steps_per_second": 7.631,
|
| 1097 |
+
"step": 230
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 115.89,
|
| 1101 |
+
"eval_accuracy": 0.8714285714285714,
|
| 1102 |
+
"eval_loss": 0.27446889877319336,
|
| 1103 |
+
"eval_runtime": 0.3931,
|
| 1104 |
+
"eval_samples_per_second": 178.076,
|
| 1105 |
+
"eval_steps_per_second": 7.632,
|
| 1106 |
+
"step": 232
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 116.89,
|
| 1110 |
+
"eval_accuracy": 0.8714285714285714,
|
| 1111 |
+
"eval_loss": 0.2735791802406311,
|
| 1112 |
+
"eval_runtime": 0.3922,
|
| 1113 |
+
"eval_samples_per_second": 178.501,
|
| 1114 |
+
"eval_steps_per_second": 7.65,
|
| 1115 |
+
"step": 234
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 117.89,
|
| 1119 |
+
"eval_accuracy": 0.8714285714285714,
|
| 1120 |
+
"eval_loss": 0.27332428097724915,
|
| 1121 |
+
"eval_runtime": 0.3911,
|
| 1122 |
+
"eval_samples_per_second": 178.974,
|
| 1123 |
+
"eval_steps_per_second": 7.67,
|
| 1124 |
+
"step": 236
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 118.89,
|
| 1128 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1129 |
+
"eval_loss": 0.2734207212924957,
|
| 1130 |
+
"eval_runtime": 0.396,
|
| 1131 |
+
"eval_samples_per_second": 176.768,
|
| 1132 |
+
"eval_steps_per_second": 7.576,
|
| 1133 |
+
"step": 238
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 119.89,
|
| 1137 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1138 |
+
"eval_loss": 0.27379804849624634,
|
| 1139 |
+
"eval_runtime": 0.3939,
|
| 1140 |
+
"eval_samples_per_second": 177.731,
|
| 1141 |
+
"eval_steps_per_second": 7.617,
|
| 1142 |
+
"step": 240
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 120.89,
|
| 1146 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1147 |
+
"eval_loss": 0.2744147777557373,
|
| 1148 |
+
"eval_runtime": 0.3947,
|
| 1149 |
+
"eval_samples_per_second": 177.362,
|
| 1150 |
+
"eval_steps_per_second": 7.601,
|
| 1151 |
+
"step": 242
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 121.89,
|
| 1155 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1156 |
+
"eval_loss": 0.2748957574367523,
|
| 1157 |
+
"eval_runtime": 0.4048,
|
| 1158 |
+
"eval_samples_per_second": 172.911,
|
| 1159 |
+
"eval_steps_per_second": 7.41,
|
| 1160 |
+
"step": 244
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 122.89,
|
| 1164 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1165 |
+
"eval_loss": 0.27517834305763245,
|
| 1166 |
+
"eval_runtime": 0.398,
|
| 1167 |
+
"eval_samples_per_second": 175.867,
|
| 1168 |
+
"eval_steps_per_second": 7.537,
|
| 1169 |
+
"step": 246
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 123.89,
|
| 1173 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1174 |
+
"eval_loss": 0.2754015624523163,
|
| 1175 |
+
"eval_runtime": 0.394,
|
| 1176 |
+
"eval_samples_per_second": 177.676,
|
| 1177 |
+
"eval_steps_per_second": 7.615,
|
| 1178 |
+
"step": 248
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 124.89,
|
| 1182 |
+
"learning_rate": 1.8518518518518519e-06,
|
| 1183 |
+
"loss": 0.2757,
|
| 1184 |
+
"step": 250
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 124.89,
|
| 1188 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1189 |
+
"eval_loss": 0.27559223771095276,
|
| 1190 |
+
"eval_runtime": 0.3999,
|
| 1191 |
+
"eval_samples_per_second": 175.064,
|
| 1192 |
+
"eval_steps_per_second": 7.503,
|
| 1193 |
+
"step": 250
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 125.89,
|
| 1197 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1198 |
+
"eval_loss": 0.27577343583106995,
|
| 1199 |
+
"eval_runtime": 0.3951,
|
| 1200 |
+
"eval_samples_per_second": 177.152,
|
| 1201 |
+
"eval_steps_per_second": 7.592,
|
| 1202 |
+
"step": 252
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 126.89,
|
| 1206 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1207 |
+
"eval_loss": 0.2756122648715973,
|
| 1208 |
+
"eval_runtime": 0.3965,
|
| 1209 |
+
"eval_samples_per_second": 176.526,
|
| 1210 |
+
"eval_steps_per_second": 7.565,
|
| 1211 |
+
"step": 254
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 127.89,
|
| 1215 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1216 |
+
"eval_loss": 0.2754615247249603,
|
| 1217 |
+
"eval_runtime": 0.3971,
|
| 1218 |
+
"eval_samples_per_second": 176.293,
|
| 1219 |
+
"eval_steps_per_second": 7.555,
|
| 1220 |
+
"step": 256
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 128.89,
|
| 1224 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1225 |
+
"eval_loss": 0.27490970492362976,
|
| 1226 |
+
"eval_runtime": 0.3981,
|
| 1227 |
+
"eval_samples_per_second": 175.828,
|
| 1228 |
+
"eval_steps_per_second": 7.535,
|
| 1229 |
+
"step": 258
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 129.89,
|
| 1233 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1234 |
+
"eval_loss": 0.27434810996055603,
|
| 1235 |
+
"eval_runtime": 0.3988,
|
| 1236 |
+
"eval_samples_per_second": 175.52,
|
| 1237 |
+
"eval_steps_per_second": 7.522,
|
| 1238 |
+
"step": 260
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 130.89,
|
| 1242 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1243 |
+
"eval_loss": 0.27393919229507446,
|
| 1244 |
+
"eval_runtime": 0.3964,
|
| 1245 |
+
"eval_samples_per_second": 176.607,
|
| 1246 |
+
"eval_steps_per_second": 7.569,
|
| 1247 |
+
"step": 262
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 131.89,
|
| 1251 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1252 |
+
"eval_loss": 0.2735598385334015,
|
| 1253 |
+
"eval_runtime": 0.3963,
|
| 1254 |
+
"eval_samples_per_second": 176.647,
|
| 1255 |
+
"eval_steps_per_second": 7.571,
|
| 1256 |
+
"step": 264
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 132.89,
|
| 1260 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1261 |
+
"eval_loss": 0.27345114946365356,
|
| 1262 |
+
"eval_runtime": 0.3943,
|
| 1263 |
+
"eval_samples_per_second": 177.525,
|
| 1264 |
+
"eval_steps_per_second": 7.608,
|
| 1265 |
+
"step": 266
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 133.89,
|
| 1269 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1270 |
+
"eval_loss": 0.27316632866859436,
|
| 1271 |
+
"eval_runtime": 0.3968,
|
| 1272 |
+
"eval_samples_per_second": 176.395,
|
| 1273 |
+
"eval_steps_per_second": 7.56,
|
| 1274 |
+
"step": 268
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 134.89,
|
| 1278 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1279 |
+
"eval_loss": 0.2731429636478424,
|
| 1280 |
+
"eval_runtime": 0.3964,
|
| 1281 |
+
"eval_samples_per_second": 176.608,
|
| 1282 |
+
"eval_steps_per_second": 7.569,
|
| 1283 |
+
"step": 270
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 135.89,
|
| 1287 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1288 |
+
"eval_loss": 0.2727990746498108,
|
| 1289 |
+
"eval_runtime": 0.3965,
|
| 1290 |
+
"eval_samples_per_second": 176.545,
|
| 1291 |
+
"eval_steps_per_second": 7.566,
|
| 1292 |
+
"step": 272
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 136.89,
|
| 1296 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1297 |
+
"eval_loss": 0.2725311815738678,
|
| 1298 |
+
"eval_runtime": 0.3934,
|
| 1299 |
+
"eval_samples_per_second": 177.937,
|
| 1300 |
+
"eval_steps_per_second": 7.626,
|
| 1301 |
+
"step": 274
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 137.44,
|
| 1305 |
+
"learning_rate": 9.259259259259259e-07,
|
| 1306 |
+
"loss": 0.2724,
|
| 1307 |
+
"step": 275
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 137.89,
|
| 1311 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1312 |
+
"eval_loss": 0.27232787013053894,
|
| 1313 |
+
"eval_runtime": 0.397,
|
| 1314 |
+
"eval_samples_per_second": 176.334,
|
| 1315 |
+
"eval_steps_per_second": 7.557,
|
| 1316 |
+
"step": 276
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 138.89,
|
| 1320 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1321 |
+
"eval_loss": 0.2722608149051666,
|
| 1322 |
+
"eval_runtime": 0.3984,
|
| 1323 |
+
"eval_samples_per_second": 175.719,
|
| 1324 |
+
"eval_steps_per_second": 7.531,
|
| 1325 |
+
"step": 278
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 139.89,
|
| 1329 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1330 |
+
"eval_loss": 0.2721470594406128,
|
| 1331 |
+
"eval_runtime": 0.3948,
|
| 1332 |
+
"eval_samples_per_second": 177.289,
|
| 1333 |
+
"eval_steps_per_second": 7.598,
|
| 1334 |
+
"step": 280
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 140.89,
|
| 1338 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1339 |
+
"eval_loss": 0.2720603346824646,
|
| 1340 |
+
"eval_runtime": 0.4008,
|
| 1341 |
+
"eval_samples_per_second": 174.65,
|
| 1342 |
+
"eval_steps_per_second": 7.485,
|
| 1343 |
+
"step": 282
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 141.89,
|
| 1347 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1348 |
+
"eval_loss": 0.27187877893447876,
|
| 1349 |
+
"eval_runtime": 0.3937,
|
| 1350 |
+
"eval_samples_per_second": 177.785,
|
| 1351 |
+
"eval_steps_per_second": 7.619,
|
| 1352 |
+
"step": 284
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 142.89,
|
| 1356 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1357 |
+
"eval_loss": 0.27186280488967896,
|
| 1358 |
+
"eval_runtime": 0.3942,
|
| 1359 |
+
"eval_samples_per_second": 177.564,
|
| 1360 |
+
"eval_steps_per_second": 7.61,
|
| 1361 |
+
"step": 286
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 143.89,
|
| 1365 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1366 |
+
"eval_loss": 0.27189135551452637,
|
| 1367 |
+
"eval_runtime": 0.3939,
|
| 1368 |
+
"eval_samples_per_second": 177.73,
|
| 1369 |
+
"eval_steps_per_second": 7.617,
|
| 1370 |
+
"step": 288
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 144.89,
|
| 1374 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1375 |
+
"eval_loss": 0.27191561460494995,
|
| 1376 |
+
"eval_runtime": 0.4045,
|
| 1377 |
+
"eval_samples_per_second": 173.067,
|
| 1378 |
+
"eval_steps_per_second": 7.417,
|
| 1379 |
+
"step": 290
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 145.89,
|
| 1383 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1384 |
+
"eval_loss": 0.2719388008117676,
|
| 1385 |
+
"eval_runtime": 0.3958,
|
| 1386 |
+
"eval_samples_per_second": 176.858,
|
| 1387 |
+
"eval_steps_per_second": 7.58,
|
| 1388 |
+
"step": 292
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 146.89,
|
| 1392 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1393 |
+
"eval_loss": 0.2719106674194336,
|
| 1394 |
+
"eval_runtime": 0.3928,
|
| 1395 |
+
"eval_samples_per_second": 178.226,
|
| 1396 |
+
"eval_steps_per_second": 7.638,
|
| 1397 |
+
"step": 294
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 147.89,
|
| 1401 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1402 |
+
"eval_loss": 0.271912157535553,
|
| 1403 |
+
"eval_runtime": 0.3948,
|
| 1404 |
+
"eval_samples_per_second": 177.319,
|
| 1405 |
+
"eval_steps_per_second": 7.599,
|
| 1406 |
+
"step": 296
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 148.89,
|
| 1410 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1411 |
+
"eval_loss": 0.271916002035141,
|
| 1412 |
+
"eval_runtime": 0.3979,
|
| 1413 |
+
"eval_samples_per_second": 175.916,
|
| 1414 |
+
"eval_steps_per_second": 7.539,
|
| 1415 |
+
"step": 298
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 149.89,
|
| 1419 |
+
"learning_rate": 0.0,
|
| 1420 |
+
"loss": 0.2614,
|
| 1421 |
+
"step": 300
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 149.89,
|
| 1425 |
+
"eval_accuracy": 0.8857142857142857,
|
| 1426 |
+
"eval_loss": 0.27192050218582153,
|
| 1427 |
+
"eval_runtime": 0.3918,
|
| 1428 |
+
"eval_samples_per_second": 178.659,
|
| 1429 |
+
"eval_steps_per_second": 7.657,
|
| 1430 |
+
"step": 300
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 149.89,
|
| 1434 |
+
"step": 300,
|
| 1435 |
+
"total_flos": 1.0435256966870508e+18,
|
| 1436 |
+
"train_loss": 0.39242326736450195,
|
| 1437 |
+
"train_runtime": 677.2441,
|
| 1438 |
+
"train_samples_per_second": 61.352,
|
| 1439 |
+
"train_steps_per_second": 0.443
|
| 1440 |
+
}
|
| 1441 |
+
],
|
| 1442 |
+
"max_steps": 300,
|
| 1443 |
+
"num_train_epochs": 150,
|
| 1444 |
+
"total_flos": 1.0435256966870508e+18,
|
| 1445 |
+
"trial_name": null,
|
| 1446 |
+
"trial_params": null
|
| 1447 |
+
}
|