End of training
Browse files- all_results.json +13 -0
- eval_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +3496 -0
all_results.json
ADDED
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 14.0,
|
3 |
+
"eval_accuracy": 0.9872761545711592,
|
4 |
+
"eval_loss": 0.0543154738843441,
|
5 |
+
"eval_runtime": 66.8631,
|
6 |
+
"eval_samples_per_second": 31.737,
|
7 |
+
"eval_steps_per_second": 1.989,
|
8 |
+
"total_flos": 1.5805887922325717e+19,
|
9 |
+
"train_loss": 0.17515284487896893,
|
10 |
+
"train_runtime": 12010.6398,
|
11 |
+
"train_samples_per_second": 7.418,
|
12 |
+
"train_steps_per_second": 0.232
|
13 |
+
}
|
eval_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 14.0,
|
3 |
+
"eval_accuracy": 0.9872761545711592,
|
4 |
+
"eval_loss": 0.0543154738843441,
|
5 |
+
"eval_runtime": 66.8631,
|
6 |
+
"eval_samples_per_second": 31.737,
|
7 |
+
"eval_steps_per_second": 1.989
|
8 |
+
}
|
train_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 14.0,
|
3 |
+
"total_flos": 1.5805887922325717e+19,
|
4 |
+
"train_loss": 0.17515284487896893,
|
5 |
+
"train_runtime": 12010.6398,
|
6 |
+
"train_samples_per_second": 7.418,
|
7 |
+
"train_steps_per_second": 0.232
|
8 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,3496 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 0.9872761545711592,
|
3 |
+
"best_model_checkpoint": "convnextv2-large-1k-224-finetuned-BreastCancer-Classification-BreakHis-AH-60-20-20-Shuffled-3rd/checkpoint-2786",
|
4 |
+
"epoch": 14.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 2786,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.03,
|
13 |
+
"learning_rate": 9.96810207336523e-08,
|
14 |
+
"loss": 0.6896,
|
15 |
+
"step": 5
|
16 |
+
},
|
17 |
+
{
|
18 |
+
"epoch": 0.05,
|
19 |
+
"learning_rate": 1.993620414673046e-07,
|
20 |
+
"loss": 0.6883,
|
21 |
+
"step": 10
|
22 |
+
},
|
23 |
+
{
|
24 |
+
"epoch": 0.08,
|
25 |
+
"learning_rate": 2.990430622009569e-07,
|
26 |
+
"loss": 0.6853,
|
27 |
+
"step": 15
|
28 |
+
},
|
29 |
+
{
|
30 |
+
"epoch": 0.1,
|
31 |
+
"learning_rate": 3.987240829346092e-07,
|
32 |
+
"loss": 0.6958,
|
33 |
+
"step": 20
|
34 |
+
},
|
35 |
+
{
|
36 |
+
"epoch": 0.13,
|
37 |
+
"learning_rate": 4.984051036682616e-07,
|
38 |
+
"loss": 0.6908,
|
39 |
+
"step": 25
|
40 |
+
},
|
41 |
+
{
|
42 |
+
"epoch": 0.15,
|
43 |
+
"learning_rate": 5.980861244019139e-07,
|
44 |
+
"loss": 0.6933,
|
45 |
+
"step": 30
|
46 |
+
},
|
47 |
+
{
|
48 |
+
"epoch": 0.18,
|
49 |
+
"learning_rate": 6.977671451355662e-07,
|
50 |
+
"loss": 0.686,
|
51 |
+
"step": 35
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.2,
|
55 |
+
"learning_rate": 7.974481658692184e-07,
|
56 |
+
"loss": 0.6897,
|
57 |
+
"step": 40
|
58 |
+
},
|
59 |
+
{
|
60 |
+
"epoch": 0.23,
|
61 |
+
"learning_rate": 8.971291866028709e-07,
|
62 |
+
"loss": 0.6842,
|
63 |
+
"step": 45
|
64 |
+
},
|
65 |
+
{
|
66 |
+
"epoch": 0.25,
|
67 |
+
"learning_rate": 9.968102073365232e-07,
|
68 |
+
"loss": 0.6886,
|
69 |
+
"step": 50
|
70 |
+
},
|
71 |
+
{
|
72 |
+
"epoch": 0.28,
|
73 |
+
"learning_rate": 1.0964912280701754e-06,
|
74 |
+
"loss": 0.6794,
|
75 |
+
"step": 55
|
76 |
+
},
|
77 |
+
{
|
78 |
+
"epoch": 0.3,
|
79 |
+
"learning_rate": 1.1961722488038277e-06,
|
80 |
+
"loss": 0.6829,
|
81 |
+
"step": 60
|
82 |
+
},
|
83 |
+
{
|
84 |
+
"epoch": 0.33,
|
85 |
+
"learning_rate": 1.2958532695374803e-06,
|
86 |
+
"loss": 0.677,
|
87 |
+
"step": 65
|
88 |
+
},
|
89 |
+
{
|
90 |
+
"epoch": 0.35,
|
91 |
+
"learning_rate": 1.3955342902711324e-06,
|
92 |
+
"loss": 0.6723,
|
93 |
+
"step": 70
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.38,
|
97 |
+
"learning_rate": 1.4952153110047847e-06,
|
98 |
+
"loss": 0.6743,
|
99 |
+
"step": 75
|
100 |
+
},
|
101 |
+
{
|
102 |
+
"epoch": 0.4,
|
103 |
+
"learning_rate": 1.5948963317384369e-06,
|
104 |
+
"loss": 0.6695,
|
105 |
+
"step": 80
|
106 |
+
},
|
107 |
+
{
|
108 |
+
"epoch": 0.43,
|
109 |
+
"learning_rate": 1.6945773524720894e-06,
|
110 |
+
"loss": 0.6711,
|
111 |
+
"step": 85
|
112 |
+
},
|
113 |
+
{
|
114 |
+
"epoch": 0.45,
|
115 |
+
"learning_rate": 1.7942583732057418e-06,
|
116 |
+
"loss": 0.6671,
|
117 |
+
"step": 90
|
118 |
+
},
|
119 |
+
{
|
120 |
+
"epoch": 0.48,
|
121 |
+
"learning_rate": 1.8939393939393941e-06,
|
122 |
+
"loss": 0.6599,
|
123 |
+
"step": 95
|
124 |
+
},
|
125 |
+
{
|
126 |
+
"epoch": 0.5,
|
127 |
+
"learning_rate": 1.9936204146730465e-06,
|
128 |
+
"loss": 0.6581,
|
129 |
+
"step": 100
|
130 |
+
},
|
131 |
+
{
|
132 |
+
"epoch": 0.53,
|
133 |
+
"learning_rate": 2.093301435406699e-06,
|
134 |
+
"loss": 0.6485,
|
135 |
+
"step": 105
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 0.55,
|
139 |
+
"learning_rate": 2.1929824561403507e-06,
|
140 |
+
"loss": 0.6466,
|
141 |
+
"step": 110
|
142 |
+
},
|
143 |
+
{
|
144 |
+
"epoch": 0.58,
|
145 |
+
"learning_rate": 2.292663476874003e-06,
|
146 |
+
"loss": 0.6416,
|
147 |
+
"step": 115
|
148 |
+
},
|
149 |
+
{
|
150 |
+
"epoch": 0.6,
|
151 |
+
"learning_rate": 2.3923444976076554e-06,
|
152 |
+
"loss": 0.6462,
|
153 |
+
"step": 120
|
154 |
+
},
|
155 |
+
{
|
156 |
+
"epoch": 0.63,
|
157 |
+
"learning_rate": 2.492025518341308e-06,
|
158 |
+
"loss": 0.6363,
|
159 |
+
"step": 125
|
160 |
+
},
|
161 |
+
{
|
162 |
+
"epoch": 0.65,
|
163 |
+
"learning_rate": 2.5917065390749605e-06,
|
164 |
+
"loss": 0.6252,
|
165 |
+
"step": 130
|
166 |
+
},
|
167 |
+
{
|
168 |
+
"epoch": 0.68,
|
169 |
+
"learning_rate": 2.691387559808613e-06,
|
170 |
+
"loss": 0.622,
|
171 |
+
"step": 135
|
172 |
+
},
|
173 |
+
{
|
174 |
+
"epoch": 0.7,
|
175 |
+
"learning_rate": 2.7910685805422648e-06,
|
176 |
+
"loss": 0.6149,
|
177 |
+
"step": 140
|
178 |
+
},
|
179 |
+
{
|
180 |
+
"epoch": 0.73,
|
181 |
+
"learning_rate": 2.890749601275917e-06,
|
182 |
+
"loss": 0.6098,
|
183 |
+
"step": 145
|
184 |
+
},
|
185 |
+
{
|
186 |
+
"epoch": 0.75,
|
187 |
+
"learning_rate": 2.9904306220095695e-06,
|
188 |
+
"loss": 0.5992,
|
189 |
+
"step": 150
|
190 |
+
},
|
191 |
+
{
|
192 |
+
"epoch": 0.78,
|
193 |
+
"learning_rate": 3.090111642743222e-06,
|
194 |
+
"loss": 0.606,
|
195 |
+
"step": 155
|
196 |
+
},
|
197 |
+
{
|
198 |
+
"epoch": 0.8,
|
199 |
+
"learning_rate": 3.1897926634768737e-06,
|
200 |
+
"loss": 0.5844,
|
201 |
+
"step": 160
|
202 |
+
},
|
203 |
+
{
|
204 |
+
"epoch": 0.83,
|
205 |
+
"learning_rate": 3.2894736842105265e-06,
|
206 |
+
"loss": 0.5673,
|
207 |
+
"step": 165
|
208 |
+
},
|
209 |
+
{
|
210 |
+
"epoch": 0.85,
|
211 |
+
"learning_rate": 3.389154704944179e-06,
|
212 |
+
"loss": 0.5577,
|
213 |
+
"step": 170
|
214 |
+
},
|
215 |
+
{
|
216 |
+
"epoch": 0.88,
|
217 |
+
"learning_rate": 3.488835725677831e-06,
|
218 |
+
"loss": 0.5659,
|
219 |
+
"step": 175
|
220 |
+
},
|
221 |
+
{
|
222 |
+
"epoch": 0.9,
|
223 |
+
"learning_rate": 3.5885167464114835e-06,
|
224 |
+
"loss": 0.5354,
|
225 |
+
"step": 180
|
226 |
+
},
|
227 |
+
{
|
228 |
+
"epoch": 0.93,
|
229 |
+
"learning_rate": 3.688197767145136e-06,
|
230 |
+
"loss": 0.5417,
|
231 |
+
"step": 185
|
232 |
+
},
|
233 |
+
{
|
234 |
+
"epoch": 0.95,
|
235 |
+
"learning_rate": 3.7878787878787882e-06,
|
236 |
+
"loss": 0.5481,
|
237 |
+
"step": 190
|
238 |
+
},
|
239 |
+
{
|
240 |
+
"epoch": 0.98,
|
241 |
+
"learning_rate": 3.88755980861244e-06,
|
242 |
+
"loss": 0.5284,
|
243 |
+
"step": 195
|
244 |
+
},
|
245 |
+
{
|
246 |
+
"epoch": 1.0,
|
247 |
+
"eval_accuracy": 0.9095193213949104,
|
248 |
+
"eval_loss": 0.5013382434844971,
|
249 |
+
"eval_runtime": 773.5063,
|
250 |
+
"eval_samples_per_second": 2.743,
|
251 |
+
"eval_steps_per_second": 0.172,
|
252 |
+
"step": 199
|
253 |
+
},
|
254 |
+
{
|
255 |
+
"epoch": 1.01,
|
256 |
+
"learning_rate": 3.987240829346093e-06,
|
257 |
+
"loss": 0.5202,
|
258 |
+
"step": 200
|
259 |
+
},
|
260 |
+
{
|
261 |
+
"epoch": 1.03,
|
262 |
+
"learning_rate": 4.086921850079746e-06,
|
263 |
+
"loss": 0.5187,
|
264 |
+
"step": 205
|
265 |
+
},
|
266 |
+
{
|
267 |
+
"epoch": 1.06,
|
268 |
+
"learning_rate": 4.186602870813398e-06,
|
269 |
+
"loss": 0.5106,
|
270 |
+
"step": 210
|
271 |
+
},
|
272 |
+
{
|
273 |
+
"epoch": 1.08,
|
274 |
+
"learning_rate": 4.2862838915470495e-06,
|
275 |
+
"loss": 0.4866,
|
276 |
+
"step": 215
|
277 |
+
},
|
278 |
+
{
|
279 |
+
"epoch": 1.11,
|
280 |
+
"learning_rate": 4.3859649122807014e-06,
|
281 |
+
"loss": 0.4893,
|
282 |
+
"step": 220
|
283 |
+
},
|
284 |
+
{
|
285 |
+
"epoch": 1.13,
|
286 |
+
"learning_rate": 4.485645933014354e-06,
|
287 |
+
"loss": 0.4467,
|
288 |
+
"step": 225
|
289 |
+
},
|
290 |
+
{
|
291 |
+
"epoch": 1.16,
|
292 |
+
"learning_rate": 4.585326953748006e-06,
|
293 |
+
"loss": 0.4381,
|
294 |
+
"step": 230
|
295 |
+
},
|
296 |
+
{
|
297 |
+
"epoch": 1.18,
|
298 |
+
"learning_rate": 4.685007974481659e-06,
|
299 |
+
"loss": 0.422,
|
300 |
+
"step": 235
|
301 |
+
},
|
302 |
+
{
|
303 |
+
"epoch": 1.21,
|
304 |
+
"learning_rate": 4.784688995215311e-06,
|
305 |
+
"loss": 0.4099,
|
306 |
+
"step": 240
|
307 |
+
},
|
308 |
+
{
|
309 |
+
"epoch": 1.23,
|
310 |
+
"learning_rate": 4.8843700159489636e-06,
|
311 |
+
"loss": 0.4288,
|
312 |
+
"step": 245
|
313 |
+
},
|
314 |
+
{
|
315 |
+
"epoch": 1.26,
|
316 |
+
"learning_rate": 4.984051036682616e-06,
|
317 |
+
"loss": 0.3929,
|
318 |
+
"step": 250
|
319 |
+
},
|
320 |
+
{
|
321 |
+
"epoch": 1.28,
|
322 |
+
"learning_rate": 5.083732057416268e-06,
|
323 |
+
"loss": 0.3995,
|
324 |
+
"step": 255
|
325 |
+
},
|
326 |
+
{
|
327 |
+
"epoch": 1.31,
|
328 |
+
"learning_rate": 5.183413078149921e-06,
|
329 |
+
"loss": 0.3699,
|
330 |
+
"step": 260
|
331 |
+
},
|
332 |
+
{
|
333 |
+
"epoch": 1.33,
|
334 |
+
"learning_rate": 5.283094098883573e-06,
|
335 |
+
"loss": 0.4035,
|
336 |
+
"step": 265
|
337 |
+
},
|
338 |
+
{
|
339 |
+
"epoch": 1.36,
|
340 |
+
"learning_rate": 5.382775119617226e-06,
|
341 |
+
"loss": 0.3549,
|
342 |
+
"step": 270
|
343 |
+
},
|
344 |
+
{
|
345 |
+
"epoch": 1.38,
|
346 |
+
"learning_rate": 5.482456140350877e-06,
|
347 |
+
"loss": 0.3314,
|
348 |
+
"step": 275
|
349 |
+
},
|
350 |
+
{
|
351 |
+
"epoch": 1.41,
|
352 |
+
"learning_rate": 5.5821371610845296e-06,
|
353 |
+
"loss": 0.352,
|
354 |
+
"step": 280
|
355 |
+
},
|
356 |
+
{
|
357 |
+
"epoch": 1.43,
|
358 |
+
"learning_rate": 5.681818181818182e-06,
|
359 |
+
"loss": 0.3343,
|
360 |
+
"step": 285
|
361 |
+
},
|
362 |
+
{
|
363 |
+
"epoch": 1.46,
|
364 |
+
"learning_rate": 5.781499202551834e-06,
|
365 |
+
"loss": 0.3346,
|
366 |
+
"step": 290
|
367 |
+
},
|
368 |
+
{
|
369 |
+
"epoch": 1.48,
|
370 |
+
"learning_rate": 5.881180223285487e-06,
|
371 |
+
"loss": 0.3272,
|
372 |
+
"step": 295
|
373 |
+
},
|
374 |
+
{
|
375 |
+
"epoch": 1.51,
|
376 |
+
"learning_rate": 5.980861244019139e-06,
|
377 |
+
"loss": 0.3045,
|
378 |
+
"step": 300
|
379 |
+
},
|
380 |
+
{
|
381 |
+
"epoch": 1.53,
|
382 |
+
"learning_rate": 6.080542264752792e-06,
|
383 |
+
"loss": 0.334,
|
384 |
+
"step": 305
|
385 |
+
},
|
386 |
+
{
|
387 |
+
"epoch": 1.56,
|
388 |
+
"learning_rate": 6.180223285486444e-06,
|
389 |
+
"loss": 0.2874,
|
390 |
+
"step": 310
|
391 |
+
},
|
392 |
+
{
|
393 |
+
"epoch": 1.58,
|
394 |
+
"learning_rate": 6.2799043062200955e-06,
|
395 |
+
"loss": 0.3253,
|
396 |
+
"step": 315
|
397 |
+
},
|
398 |
+
{
|
399 |
+
"epoch": 1.61,
|
400 |
+
"learning_rate": 6.3795853269537475e-06,
|
401 |
+
"loss": 0.3083,
|
402 |
+
"step": 320
|
403 |
+
},
|
404 |
+
{
|
405 |
+
"epoch": 1.63,
|
406 |
+
"learning_rate": 6.4792663476874e-06,
|
407 |
+
"loss": 0.2564,
|
408 |
+
"step": 325
|
409 |
+
},
|
410 |
+
{
|
411 |
+
"epoch": 1.66,
|
412 |
+
"learning_rate": 6.578947368421053e-06,
|
413 |
+
"loss": 0.3236,
|
414 |
+
"step": 330
|
415 |
+
},
|
416 |
+
{
|
417 |
+
"epoch": 1.68,
|
418 |
+
"learning_rate": 6.678628389154705e-06,
|
419 |
+
"loss": 0.3053,
|
420 |
+
"step": 335
|
421 |
+
},
|
422 |
+
{
|
423 |
+
"epoch": 1.71,
|
424 |
+
"learning_rate": 6.778309409888358e-06,
|
425 |
+
"loss": 0.3211,
|
426 |
+
"step": 340
|
427 |
+
},
|
428 |
+
{
|
429 |
+
"epoch": 1.73,
|
430 |
+
"learning_rate": 6.87799043062201e-06,
|
431 |
+
"loss": 0.2978,
|
432 |
+
"step": 345
|
433 |
+
},
|
434 |
+
{
|
435 |
+
"epoch": 1.76,
|
436 |
+
"learning_rate": 6.977671451355662e-06,
|
437 |
+
"loss": 0.283,
|
438 |
+
"step": 350
|
439 |
+
},
|
440 |
+
{
|
441 |
+
"epoch": 1.78,
|
442 |
+
"learning_rate": 7.077352472089314e-06,
|
443 |
+
"loss": 0.3428,
|
444 |
+
"step": 355
|
445 |
+
},
|
446 |
+
{
|
447 |
+
"epoch": 1.81,
|
448 |
+
"learning_rate": 7.177033492822967e-06,
|
449 |
+
"loss": 0.2284,
|
450 |
+
"step": 360
|
451 |
+
},
|
452 |
+
{
|
453 |
+
"epoch": 1.83,
|
454 |
+
"learning_rate": 7.276714513556619e-06,
|
455 |
+
"loss": 0.2762,
|
456 |
+
"step": 365
|
457 |
+
},
|
458 |
+
{
|
459 |
+
"epoch": 1.86,
|
460 |
+
"learning_rate": 7.376395534290272e-06,
|
461 |
+
"loss": 0.2744,
|
462 |
+
"step": 370
|
463 |
+
},
|
464 |
+
{
|
465 |
+
"epoch": 1.88,
|
466 |
+
"learning_rate": 7.476076555023924e-06,
|
467 |
+
"loss": 0.2261,
|
468 |
+
"step": 375
|
469 |
+
},
|
470 |
+
{
|
471 |
+
"epoch": 1.91,
|
472 |
+
"learning_rate": 7.5757575757575764e-06,
|
473 |
+
"loss": 0.2657,
|
474 |
+
"step": 380
|
475 |
+
},
|
476 |
+
{
|
477 |
+
"epoch": 1.93,
|
478 |
+
"learning_rate": 7.675438596491228e-06,
|
479 |
+
"loss": 0.2806,
|
480 |
+
"step": 385
|
481 |
+
},
|
482 |
+
{
|
483 |
+
"epoch": 1.96,
|
484 |
+
"learning_rate": 7.77511961722488e-06,
|
485 |
+
"loss": 0.2418,
|
486 |
+
"step": 390
|
487 |
+
},
|
488 |
+
{
|
489 |
+
"epoch": 1.98,
|
490 |
+
"learning_rate": 7.874800637958534e-06,
|
491 |
+
"loss": 0.2084,
|
492 |
+
"step": 395
|
493 |
+
},
|
494 |
+
{
|
495 |
+
"epoch": 2.0,
|
496 |
+
"eval_accuracy": 0.9524033930254477,
|
497 |
+
"eval_loss": 0.2076038420200348,
|
498 |
+
"eval_runtime": 66.1775,
|
499 |
+
"eval_samples_per_second": 32.065,
|
500 |
+
"eval_steps_per_second": 2.01,
|
501 |
+
"step": 398
|
502 |
+
},
|
503 |
+
{
|
504 |
+
"epoch": 2.01,
|
505 |
+
"learning_rate": 7.974481658692186e-06,
|
506 |
+
"loss": 0.1872,
|
507 |
+
"step": 400
|
508 |
+
},
|
509 |
+
{
|
510 |
+
"epoch": 2.04,
|
511 |
+
"learning_rate": 8.074162679425838e-06,
|
512 |
+
"loss": 0.2704,
|
513 |
+
"step": 405
|
514 |
+
},
|
515 |
+
{
|
516 |
+
"epoch": 2.06,
|
517 |
+
"learning_rate": 8.173843700159491e-06,
|
518 |
+
"loss": 0.2772,
|
519 |
+
"step": 410
|
520 |
+
},
|
521 |
+
{
|
522 |
+
"epoch": 2.09,
|
523 |
+
"learning_rate": 8.273524720893143e-06,
|
524 |
+
"loss": 0.2041,
|
525 |
+
"step": 415
|
526 |
+
},
|
527 |
+
{
|
528 |
+
"epoch": 2.11,
|
529 |
+
"learning_rate": 8.373205741626795e-06,
|
530 |
+
"loss": 0.1876,
|
531 |
+
"step": 420
|
532 |
+
},
|
533 |
+
{
|
534 |
+
"epoch": 2.14,
|
535 |
+
"learning_rate": 8.472886762360447e-06,
|
536 |
+
"loss": 0.2421,
|
537 |
+
"step": 425
|
538 |
+
},
|
539 |
+
{
|
540 |
+
"epoch": 2.16,
|
541 |
+
"learning_rate": 8.572567783094099e-06,
|
542 |
+
"loss": 0.1965,
|
543 |
+
"step": 430
|
544 |
+
},
|
545 |
+
{
|
546 |
+
"epoch": 2.19,
|
547 |
+
"learning_rate": 8.672248803827751e-06,
|
548 |
+
"loss": 0.2317,
|
549 |
+
"step": 435
|
550 |
+
},
|
551 |
+
{
|
552 |
+
"epoch": 2.21,
|
553 |
+
"learning_rate": 8.771929824561403e-06,
|
554 |
+
"loss": 0.237,
|
555 |
+
"step": 440
|
556 |
+
},
|
557 |
+
{
|
558 |
+
"epoch": 2.24,
|
559 |
+
"learning_rate": 8.871610845295056e-06,
|
560 |
+
"loss": 0.1576,
|
561 |
+
"step": 445
|
562 |
+
},
|
563 |
+
{
|
564 |
+
"epoch": 2.26,
|
565 |
+
"learning_rate": 8.971291866028708e-06,
|
566 |
+
"loss": 0.2396,
|
567 |
+
"step": 450
|
568 |
+
},
|
569 |
+
{
|
570 |
+
"epoch": 2.29,
|
571 |
+
"learning_rate": 9.07097288676236e-06,
|
572 |
+
"loss": 0.1799,
|
573 |
+
"step": 455
|
574 |
+
},
|
575 |
+
{
|
576 |
+
"epoch": 2.31,
|
577 |
+
"learning_rate": 9.170653907496012e-06,
|
578 |
+
"loss": 0.2594,
|
579 |
+
"step": 460
|
580 |
+
},
|
581 |
+
{
|
582 |
+
"epoch": 2.34,
|
583 |
+
"learning_rate": 9.270334928229666e-06,
|
584 |
+
"loss": 0.1516,
|
585 |
+
"step": 465
|
586 |
+
},
|
587 |
+
{
|
588 |
+
"epoch": 2.36,
|
589 |
+
"learning_rate": 9.370015948963318e-06,
|
590 |
+
"loss": 0.1638,
|
591 |
+
"step": 470
|
592 |
+
},
|
593 |
+
{
|
594 |
+
"epoch": 2.39,
|
595 |
+
"learning_rate": 9.46969696969697e-06,
|
596 |
+
"loss": 0.211,
|
597 |
+
"step": 475
|
598 |
+
},
|
599 |
+
{
|
600 |
+
"epoch": 2.41,
|
601 |
+
"learning_rate": 9.569377990430622e-06,
|
602 |
+
"loss": 0.2195,
|
603 |
+
"step": 480
|
604 |
+
},
|
605 |
+
{
|
606 |
+
"epoch": 2.44,
|
607 |
+
"learning_rate": 9.669059011164275e-06,
|
608 |
+
"loss": 0.2009,
|
609 |
+
"step": 485
|
610 |
+
},
|
611 |
+
{
|
612 |
+
"epoch": 2.46,
|
613 |
+
"learning_rate": 9.768740031897927e-06,
|
614 |
+
"loss": 0.2175,
|
615 |
+
"step": 490
|
616 |
+
},
|
617 |
+
{
|
618 |
+
"epoch": 2.49,
|
619 |
+
"learning_rate": 9.868421052631579e-06,
|
620 |
+
"loss": 0.2132,
|
621 |
+
"step": 495
|
622 |
+
},
|
623 |
+
{
|
624 |
+
"epoch": 2.51,
|
625 |
+
"learning_rate": 9.968102073365233e-06,
|
626 |
+
"loss": 0.172,
|
627 |
+
"step": 500
|
628 |
+
},
|
629 |
+
{
|
630 |
+
"epoch": 2.54,
|
631 |
+
"learning_rate": 1.0067783094098885e-05,
|
632 |
+
"loss": 0.2102,
|
633 |
+
"step": 505
|
634 |
+
},
|
635 |
+
{
|
636 |
+
"epoch": 2.56,
|
637 |
+
"learning_rate": 1.0167464114832537e-05,
|
638 |
+
"loss": 0.2038,
|
639 |
+
"step": 510
|
640 |
+
},
|
641 |
+
{
|
642 |
+
"epoch": 2.59,
|
643 |
+
"learning_rate": 1.0267145135566188e-05,
|
644 |
+
"loss": 0.1683,
|
645 |
+
"step": 515
|
646 |
+
},
|
647 |
+
{
|
648 |
+
"epoch": 2.61,
|
649 |
+
"learning_rate": 1.0366826156299842e-05,
|
650 |
+
"loss": 0.2149,
|
651 |
+
"step": 520
|
652 |
+
},
|
653 |
+
{
|
654 |
+
"epoch": 2.64,
|
655 |
+
"learning_rate": 1.0466507177033494e-05,
|
656 |
+
"loss": 0.2511,
|
657 |
+
"step": 525
|
658 |
+
},
|
659 |
+
{
|
660 |
+
"epoch": 2.66,
|
661 |
+
"learning_rate": 1.0566188197767146e-05,
|
662 |
+
"loss": 0.182,
|
663 |
+
"step": 530
|
664 |
+
},
|
665 |
+
{
|
666 |
+
"epoch": 2.69,
|
667 |
+
"learning_rate": 1.0665869218500798e-05,
|
668 |
+
"loss": 0.1928,
|
669 |
+
"step": 535
|
670 |
+
},
|
671 |
+
{
|
672 |
+
"epoch": 2.71,
|
673 |
+
"learning_rate": 1.0765550239234451e-05,
|
674 |
+
"loss": 0.2304,
|
675 |
+
"step": 540
|
676 |
+
},
|
677 |
+
{
|
678 |
+
"epoch": 2.74,
|
679 |
+
"learning_rate": 1.0865231259968102e-05,
|
680 |
+
"loss": 0.2004,
|
681 |
+
"step": 545
|
682 |
+
},
|
683 |
+
{
|
684 |
+
"epoch": 2.76,
|
685 |
+
"learning_rate": 1.0964912280701754e-05,
|
686 |
+
"loss": 0.1663,
|
687 |
+
"step": 550
|
688 |
+
},
|
689 |
+
{
|
690 |
+
"epoch": 2.79,
|
691 |
+
"learning_rate": 1.1064593301435407e-05,
|
692 |
+
"loss": 0.2223,
|
693 |
+
"step": 555
|
694 |
+
},
|
695 |
+
{
|
696 |
+
"epoch": 2.81,
|
697 |
+
"learning_rate": 1.1164274322169059e-05,
|
698 |
+
"loss": 0.173,
|
699 |
+
"step": 560
|
700 |
+
},
|
701 |
+
{
|
702 |
+
"epoch": 2.84,
|
703 |
+
"learning_rate": 1.1263955342902711e-05,
|
704 |
+
"loss": 0.1882,
|
705 |
+
"step": 565
|
706 |
+
},
|
707 |
+
{
|
708 |
+
"epoch": 2.86,
|
709 |
+
"learning_rate": 1.1363636363636365e-05,
|
710 |
+
"loss": 0.2424,
|
711 |
+
"step": 570
|
712 |
+
},
|
713 |
+
{
|
714 |
+
"epoch": 2.89,
|
715 |
+
"learning_rate": 1.1463317384370017e-05,
|
716 |
+
"loss": 0.1171,
|
717 |
+
"step": 575
|
718 |
+
},
|
719 |
+
{
|
720 |
+
"epoch": 2.91,
|
721 |
+
"learning_rate": 1.1562998405103668e-05,
|
722 |
+
"loss": 0.1704,
|
723 |
+
"step": 580
|
724 |
+
},
|
725 |
+
{
|
726 |
+
"epoch": 2.94,
|
727 |
+
"learning_rate": 1.166267942583732e-05,
|
728 |
+
"loss": 0.1806,
|
729 |
+
"step": 585
|
730 |
+
},
|
731 |
+
{
|
732 |
+
"epoch": 2.96,
|
733 |
+
"learning_rate": 1.1762360446570974e-05,
|
734 |
+
"loss": 0.214,
|
735 |
+
"step": 590
|
736 |
+
},
|
737 |
+
{
|
738 |
+
"epoch": 2.99,
|
739 |
+
"learning_rate": 1.1862041467304626e-05,
|
740 |
+
"loss": 0.1274,
|
741 |
+
"step": 595
|
742 |
+
},
|
743 |
+
{
|
744 |
+
"epoch": 3.0,
|
745 |
+
"eval_accuracy": 0.9566446748350612,
|
746 |
+
"eval_loss": 0.1459183692932129,
|
747 |
+
"eval_runtime": 66.1612,
|
748 |
+
"eval_samples_per_second": 32.073,
|
749 |
+
"eval_steps_per_second": 2.01,
|
750 |
+
"step": 597
|
751 |
+
},
|
752 |
+
{
|
753 |
+
"epoch": 3.02,
|
754 |
+
"learning_rate": 1.1961722488038278e-05,
|
755 |
+
"loss": 0.2133,
|
756 |
+
"step": 600
|
757 |
+
},
|
758 |
+
{
|
759 |
+
"epoch": 3.04,
|
760 |
+
"learning_rate": 1.206140350877193e-05,
|
761 |
+
"loss": 0.181,
|
762 |
+
"step": 605
|
763 |
+
},
|
764 |
+
{
|
765 |
+
"epoch": 3.07,
|
766 |
+
"learning_rate": 1.2161084529505583e-05,
|
767 |
+
"loss": 0.1283,
|
768 |
+
"step": 610
|
769 |
+
},
|
770 |
+
{
|
771 |
+
"epoch": 3.09,
|
772 |
+
"learning_rate": 1.2260765550239235e-05,
|
773 |
+
"loss": 0.13,
|
774 |
+
"step": 615
|
775 |
+
},
|
776 |
+
{
|
777 |
+
"epoch": 3.12,
|
778 |
+
"learning_rate": 1.2360446570972887e-05,
|
779 |
+
"loss": 0.1608,
|
780 |
+
"step": 620
|
781 |
+
},
|
782 |
+
{
|
783 |
+
"epoch": 3.14,
|
784 |
+
"learning_rate": 1.246012759170654e-05,
|
785 |
+
"loss": 0.2181,
|
786 |
+
"step": 625
|
787 |
+
},
|
788 |
+
{
|
789 |
+
"epoch": 3.17,
|
790 |
+
"learning_rate": 1.2559808612440191e-05,
|
791 |
+
"loss": 0.232,
|
792 |
+
"step": 630
|
793 |
+
},
|
794 |
+
{
|
795 |
+
"epoch": 3.19,
|
796 |
+
"learning_rate": 1.2659489633173843e-05,
|
797 |
+
"loss": 0.1208,
|
798 |
+
"step": 635
|
799 |
+
},
|
800 |
+
{
|
801 |
+
"epoch": 3.22,
|
802 |
+
"learning_rate": 1.2759170653907495e-05,
|
803 |
+
"loss": 0.269,
|
804 |
+
"step": 640
|
805 |
+
},
|
806 |
+
{
|
807 |
+
"epoch": 3.24,
|
808 |
+
"learning_rate": 1.2858851674641149e-05,
|
809 |
+
"loss": 0.1324,
|
810 |
+
"step": 645
|
811 |
+
},
|
812 |
+
{
|
813 |
+
"epoch": 3.27,
|
814 |
+
"learning_rate": 1.29585326953748e-05,
|
815 |
+
"loss": 0.2421,
|
816 |
+
"step": 650
|
817 |
+
},
|
818 |
+
{
|
819 |
+
"epoch": 3.29,
|
820 |
+
"learning_rate": 1.3058213716108452e-05,
|
821 |
+
"loss": 0.1482,
|
822 |
+
"step": 655
|
823 |
+
},
|
824 |
+
{
|
825 |
+
"epoch": 3.32,
|
826 |
+
"learning_rate": 1.3157894736842106e-05,
|
827 |
+
"loss": 0.2451,
|
828 |
+
"step": 660
|
829 |
+
},
|
830 |
+
{
|
831 |
+
"epoch": 3.34,
|
832 |
+
"learning_rate": 1.3257575757575758e-05,
|
833 |
+
"loss": 0.2553,
|
834 |
+
"step": 665
|
835 |
+
},
|
836 |
+
{
|
837 |
+
"epoch": 3.37,
|
838 |
+
"learning_rate": 1.335725677830941e-05,
|
839 |
+
"loss": 0.1594,
|
840 |
+
"step": 670
|
841 |
+
},
|
842 |
+
{
|
843 |
+
"epoch": 3.39,
|
844 |
+
"learning_rate": 1.3456937799043062e-05,
|
845 |
+
"loss": 0.066,
|
846 |
+
"step": 675
|
847 |
+
},
|
848 |
+
{
|
849 |
+
"epoch": 3.42,
|
850 |
+
"learning_rate": 1.3556618819776715e-05,
|
851 |
+
"loss": 0.1135,
|
852 |
+
"step": 680
|
853 |
+
},
|
854 |
+
{
|
855 |
+
"epoch": 3.44,
|
856 |
+
"learning_rate": 1.3656299840510367e-05,
|
857 |
+
"loss": 0.2081,
|
858 |
+
"step": 685
|
859 |
+
},
|
860 |
+
{
|
861 |
+
"epoch": 3.47,
|
862 |
+
"learning_rate": 1.375598086124402e-05,
|
863 |
+
"loss": 0.1919,
|
864 |
+
"step": 690
|
865 |
+
},
|
866 |
+
{
|
867 |
+
"epoch": 3.49,
|
868 |
+
"learning_rate": 1.3855661881977671e-05,
|
869 |
+
"loss": 0.1001,
|
870 |
+
"step": 695
|
871 |
+
},
|
872 |
+
{
|
873 |
+
"epoch": 3.52,
|
874 |
+
"learning_rate": 1.3955342902711325e-05,
|
875 |
+
"loss": 0.1908,
|
876 |
+
"step": 700
|
877 |
+
},
|
878 |
+
{
|
879 |
+
"epoch": 3.54,
|
880 |
+
"learning_rate": 1.4055023923444977e-05,
|
881 |
+
"loss": 0.1767,
|
882 |
+
"step": 705
|
883 |
+
},
|
884 |
+
{
|
885 |
+
"epoch": 3.57,
|
886 |
+
"learning_rate": 1.4154704944178629e-05,
|
887 |
+
"loss": 0.1658,
|
888 |
+
"step": 710
|
889 |
+
},
|
890 |
+
{
|
891 |
+
"epoch": 3.59,
|
892 |
+
"learning_rate": 1.425438596491228e-05,
|
893 |
+
"loss": 0.1454,
|
894 |
+
"step": 715
|
895 |
+
},
|
896 |
+
{
|
897 |
+
"epoch": 3.62,
|
898 |
+
"learning_rate": 1.4354066985645934e-05,
|
899 |
+
"loss": 0.1557,
|
900 |
+
"step": 720
|
901 |
+
},
|
902 |
+
{
|
903 |
+
"epoch": 3.64,
|
904 |
+
"learning_rate": 1.4453748006379586e-05,
|
905 |
+
"loss": 0.1802,
|
906 |
+
"step": 725
|
907 |
+
},
|
908 |
+
{
|
909 |
+
"epoch": 3.67,
|
910 |
+
"learning_rate": 1.4553429027113238e-05,
|
911 |
+
"loss": 0.1028,
|
912 |
+
"step": 730
|
913 |
+
},
|
914 |
+
{
|
915 |
+
"epoch": 3.69,
|
916 |
+
"learning_rate": 1.4653110047846892e-05,
|
917 |
+
"loss": 0.1706,
|
918 |
+
"step": 735
|
919 |
+
},
|
920 |
+
{
|
921 |
+
"epoch": 3.72,
|
922 |
+
"learning_rate": 1.4752791068580543e-05,
|
923 |
+
"loss": 0.1371,
|
924 |
+
"step": 740
|
925 |
+
},
|
926 |
+
{
|
927 |
+
"epoch": 3.74,
|
928 |
+
"learning_rate": 1.4852472089314195e-05,
|
929 |
+
"loss": 0.1392,
|
930 |
+
"step": 745
|
931 |
+
},
|
932 |
+
{
|
933 |
+
"epoch": 3.77,
|
934 |
+
"learning_rate": 1.4952153110047847e-05,
|
935 |
+
"loss": 0.1314,
|
936 |
+
"step": 750
|
937 |
+
},
|
938 |
+
{
|
939 |
+
"epoch": 3.79,
|
940 |
+
"learning_rate": 1.5051834130781501e-05,
|
941 |
+
"loss": 0.2435,
|
942 |
+
"step": 755
|
943 |
+
},
|
944 |
+
{
|
945 |
+
"epoch": 3.82,
|
946 |
+
"learning_rate": 1.5151515151515153e-05,
|
947 |
+
"loss": 0.1421,
|
948 |
+
"step": 760
|
949 |
+
},
|
950 |
+
{
|
951 |
+
"epoch": 3.84,
|
952 |
+
"learning_rate": 1.5251196172248805e-05,
|
953 |
+
"loss": 0.1292,
|
954 |
+
"step": 765
|
955 |
+
},
|
956 |
+
{
|
957 |
+
"epoch": 3.87,
|
958 |
+
"learning_rate": 1.5350877192982457e-05,
|
959 |
+
"loss": 0.1207,
|
960 |
+
"step": 770
|
961 |
+
},
|
962 |
+
{
|
963 |
+
"epoch": 3.89,
|
964 |
+
"learning_rate": 1.545055821371611e-05,
|
965 |
+
"loss": 0.2106,
|
966 |
+
"step": 775
|
967 |
+
},
|
968 |
+
{
|
969 |
+
"epoch": 3.92,
|
970 |
+
"learning_rate": 1.555023923444976e-05,
|
971 |
+
"loss": 0.1644,
|
972 |
+
"step": 780
|
973 |
+
},
|
974 |
+
{
|
975 |
+
"epoch": 3.94,
|
976 |
+
"learning_rate": 1.5649920255183416e-05,
|
977 |
+
"loss": 0.1659,
|
978 |
+
"step": 785
|
979 |
+
},
|
980 |
+
{
|
981 |
+
"epoch": 3.97,
|
982 |
+
"learning_rate": 1.5749601275917068e-05,
|
983 |
+
"loss": 0.096,
|
984 |
+
"step": 790
|
985 |
+
},
|
986 |
+
{
|
987 |
+
"epoch": 3.99,
|
988 |
+
"learning_rate": 1.584928229665072e-05,
|
989 |
+
"loss": 0.1618,
|
990 |
+
"step": 795
|
991 |
+
},
|
992 |
+
{
|
993 |
+
"epoch": 4.0,
|
994 |
+
"eval_accuracy": 0.9382657869934025,
|
995 |
+
"eval_loss": 0.1533595770597458,
|
996 |
+
"eval_runtime": 66.239,
|
997 |
+
"eval_samples_per_second": 32.035,
|
998 |
+
"eval_steps_per_second": 2.008,
|
999 |
+
"step": 796
|
1000 |
+
},
|
1001 |
+
{
|
1002 |
+
"epoch": 4.02,
|
1003 |
+
"learning_rate": 1.594896331738437e-05,
|
1004 |
+
"loss": 0.1428,
|
1005 |
+
"step": 800
|
1006 |
+
},
|
1007 |
+
{
|
1008 |
+
"epoch": 4.05,
|
1009 |
+
"learning_rate": 1.6048644338118024e-05,
|
1010 |
+
"loss": 0.1488,
|
1011 |
+
"step": 805
|
1012 |
+
},
|
1013 |
+
{
|
1014 |
+
"epoch": 4.07,
|
1015 |
+
"learning_rate": 1.6148325358851675e-05,
|
1016 |
+
"loss": 0.0972,
|
1017 |
+
"step": 810
|
1018 |
+
},
|
1019 |
+
{
|
1020 |
+
"epoch": 4.1,
|
1021 |
+
"learning_rate": 1.6248006379585327e-05,
|
1022 |
+
"loss": 0.1187,
|
1023 |
+
"step": 815
|
1024 |
+
},
|
1025 |
+
{
|
1026 |
+
"epoch": 4.12,
|
1027 |
+
"learning_rate": 1.6347687400318983e-05,
|
1028 |
+
"loss": 0.0871,
|
1029 |
+
"step": 820
|
1030 |
+
},
|
1031 |
+
{
|
1032 |
+
"epoch": 4.15,
|
1033 |
+
"learning_rate": 1.6447368421052635e-05,
|
1034 |
+
"loss": 0.1333,
|
1035 |
+
"step": 825
|
1036 |
+
},
|
1037 |
+
{
|
1038 |
+
"epoch": 4.17,
|
1039 |
+
"learning_rate": 1.6547049441786287e-05,
|
1040 |
+
"loss": 0.148,
|
1041 |
+
"step": 830
|
1042 |
+
},
|
1043 |
+
{
|
1044 |
+
"epoch": 4.2,
|
1045 |
+
"learning_rate": 1.664673046251994e-05,
|
1046 |
+
"loss": 0.1552,
|
1047 |
+
"step": 835
|
1048 |
+
},
|
1049 |
+
{
|
1050 |
+
"epoch": 4.22,
|
1051 |
+
"learning_rate": 1.674641148325359e-05,
|
1052 |
+
"loss": 0.1198,
|
1053 |
+
"step": 840
|
1054 |
+
},
|
1055 |
+
{
|
1056 |
+
"epoch": 4.25,
|
1057 |
+
"learning_rate": 1.684609250398724e-05,
|
1058 |
+
"loss": 0.1764,
|
1059 |
+
"step": 845
|
1060 |
+
},
|
1061 |
+
{
|
1062 |
+
"epoch": 4.27,
|
1063 |
+
"learning_rate": 1.6945773524720894e-05,
|
1064 |
+
"loss": 0.1048,
|
1065 |
+
"step": 850
|
1066 |
+
},
|
1067 |
+
{
|
1068 |
+
"epoch": 4.3,
|
1069 |
+
"learning_rate": 1.7045454545454546e-05,
|
1070 |
+
"loss": 0.065,
|
1071 |
+
"step": 855
|
1072 |
+
},
|
1073 |
+
{
|
1074 |
+
"epoch": 4.32,
|
1075 |
+
"learning_rate": 1.7145135566188198e-05,
|
1076 |
+
"loss": 0.1051,
|
1077 |
+
"step": 860
|
1078 |
+
},
|
1079 |
+
{
|
1080 |
+
"epoch": 4.35,
|
1081 |
+
"learning_rate": 1.724481658692185e-05,
|
1082 |
+
"loss": 0.1513,
|
1083 |
+
"step": 865
|
1084 |
+
},
|
1085 |
+
{
|
1086 |
+
"epoch": 4.37,
|
1087 |
+
"learning_rate": 1.7344497607655502e-05,
|
1088 |
+
"loss": 0.173,
|
1089 |
+
"step": 870
|
1090 |
+
},
|
1091 |
+
{
|
1092 |
+
"epoch": 4.4,
|
1093 |
+
"learning_rate": 1.7444178628389154e-05,
|
1094 |
+
"loss": 0.2399,
|
1095 |
+
"step": 875
|
1096 |
+
},
|
1097 |
+
{
|
1098 |
+
"epoch": 4.42,
|
1099 |
+
"learning_rate": 1.7543859649122806e-05,
|
1100 |
+
"loss": 0.1987,
|
1101 |
+
"step": 880
|
1102 |
+
},
|
1103 |
+
{
|
1104 |
+
"epoch": 4.45,
|
1105 |
+
"learning_rate": 1.7643540669856458e-05,
|
1106 |
+
"loss": 0.16,
|
1107 |
+
"step": 885
|
1108 |
+
},
|
1109 |
+
{
|
1110 |
+
"epoch": 4.47,
|
1111 |
+
"learning_rate": 1.7743221690590113e-05,
|
1112 |
+
"loss": 0.0938,
|
1113 |
+
"step": 890
|
1114 |
+
},
|
1115 |
+
{
|
1116 |
+
"epoch": 4.5,
|
1117 |
+
"learning_rate": 1.7842902711323765e-05,
|
1118 |
+
"loss": 0.1978,
|
1119 |
+
"step": 895
|
1120 |
+
},
|
1121 |
+
{
|
1122 |
+
"epoch": 4.52,
|
1123 |
+
"learning_rate": 1.7942583732057417e-05,
|
1124 |
+
"loss": 0.1199,
|
1125 |
+
"step": 900
|
1126 |
+
},
|
1127 |
+
{
|
1128 |
+
"epoch": 4.55,
|
1129 |
+
"learning_rate": 1.804226475279107e-05,
|
1130 |
+
"loss": 0.1267,
|
1131 |
+
"step": 905
|
1132 |
+
},
|
1133 |
+
{
|
1134 |
+
"epoch": 4.57,
|
1135 |
+
"learning_rate": 1.814194577352472e-05,
|
1136 |
+
"loss": 0.0988,
|
1137 |
+
"step": 910
|
1138 |
+
},
|
1139 |
+
{
|
1140 |
+
"epoch": 4.6,
|
1141 |
+
"learning_rate": 1.8241626794258373e-05,
|
1142 |
+
"loss": 0.2077,
|
1143 |
+
"step": 915
|
1144 |
+
},
|
1145 |
+
{
|
1146 |
+
"epoch": 4.62,
|
1147 |
+
"learning_rate": 1.8341307814992024e-05,
|
1148 |
+
"loss": 0.0723,
|
1149 |
+
"step": 920
|
1150 |
+
},
|
1151 |
+
{
|
1152 |
+
"epoch": 4.65,
|
1153 |
+
"learning_rate": 1.844098883572568e-05,
|
1154 |
+
"loss": 0.107,
|
1155 |
+
"step": 925
|
1156 |
+
},
|
1157 |
+
{
|
1158 |
+
"epoch": 4.67,
|
1159 |
+
"learning_rate": 1.8540669856459332e-05,
|
1160 |
+
"loss": 0.0574,
|
1161 |
+
"step": 930
|
1162 |
+
},
|
1163 |
+
{
|
1164 |
+
"epoch": 4.7,
|
1165 |
+
"learning_rate": 1.8640350877192984e-05,
|
1166 |
+
"loss": 0.1195,
|
1167 |
+
"step": 935
|
1168 |
+
},
|
1169 |
+
{
|
1170 |
+
"epoch": 4.72,
|
1171 |
+
"learning_rate": 1.8740031897926636e-05,
|
1172 |
+
"loss": 0.0668,
|
1173 |
+
"step": 940
|
1174 |
+
},
|
1175 |
+
{
|
1176 |
+
"epoch": 4.75,
|
1177 |
+
"learning_rate": 1.8839712918660287e-05,
|
1178 |
+
"loss": 0.1729,
|
1179 |
+
"step": 945
|
1180 |
+
},
|
1181 |
+
{
|
1182 |
+
"epoch": 4.77,
|
1183 |
+
"learning_rate": 1.893939393939394e-05,
|
1184 |
+
"loss": 0.1988,
|
1185 |
+
"step": 950
|
1186 |
+
},
|
1187 |
+
{
|
1188 |
+
"epoch": 4.8,
|
1189 |
+
"learning_rate": 1.903907496012759e-05,
|
1190 |
+
"loss": 0.1572,
|
1191 |
+
"step": 955
|
1192 |
+
},
|
1193 |
+
{
|
1194 |
+
"epoch": 4.82,
|
1195 |
+
"learning_rate": 1.9138755980861243e-05,
|
1196 |
+
"loss": 0.1638,
|
1197 |
+
"step": 960
|
1198 |
+
},
|
1199 |
+
{
|
1200 |
+
"epoch": 4.85,
|
1201 |
+
"learning_rate": 1.92384370015949e-05,
|
1202 |
+
"loss": 0.0853,
|
1203 |
+
"step": 965
|
1204 |
+
},
|
1205 |
+
{
|
1206 |
+
"epoch": 4.87,
|
1207 |
+
"learning_rate": 1.933811802232855e-05,
|
1208 |
+
"loss": 0.1221,
|
1209 |
+
"step": 970
|
1210 |
+
},
|
1211 |
+
{
|
1212 |
+
"epoch": 4.9,
|
1213 |
+
"learning_rate": 1.9437799043062202e-05,
|
1214 |
+
"loss": 0.2462,
|
1215 |
+
"step": 975
|
1216 |
+
},
|
1217 |
+
{
|
1218 |
+
"epoch": 4.92,
|
1219 |
+
"learning_rate": 1.9537480063795854e-05,
|
1220 |
+
"loss": 0.3388,
|
1221 |
+
"step": 980
|
1222 |
+
},
|
1223 |
+
{
|
1224 |
+
"epoch": 4.95,
|
1225 |
+
"learning_rate": 1.9637161084529506e-05,
|
1226 |
+
"loss": 0.1064,
|
1227 |
+
"step": 985
|
1228 |
+
},
|
1229 |
+
{
|
1230 |
+
"epoch": 4.97,
|
1231 |
+
"learning_rate": 1.9736842105263158e-05,
|
1232 |
+
"loss": 0.0871,
|
1233 |
+
"step": 990
|
1234 |
+
},
|
1235 |
+
{
|
1236 |
+
"epoch": 5.0,
|
1237 |
+
"learning_rate": 1.983652312599681e-05,
|
1238 |
+
"loss": 0.2118,
|
1239 |
+
"step": 995
|
1240 |
+
},
|
1241 |
+
{
|
1242 |
+
"epoch": 5.0,
|
1243 |
+
"eval_accuracy": 0.9726672950047125,
|
1244 |
+
"eval_loss": 0.08765223622322083,
|
1245 |
+
"eval_runtime": 66.1349,
|
1246 |
+
"eval_samples_per_second": 32.086,
|
1247 |
+
"eval_steps_per_second": 2.011,
|
1248 |
+
"step": 995
|
1249 |
+
},
|
1250 |
+
{
|
1251 |
+
"epoch": 5.03,
|
1252 |
+
"learning_rate": 1.9936204146730465e-05,
|
1253 |
+
"loss": 0.221,
|
1254 |
+
"step": 1000
|
1255 |
+
},
|
1256 |
+
{
|
1257 |
+
"epoch": 5.05,
|
1258 |
+
"learning_rate": 2.0035885167464117e-05,
|
1259 |
+
"loss": 0.1202,
|
1260 |
+
"step": 1005
|
1261 |
+
},
|
1262 |
+
{
|
1263 |
+
"epoch": 5.08,
|
1264 |
+
"learning_rate": 2.013556618819777e-05,
|
1265 |
+
"loss": 0.2251,
|
1266 |
+
"step": 1010
|
1267 |
+
},
|
1268 |
+
{
|
1269 |
+
"epoch": 5.1,
|
1270 |
+
"learning_rate": 2.023524720893142e-05,
|
1271 |
+
"loss": 0.0988,
|
1272 |
+
"step": 1015
|
1273 |
+
},
|
1274 |
+
{
|
1275 |
+
"epoch": 5.13,
|
1276 |
+
"learning_rate": 2.0334928229665073e-05,
|
1277 |
+
"loss": 0.1135,
|
1278 |
+
"step": 1020
|
1279 |
+
},
|
1280 |
+
{
|
1281 |
+
"epoch": 5.15,
|
1282 |
+
"learning_rate": 2.0434609250398725e-05,
|
1283 |
+
"loss": 0.312,
|
1284 |
+
"step": 1025
|
1285 |
+
},
|
1286 |
+
{
|
1287 |
+
"epoch": 5.18,
|
1288 |
+
"learning_rate": 2.0534290271132377e-05,
|
1289 |
+
"loss": 0.1456,
|
1290 |
+
"step": 1030
|
1291 |
+
},
|
1292 |
+
{
|
1293 |
+
"epoch": 5.2,
|
1294 |
+
"learning_rate": 2.0633971291866032e-05,
|
1295 |
+
"loss": 0.0716,
|
1296 |
+
"step": 1035
|
1297 |
+
},
|
1298 |
+
{
|
1299 |
+
"epoch": 5.23,
|
1300 |
+
"learning_rate": 2.0733652312599684e-05,
|
1301 |
+
"loss": 0.1508,
|
1302 |
+
"step": 1040
|
1303 |
+
},
|
1304 |
+
{
|
1305 |
+
"epoch": 5.25,
|
1306 |
+
"learning_rate": 2.0833333333333336e-05,
|
1307 |
+
"loss": 0.2162,
|
1308 |
+
"step": 1045
|
1309 |
+
},
|
1310 |
+
{
|
1311 |
+
"epoch": 5.28,
|
1312 |
+
"learning_rate": 2.0933014354066988e-05,
|
1313 |
+
"loss": 0.2184,
|
1314 |
+
"step": 1050
|
1315 |
+
},
|
1316 |
+
{
|
1317 |
+
"epoch": 5.3,
|
1318 |
+
"learning_rate": 2.103269537480064e-05,
|
1319 |
+
"loss": 0.1873,
|
1320 |
+
"step": 1055
|
1321 |
+
},
|
1322 |
+
{
|
1323 |
+
"epoch": 5.33,
|
1324 |
+
"learning_rate": 2.1132376395534292e-05,
|
1325 |
+
"loss": 0.1089,
|
1326 |
+
"step": 1060
|
1327 |
+
},
|
1328 |
+
{
|
1329 |
+
"epoch": 5.35,
|
1330 |
+
"learning_rate": 2.1232057416267944e-05,
|
1331 |
+
"loss": 0.1074,
|
1332 |
+
"step": 1065
|
1333 |
+
},
|
1334 |
+
{
|
1335 |
+
"epoch": 5.38,
|
1336 |
+
"learning_rate": 2.1331738437001596e-05,
|
1337 |
+
"loss": 0.1286,
|
1338 |
+
"step": 1070
|
1339 |
+
},
|
1340 |
+
{
|
1341 |
+
"epoch": 5.4,
|
1342 |
+
"learning_rate": 2.143141945773525e-05,
|
1343 |
+
"loss": 0.1686,
|
1344 |
+
"step": 1075
|
1345 |
+
},
|
1346 |
+
{
|
1347 |
+
"epoch": 5.43,
|
1348 |
+
"learning_rate": 2.1531100478468903e-05,
|
1349 |
+
"loss": 0.1312,
|
1350 |
+
"step": 1080
|
1351 |
+
},
|
1352 |
+
{
|
1353 |
+
"epoch": 5.45,
|
1354 |
+
"learning_rate": 2.1630781499202555e-05,
|
1355 |
+
"loss": 0.1211,
|
1356 |
+
"step": 1085
|
1357 |
+
},
|
1358 |
+
{
|
1359 |
+
"epoch": 5.48,
|
1360 |
+
"learning_rate": 2.1730462519936203e-05,
|
1361 |
+
"loss": 0.0526,
|
1362 |
+
"step": 1090
|
1363 |
+
},
|
1364 |
+
{
|
1365 |
+
"epoch": 5.5,
|
1366 |
+
"learning_rate": 2.1830143540669855e-05,
|
1367 |
+
"loss": 0.0628,
|
1368 |
+
"step": 1095
|
1369 |
+
},
|
1370 |
+
{
|
1371 |
+
"epoch": 5.53,
|
1372 |
+
"learning_rate": 2.1929824561403507e-05,
|
1373 |
+
"loss": 0.0398,
|
1374 |
+
"step": 1100
|
1375 |
+
},
|
1376 |
+
{
|
1377 |
+
"epoch": 5.55,
|
1378 |
+
"learning_rate": 2.2029505582137162e-05,
|
1379 |
+
"loss": 0.2401,
|
1380 |
+
"step": 1105
|
1381 |
+
},
|
1382 |
+
{
|
1383 |
+
"epoch": 5.58,
|
1384 |
+
"learning_rate": 2.2129186602870814e-05,
|
1385 |
+
"loss": 0.1126,
|
1386 |
+
"step": 1110
|
1387 |
+
},
|
1388 |
+
{
|
1389 |
+
"epoch": 5.6,
|
1390 |
+
"learning_rate": 2.2228867623604466e-05,
|
1391 |
+
"loss": 0.1391,
|
1392 |
+
"step": 1115
|
1393 |
+
},
|
1394 |
+
{
|
1395 |
+
"epoch": 5.63,
|
1396 |
+
"learning_rate": 2.2328548644338118e-05,
|
1397 |
+
"loss": 0.1835,
|
1398 |
+
"step": 1120
|
1399 |
+
},
|
1400 |
+
{
|
1401 |
+
"epoch": 5.65,
|
1402 |
+
"learning_rate": 2.242822966507177e-05,
|
1403 |
+
"loss": 0.2572,
|
1404 |
+
"step": 1125
|
1405 |
+
},
|
1406 |
+
{
|
1407 |
+
"epoch": 5.68,
|
1408 |
+
"learning_rate": 2.2527910685805422e-05,
|
1409 |
+
"loss": 0.1638,
|
1410 |
+
"step": 1130
|
1411 |
+
},
|
1412 |
+
{
|
1413 |
+
"epoch": 5.7,
|
1414 |
+
"learning_rate": 2.2627591706539074e-05,
|
1415 |
+
"loss": 0.1415,
|
1416 |
+
"step": 1135
|
1417 |
+
},
|
1418 |
+
{
|
1419 |
+
"epoch": 5.73,
|
1420 |
+
"learning_rate": 2.272727272727273e-05,
|
1421 |
+
"loss": 0.1298,
|
1422 |
+
"step": 1140
|
1423 |
+
},
|
1424 |
+
{
|
1425 |
+
"epoch": 5.75,
|
1426 |
+
"learning_rate": 2.282695374800638e-05,
|
1427 |
+
"loss": 0.102,
|
1428 |
+
"step": 1145
|
1429 |
+
},
|
1430 |
+
{
|
1431 |
+
"epoch": 5.78,
|
1432 |
+
"learning_rate": 2.2926634768740033e-05,
|
1433 |
+
"loss": 0.1355,
|
1434 |
+
"step": 1150
|
1435 |
+
},
|
1436 |
+
{
|
1437 |
+
"epoch": 5.8,
|
1438 |
+
"learning_rate": 2.3026315789473685e-05,
|
1439 |
+
"loss": 0.2175,
|
1440 |
+
"step": 1155
|
1441 |
+
},
|
1442 |
+
{
|
1443 |
+
"epoch": 5.83,
|
1444 |
+
"learning_rate": 2.3125996810207337e-05,
|
1445 |
+
"loss": 0.1042,
|
1446 |
+
"step": 1160
|
1447 |
+
},
|
1448 |
+
{
|
1449 |
+
"epoch": 5.85,
|
1450 |
+
"learning_rate": 2.322567783094099e-05,
|
1451 |
+
"loss": 0.1658,
|
1452 |
+
"step": 1165
|
1453 |
+
},
|
1454 |
+
{
|
1455 |
+
"epoch": 5.88,
|
1456 |
+
"learning_rate": 2.332535885167464e-05,
|
1457 |
+
"loss": 0.1317,
|
1458 |
+
"step": 1170
|
1459 |
+
},
|
1460 |
+
{
|
1461 |
+
"epoch": 5.9,
|
1462 |
+
"learning_rate": 2.3425039872408293e-05,
|
1463 |
+
"loss": 0.1368,
|
1464 |
+
"step": 1175
|
1465 |
+
},
|
1466 |
+
{
|
1467 |
+
"epoch": 5.93,
|
1468 |
+
"learning_rate": 2.3524720893141948e-05,
|
1469 |
+
"loss": 0.0984,
|
1470 |
+
"step": 1180
|
1471 |
+
},
|
1472 |
+
{
|
1473 |
+
"epoch": 5.95,
|
1474 |
+
"learning_rate": 2.36244019138756e-05,
|
1475 |
+
"loss": 0.1546,
|
1476 |
+
"step": 1185
|
1477 |
+
},
|
1478 |
+
{
|
1479 |
+
"epoch": 5.98,
|
1480 |
+
"learning_rate": 2.3724082934609252e-05,
|
1481 |
+
"loss": 0.0306,
|
1482 |
+
"step": 1190
|
1483 |
+
},
|
1484 |
+
{
|
1485 |
+
"epoch": 6.0,
|
1486 |
+
"eval_accuracy": 0.96559849198869,
|
1487 |
+
"eval_loss": 0.10479557514190674,
|
1488 |
+
"eval_runtime": 66.2517,
|
1489 |
+
"eval_samples_per_second": 32.029,
|
1490 |
+
"eval_steps_per_second": 2.007,
|
1491 |
+
"step": 1194
|
1492 |
+
},
|
1493 |
+
{
|
1494 |
+
"epoch": 6.01,
|
1495 |
+
"learning_rate": 2.3823763955342904e-05,
|
1496 |
+
"loss": 0.2805,
|
1497 |
+
"step": 1195
|
1498 |
+
},
|
1499 |
+
{
|
1500 |
+
"epoch": 6.03,
|
1501 |
+
"learning_rate": 2.3923444976076556e-05,
|
1502 |
+
"loss": 0.1948,
|
1503 |
+
"step": 1200
|
1504 |
+
},
|
1505 |
+
{
|
1506 |
+
"epoch": 6.06,
|
1507 |
+
"learning_rate": 2.4023125996810208e-05,
|
1508 |
+
"loss": 0.1294,
|
1509 |
+
"step": 1205
|
1510 |
+
},
|
1511 |
+
{
|
1512 |
+
"epoch": 6.08,
|
1513 |
+
"learning_rate": 2.412280701754386e-05,
|
1514 |
+
"loss": 0.131,
|
1515 |
+
"step": 1210
|
1516 |
+
},
|
1517 |
+
{
|
1518 |
+
"epoch": 6.11,
|
1519 |
+
"learning_rate": 2.4222488038277515e-05,
|
1520 |
+
"loss": 0.0771,
|
1521 |
+
"step": 1215
|
1522 |
+
},
|
1523 |
+
{
|
1524 |
+
"epoch": 6.13,
|
1525 |
+
"learning_rate": 2.4322169059011167e-05,
|
1526 |
+
"loss": 0.056,
|
1527 |
+
"step": 1220
|
1528 |
+
},
|
1529 |
+
{
|
1530 |
+
"epoch": 6.16,
|
1531 |
+
"learning_rate": 2.442185007974482e-05,
|
1532 |
+
"loss": 0.0424,
|
1533 |
+
"step": 1225
|
1534 |
+
},
|
1535 |
+
{
|
1536 |
+
"epoch": 6.18,
|
1537 |
+
"learning_rate": 2.452153110047847e-05,
|
1538 |
+
"loss": 0.1259,
|
1539 |
+
"step": 1230
|
1540 |
+
},
|
1541 |
+
{
|
1542 |
+
"epoch": 6.21,
|
1543 |
+
"learning_rate": 2.4621212121212123e-05,
|
1544 |
+
"loss": 0.1575,
|
1545 |
+
"step": 1235
|
1546 |
+
},
|
1547 |
+
{
|
1548 |
+
"epoch": 6.23,
|
1549 |
+
"learning_rate": 2.4720893141945774e-05,
|
1550 |
+
"loss": 0.1067,
|
1551 |
+
"step": 1240
|
1552 |
+
},
|
1553 |
+
{
|
1554 |
+
"epoch": 6.26,
|
1555 |
+
"learning_rate": 2.4820574162679426e-05,
|
1556 |
+
"loss": 0.0743,
|
1557 |
+
"step": 1245
|
1558 |
+
},
|
1559 |
+
{
|
1560 |
+
"epoch": 6.28,
|
1561 |
+
"learning_rate": 2.492025518341308e-05,
|
1562 |
+
"loss": 0.1293,
|
1563 |
+
"step": 1250
|
1564 |
+
},
|
1565 |
+
{
|
1566 |
+
"epoch": 6.31,
|
1567 |
+
"learning_rate": 2.5019936204146734e-05,
|
1568 |
+
"loss": 0.0878,
|
1569 |
+
"step": 1255
|
1570 |
+
},
|
1571 |
+
{
|
1572 |
+
"epoch": 6.33,
|
1573 |
+
"learning_rate": 2.5119617224880382e-05,
|
1574 |
+
"loss": 0.0561,
|
1575 |
+
"step": 1260
|
1576 |
+
},
|
1577 |
+
{
|
1578 |
+
"epoch": 6.36,
|
1579 |
+
"learning_rate": 2.5219298245614037e-05,
|
1580 |
+
"loss": 0.1468,
|
1581 |
+
"step": 1265
|
1582 |
+
},
|
1583 |
+
{
|
1584 |
+
"epoch": 6.38,
|
1585 |
+
"learning_rate": 2.5318979266347686e-05,
|
1586 |
+
"loss": 0.1016,
|
1587 |
+
"step": 1270
|
1588 |
+
},
|
1589 |
+
{
|
1590 |
+
"epoch": 6.41,
|
1591 |
+
"learning_rate": 2.541866028708134e-05,
|
1592 |
+
"loss": 0.1267,
|
1593 |
+
"step": 1275
|
1594 |
+
},
|
1595 |
+
{
|
1596 |
+
"epoch": 6.43,
|
1597 |
+
"learning_rate": 2.551834130781499e-05,
|
1598 |
+
"loss": 0.0815,
|
1599 |
+
"step": 1280
|
1600 |
+
},
|
1601 |
+
{
|
1602 |
+
"epoch": 6.46,
|
1603 |
+
"learning_rate": 2.5618022328548645e-05,
|
1604 |
+
"loss": 0.1103,
|
1605 |
+
"step": 1285
|
1606 |
+
},
|
1607 |
+
{
|
1608 |
+
"epoch": 6.48,
|
1609 |
+
"learning_rate": 2.5717703349282297e-05,
|
1610 |
+
"loss": 0.0895,
|
1611 |
+
"step": 1290
|
1612 |
+
},
|
1613 |
+
{
|
1614 |
+
"epoch": 6.51,
|
1615 |
+
"learning_rate": 2.5817384370015952e-05,
|
1616 |
+
"loss": 0.2504,
|
1617 |
+
"step": 1295
|
1618 |
+
},
|
1619 |
+
{
|
1620 |
+
"epoch": 6.53,
|
1621 |
+
"learning_rate": 2.59170653907496e-05,
|
1622 |
+
"loss": 0.078,
|
1623 |
+
"step": 1300
|
1624 |
+
},
|
1625 |
+
{
|
1626 |
+
"epoch": 6.56,
|
1627 |
+
"learning_rate": 2.6016746411483256e-05,
|
1628 |
+
"loss": 0.0883,
|
1629 |
+
"step": 1305
|
1630 |
+
},
|
1631 |
+
{
|
1632 |
+
"epoch": 6.58,
|
1633 |
+
"learning_rate": 2.6116427432216905e-05,
|
1634 |
+
"loss": 0.1765,
|
1635 |
+
"step": 1310
|
1636 |
+
},
|
1637 |
+
{
|
1638 |
+
"epoch": 6.61,
|
1639 |
+
"learning_rate": 2.621610845295056e-05,
|
1640 |
+
"loss": 0.0785,
|
1641 |
+
"step": 1315
|
1642 |
+
},
|
1643 |
+
{
|
1644 |
+
"epoch": 6.63,
|
1645 |
+
"learning_rate": 2.6315789473684212e-05,
|
1646 |
+
"loss": 0.0979,
|
1647 |
+
"step": 1320
|
1648 |
+
},
|
1649 |
+
{
|
1650 |
+
"epoch": 6.66,
|
1651 |
+
"learning_rate": 2.6415470494417864e-05,
|
1652 |
+
"loss": 0.0959,
|
1653 |
+
"step": 1325
|
1654 |
+
},
|
1655 |
+
{
|
1656 |
+
"epoch": 6.68,
|
1657 |
+
"learning_rate": 2.6515151515151516e-05,
|
1658 |
+
"loss": 0.0935,
|
1659 |
+
"step": 1330
|
1660 |
+
},
|
1661 |
+
{
|
1662 |
+
"epoch": 6.71,
|
1663 |
+
"learning_rate": 2.661483253588517e-05,
|
1664 |
+
"loss": 0.0987,
|
1665 |
+
"step": 1335
|
1666 |
+
},
|
1667 |
+
{
|
1668 |
+
"epoch": 6.73,
|
1669 |
+
"learning_rate": 2.671451355661882e-05,
|
1670 |
+
"loss": 0.1285,
|
1671 |
+
"step": 1340
|
1672 |
+
},
|
1673 |
+
{
|
1674 |
+
"epoch": 6.76,
|
1675 |
+
"learning_rate": 2.6814194577352475e-05,
|
1676 |
+
"loss": 0.0481,
|
1677 |
+
"step": 1345
|
1678 |
+
},
|
1679 |
+
{
|
1680 |
+
"epoch": 6.78,
|
1681 |
+
"learning_rate": 2.6913875598086123e-05,
|
1682 |
+
"loss": 0.0909,
|
1683 |
+
"step": 1350
|
1684 |
+
},
|
1685 |
+
{
|
1686 |
+
"epoch": 6.81,
|
1687 |
+
"learning_rate": 2.701355661881978e-05,
|
1688 |
+
"loss": 0.0919,
|
1689 |
+
"step": 1355
|
1690 |
+
},
|
1691 |
+
{
|
1692 |
+
"epoch": 6.83,
|
1693 |
+
"learning_rate": 2.711323763955343e-05,
|
1694 |
+
"loss": 0.1335,
|
1695 |
+
"step": 1360
|
1696 |
+
},
|
1697 |
+
{
|
1698 |
+
"epoch": 6.86,
|
1699 |
+
"learning_rate": 2.7212918660287086e-05,
|
1700 |
+
"loss": 0.1727,
|
1701 |
+
"step": 1365
|
1702 |
+
},
|
1703 |
+
{
|
1704 |
+
"epoch": 6.88,
|
1705 |
+
"learning_rate": 2.7312599681020735e-05,
|
1706 |
+
"loss": 0.1956,
|
1707 |
+
"step": 1370
|
1708 |
+
},
|
1709 |
+
{
|
1710 |
+
"epoch": 6.91,
|
1711 |
+
"learning_rate": 2.741228070175439e-05,
|
1712 |
+
"loss": 0.0489,
|
1713 |
+
"step": 1375
|
1714 |
+
},
|
1715 |
+
{
|
1716 |
+
"epoch": 6.93,
|
1717 |
+
"learning_rate": 2.751196172248804e-05,
|
1718 |
+
"loss": 0.0656,
|
1719 |
+
"step": 1380
|
1720 |
+
},
|
1721 |
+
{
|
1722 |
+
"epoch": 6.96,
|
1723 |
+
"learning_rate": 2.7611642743221694e-05,
|
1724 |
+
"loss": 0.0851,
|
1725 |
+
"step": 1385
|
1726 |
+
},
|
1727 |
+
{
|
1728 |
+
"epoch": 6.98,
|
1729 |
+
"learning_rate": 2.7711323763955342e-05,
|
1730 |
+
"loss": 0.1012,
|
1731 |
+
"step": 1390
|
1732 |
+
},
|
1733 |
+
{
|
1734 |
+
"epoch": 7.0,
|
1735 |
+
"eval_accuracy": 0.9754948162111216,
|
1736 |
+
"eval_loss": 0.06740697473287582,
|
1737 |
+
"eval_runtime": 66.4497,
|
1738 |
+
"eval_samples_per_second": 31.934,
|
1739 |
+
"eval_steps_per_second": 2.002,
|
1740 |
+
"step": 1393
|
1741 |
+
},
|
1742 |
+
{
|
1743 |
+
"epoch": 7.01,
|
1744 |
+
"learning_rate": 2.7811004784688998e-05,
|
1745 |
+
"loss": 0.1534,
|
1746 |
+
"step": 1395
|
1747 |
+
},
|
1748 |
+
{
|
1749 |
+
"epoch": 7.04,
|
1750 |
+
"learning_rate": 2.791068580542265e-05,
|
1751 |
+
"loss": 0.0807,
|
1752 |
+
"step": 1400
|
1753 |
+
},
|
1754 |
+
{
|
1755 |
+
"epoch": 7.06,
|
1756 |
+
"learning_rate": 2.8010366826156305e-05,
|
1757 |
+
"loss": 0.0988,
|
1758 |
+
"step": 1405
|
1759 |
+
},
|
1760 |
+
{
|
1761 |
+
"epoch": 7.09,
|
1762 |
+
"learning_rate": 2.8110047846889953e-05,
|
1763 |
+
"loss": 0.1152,
|
1764 |
+
"step": 1410
|
1765 |
+
},
|
1766 |
+
{
|
1767 |
+
"epoch": 7.11,
|
1768 |
+
"learning_rate": 2.820972886762361e-05,
|
1769 |
+
"loss": 0.2121,
|
1770 |
+
"step": 1415
|
1771 |
+
},
|
1772 |
+
{
|
1773 |
+
"epoch": 7.14,
|
1774 |
+
"learning_rate": 2.8309409888357257e-05,
|
1775 |
+
"loss": 0.0698,
|
1776 |
+
"step": 1420
|
1777 |
+
},
|
1778 |
+
{
|
1779 |
+
"epoch": 7.16,
|
1780 |
+
"learning_rate": 2.8409090909090912e-05,
|
1781 |
+
"loss": 0.146,
|
1782 |
+
"step": 1425
|
1783 |
+
},
|
1784 |
+
{
|
1785 |
+
"epoch": 7.19,
|
1786 |
+
"learning_rate": 2.850877192982456e-05,
|
1787 |
+
"loss": 0.0627,
|
1788 |
+
"step": 1430
|
1789 |
+
},
|
1790 |
+
{
|
1791 |
+
"epoch": 7.21,
|
1792 |
+
"learning_rate": 2.8608452950558216e-05,
|
1793 |
+
"loss": 0.0741,
|
1794 |
+
"step": 1435
|
1795 |
+
},
|
1796 |
+
{
|
1797 |
+
"epoch": 7.24,
|
1798 |
+
"learning_rate": 2.8708133971291868e-05,
|
1799 |
+
"loss": 0.1172,
|
1800 |
+
"step": 1440
|
1801 |
+
},
|
1802 |
+
{
|
1803 |
+
"epoch": 7.26,
|
1804 |
+
"learning_rate": 2.8807814992025524e-05,
|
1805 |
+
"loss": 0.0642,
|
1806 |
+
"step": 1445
|
1807 |
+
},
|
1808 |
+
{
|
1809 |
+
"epoch": 7.29,
|
1810 |
+
"learning_rate": 2.8907496012759172e-05,
|
1811 |
+
"loss": 0.0695,
|
1812 |
+
"step": 1450
|
1813 |
+
},
|
1814 |
+
{
|
1815 |
+
"epoch": 7.31,
|
1816 |
+
"learning_rate": 2.900717703349282e-05,
|
1817 |
+
"loss": 0.0517,
|
1818 |
+
"step": 1455
|
1819 |
+
},
|
1820 |
+
{
|
1821 |
+
"epoch": 7.34,
|
1822 |
+
"learning_rate": 2.9106858054226476e-05,
|
1823 |
+
"loss": 0.1438,
|
1824 |
+
"step": 1460
|
1825 |
+
},
|
1826 |
+
{
|
1827 |
+
"epoch": 7.36,
|
1828 |
+
"learning_rate": 2.9206539074960128e-05,
|
1829 |
+
"loss": 0.1856,
|
1830 |
+
"step": 1465
|
1831 |
+
},
|
1832 |
+
{
|
1833 |
+
"epoch": 7.39,
|
1834 |
+
"learning_rate": 2.9306220095693783e-05,
|
1835 |
+
"loss": 0.1129,
|
1836 |
+
"step": 1470
|
1837 |
+
},
|
1838 |
+
{
|
1839 |
+
"epoch": 7.41,
|
1840 |
+
"learning_rate": 2.940590111642743e-05,
|
1841 |
+
"loss": 0.1274,
|
1842 |
+
"step": 1475
|
1843 |
+
},
|
1844 |
+
{
|
1845 |
+
"epoch": 7.44,
|
1846 |
+
"learning_rate": 2.9505582137161087e-05,
|
1847 |
+
"loss": 0.066,
|
1848 |
+
"step": 1480
|
1849 |
+
},
|
1850 |
+
{
|
1851 |
+
"epoch": 7.46,
|
1852 |
+
"learning_rate": 2.9605263157894735e-05,
|
1853 |
+
"loss": 0.0464,
|
1854 |
+
"step": 1485
|
1855 |
+
},
|
1856 |
+
{
|
1857 |
+
"epoch": 7.49,
|
1858 |
+
"learning_rate": 2.970494417862839e-05,
|
1859 |
+
"loss": 0.0444,
|
1860 |
+
"step": 1490
|
1861 |
+
},
|
1862 |
+
{
|
1863 |
+
"epoch": 7.51,
|
1864 |
+
"learning_rate": 2.980462519936204e-05,
|
1865 |
+
"loss": 0.1212,
|
1866 |
+
"step": 1495
|
1867 |
+
},
|
1868 |
+
{
|
1869 |
+
"epoch": 7.54,
|
1870 |
+
"learning_rate": 2.9904306220095695e-05,
|
1871 |
+
"loss": 0.0917,
|
1872 |
+
"step": 1500
|
1873 |
+
},
|
1874 |
+
{
|
1875 |
+
"epoch": 7.56,
|
1876 |
+
"learning_rate": 3.0003987240829347e-05,
|
1877 |
+
"loss": 0.0964,
|
1878 |
+
"step": 1505
|
1879 |
+
},
|
1880 |
+
{
|
1881 |
+
"epoch": 7.59,
|
1882 |
+
"learning_rate": 3.0103668261563002e-05,
|
1883 |
+
"loss": 0.1538,
|
1884 |
+
"step": 1510
|
1885 |
+
},
|
1886 |
+
{
|
1887 |
+
"epoch": 7.61,
|
1888 |
+
"learning_rate": 3.020334928229665e-05,
|
1889 |
+
"loss": 0.2752,
|
1890 |
+
"step": 1515
|
1891 |
+
},
|
1892 |
+
{
|
1893 |
+
"epoch": 7.64,
|
1894 |
+
"learning_rate": 3.0303030303030306e-05,
|
1895 |
+
"loss": 0.1428,
|
1896 |
+
"step": 1520
|
1897 |
+
},
|
1898 |
+
{
|
1899 |
+
"epoch": 7.66,
|
1900 |
+
"learning_rate": 3.0402711323763954e-05,
|
1901 |
+
"loss": 0.0793,
|
1902 |
+
"step": 1525
|
1903 |
+
},
|
1904 |
+
{
|
1905 |
+
"epoch": 7.69,
|
1906 |
+
"learning_rate": 3.050239234449761e-05,
|
1907 |
+
"loss": 0.14,
|
1908 |
+
"step": 1530
|
1909 |
+
},
|
1910 |
+
{
|
1911 |
+
"epoch": 7.71,
|
1912 |
+
"learning_rate": 3.060207336523126e-05,
|
1913 |
+
"loss": 0.0503,
|
1914 |
+
"step": 1535
|
1915 |
+
},
|
1916 |
+
{
|
1917 |
+
"epoch": 7.74,
|
1918 |
+
"learning_rate": 3.0701754385964913e-05,
|
1919 |
+
"loss": 0.0705,
|
1920 |
+
"step": 1540
|
1921 |
+
},
|
1922 |
+
{
|
1923 |
+
"epoch": 7.76,
|
1924 |
+
"learning_rate": 3.080143540669856e-05,
|
1925 |
+
"loss": 0.0764,
|
1926 |
+
"step": 1545
|
1927 |
+
},
|
1928 |
+
{
|
1929 |
+
"epoch": 7.79,
|
1930 |
+
"learning_rate": 3.090111642743222e-05,
|
1931 |
+
"loss": 0.0533,
|
1932 |
+
"step": 1550
|
1933 |
+
},
|
1934 |
+
{
|
1935 |
+
"epoch": 7.81,
|
1936 |
+
"learning_rate": 3.100079744816587e-05,
|
1937 |
+
"loss": 0.234,
|
1938 |
+
"step": 1555
|
1939 |
+
},
|
1940 |
+
{
|
1941 |
+
"epoch": 7.84,
|
1942 |
+
"learning_rate": 3.110047846889952e-05,
|
1943 |
+
"loss": 0.0946,
|
1944 |
+
"step": 1560
|
1945 |
+
},
|
1946 |
+
{
|
1947 |
+
"epoch": 7.86,
|
1948 |
+
"learning_rate": 3.1200159489633176e-05,
|
1949 |
+
"loss": 0.1723,
|
1950 |
+
"step": 1565
|
1951 |
+
},
|
1952 |
+
{
|
1953 |
+
"epoch": 7.89,
|
1954 |
+
"learning_rate": 3.129984051036683e-05,
|
1955 |
+
"loss": 0.0569,
|
1956 |
+
"step": 1570
|
1957 |
+
},
|
1958 |
+
{
|
1959 |
+
"epoch": 7.91,
|
1960 |
+
"learning_rate": 3.139952153110048e-05,
|
1961 |
+
"loss": 0.1702,
|
1962 |
+
"step": 1575
|
1963 |
+
},
|
1964 |
+
{
|
1965 |
+
"epoch": 7.94,
|
1966 |
+
"learning_rate": 3.1499202551834136e-05,
|
1967 |
+
"loss": 0.2412,
|
1968 |
+
"step": 1580
|
1969 |
+
},
|
1970 |
+
{
|
1971 |
+
"epoch": 7.96,
|
1972 |
+
"learning_rate": 3.1598883572567784e-05,
|
1973 |
+
"loss": 0.1177,
|
1974 |
+
"step": 1585
|
1975 |
+
},
|
1976 |
+
{
|
1977 |
+
"epoch": 7.99,
|
1978 |
+
"learning_rate": 3.169856459330144e-05,
|
1979 |
+
"loss": 0.2079,
|
1980 |
+
"step": 1590
|
1981 |
+
},
|
1982 |
+
{
|
1983 |
+
"epoch": 8.0,
|
1984 |
+
"eval_accuracy": 0.9731385485391141,
|
1985 |
+
"eval_loss": 0.0661500096321106,
|
1986 |
+
"eval_runtime": 66.1809,
|
1987 |
+
"eval_samples_per_second": 32.064,
|
1988 |
+
"eval_steps_per_second": 2.01,
|
1989 |
+
"step": 1592
|
1990 |
+
},
|
1991 |
+
{
|
1992 |
+
"epoch": 8.02,
|
1993 |
+
"learning_rate": 3.179824561403509e-05,
|
1994 |
+
"loss": 0.0555,
|
1995 |
+
"step": 1595
|
1996 |
+
},
|
1997 |
+
{
|
1998 |
+
"epoch": 8.04,
|
1999 |
+
"learning_rate": 3.189792663476874e-05,
|
2000 |
+
"loss": 0.0337,
|
2001 |
+
"step": 1600
|
2002 |
+
},
|
2003 |
+
{
|
2004 |
+
"epoch": 8.07,
|
2005 |
+
"learning_rate": 3.199760765550239e-05,
|
2006 |
+
"loss": 0.1281,
|
2007 |
+
"step": 1605
|
2008 |
+
},
|
2009 |
+
{
|
2010 |
+
"epoch": 8.09,
|
2011 |
+
"learning_rate": 3.209728867623605e-05,
|
2012 |
+
"loss": 0.1517,
|
2013 |
+
"step": 1610
|
2014 |
+
},
|
2015 |
+
{
|
2016 |
+
"epoch": 8.12,
|
2017 |
+
"learning_rate": 3.2196969696969696e-05,
|
2018 |
+
"loss": 0.0803,
|
2019 |
+
"step": 1615
|
2020 |
+
},
|
2021 |
+
{
|
2022 |
+
"epoch": 8.14,
|
2023 |
+
"learning_rate": 3.229665071770335e-05,
|
2024 |
+
"loss": 0.1442,
|
2025 |
+
"step": 1620
|
2026 |
+
},
|
2027 |
+
{
|
2028 |
+
"epoch": 8.17,
|
2029 |
+
"learning_rate": 3.2396331738437e-05,
|
2030 |
+
"loss": 0.2514,
|
2031 |
+
"step": 1625
|
2032 |
+
},
|
2033 |
+
{
|
2034 |
+
"epoch": 8.19,
|
2035 |
+
"learning_rate": 3.2496012759170655e-05,
|
2036 |
+
"loss": 0.0401,
|
2037 |
+
"step": 1630
|
2038 |
+
},
|
2039 |
+
{
|
2040 |
+
"epoch": 8.22,
|
2041 |
+
"learning_rate": 3.259569377990431e-05,
|
2042 |
+
"loss": 0.0642,
|
2043 |
+
"step": 1635
|
2044 |
+
},
|
2045 |
+
{
|
2046 |
+
"epoch": 8.24,
|
2047 |
+
"learning_rate": 3.2695374800637965e-05,
|
2048 |
+
"loss": 0.18,
|
2049 |
+
"step": 1640
|
2050 |
+
},
|
2051 |
+
{
|
2052 |
+
"epoch": 8.27,
|
2053 |
+
"learning_rate": 3.2795055821371614e-05,
|
2054 |
+
"loss": 0.0759,
|
2055 |
+
"step": 1645
|
2056 |
+
},
|
2057 |
+
{
|
2058 |
+
"epoch": 8.29,
|
2059 |
+
"learning_rate": 3.289473684210527e-05,
|
2060 |
+
"loss": 0.3765,
|
2061 |
+
"step": 1650
|
2062 |
+
},
|
2063 |
+
{
|
2064 |
+
"epoch": 8.32,
|
2065 |
+
"learning_rate": 3.299441786283892e-05,
|
2066 |
+
"loss": 0.259,
|
2067 |
+
"step": 1655
|
2068 |
+
},
|
2069 |
+
{
|
2070 |
+
"epoch": 8.34,
|
2071 |
+
"learning_rate": 3.309409888357257e-05,
|
2072 |
+
"loss": 0.0904,
|
2073 |
+
"step": 1660
|
2074 |
+
},
|
2075 |
+
{
|
2076 |
+
"epoch": 8.37,
|
2077 |
+
"learning_rate": 3.319377990430622e-05,
|
2078 |
+
"loss": 0.1401,
|
2079 |
+
"step": 1665
|
2080 |
+
},
|
2081 |
+
{
|
2082 |
+
"epoch": 8.39,
|
2083 |
+
"learning_rate": 3.329346092503988e-05,
|
2084 |
+
"loss": 0.0365,
|
2085 |
+
"step": 1670
|
2086 |
+
},
|
2087 |
+
{
|
2088 |
+
"epoch": 8.42,
|
2089 |
+
"learning_rate": 3.3393141945773525e-05,
|
2090 |
+
"loss": 0.0753,
|
2091 |
+
"step": 1675
|
2092 |
+
},
|
2093 |
+
{
|
2094 |
+
"epoch": 8.44,
|
2095 |
+
"learning_rate": 3.349282296650718e-05,
|
2096 |
+
"loss": 0.1211,
|
2097 |
+
"step": 1680
|
2098 |
+
},
|
2099 |
+
{
|
2100 |
+
"epoch": 8.47,
|
2101 |
+
"learning_rate": 3.359250398724083e-05,
|
2102 |
+
"loss": 0.1027,
|
2103 |
+
"step": 1685
|
2104 |
+
},
|
2105 |
+
{
|
2106 |
+
"epoch": 8.49,
|
2107 |
+
"learning_rate": 3.369218500797448e-05,
|
2108 |
+
"loss": 0.1879,
|
2109 |
+
"step": 1690
|
2110 |
+
},
|
2111 |
+
{
|
2112 |
+
"epoch": 8.52,
|
2113 |
+
"learning_rate": 3.379186602870813e-05,
|
2114 |
+
"loss": 0.1507,
|
2115 |
+
"step": 1695
|
2116 |
+
},
|
2117 |
+
{
|
2118 |
+
"epoch": 8.54,
|
2119 |
+
"learning_rate": 3.389154704944179e-05,
|
2120 |
+
"loss": 0.0764,
|
2121 |
+
"step": 1700
|
2122 |
+
},
|
2123 |
+
{
|
2124 |
+
"epoch": 8.57,
|
2125 |
+
"learning_rate": 3.3991228070175444e-05,
|
2126 |
+
"loss": 0.0566,
|
2127 |
+
"step": 1705
|
2128 |
+
},
|
2129 |
+
{
|
2130 |
+
"epoch": 8.59,
|
2131 |
+
"learning_rate": 3.409090909090909e-05,
|
2132 |
+
"loss": 0.1513,
|
2133 |
+
"step": 1710
|
2134 |
+
},
|
2135 |
+
{
|
2136 |
+
"epoch": 8.62,
|
2137 |
+
"learning_rate": 3.419059011164275e-05,
|
2138 |
+
"loss": 0.0502,
|
2139 |
+
"step": 1715
|
2140 |
+
},
|
2141 |
+
{
|
2142 |
+
"epoch": 8.64,
|
2143 |
+
"learning_rate": 3.4290271132376396e-05,
|
2144 |
+
"loss": 0.0252,
|
2145 |
+
"step": 1720
|
2146 |
+
},
|
2147 |
+
{
|
2148 |
+
"epoch": 8.67,
|
2149 |
+
"learning_rate": 3.438995215311005e-05,
|
2150 |
+
"loss": 0.168,
|
2151 |
+
"step": 1725
|
2152 |
+
},
|
2153 |
+
{
|
2154 |
+
"epoch": 8.69,
|
2155 |
+
"learning_rate": 3.44896331738437e-05,
|
2156 |
+
"loss": 0.0836,
|
2157 |
+
"step": 1730
|
2158 |
+
},
|
2159 |
+
{
|
2160 |
+
"epoch": 8.72,
|
2161 |
+
"learning_rate": 3.4589314194577355e-05,
|
2162 |
+
"loss": 0.0736,
|
2163 |
+
"step": 1735
|
2164 |
+
},
|
2165 |
+
{
|
2166 |
+
"epoch": 8.74,
|
2167 |
+
"learning_rate": 3.4688995215311004e-05,
|
2168 |
+
"loss": 0.0712,
|
2169 |
+
"step": 1740
|
2170 |
+
},
|
2171 |
+
{
|
2172 |
+
"epoch": 8.77,
|
2173 |
+
"learning_rate": 3.478867623604466e-05,
|
2174 |
+
"loss": 0.103,
|
2175 |
+
"step": 1745
|
2176 |
+
},
|
2177 |
+
{
|
2178 |
+
"epoch": 8.79,
|
2179 |
+
"learning_rate": 3.488835725677831e-05,
|
2180 |
+
"loss": 0.1001,
|
2181 |
+
"step": 1750
|
2182 |
+
},
|
2183 |
+
{
|
2184 |
+
"epoch": 8.82,
|
2185 |
+
"learning_rate": 3.498803827751196e-05,
|
2186 |
+
"loss": 0.1186,
|
2187 |
+
"step": 1755
|
2188 |
+
},
|
2189 |
+
{
|
2190 |
+
"epoch": 8.84,
|
2191 |
+
"learning_rate": 3.508771929824561e-05,
|
2192 |
+
"loss": 0.0598,
|
2193 |
+
"step": 1760
|
2194 |
+
},
|
2195 |
+
{
|
2196 |
+
"epoch": 8.87,
|
2197 |
+
"learning_rate": 3.518740031897927e-05,
|
2198 |
+
"loss": 0.0643,
|
2199 |
+
"step": 1765
|
2200 |
+
},
|
2201 |
+
{
|
2202 |
+
"epoch": 8.89,
|
2203 |
+
"learning_rate": 3.5287081339712915e-05,
|
2204 |
+
"loss": 0.1206,
|
2205 |
+
"step": 1770
|
2206 |
+
},
|
2207 |
+
{
|
2208 |
+
"epoch": 8.92,
|
2209 |
+
"learning_rate": 3.538676236044657e-05,
|
2210 |
+
"loss": 0.0932,
|
2211 |
+
"step": 1775
|
2212 |
+
},
|
2213 |
+
{
|
2214 |
+
"epoch": 8.94,
|
2215 |
+
"learning_rate": 3.5486443381180226e-05,
|
2216 |
+
"loss": 0.1756,
|
2217 |
+
"step": 1780
|
2218 |
+
},
|
2219 |
+
{
|
2220 |
+
"epoch": 8.97,
|
2221 |
+
"learning_rate": 3.558612440191388e-05,
|
2222 |
+
"loss": 0.0591,
|
2223 |
+
"step": 1785
|
2224 |
+
},
|
2225 |
+
{
|
2226 |
+
"epoch": 8.99,
|
2227 |
+
"learning_rate": 3.568580542264753e-05,
|
2228 |
+
"loss": 0.087,
|
2229 |
+
"step": 1790
|
2230 |
+
},
|
2231 |
+
{
|
2232 |
+
"epoch": 9.0,
|
2233 |
+
"eval_accuracy": 0.9458058435438266,
|
2234 |
+
"eval_loss": 0.11827990412712097,
|
2235 |
+
"eval_runtime": 66.1948,
|
2236 |
+
"eval_samples_per_second": 32.057,
|
2237 |
+
"eval_steps_per_second": 2.009,
|
2238 |
+
"step": 1791
|
2239 |
+
},
|
2240 |
+
{
|
2241 |
+
"epoch": 9.02,
|
2242 |
+
"learning_rate": 3.5785486443381185e-05,
|
2243 |
+
"loss": 0.187,
|
2244 |
+
"step": 1795
|
2245 |
+
},
|
2246 |
+
{
|
2247 |
+
"epoch": 9.05,
|
2248 |
+
"learning_rate": 3.5885167464114834e-05,
|
2249 |
+
"loss": 0.107,
|
2250 |
+
"step": 1800
|
2251 |
+
},
|
2252 |
+
{
|
2253 |
+
"epoch": 9.07,
|
2254 |
+
"learning_rate": 3.598484848484849e-05,
|
2255 |
+
"loss": 0.0999,
|
2256 |
+
"step": 1805
|
2257 |
+
},
|
2258 |
+
{
|
2259 |
+
"epoch": 9.1,
|
2260 |
+
"learning_rate": 3.608452950558214e-05,
|
2261 |
+
"loss": 0.0598,
|
2262 |
+
"step": 1810
|
2263 |
+
},
|
2264 |
+
{
|
2265 |
+
"epoch": 9.12,
|
2266 |
+
"learning_rate": 3.618421052631579e-05,
|
2267 |
+
"loss": 0.1376,
|
2268 |
+
"step": 1815
|
2269 |
+
},
|
2270 |
+
{
|
2271 |
+
"epoch": 9.15,
|
2272 |
+
"learning_rate": 3.628389154704944e-05,
|
2273 |
+
"loss": 0.0898,
|
2274 |
+
"step": 1820
|
2275 |
+
},
|
2276 |
+
{
|
2277 |
+
"epoch": 9.17,
|
2278 |
+
"learning_rate": 3.6383572567783097e-05,
|
2279 |
+
"loss": 0.0416,
|
2280 |
+
"step": 1825
|
2281 |
+
},
|
2282 |
+
{
|
2283 |
+
"epoch": 9.2,
|
2284 |
+
"learning_rate": 3.6483253588516745e-05,
|
2285 |
+
"loss": 0.0715,
|
2286 |
+
"step": 1830
|
2287 |
+
},
|
2288 |
+
{
|
2289 |
+
"epoch": 9.22,
|
2290 |
+
"learning_rate": 3.65829346092504e-05,
|
2291 |
+
"loss": 0.0467,
|
2292 |
+
"step": 1835
|
2293 |
+
},
|
2294 |
+
{
|
2295 |
+
"epoch": 9.25,
|
2296 |
+
"learning_rate": 3.668261562998405e-05,
|
2297 |
+
"loss": 0.0446,
|
2298 |
+
"step": 1840
|
2299 |
+
},
|
2300 |
+
{
|
2301 |
+
"epoch": 9.27,
|
2302 |
+
"learning_rate": 3.6782296650717704e-05,
|
2303 |
+
"loss": 0.1276,
|
2304 |
+
"step": 1845
|
2305 |
+
},
|
2306 |
+
{
|
2307 |
+
"epoch": 9.3,
|
2308 |
+
"learning_rate": 3.688197767145136e-05,
|
2309 |
+
"loss": 0.1263,
|
2310 |
+
"step": 1850
|
2311 |
+
},
|
2312 |
+
{
|
2313 |
+
"epoch": 9.32,
|
2314 |
+
"learning_rate": 3.6981658692185015e-05,
|
2315 |
+
"loss": 0.2921,
|
2316 |
+
"step": 1855
|
2317 |
+
},
|
2318 |
+
{
|
2319 |
+
"epoch": 9.35,
|
2320 |
+
"learning_rate": 3.7081339712918663e-05,
|
2321 |
+
"loss": 0.2148,
|
2322 |
+
"step": 1860
|
2323 |
+
},
|
2324 |
+
{
|
2325 |
+
"epoch": 9.37,
|
2326 |
+
"learning_rate": 3.718102073365232e-05,
|
2327 |
+
"loss": 0.343,
|
2328 |
+
"step": 1865
|
2329 |
+
},
|
2330 |
+
{
|
2331 |
+
"epoch": 9.4,
|
2332 |
+
"learning_rate": 3.728070175438597e-05,
|
2333 |
+
"loss": 0.1294,
|
2334 |
+
"step": 1870
|
2335 |
+
},
|
2336 |
+
{
|
2337 |
+
"epoch": 9.42,
|
2338 |
+
"learning_rate": 3.738038277511962e-05,
|
2339 |
+
"loss": 0.1261,
|
2340 |
+
"step": 1875
|
2341 |
+
},
|
2342 |
+
{
|
2343 |
+
"epoch": 9.45,
|
2344 |
+
"learning_rate": 3.748006379585327e-05,
|
2345 |
+
"loss": 0.0339,
|
2346 |
+
"step": 1880
|
2347 |
+
},
|
2348 |
+
{
|
2349 |
+
"epoch": 9.47,
|
2350 |
+
"learning_rate": 3.7579744816586926e-05,
|
2351 |
+
"loss": 0.0713,
|
2352 |
+
"step": 1885
|
2353 |
+
},
|
2354 |
+
{
|
2355 |
+
"epoch": 9.5,
|
2356 |
+
"learning_rate": 3.7679425837320575e-05,
|
2357 |
+
"loss": 0.1052,
|
2358 |
+
"step": 1890
|
2359 |
+
},
|
2360 |
+
{
|
2361 |
+
"epoch": 9.52,
|
2362 |
+
"learning_rate": 3.777910685805423e-05,
|
2363 |
+
"loss": 0.0637,
|
2364 |
+
"step": 1895
|
2365 |
+
},
|
2366 |
+
{
|
2367 |
+
"epoch": 9.55,
|
2368 |
+
"learning_rate": 3.787878787878788e-05,
|
2369 |
+
"loss": 0.0649,
|
2370 |
+
"step": 1900
|
2371 |
+
},
|
2372 |
+
{
|
2373 |
+
"epoch": 9.57,
|
2374 |
+
"learning_rate": 3.7978468899521534e-05,
|
2375 |
+
"loss": 0.1127,
|
2376 |
+
"step": 1905
|
2377 |
+
},
|
2378 |
+
{
|
2379 |
+
"epoch": 9.6,
|
2380 |
+
"learning_rate": 3.807814992025518e-05,
|
2381 |
+
"loss": 0.0524,
|
2382 |
+
"step": 1910
|
2383 |
+
},
|
2384 |
+
{
|
2385 |
+
"epoch": 9.62,
|
2386 |
+
"learning_rate": 3.817783094098884e-05,
|
2387 |
+
"loss": 0.1178,
|
2388 |
+
"step": 1915
|
2389 |
+
},
|
2390 |
+
{
|
2391 |
+
"epoch": 9.65,
|
2392 |
+
"learning_rate": 3.8277511961722486e-05,
|
2393 |
+
"loss": 0.0226,
|
2394 |
+
"step": 1920
|
2395 |
+
},
|
2396 |
+
{
|
2397 |
+
"epoch": 9.67,
|
2398 |
+
"learning_rate": 3.837719298245614e-05,
|
2399 |
+
"loss": 0.1353,
|
2400 |
+
"step": 1925
|
2401 |
+
},
|
2402 |
+
{
|
2403 |
+
"epoch": 9.7,
|
2404 |
+
"learning_rate": 3.84768740031898e-05,
|
2405 |
+
"loss": 0.1414,
|
2406 |
+
"step": 1930
|
2407 |
+
},
|
2408 |
+
{
|
2409 |
+
"epoch": 9.72,
|
2410 |
+
"learning_rate": 3.8576555023923446e-05,
|
2411 |
+
"loss": 0.2608,
|
2412 |
+
"step": 1935
|
2413 |
+
},
|
2414 |
+
{
|
2415 |
+
"epoch": 9.75,
|
2416 |
+
"learning_rate": 3.86762360446571e-05,
|
2417 |
+
"loss": 0.1566,
|
2418 |
+
"step": 1940
|
2419 |
+
},
|
2420 |
+
{
|
2421 |
+
"epoch": 9.77,
|
2422 |
+
"learning_rate": 3.877591706539075e-05,
|
2423 |
+
"loss": 0.0824,
|
2424 |
+
"step": 1945
|
2425 |
+
},
|
2426 |
+
{
|
2427 |
+
"epoch": 9.8,
|
2428 |
+
"learning_rate": 3.8875598086124405e-05,
|
2429 |
+
"loss": 0.091,
|
2430 |
+
"step": 1950
|
2431 |
+
},
|
2432 |
+
{
|
2433 |
+
"epoch": 9.82,
|
2434 |
+
"learning_rate": 3.897527910685805e-05,
|
2435 |
+
"loss": 0.1284,
|
2436 |
+
"step": 1955
|
2437 |
+
},
|
2438 |
+
{
|
2439 |
+
"epoch": 9.85,
|
2440 |
+
"learning_rate": 3.907496012759171e-05,
|
2441 |
+
"loss": 0.1011,
|
2442 |
+
"step": 1960
|
2443 |
+
},
|
2444 |
+
{
|
2445 |
+
"epoch": 9.87,
|
2446 |
+
"learning_rate": 3.917464114832536e-05,
|
2447 |
+
"loss": 0.0626,
|
2448 |
+
"step": 1965
|
2449 |
+
},
|
2450 |
+
{
|
2451 |
+
"epoch": 9.9,
|
2452 |
+
"learning_rate": 3.927432216905901e-05,
|
2453 |
+
"loss": 0.2002,
|
2454 |
+
"step": 1970
|
2455 |
+
},
|
2456 |
+
{
|
2457 |
+
"epoch": 9.92,
|
2458 |
+
"learning_rate": 3.937400318979266e-05,
|
2459 |
+
"loss": 0.1474,
|
2460 |
+
"step": 1975
|
2461 |
+
},
|
2462 |
+
{
|
2463 |
+
"epoch": 9.95,
|
2464 |
+
"learning_rate": 3.9473684210526316e-05,
|
2465 |
+
"loss": 0.0989,
|
2466 |
+
"step": 1980
|
2467 |
+
},
|
2468 |
+
{
|
2469 |
+
"epoch": 9.97,
|
2470 |
+
"learning_rate": 3.9573365231259965e-05,
|
2471 |
+
"loss": 0.1807,
|
2472 |
+
"step": 1985
|
2473 |
+
},
|
2474 |
+
{
|
2475 |
+
"epoch": 10.0,
|
2476 |
+
"learning_rate": 3.967304625199362e-05,
|
2477 |
+
"loss": 0.1543,
|
2478 |
+
"step": 1990
|
2479 |
+
},
|
2480 |
+
{
|
2481 |
+
"epoch": 10.0,
|
2482 |
+
"eval_accuracy": 0.9839773798303487,
|
2483 |
+
"eval_loss": 0.06045444682240486,
|
2484 |
+
"eval_runtime": 66.0471,
|
2485 |
+
"eval_samples_per_second": 32.129,
|
2486 |
+
"eval_steps_per_second": 2.014,
|
2487 |
+
"step": 1990
|
2488 |
+
},
|
2489 |
+
{
|
2490 |
+
"epoch": 10.03,
|
2491 |
+
"learning_rate": 3.9772727272727275e-05,
|
2492 |
+
"loss": 0.1179,
|
2493 |
+
"step": 1995
|
2494 |
+
},
|
2495 |
+
{
|
2496 |
+
"epoch": 10.05,
|
2497 |
+
"learning_rate": 3.987240829346093e-05,
|
2498 |
+
"loss": 0.0731,
|
2499 |
+
"step": 2000
|
2500 |
+
},
|
2501 |
+
{
|
2502 |
+
"epoch": 10.08,
|
2503 |
+
"learning_rate": 3.997208931419458e-05,
|
2504 |
+
"loss": 0.0752,
|
2505 |
+
"step": 2005
|
2506 |
+
},
|
2507 |
+
{
|
2508 |
+
"epoch": 10.1,
|
2509 |
+
"learning_rate": 4.0071770334928235e-05,
|
2510 |
+
"loss": 0.1055,
|
2511 |
+
"step": 2010
|
2512 |
+
},
|
2513 |
+
{
|
2514 |
+
"epoch": 10.13,
|
2515 |
+
"learning_rate": 4.017145135566188e-05,
|
2516 |
+
"loss": 0.0622,
|
2517 |
+
"step": 2015
|
2518 |
+
},
|
2519 |
+
{
|
2520 |
+
"epoch": 10.15,
|
2521 |
+
"learning_rate": 4.027113237639554e-05,
|
2522 |
+
"loss": 0.0227,
|
2523 |
+
"step": 2020
|
2524 |
+
},
|
2525 |
+
{
|
2526 |
+
"epoch": 10.18,
|
2527 |
+
"learning_rate": 4.037081339712919e-05,
|
2528 |
+
"loss": 0.105,
|
2529 |
+
"step": 2025
|
2530 |
+
},
|
2531 |
+
{
|
2532 |
+
"epoch": 10.2,
|
2533 |
+
"learning_rate": 4.047049441786284e-05,
|
2534 |
+
"loss": 0.107,
|
2535 |
+
"step": 2030
|
2536 |
+
},
|
2537 |
+
{
|
2538 |
+
"epoch": 10.23,
|
2539 |
+
"learning_rate": 4.057017543859649e-05,
|
2540 |
+
"loss": 0.1032,
|
2541 |
+
"step": 2035
|
2542 |
+
},
|
2543 |
+
{
|
2544 |
+
"epoch": 10.25,
|
2545 |
+
"learning_rate": 4.0669856459330146e-05,
|
2546 |
+
"loss": 0.1356,
|
2547 |
+
"step": 2040
|
2548 |
+
},
|
2549 |
+
{
|
2550 |
+
"epoch": 10.28,
|
2551 |
+
"learning_rate": 4.0769537480063795e-05,
|
2552 |
+
"loss": 0.1786,
|
2553 |
+
"step": 2045
|
2554 |
+
},
|
2555 |
+
{
|
2556 |
+
"epoch": 10.3,
|
2557 |
+
"learning_rate": 4.086921850079745e-05,
|
2558 |
+
"loss": 0.1046,
|
2559 |
+
"step": 2050
|
2560 |
+
},
|
2561 |
+
{
|
2562 |
+
"epoch": 10.33,
|
2563 |
+
"learning_rate": 4.09688995215311e-05,
|
2564 |
+
"loss": 0.0877,
|
2565 |
+
"step": 2055
|
2566 |
+
},
|
2567 |
+
{
|
2568 |
+
"epoch": 10.35,
|
2569 |
+
"learning_rate": 4.1068580542264754e-05,
|
2570 |
+
"loss": 0.1164,
|
2571 |
+
"step": 2060
|
2572 |
+
},
|
2573 |
+
{
|
2574 |
+
"epoch": 10.38,
|
2575 |
+
"learning_rate": 4.116826156299841e-05,
|
2576 |
+
"loss": 0.072,
|
2577 |
+
"step": 2065
|
2578 |
+
},
|
2579 |
+
{
|
2580 |
+
"epoch": 10.4,
|
2581 |
+
"learning_rate": 4.1267942583732064e-05,
|
2582 |
+
"loss": 0.041,
|
2583 |
+
"step": 2070
|
2584 |
+
},
|
2585 |
+
{
|
2586 |
+
"epoch": 10.43,
|
2587 |
+
"learning_rate": 4.136762360446571e-05,
|
2588 |
+
"loss": 0.0274,
|
2589 |
+
"step": 2075
|
2590 |
+
},
|
2591 |
+
{
|
2592 |
+
"epoch": 10.45,
|
2593 |
+
"learning_rate": 4.146730462519937e-05,
|
2594 |
+
"loss": 0.1184,
|
2595 |
+
"step": 2080
|
2596 |
+
},
|
2597 |
+
{
|
2598 |
+
"epoch": 10.48,
|
2599 |
+
"learning_rate": 4.156698564593302e-05,
|
2600 |
+
"loss": 0.2738,
|
2601 |
+
"step": 2085
|
2602 |
+
},
|
2603 |
+
{
|
2604 |
+
"epoch": 10.5,
|
2605 |
+
"learning_rate": 4.166666666666667e-05,
|
2606 |
+
"loss": 0.0581,
|
2607 |
+
"step": 2090
|
2608 |
+
},
|
2609 |
+
{
|
2610 |
+
"epoch": 10.53,
|
2611 |
+
"learning_rate": 4.176634768740032e-05,
|
2612 |
+
"loss": 0.0789,
|
2613 |
+
"step": 2095
|
2614 |
+
},
|
2615 |
+
{
|
2616 |
+
"epoch": 10.55,
|
2617 |
+
"learning_rate": 4.1866028708133976e-05,
|
2618 |
+
"loss": 0.1123,
|
2619 |
+
"step": 2100
|
2620 |
+
},
|
2621 |
+
{
|
2622 |
+
"epoch": 10.58,
|
2623 |
+
"learning_rate": 4.1965709728867624e-05,
|
2624 |
+
"loss": 0.0608,
|
2625 |
+
"step": 2105
|
2626 |
+
},
|
2627 |
+
{
|
2628 |
+
"epoch": 10.6,
|
2629 |
+
"learning_rate": 4.206539074960128e-05,
|
2630 |
+
"loss": 0.138,
|
2631 |
+
"step": 2110
|
2632 |
+
},
|
2633 |
+
{
|
2634 |
+
"epoch": 10.63,
|
2635 |
+
"learning_rate": 4.216507177033493e-05,
|
2636 |
+
"loss": 0.1384,
|
2637 |
+
"step": 2115
|
2638 |
+
},
|
2639 |
+
{
|
2640 |
+
"epoch": 10.65,
|
2641 |
+
"learning_rate": 4.2264752791068584e-05,
|
2642 |
+
"loss": 0.1021,
|
2643 |
+
"step": 2120
|
2644 |
+
},
|
2645 |
+
{
|
2646 |
+
"epoch": 10.68,
|
2647 |
+
"learning_rate": 4.236443381180223e-05,
|
2648 |
+
"loss": 0.1238,
|
2649 |
+
"step": 2125
|
2650 |
+
},
|
2651 |
+
{
|
2652 |
+
"epoch": 10.7,
|
2653 |
+
"learning_rate": 4.246411483253589e-05,
|
2654 |
+
"loss": 0.0837,
|
2655 |
+
"step": 2130
|
2656 |
+
},
|
2657 |
+
{
|
2658 |
+
"epoch": 10.73,
|
2659 |
+
"learning_rate": 4.2563795853269536e-05,
|
2660 |
+
"loss": 0.0611,
|
2661 |
+
"step": 2135
|
2662 |
+
},
|
2663 |
+
{
|
2664 |
+
"epoch": 10.75,
|
2665 |
+
"learning_rate": 4.266347687400319e-05,
|
2666 |
+
"loss": 0.0489,
|
2667 |
+
"step": 2140
|
2668 |
+
},
|
2669 |
+
{
|
2670 |
+
"epoch": 10.78,
|
2671 |
+
"learning_rate": 4.2763157894736847e-05,
|
2672 |
+
"loss": 0.1189,
|
2673 |
+
"step": 2145
|
2674 |
+
},
|
2675 |
+
{
|
2676 |
+
"epoch": 10.8,
|
2677 |
+
"learning_rate": 4.28628389154705e-05,
|
2678 |
+
"loss": 0.1002,
|
2679 |
+
"step": 2150
|
2680 |
+
},
|
2681 |
+
{
|
2682 |
+
"epoch": 10.83,
|
2683 |
+
"learning_rate": 4.296251993620415e-05,
|
2684 |
+
"loss": 0.0508,
|
2685 |
+
"step": 2155
|
2686 |
+
},
|
2687 |
+
{
|
2688 |
+
"epoch": 10.85,
|
2689 |
+
"learning_rate": 4.3062200956937806e-05,
|
2690 |
+
"loss": 0.1133,
|
2691 |
+
"step": 2160
|
2692 |
+
},
|
2693 |
+
{
|
2694 |
+
"epoch": 10.88,
|
2695 |
+
"learning_rate": 4.3161881977671454e-05,
|
2696 |
+
"loss": 0.125,
|
2697 |
+
"step": 2165
|
2698 |
+
},
|
2699 |
+
{
|
2700 |
+
"epoch": 10.9,
|
2701 |
+
"learning_rate": 4.326156299840511e-05,
|
2702 |
+
"loss": 0.0678,
|
2703 |
+
"step": 2170
|
2704 |
+
},
|
2705 |
+
{
|
2706 |
+
"epoch": 10.93,
|
2707 |
+
"learning_rate": 4.336124401913876e-05,
|
2708 |
+
"loss": 0.0494,
|
2709 |
+
"step": 2175
|
2710 |
+
},
|
2711 |
+
{
|
2712 |
+
"epoch": 10.95,
|
2713 |
+
"learning_rate": 4.346092503987241e-05,
|
2714 |
+
"loss": 0.018,
|
2715 |
+
"step": 2180
|
2716 |
+
},
|
2717 |
+
{
|
2718 |
+
"epoch": 10.98,
|
2719 |
+
"learning_rate": 4.356060606060606e-05,
|
2720 |
+
"loss": 0.0788,
|
2721 |
+
"step": 2185
|
2722 |
+
},
|
2723 |
+
{
|
2724 |
+
"epoch": 11.0,
|
2725 |
+
"eval_accuracy": 0.9868049010367578,
|
2726 |
+
"eval_loss": 0.05569874495267868,
|
2727 |
+
"eval_runtime": 66.0864,
|
2728 |
+
"eval_samples_per_second": 32.109,
|
2729 |
+
"eval_steps_per_second": 2.013,
|
2730 |
+
"step": 2189
|
2731 |
+
},
|
2732 |
+
{
|
2733 |
+
"epoch": 11.01,
|
2734 |
+
"learning_rate": 4.366028708133971e-05,
|
2735 |
+
"loss": 0.0769,
|
2736 |
+
"step": 2190
|
2737 |
+
},
|
2738 |
+
{
|
2739 |
+
"epoch": 11.03,
|
2740 |
+
"learning_rate": 4.3759968102073366e-05,
|
2741 |
+
"loss": 0.0467,
|
2742 |
+
"step": 2195
|
2743 |
+
},
|
2744 |
+
{
|
2745 |
+
"epoch": 11.06,
|
2746 |
+
"learning_rate": 4.3859649122807014e-05,
|
2747 |
+
"loss": 0.0651,
|
2748 |
+
"step": 2200
|
2749 |
+
},
|
2750 |
+
{
|
2751 |
+
"epoch": 11.08,
|
2752 |
+
"learning_rate": 4.395933014354067e-05,
|
2753 |
+
"loss": 0.0585,
|
2754 |
+
"step": 2205
|
2755 |
+
},
|
2756 |
+
{
|
2757 |
+
"epoch": 11.11,
|
2758 |
+
"learning_rate": 4.4059011164274325e-05,
|
2759 |
+
"loss": 0.0348,
|
2760 |
+
"step": 2210
|
2761 |
+
},
|
2762 |
+
{
|
2763 |
+
"epoch": 11.13,
|
2764 |
+
"learning_rate": 4.415869218500798e-05,
|
2765 |
+
"loss": 0.0733,
|
2766 |
+
"step": 2215
|
2767 |
+
},
|
2768 |
+
{
|
2769 |
+
"epoch": 11.16,
|
2770 |
+
"learning_rate": 4.425837320574163e-05,
|
2771 |
+
"loss": 0.0373,
|
2772 |
+
"step": 2220
|
2773 |
+
},
|
2774 |
+
{
|
2775 |
+
"epoch": 11.18,
|
2776 |
+
"learning_rate": 4.4358054226475284e-05,
|
2777 |
+
"loss": 0.1054,
|
2778 |
+
"step": 2225
|
2779 |
+
},
|
2780 |
+
{
|
2781 |
+
"epoch": 11.21,
|
2782 |
+
"learning_rate": 4.445773524720893e-05,
|
2783 |
+
"loss": 0.0776,
|
2784 |
+
"step": 2230
|
2785 |
+
},
|
2786 |
+
{
|
2787 |
+
"epoch": 11.23,
|
2788 |
+
"learning_rate": 4.455741626794259e-05,
|
2789 |
+
"loss": 0.1979,
|
2790 |
+
"step": 2235
|
2791 |
+
},
|
2792 |
+
{
|
2793 |
+
"epoch": 11.26,
|
2794 |
+
"learning_rate": 4.4657097288676236e-05,
|
2795 |
+
"loss": 0.0852,
|
2796 |
+
"step": 2240
|
2797 |
+
},
|
2798 |
+
{
|
2799 |
+
"epoch": 11.28,
|
2800 |
+
"learning_rate": 4.475677830940989e-05,
|
2801 |
+
"loss": 0.1613,
|
2802 |
+
"step": 2245
|
2803 |
+
},
|
2804 |
+
{
|
2805 |
+
"epoch": 11.31,
|
2806 |
+
"learning_rate": 4.485645933014354e-05,
|
2807 |
+
"loss": 0.0458,
|
2808 |
+
"step": 2250
|
2809 |
+
},
|
2810 |
+
{
|
2811 |
+
"epoch": 11.33,
|
2812 |
+
"learning_rate": 4.4956140350877196e-05,
|
2813 |
+
"loss": 0.1051,
|
2814 |
+
"step": 2255
|
2815 |
+
},
|
2816 |
+
{
|
2817 |
+
"epoch": 11.36,
|
2818 |
+
"learning_rate": 4.5055821371610844e-05,
|
2819 |
+
"loss": 0.0443,
|
2820 |
+
"step": 2260
|
2821 |
+
},
|
2822 |
+
{
|
2823 |
+
"epoch": 11.38,
|
2824 |
+
"learning_rate": 4.51555023923445e-05,
|
2825 |
+
"loss": 0.0704,
|
2826 |
+
"step": 2265
|
2827 |
+
},
|
2828 |
+
{
|
2829 |
+
"epoch": 11.41,
|
2830 |
+
"learning_rate": 4.525518341307815e-05,
|
2831 |
+
"loss": 0.0637,
|
2832 |
+
"step": 2270
|
2833 |
+
},
|
2834 |
+
{
|
2835 |
+
"epoch": 11.43,
|
2836 |
+
"learning_rate": 4.53548644338118e-05,
|
2837 |
+
"loss": 0.1003,
|
2838 |
+
"step": 2275
|
2839 |
+
},
|
2840 |
+
{
|
2841 |
+
"epoch": 11.46,
|
2842 |
+
"learning_rate": 4.545454545454546e-05,
|
2843 |
+
"loss": 0.0275,
|
2844 |
+
"step": 2280
|
2845 |
+
},
|
2846 |
+
{
|
2847 |
+
"epoch": 11.48,
|
2848 |
+
"learning_rate": 4.555422647527911e-05,
|
2849 |
+
"loss": 0.0417,
|
2850 |
+
"step": 2285
|
2851 |
+
},
|
2852 |
+
{
|
2853 |
+
"epoch": 11.51,
|
2854 |
+
"learning_rate": 4.565390749601276e-05,
|
2855 |
+
"loss": 0.0953,
|
2856 |
+
"step": 2290
|
2857 |
+
},
|
2858 |
+
{
|
2859 |
+
"epoch": 11.53,
|
2860 |
+
"learning_rate": 4.575358851674642e-05,
|
2861 |
+
"loss": 0.2359,
|
2862 |
+
"step": 2295
|
2863 |
+
},
|
2864 |
+
{
|
2865 |
+
"epoch": 11.56,
|
2866 |
+
"learning_rate": 4.5853269537480066e-05,
|
2867 |
+
"loss": 0.0955,
|
2868 |
+
"step": 2300
|
2869 |
+
},
|
2870 |
+
{
|
2871 |
+
"epoch": 11.58,
|
2872 |
+
"learning_rate": 4.595295055821372e-05,
|
2873 |
+
"loss": 0.0316,
|
2874 |
+
"step": 2305
|
2875 |
+
},
|
2876 |
+
{
|
2877 |
+
"epoch": 11.61,
|
2878 |
+
"learning_rate": 4.605263157894737e-05,
|
2879 |
+
"loss": 0.1141,
|
2880 |
+
"step": 2310
|
2881 |
+
},
|
2882 |
+
{
|
2883 |
+
"epoch": 11.63,
|
2884 |
+
"learning_rate": 4.6152312599681025e-05,
|
2885 |
+
"loss": 0.0744,
|
2886 |
+
"step": 2315
|
2887 |
+
},
|
2888 |
+
{
|
2889 |
+
"epoch": 11.66,
|
2890 |
+
"learning_rate": 4.6251993620414674e-05,
|
2891 |
+
"loss": 0.0027,
|
2892 |
+
"step": 2320
|
2893 |
+
},
|
2894 |
+
{
|
2895 |
+
"epoch": 11.68,
|
2896 |
+
"learning_rate": 4.635167464114833e-05,
|
2897 |
+
"loss": 0.1356,
|
2898 |
+
"step": 2325
|
2899 |
+
},
|
2900 |
+
{
|
2901 |
+
"epoch": 11.71,
|
2902 |
+
"learning_rate": 4.645135566188198e-05,
|
2903 |
+
"loss": 0.1447,
|
2904 |
+
"step": 2330
|
2905 |
+
},
|
2906 |
+
{
|
2907 |
+
"epoch": 11.73,
|
2908 |
+
"learning_rate": 4.655103668261563e-05,
|
2909 |
+
"loss": 0.1148,
|
2910 |
+
"step": 2335
|
2911 |
+
},
|
2912 |
+
{
|
2913 |
+
"epoch": 11.76,
|
2914 |
+
"learning_rate": 4.665071770334928e-05,
|
2915 |
+
"loss": 0.0077,
|
2916 |
+
"step": 2340
|
2917 |
+
},
|
2918 |
+
{
|
2919 |
+
"epoch": 11.78,
|
2920 |
+
"learning_rate": 4.675039872408294e-05,
|
2921 |
+
"loss": 0.0557,
|
2922 |
+
"step": 2345
|
2923 |
+
},
|
2924 |
+
{
|
2925 |
+
"epoch": 11.81,
|
2926 |
+
"learning_rate": 4.6850079744816585e-05,
|
2927 |
+
"loss": 0.1106,
|
2928 |
+
"step": 2350
|
2929 |
+
},
|
2930 |
+
{
|
2931 |
+
"epoch": 11.83,
|
2932 |
+
"learning_rate": 4.694976076555024e-05,
|
2933 |
+
"loss": 0.1177,
|
2934 |
+
"step": 2355
|
2935 |
+
},
|
2936 |
+
{
|
2937 |
+
"epoch": 11.86,
|
2938 |
+
"learning_rate": 4.7049441786283896e-05,
|
2939 |
+
"loss": 0.0267,
|
2940 |
+
"step": 2360
|
2941 |
+
},
|
2942 |
+
{
|
2943 |
+
"epoch": 11.88,
|
2944 |
+
"learning_rate": 4.714912280701755e-05,
|
2945 |
+
"loss": 0.1233,
|
2946 |
+
"step": 2365
|
2947 |
+
},
|
2948 |
+
{
|
2949 |
+
"epoch": 11.91,
|
2950 |
+
"learning_rate": 4.72488038277512e-05,
|
2951 |
+
"loss": 0.0681,
|
2952 |
+
"step": 2370
|
2953 |
+
},
|
2954 |
+
{
|
2955 |
+
"epoch": 11.93,
|
2956 |
+
"learning_rate": 4.7348484848484855e-05,
|
2957 |
+
"loss": 0.0687,
|
2958 |
+
"step": 2375
|
2959 |
+
},
|
2960 |
+
{
|
2961 |
+
"epoch": 11.96,
|
2962 |
+
"learning_rate": 4.7448165869218504e-05,
|
2963 |
+
"loss": 0.0903,
|
2964 |
+
"step": 2380
|
2965 |
+
},
|
2966 |
+
{
|
2967 |
+
"epoch": 11.98,
|
2968 |
+
"learning_rate": 4.754784688995216e-05,
|
2969 |
+
"loss": 0.0604,
|
2970 |
+
"step": 2385
|
2971 |
+
},
|
2972 |
+
{
|
2973 |
+
"epoch": 12.0,
|
2974 |
+
"eval_accuracy": 0.9868049010367578,
|
2975 |
+
"eval_loss": 0.04614832252264023,
|
2976 |
+
"eval_runtime": 66.1088,
|
2977 |
+
"eval_samples_per_second": 32.099,
|
2978 |
+
"eval_steps_per_second": 2.012,
|
2979 |
+
"step": 2388
|
2980 |
+
},
|
2981 |
+
{
|
2982 |
+
"epoch": 12.01,
|
2983 |
+
"learning_rate": 4.764752791068581e-05,
|
2984 |
+
"loss": 0.0806,
|
2985 |
+
"step": 2390
|
2986 |
+
},
|
2987 |
+
{
|
2988 |
+
"epoch": 12.04,
|
2989 |
+
"learning_rate": 4.774720893141946e-05,
|
2990 |
+
"loss": 0.1088,
|
2991 |
+
"step": 2395
|
2992 |
+
},
|
2993 |
+
{
|
2994 |
+
"epoch": 12.06,
|
2995 |
+
"learning_rate": 4.784688995215311e-05,
|
2996 |
+
"loss": 0.0385,
|
2997 |
+
"step": 2400
|
2998 |
+
},
|
2999 |
+
{
|
3000 |
+
"epoch": 12.09,
|
3001 |
+
"learning_rate": 4.794657097288677e-05,
|
3002 |
+
"loss": 0.0434,
|
3003 |
+
"step": 2405
|
3004 |
+
},
|
3005 |
+
{
|
3006 |
+
"epoch": 12.11,
|
3007 |
+
"learning_rate": 4.8046251993620415e-05,
|
3008 |
+
"loss": 0.004,
|
3009 |
+
"step": 2410
|
3010 |
+
},
|
3011 |
+
{
|
3012 |
+
"epoch": 12.14,
|
3013 |
+
"learning_rate": 4.8145933014354064e-05,
|
3014 |
+
"loss": 0.0189,
|
3015 |
+
"step": 2415
|
3016 |
+
},
|
3017 |
+
{
|
3018 |
+
"epoch": 12.16,
|
3019 |
+
"learning_rate": 4.824561403508772e-05,
|
3020 |
+
"loss": 0.0846,
|
3021 |
+
"step": 2420
|
3022 |
+
},
|
3023 |
+
{
|
3024 |
+
"epoch": 12.19,
|
3025 |
+
"learning_rate": 4.8345295055821374e-05,
|
3026 |
+
"loss": 0.3582,
|
3027 |
+
"step": 2425
|
3028 |
+
},
|
3029 |
+
{
|
3030 |
+
"epoch": 12.21,
|
3031 |
+
"learning_rate": 4.844497607655503e-05,
|
3032 |
+
"loss": 0.2713,
|
3033 |
+
"step": 2430
|
3034 |
+
},
|
3035 |
+
{
|
3036 |
+
"epoch": 12.24,
|
3037 |
+
"learning_rate": 4.854465709728868e-05,
|
3038 |
+
"loss": 0.2434,
|
3039 |
+
"step": 2435
|
3040 |
+
},
|
3041 |
+
{
|
3042 |
+
"epoch": 12.26,
|
3043 |
+
"learning_rate": 4.8644338118022334e-05,
|
3044 |
+
"loss": 0.0674,
|
3045 |
+
"step": 2440
|
3046 |
+
},
|
3047 |
+
{
|
3048 |
+
"epoch": 12.29,
|
3049 |
+
"learning_rate": 4.874401913875598e-05,
|
3050 |
+
"loss": 0.1085,
|
3051 |
+
"step": 2445
|
3052 |
+
},
|
3053 |
+
{
|
3054 |
+
"epoch": 12.31,
|
3055 |
+
"learning_rate": 4.884370015948964e-05,
|
3056 |
+
"loss": 0.0215,
|
3057 |
+
"step": 2450
|
3058 |
+
},
|
3059 |
+
{
|
3060 |
+
"epoch": 12.34,
|
3061 |
+
"learning_rate": 4.8943381180223286e-05,
|
3062 |
+
"loss": 0.0041,
|
3063 |
+
"step": 2455
|
3064 |
+
},
|
3065 |
+
{
|
3066 |
+
"epoch": 12.36,
|
3067 |
+
"learning_rate": 4.904306220095694e-05,
|
3068 |
+
"loss": 0.1379,
|
3069 |
+
"step": 2460
|
3070 |
+
},
|
3071 |
+
{
|
3072 |
+
"epoch": 12.39,
|
3073 |
+
"learning_rate": 4.914274322169059e-05,
|
3074 |
+
"loss": 0.111,
|
3075 |
+
"step": 2465
|
3076 |
+
},
|
3077 |
+
{
|
3078 |
+
"epoch": 12.41,
|
3079 |
+
"learning_rate": 4.9242424242424245e-05,
|
3080 |
+
"loss": 0.0911,
|
3081 |
+
"step": 2470
|
3082 |
+
},
|
3083 |
+
{
|
3084 |
+
"epoch": 12.44,
|
3085 |
+
"learning_rate": 4.9342105263157894e-05,
|
3086 |
+
"loss": 0.0739,
|
3087 |
+
"step": 2475
|
3088 |
+
},
|
3089 |
+
{
|
3090 |
+
"epoch": 12.46,
|
3091 |
+
"learning_rate": 4.944178628389155e-05,
|
3092 |
+
"loss": 0.094,
|
3093 |
+
"step": 2480
|
3094 |
+
},
|
3095 |
+
{
|
3096 |
+
"epoch": 12.49,
|
3097 |
+
"learning_rate": 4.95414673046252e-05,
|
3098 |
+
"loss": 0.1502,
|
3099 |
+
"step": 2485
|
3100 |
+
},
|
3101 |
+
{
|
3102 |
+
"epoch": 12.51,
|
3103 |
+
"learning_rate": 4.964114832535885e-05,
|
3104 |
+
"loss": 0.1727,
|
3105 |
+
"step": 2490
|
3106 |
+
},
|
3107 |
+
{
|
3108 |
+
"epoch": 12.54,
|
3109 |
+
"learning_rate": 4.97408293460925e-05,
|
3110 |
+
"loss": 0.1593,
|
3111 |
+
"step": 2495
|
3112 |
+
},
|
3113 |
+
{
|
3114 |
+
"epoch": 12.56,
|
3115 |
+
"learning_rate": 4.984051036682616e-05,
|
3116 |
+
"loss": 0.1629,
|
3117 |
+
"step": 2500
|
3118 |
+
},
|
3119 |
+
{
|
3120 |
+
"epoch": 12.59,
|
3121 |
+
"learning_rate": 4.994019138755981e-05,
|
3122 |
+
"loss": 0.1052,
|
3123 |
+
"step": 2505
|
3124 |
+
},
|
3125 |
+
{
|
3126 |
+
"epoch": 12.61,
|
3127 |
+
"learning_rate": 4.964028776978418e-05,
|
3128 |
+
"loss": 0.0405,
|
3129 |
+
"step": 2510
|
3130 |
+
},
|
3131 |
+
{
|
3132 |
+
"epoch": 12.64,
|
3133 |
+
"learning_rate": 4.874100719424461e-05,
|
3134 |
+
"loss": 0.1163,
|
3135 |
+
"step": 2515
|
3136 |
+
},
|
3137 |
+
{
|
3138 |
+
"epoch": 12.66,
|
3139 |
+
"learning_rate": 4.784172661870504e-05,
|
3140 |
+
"loss": 0.1759,
|
3141 |
+
"step": 2520
|
3142 |
+
},
|
3143 |
+
{
|
3144 |
+
"epoch": 12.69,
|
3145 |
+
"learning_rate": 4.6942446043165475e-05,
|
3146 |
+
"loss": 0.0619,
|
3147 |
+
"step": 2525
|
3148 |
+
},
|
3149 |
+
{
|
3150 |
+
"epoch": 12.71,
|
3151 |
+
"learning_rate": 4.60431654676259e-05,
|
3152 |
+
"loss": 0.2245,
|
3153 |
+
"step": 2530
|
3154 |
+
},
|
3155 |
+
{
|
3156 |
+
"epoch": 12.74,
|
3157 |
+
"learning_rate": 4.5143884892086334e-05,
|
3158 |
+
"loss": 0.4108,
|
3159 |
+
"step": 2535
|
3160 |
+
},
|
3161 |
+
{
|
3162 |
+
"epoch": 12.76,
|
3163 |
+
"learning_rate": 4.424460431654677e-05,
|
3164 |
+
"loss": 0.1267,
|
3165 |
+
"step": 2540
|
3166 |
+
},
|
3167 |
+
{
|
3168 |
+
"epoch": 12.79,
|
3169 |
+
"learning_rate": 4.33453237410072e-05,
|
3170 |
+
"loss": 0.0401,
|
3171 |
+
"step": 2545
|
3172 |
+
},
|
3173 |
+
{
|
3174 |
+
"epoch": 12.81,
|
3175 |
+
"learning_rate": 4.244604316546763e-05,
|
3176 |
+
"loss": 0.0667,
|
3177 |
+
"step": 2550
|
3178 |
+
},
|
3179 |
+
{
|
3180 |
+
"epoch": 12.84,
|
3181 |
+
"learning_rate": 4.154676258992806e-05,
|
3182 |
+
"loss": 0.2116,
|
3183 |
+
"step": 2555
|
3184 |
+
},
|
3185 |
+
{
|
3186 |
+
"epoch": 12.86,
|
3187 |
+
"learning_rate": 4.064748201438849e-05,
|
3188 |
+
"loss": 0.2168,
|
3189 |
+
"step": 2560
|
3190 |
+
},
|
3191 |
+
{
|
3192 |
+
"epoch": 12.89,
|
3193 |
+
"learning_rate": 3.9748201438848925e-05,
|
3194 |
+
"loss": 0.1982,
|
3195 |
+
"step": 2565
|
3196 |
+
},
|
3197 |
+
{
|
3198 |
+
"epoch": 12.91,
|
3199 |
+
"learning_rate": 3.884892086330936e-05,
|
3200 |
+
"loss": 0.1961,
|
3201 |
+
"step": 2570
|
3202 |
+
},
|
3203 |
+
{
|
3204 |
+
"epoch": 12.94,
|
3205 |
+
"learning_rate": 3.794964028776979e-05,
|
3206 |
+
"loss": 0.1808,
|
3207 |
+
"step": 2575
|
3208 |
+
},
|
3209 |
+
{
|
3210 |
+
"epoch": 12.96,
|
3211 |
+
"learning_rate": 3.7050359712230217e-05,
|
3212 |
+
"loss": 0.1237,
|
3213 |
+
"step": 2580
|
3214 |
+
},
|
3215 |
+
{
|
3216 |
+
"epoch": 12.99,
|
3217 |
+
"learning_rate": 3.615107913669065e-05,
|
3218 |
+
"loss": 0.0306,
|
3219 |
+
"step": 2585
|
3220 |
+
},
|
3221 |
+
{
|
3222 |
+
"epoch": 13.0,
|
3223 |
+
"eval_accuracy": 0.9853911404335532,
|
3224 |
+
"eval_loss": 0.04759230092167854,
|
3225 |
+
"eval_runtime": 66.3415,
|
3226 |
+
"eval_samples_per_second": 31.986,
|
3227 |
+
"eval_steps_per_second": 2.005,
|
3228 |
+
"step": 2587
|
3229 |
+
},
|
3230 |
+
{
|
3231 |
+
"epoch": 13.02,
|
3232 |
+
"learning_rate": 3.5251798561151075e-05,
|
3233 |
+
"loss": 0.0663,
|
3234 |
+
"step": 2590
|
3235 |
+
},
|
3236 |
+
{
|
3237 |
+
"epoch": 13.04,
|
3238 |
+
"learning_rate": 3.435251798561151e-05,
|
3239 |
+
"loss": 0.069,
|
3240 |
+
"step": 2595
|
3241 |
+
},
|
3242 |
+
{
|
3243 |
+
"epoch": 13.07,
|
3244 |
+
"learning_rate": 3.345323741007194e-05,
|
3245 |
+
"loss": 0.0514,
|
3246 |
+
"step": 2600
|
3247 |
+
},
|
3248 |
+
{
|
3249 |
+
"epoch": 13.09,
|
3250 |
+
"learning_rate": 3.2553956834532374e-05,
|
3251 |
+
"loss": 0.037,
|
3252 |
+
"step": 2605
|
3253 |
+
},
|
3254 |
+
{
|
3255 |
+
"epoch": 13.12,
|
3256 |
+
"learning_rate": 3.165467625899281e-05,
|
3257 |
+
"loss": 0.0133,
|
3258 |
+
"step": 2610
|
3259 |
+
},
|
3260 |
+
{
|
3261 |
+
"epoch": 13.14,
|
3262 |
+
"learning_rate": 3.075539568345324e-05,
|
3263 |
+
"loss": 0.0275,
|
3264 |
+
"step": 2615
|
3265 |
+
},
|
3266 |
+
{
|
3267 |
+
"epoch": 13.17,
|
3268 |
+
"learning_rate": 2.985611510791367e-05,
|
3269 |
+
"loss": 0.0972,
|
3270 |
+
"step": 2620
|
3271 |
+
},
|
3272 |
+
{
|
3273 |
+
"epoch": 13.19,
|
3274 |
+
"learning_rate": 2.89568345323741e-05,
|
3275 |
+
"loss": 0.0956,
|
3276 |
+
"step": 2625
|
3277 |
+
},
|
3278 |
+
{
|
3279 |
+
"epoch": 13.22,
|
3280 |
+
"learning_rate": 2.805755395683453e-05,
|
3281 |
+
"loss": 0.0767,
|
3282 |
+
"step": 2630
|
3283 |
+
},
|
3284 |
+
{
|
3285 |
+
"epoch": 13.24,
|
3286 |
+
"learning_rate": 2.7158273381294964e-05,
|
3287 |
+
"loss": 0.15,
|
3288 |
+
"step": 2635
|
3289 |
+
},
|
3290 |
+
{
|
3291 |
+
"epoch": 13.27,
|
3292 |
+
"learning_rate": 2.6258992805755394e-05,
|
3293 |
+
"loss": 0.0419,
|
3294 |
+
"step": 2640
|
3295 |
+
},
|
3296 |
+
{
|
3297 |
+
"epoch": 13.29,
|
3298 |
+
"learning_rate": 2.5359712230215827e-05,
|
3299 |
+
"loss": 0.0507,
|
3300 |
+
"step": 2645
|
3301 |
+
},
|
3302 |
+
{
|
3303 |
+
"epoch": 13.32,
|
3304 |
+
"learning_rate": 2.446043165467626e-05,
|
3305 |
+
"loss": 0.0616,
|
3306 |
+
"step": 2650
|
3307 |
+
},
|
3308 |
+
{
|
3309 |
+
"epoch": 13.34,
|
3310 |
+
"learning_rate": 2.3561151079136692e-05,
|
3311 |
+
"loss": 0.044,
|
3312 |
+
"step": 2655
|
3313 |
+
},
|
3314 |
+
{
|
3315 |
+
"epoch": 13.37,
|
3316 |
+
"learning_rate": 2.2661870503597125e-05,
|
3317 |
+
"loss": 0.0768,
|
3318 |
+
"step": 2660
|
3319 |
+
},
|
3320 |
+
{
|
3321 |
+
"epoch": 13.39,
|
3322 |
+
"learning_rate": 2.1762589928057555e-05,
|
3323 |
+
"loss": 0.0692,
|
3324 |
+
"step": 2665
|
3325 |
+
},
|
3326 |
+
{
|
3327 |
+
"epoch": 13.42,
|
3328 |
+
"learning_rate": 2.0863309352517988e-05,
|
3329 |
+
"loss": 0.0125,
|
3330 |
+
"step": 2670
|
3331 |
+
},
|
3332 |
+
{
|
3333 |
+
"epoch": 13.44,
|
3334 |
+
"learning_rate": 1.996402877697842e-05,
|
3335 |
+
"loss": 0.0041,
|
3336 |
+
"step": 2675
|
3337 |
+
},
|
3338 |
+
{
|
3339 |
+
"epoch": 13.47,
|
3340 |
+
"learning_rate": 1.906474820143885e-05,
|
3341 |
+
"loss": 0.0146,
|
3342 |
+
"step": 2680
|
3343 |
+
},
|
3344 |
+
{
|
3345 |
+
"epoch": 13.49,
|
3346 |
+
"learning_rate": 1.8165467625899283e-05,
|
3347 |
+
"loss": 0.0644,
|
3348 |
+
"step": 2685
|
3349 |
+
},
|
3350 |
+
{
|
3351 |
+
"epoch": 13.52,
|
3352 |
+
"learning_rate": 1.7266187050359716e-05,
|
3353 |
+
"loss": 0.1609,
|
3354 |
+
"step": 2690
|
3355 |
+
},
|
3356 |
+
{
|
3357 |
+
"epoch": 13.54,
|
3358 |
+
"learning_rate": 1.6366906474820145e-05,
|
3359 |
+
"loss": 0.0555,
|
3360 |
+
"step": 2695
|
3361 |
+
},
|
3362 |
+
{
|
3363 |
+
"epoch": 13.57,
|
3364 |
+
"learning_rate": 1.5467625899280578e-05,
|
3365 |
+
"loss": 0.0434,
|
3366 |
+
"step": 2700
|
3367 |
+
},
|
3368 |
+
{
|
3369 |
+
"epoch": 13.59,
|
3370 |
+
"learning_rate": 1.4568345323741009e-05,
|
3371 |
+
"loss": 0.0426,
|
3372 |
+
"step": 2705
|
3373 |
+
},
|
3374 |
+
{
|
3375 |
+
"epoch": 13.62,
|
3376 |
+
"learning_rate": 1.366906474820144e-05,
|
3377 |
+
"loss": 0.0058,
|
3378 |
+
"step": 2710
|
3379 |
+
},
|
3380 |
+
{
|
3381 |
+
"epoch": 13.64,
|
3382 |
+
"learning_rate": 1.2769784172661871e-05,
|
3383 |
+
"loss": 0.0529,
|
3384 |
+
"step": 2715
|
3385 |
+
},
|
3386 |
+
{
|
3387 |
+
"epoch": 13.67,
|
3388 |
+
"learning_rate": 1.1870503597122303e-05,
|
3389 |
+
"loss": 0.0317,
|
3390 |
+
"step": 2720
|
3391 |
+
},
|
3392 |
+
{
|
3393 |
+
"epoch": 13.69,
|
3394 |
+
"learning_rate": 1.0971223021582734e-05,
|
3395 |
+
"loss": 0.0088,
|
3396 |
+
"step": 2725
|
3397 |
+
},
|
3398 |
+
{
|
3399 |
+
"epoch": 13.72,
|
3400 |
+
"learning_rate": 1.0071942446043167e-05,
|
3401 |
+
"loss": 0.0026,
|
3402 |
+
"step": 2730
|
3403 |
+
},
|
3404 |
+
{
|
3405 |
+
"epoch": 13.74,
|
3406 |
+
"learning_rate": 9.172661870503598e-06,
|
3407 |
+
"loss": 0.0342,
|
3408 |
+
"step": 2735
|
3409 |
+
},
|
3410 |
+
{
|
3411 |
+
"epoch": 13.77,
|
3412 |
+
"learning_rate": 8.273381294964029e-06,
|
3413 |
+
"loss": 0.0583,
|
3414 |
+
"step": 2740
|
3415 |
+
},
|
3416 |
+
{
|
3417 |
+
"epoch": 13.79,
|
3418 |
+
"learning_rate": 7.374100719424461e-06,
|
3419 |
+
"loss": 0.0081,
|
3420 |
+
"step": 2745
|
3421 |
+
},
|
3422 |
+
{
|
3423 |
+
"epoch": 13.82,
|
3424 |
+
"learning_rate": 6.474820143884892e-06,
|
3425 |
+
"loss": 0.0015,
|
3426 |
+
"step": 2750
|
3427 |
+
},
|
3428 |
+
{
|
3429 |
+
"epoch": 13.84,
|
3430 |
+
"learning_rate": 5.575539568345324e-06,
|
3431 |
+
"loss": 0.1185,
|
3432 |
+
"step": 2755
|
3433 |
+
},
|
3434 |
+
{
|
3435 |
+
"epoch": 13.87,
|
3436 |
+
"learning_rate": 4.676258992805756e-06,
|
3437 |
+
"loss": 0.0596,
|
3438 |
+
"step": 2760
|
3439 |
+
},
|
3440 |
+
{
|
3441 |
+
"epoch": 13.89,
|
3442 |
+
"learning_rate": 3.7769784172661873e-06,
|
3443 |
+
"loss": 0.0218,
|
3444 |
+
"step": 2765
|
3445 |
+
},
|
3446 |
+
{
|
3447 |
+
"epoch": 13.92,
|
3448 |
+
"learning_rate": 2.877697841726619e-06,
|
3449 |
+
"loss": 0.1141,
|
3450 |
+
"step": 2770
|
3451 |
+
},
|
3452 |
+
{
|
3453 |
+
"epoch": 13.94,
|
3454 |
+
"learning_rate": 1.9784172661870504e-06,
|
3455 |
+
"loss": 0.0319,
|
3456 |
+
"step": 2775
|
3457 |
+
},
|
3458 |
+
{
|
3459 |
+
"epoch": 13.97,
|
3460 |
+
"learning_rate": 1.0791366906474822e-06,
|
3461 |
+
"loss": 0.0341,
|
3462 |
+
"step": 2780
|
3463 |
+
},
|
3464 |
+
{
|
3465 |
+
"epoch": 13.99,
|
3466 |
+
"learning_rate": 1.7985611510791368e-07,
|
3467 |
+
"loss": 0.0365,
|
3468 |
+
"step": 2785
|
3469 |
+
},
|
3470 |
+
{
|
3471 |
+
"epoch": 14.0,
|
3472 |
+
"eval_accuracy": 0.9872761545711592,
|
3473 |
+
"eval_loss": 0.0543154738843441,
|
3474 |
+
"eval_runtime": 65.952,
|
3475 |
+
"eval_samples_per_second": 32.175,
|
3476 |
+
"eval_steps_per_second": 2.017,
|
3477 |
+
"step": 2786
|
3478 |
+
},
|
3479 |
+
{
|
3480 |
+
"epoch": 14.0,
|
3481 |
+
"step": 2786,
|
3482 |
+
"total_flos": 1.5805887922325717e+19,
|
3483 |
+
"train_loss": 0.17515284487896893,
|
3484 |
+
"train_runtime": 12010.6398,
|
3485 |
+
"train_samples_per_second": 7.418,
|
3486 |
+
"train_steps_per_second": 0.232
|
3487 |
+
}
|
3488 |
+
],
|
3489 |
+
"logging_steps": 5,
|
3490 |
+
"max_steps": 2786,
|
3491 |
+
"num_train_epochs": 14,
|
3492 |
+
"save_steps": 500,
|
3493 |
+
"total_flos": 1.5805887922325717e+19,
|
3494 |
+
"trial_name": null,
|
3495 |
+
"trial_params": null
|
3496 |
+
}
|