End of training
Browse files- all_results.json +13 -0
- eval_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +1657 -0
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 6.99,
|
| 3 |
+
"eval_accuracy": 0.992469545957918,
|
| 4 |
+
"eval_loss": 0.0190800242125988,
|
| 5 |
+
"eval_runtime": 190.8736,
|
| 6 |
+
"eval_samples_per_second": 23.654,
|
| 7 |
+
"eval_steps_per_second": 2.96,
|
| 8 |
+
"total_flos": 9.636137349860819e+18,
|
| 9 |
+
"train_loss": 0.2139335812344817,
|
| 10 |
+
"train_runtime": 11717.3663,
|
| 11 |
+
"train_samples_per_second": 3.575,
|
| 12 |
+
"train_steps_per_second": 0.112
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 6.99,
|
| 3 |
+
"eval_accuracy": 0.992469545957918,
|
| 4 |
+
"eval_loss": 0.0190800242125988,
|
| 5 |
+
"eval_runtime": 190.8736,
|
| 6 |
+
"eval_samples_per_second": 23.654,
|
| 7 |
+
"eval_steps_per_second": 2.96
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 6.99,
|
| 3 |
+
"total_flos": 9.636137349860819e+18,
|
| 4 |
+
"train_loss": 0.2139335812344817,
|
| 5 |
+
"train_runtime": 11717.3663,
|
| 6 |
+
"train_samples_per_second": 3.575,
|
| 7 |
+
"train_steps_per_second": 0.112
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1657 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.992469545957918,
|
| 3 |
+
"best_model_checkpoint": "swinv2-large-patch4-window12to16-192to256-22kto1k-ft-finetuned-LungCancer-LC25000-AH-40-30-30-3/checkpoint-1309",
|
| 4 |
+
"epoch": 6.990654205607477,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 1309,
|
| 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": 3.816793893129771e-06,
|
| 14 |
+
"loss": 1.1292,
|
| 15 |
+
"step": 5
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.05,
|
| 19 |
+
"learning_rate": 7.633587786259541e-06,
|
| 20 |
+
"loss": 0.9838,
|
| 21 |
+
"step": 10
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.08,
|
| 25 |
+
"learning_rate": 1.1450381679389314e-05,
|
| 26 |
+
"loss": 0.689,
|
| 27 |
+
"step": 15
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.11,
|
| 31 |
+
"learning_rate": 1.5267175572519083e-05,
|
| 32 |
+
"loss": 0.4829,
|
| 33 |
+
"step": 20
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.13,
|
| 37 |
+
"learning_rate": 1.9083969465648855e-05,
|
| 38 |
+
"loss": 0.3431,
|
| 39 |
+
"step": 25
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.16,
|
| 43 |
+
"learning_rate": 2.2900763358778628e-05,
|
| 44 |
+
"loss": 0.3189,
|
| 45 |
+
"step": 30
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.19,
|
| 49 |
+
"learning_rate": 2.6717557251908397e-05,
|
| 50 |
+
"loss": 0.1807,
|
| 51 |
+
"step": 35
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.21,
|
| 55 |
+
"learning_rate": 3.0534351145038166e-05,
|
| 56 |
+
"loss": 0.3006,
|
| 57 |
+
"step": 40
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.24,
|
| 61 |
+
"learning_rate": 3.435114503816794e-05,
|
| 62 |
+
"loss": 0.3244,
|
| 63 |
+
"step": 45
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.27,
|
| 67 |
+
"learning_rate": 3.816793893129771e-05,
|
| 68 |
+
"loss": 0.2128,
|
| 69 |
+
"step": 50
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.29,
|
| 73 |
+
"learning_rate": 4.198473282442748e-05,
|
| 74 |
+
"loss": 0.2781,
|
| 75 |
+
"step": 55
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.32,
|
| 79 |
+
"learning_rate": 4.5801526717557256e-05,
|
| 80 |
+
"loss": 0.181,
|
| 81 |
+
"step": 60
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.35,
|
| 85 |
+
"learning_rate": 4.9618320610687025e-05,
|
| 86 |
+
"loss": 0.1957,
|
| 87 |
+
"step": 65
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.37,
|
| 91 |
+
"learning_rate": 5.3435114503816794e-05,
|
| 92 |
+
"loss": 0.1996,
|
| 93 |
+
"step": 70
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.4,
|
| 97 |
+
"learning_rate": 5.725190839694656e-05,
|
| 98 |
+
"loss": 0.1654,
|
| 99 |
+
"step": 75
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.43,
|
| 103 |
+
"learning_rate": 6.106870229007633e-05,
|
| 104 |
+
"loss": 0.1993,
|
| 105 |
+
"step": 80
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.45,
|
| 109 |
+
"learning_rate": 6.488549618320611e-05,
|
| 110 |
+
"loss": 0.1601,
|
| 111 |
+
"step": 85
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.48,
|
| 115 |
+
"learning_rate": 6.870229007633588e-05,
|
| 116 |
+
"loss": 0.2335,
|
| 117 |
+
"step": 90
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.51,
|
| 121 |
+
"learning_rate": 7.251908396946565e-05,
|
| 122 |
+
"loss": 0.1595,
|
| 123 |
+
"step": 95
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.53,
|
| 127 |
+
"learning_rate": 7.633587786259542e-05,
|
| 128 |
+
"loss": 0.3264,
|
| 129 |
+
"step": 100
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.56,
|
| 133 |
+
"learning_rate": 8.015267175572518e-05,
|
| 134 |
+
"loss": 0.261,
|
| 135 |
+
"step": 105
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.59,
|
| 139 |
+
"learning_rate": 8.396946564885496e-05,
|
| 140 |
+
"loss": 0.1172,
|
| 141 |
+
"step": 110
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.61,
|
| 145 |
+
"learning_rate": 8.778625954198472e-05,
|
| 146 |
+
"loss": 0.2057,
|
| 147 |
+
"step": 115
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.64,
|
| 151 |
+
"learning_rate": 9.160305343511451e-05,
|
| 152 |
+
"loss": 0.4153,
|
| 153 |
+
"step": 120
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.67,
|
| 157 |
+
"learning_rate": 9.541984732824429e-05,
|
| 158 |
+
"loss": 0.4453,
|
| 159 |
+
"step": 125
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.69,
|
| 163 |
+
"learning_rate": 9.923664122137405e-05,
|
| 164 |
+
"loss": 0.2069,
|
| 165 |
+
"step": 130
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.72,
|
| 169 |
+
"learning_rate": 0.00010305343511450383,
|
| 170 |
+
"loss": 0.1238,
|
| 171 |
+
"step": 135
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.75,
|
| 175 |
+
"learning_rate": 0.00010687022900763359,
|
| 176 |
+
"loss": 0.3154,
|
| 177 |
+
"step": 140
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.77,
|
| 181 |
+
"learning_rate": 0.00011068702290076336,
|
| 182 |
+
"loss": 0.2941,
|
| 183 |
+
"step": 145
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 0.8,
|
| 187 |
+
"learning_rate": 0.00011450381679389313,
|
| 188 |
+
"loss": 0.3083,
|
| 189 |
+
"step": 150
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.83,
|
| 193 |
+
"learning_rate": 0.0001183206106870229,
|
| 194 |
+
"loss": 0.3601,
|
| 195 |
+
"step": 155
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 0.85,
|
| 199 |
+
"learning_rate": 0.00012213740458015266,
|
| 200 |
+
"loss": 0.3151,
|
| 201 |
+
"step": 160
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.88,
|
| 205 |
+
"learning_rate": 0.00012595419847328244,
|
| 206 |
+
"loss": 0.284,
|
| 207 |
+
"step": 165
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.91,
|
| 211 |
+
"learning_rate": 0.00012977099236641222,
|
| 212 |
+
"loss": 0.3097,
|
| 213 |
+
"step": 170
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.93,
|
| 217 |
+
"learning_rate": 0.000133587786259542,
|
| 218 |
+
"loss": 0.2929,
|
| 219 |
+
"step": 175
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.96,
|
| 223 |
+
"learning_rate": 0.00013740458015267177,
|
| 224 |
+
"loss": 0.1706,
|
| 225 |
+
"step": 180
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.99,
|
| 229 |
+
"learning_rate": 0.00014122137404580154,
|
| 230 |
+
"loss": 0.2297,
|
| 231 |
+
"step": 185
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 1.0,
|
| 235 |
+
"eval_accuracy": 0.9231450719822812,
|
| 236 |
+
"eval_loss": 0.3400990664958954,
|
| 237 |
+
"eval_runtime": 2470.4126,
|
| 238 |
+
"eval_samples_per_second": 1.828,
|
| 239 |
+
"eval_steps_per_second": 0.229,
|
| 240 |
+
"step": 187
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 1.01,
|
| 244 |
+
"learning_rate": 0.0001450381679389313,
|
| 245 |
+
"loss": 0.3795,
|
| 246 |
+
"step": 190
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 1.04,
|
| 250 |
+
"learning_rate": 0.00014885496183206107,
|
| 251 |
+
"loss": 0.8246,
|
| 252 |
+
"step": 195
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 1.07,
|
| 256 |
+
"learning_rate": 0.00015267175572519084,
|
| 257 |
+
"loss": 0.4403,
|
| 258 |
+
"step": 200
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 1.09,
|
| 262 |
+
"learning_rate": 0.00015648854961832062,
|
| 263 |
+
"loss": 0.1943,
|
| 264 |
+
"step": 205
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 1.12,
|
| 268 |
+
"learning_rate": 0.00016030534351145037,
|
| 269 |
+
"loss": 0.3229,
|
| 270 |
+
"step": 210
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 1.15,
|
| 274 |
+
"learning_rate": 0.00016412213740458014,
|
| 275 |
+
"loss": 0.262,
|
| 276 |
+
"step": 215
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 1.17,
|
| 280 |
+
"learning_rate": 0.00016793893129770992,
|
| 281 |
+
"loss": 0.2737,
|
| 282 |
+
"step": 220
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 1.2,
|
| 286 |
+
"learning_rate": 0.0001717557251908397,
|
| 287 |
+
"loss": 0.1592,
|
| 288 |
+
"step": 225
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 1.23,
|
| 292 |
+
"learning_rate": 0.00017557251908396944,
|
| 293 |
+
"loss": 0.253,
|
| 294 |
+
"step": 230
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 1.26,
|
| 298 |
+
"learning_rate": 0.00017938931297709925,
|
| 299 |
+
"loss": 0.2214,
|
| 300 |
+
"step": 235
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 1.28,
|
| 304 |
+
"learning_rate": 0.00018320610687022902,
|
| 305 |
+
"loss": 0.3303,
|
| 306 |
+
"step": 240
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 1.31,
|
| 310 |
+
"learning_rate": 0.0001870229007633588,
|
| 311 |
+
"loss": 0.3262,
|
| 312 |
+
"step": 245
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 1.34,
|
| 316 |
+
"learning_rate": 0.00019083969465648857,
|
| 317 |
+
"loss": 0.2077,
|
| 318 |
+
"step": 250
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 1.36,
|
| 322 |
+
"learning_rate": 0.00019465648854961832,
|
| 323 |
+
"loss": 0.2725,
|
| 324 |
+
"step": 255
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 1.39,
|
| 328 |
+
"learning_rate": 0.0001984732824427481,
|
| 329 |
+
"loss": 0.1802,
|
| 330 |
+
"step": 260
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 1.42,
|
| 334 |
+
"learning_rate": 0.00020229007633587788,
|
| 335 |
+
"loss": 0.2252,
|
| 336 |
+
"step": 265
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 1.44,
|
| 340 |
+
"learning_rate": 0.00020610687022900765,
|
| 341 |
+
"loss": 0.2548,
|
| 342 |
+
"step": 270
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 1.47,
|
| 346 |
+
"learning_rate": 0.0002099236641221374,
|
| 347 |
+
"loss": 0.2303,
|
| 348 |
+
"step": 275
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 1.5,
|
| 352 |
+
"learning_rate": 0.00021374045801526718,
|
| 353 |
+
"loss": 0.5035,
|
| 354 |
+
"step": 280
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 1.52,
|
| 358 |
+
"learning_rate": 0.00021755725190839695,
|
| 359 |
+
"loss": 0.4469,
|
| 360 |
+
"step": 285
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 1.55,
|
| 364 |
+
"learning_rate": 0.00022137404580152673,
|
| 365 |
+
"loss": 0.3752,
|
| 366 |
+
"step": 290
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 1.58,
|
| 370 |
+
"learning_rate": 0.00022519083969465648,
|
| 371 |
+
"loss": 0.1244,
|
| 372 |
+
"step": 295
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 1.6,
|
| 376 |
+
"learning_rate": 0.00022900763358778625,
|
| 377 |
+
"loss": 0.1889,
|
| 378 |
+
"step": 300
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 1.63,
|
| 382 |
+
"learning_rate": 0.00023282442748091603,
|
| 383 |
+
"loss": 0.1583,
|
| 384 |
+
"step": 305
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 1.66,
|
| 388 |
+
"learning_rate": 0.0002366412213740458,
|
| 389 |
+
"loss": 0.2542,
|
| 390 |
+
"step": 310
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 1.68,
|
| 394 |
+
"learning_rate": 0.00024045801526717558,
|
| 395 |
+
"loss": 0.2861,
|
| 396 |
+
"step": 315
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 1.71,
|
| 400 |
+
"learning_rate": 0.00024427480916030533,
|
| 401 |
+
"loss": 0.1671,
|
| 402 |
+
"step": 320
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 1.74,
|
| 406 |
+
"learning_rate": 0.00024809160305343513,
|
| 407 |
+
"loss": 0.2157,
|
| 408 |
+
"step": 325
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 1.76,
|
| 412 |
+
"learning_rate": 0.0002519083969465649,
|
| 413 |
+
"loss": 0.2092,
|
| 414 |
+
"step": 330
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 1.79,
|
| 418 |
+
"learning_rate": 0.00025572519083969463,
|
| 419 |
+
"loss": 0.172,
|
| 420 |
+
"step": 335
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 1.82,
|
| 424 |
+
"learning_rate": 0.00025954198473282443,
|
| 425 |
+
"loss": 0.4118,
|
| 426 |
+
"step": 340
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 1.84,
|
| 430 |
+
"learning_rate": 0.0002633587786259542,
|
| 431 |
+
"loss": 0.2941,
|
| 432 |
+
"step": 345
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 1.87,
|
| 436 |
+
"learning_rate": 0.000267175572519084,
|
| 437 |
+
"loss": 0.1484,
|
| 438 |
+
"step": 350
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 1.9,
|
| 442 |
+
"learning_rate": 0.00027099236641221373,
|
| 443 |
+
"loss": 0.287,
|
| 444 |
+
"step": 355
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 1.92,
|
| 448 |
+
"learning_rate": 0.00027480916030534353,
|
| 449 |
+
"loss": 0.279,
|
| 450 |
+
"step": 360
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 1.95,
|
| 454 |
+
"learning_rate": 0.0002786259541984733,
|
| 455 |
+
"loss": 0.2117,
|
| 456 |
+
"step": 365
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 1.98,
|
| 460 |
+
"learning_rate": 0.0002824427480916031,
|
| 461 |
+
"loss": 0.3255,
|
| 462 |
+
"step": 370
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 2.0,
|
| 466 |
+
"eval_accuracy": 0.9643410852713178,
|
| 467 |
+
"eval_loss": 0.1010039672255516,
|
| 468 |
+
"eval_runtime": 193.6164,
|
| 469 |
+
"eval_samples_per_second": 23.319,
|
| 470 |
+
"eval_steps_per_second": 2.918,
|
| 471 |
+
"step": 374
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 2.0,
|
| 475 |
+
"learning_rate": 0.0002862595419847328,
|
| 476 |
+
"loss": 0.193,
|
| 477 |
+
"step": 375
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 2.03,
|
| 481 |
+
"learning_rate": 0.0002900763358778626,
|
| 482 |
+
"loss": 0.3363,
|
| 483 |
+
"step": 380
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 2.06,
|
| 487 |
+
"learning_rate": 0.0002938931297709924,
|
| 488 |
+
"loss": 0.2783,
|
| 489 |
+
"step": 385
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 2.08,
|
| 493 |
+
"learning_rate": 0.00029770992366412214,
|
| 494 |
+
"loss": 0.1567,
|
| 495 |
+
"step": 390
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 2.11,
|
| 499 |
+
"learning_rate": 0.00030152671755725194,
|
| 500 |
+
"loss": 0.3207,
|
| 501 |
+
"step": 395
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 2.14,
|
| 505 |
+
"learning_rate": 0.0003053435114503817,
|
| 506 |
+
"loss": 0.2227,
|
| 507 |
+
"step": 400
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 2.16,
|
| 511 |
+
"learning_rate": 0.0003091603053435115,
|
| 512 |
+
"loss": 0.2319,
|
| 513 |
+
"step": 405
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 2.19,
|
| 517 |
+
"learning_rate": 0.00031297709923664124,
|
| 518 |
+
"loss": 0.2516,
|
| 519 |
+
"step": 410
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 2.22,
|
| 523 |
+
"learning_rate": 0.000316793893129771,
|
| 524 |
+
"loss": 0.198,
|
| 525 |
+
"step": 415
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 2.24,
|
| 529 |
+
"learning_rate": 0.00032061068702290074,
|
| 530 |
+
"loss": 0.3153,
|
| 531 |
+
"step": 420
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 2.27,
|
| 535 |
+
"learning_rate": 0.00032442748091603054,
|
| 536 |
+
"loss": 0.3282,
|
| 537 |
+
"step": 425
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 2.3,
|
| 541 |
+
"learning_rate": 0.0003282442748091603,
|
| 542 |
+
"loss": 0.305,
|
| 543 |
+
"step": 430
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 2.32,
|
| 547 |
+
"learning_rate": 0.0003320610687022901,
|
| 548 |
+
"loss": 0.4627,
|
| 549 |
+
"step": 435
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 2.35,
|
| 553 |
+
"learning_rate": 0.00033587786259541984,
|
| 554 |
+
"loss": 0.2607,
|
| 555 |
+
"step": 440
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 2.38,
|
| 559 |
+
"learning_rate": 0.00033969465648854964,
|
| 560 |
+
"loss": 0.235,
|
| 561 |
+
"step": 445
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 2.4,
|
| 565 |
+
"learning_rate": 0.0003435114503816794,
|
| 566 |
+
"loss": 0.2973,
|
| 567 |
+
"step": 450
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 2.43,
|
| 571 |
+
"learning_rate": 0.0003473282442748092,
|
| 572 |
+
"loss": 0.3968,
|
| 573 |
+
"step": 455
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 2.46,
|
| 577 |
+
"learning_rate": 0.0003511450381679389,
|
| 578 |
+
"loss": 0.2779,
|
| 579 |
+
"step": 460
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 2.48,
|
| 583 |
+
"learning_rate": 0.0003549618320610687,
|
| 584 |
+
"loss": 0.242,
|
| 585 |
+
"step": 465
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 2.51,
|
| 589 |
+
"learning_rate": 0.0003587786259541985,
|
| 590 |
+
"loss": 0.1816,
|
| 591 |
+
"step": 470
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 2.54,
|
| 595 |
+
"learning_rate": 0.00036259541984732824,
|
| 596 |
+
"loss": 0.2947,
|
| 597 |
+
"step": 475
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 2.56,
|
| 601 |
+
"learning_rate": 0.00036641221374045805,
|
| 602 |
+
"loss": 0.2522,
|
| 603 |
+
"step": 480
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 2.59,
|
| 607 |
+
"learning_rate": 0.0003702290076335878,
|
| 608 |
+
"loss": 0.4026,
|
| 609 |
+
"step": 485
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 2.62,
|
| 613 |
+
"learning_rate": 0.0003740458015267176,
|
| 614 |
+
"loss": 0.3203,
|
| 615 |
+
"step": 490
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 2.64,
|
| 619 |
+
"learning_rate": 0.00037786259541984735,
|
| 620 |
+
"loss": 0.1702,
|
| 621 |
+
"step": 495
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 2.67,
|
| 625 |
+
"learning_rate": 0.00038167938931297715,
|
| 626 |
+
"loss": 0.2985,
|
| 627 |
+
"step": 500
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 2.7,
|
| 631 |
+
"learning_rate": 0.00038549618320610684,
|
| 632 |
+
"loss": 0.3186,
|
| 633 |
+
"step": 505
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 2.72,
|
| 637 |
+
"learning_rate": 0.00038931297709923665,
|
| 638 |
+
"loss": 0.1724,
|
| 639 |
+
"step": 510
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 2.75,
|
| 643 |
+
"learning_rate": 0.0003931297709923664,
|
| 644 |
+
"loss": 0.3203,
|
| 645 |
+
"step": 515
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 2.78,
|
| 649 |
+
"learning_rate": 0.0003969465648854962,
|
| 650 |
+
"loss": 0.4068,
|
| 651 |
+
"step": 520
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 2.8,
|
| 655 |
+
"learning_rate": 0.00040076335877862595,
|
| 656 |
+
"loss": 0.247,
|
| 657 |
+
"step": 525
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 2.83,
|
| 661 |
+
"learning_rate": 0.00040458015267175575,
|
| 662 |
+
"loss": 0.2256,
|
| 663 |
+
"step": 530
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 2.86,
|
| 667 |
+
"learning_rate": 0.0004083969465648855,
|
| 668 |
+
"loss": 0.2111,
|
| 669 |
+
"step": 535
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 2.88,
|
| 673 |
+
"learning_rate": 0.0004122137404580153,
|
| 674 |
+
"loss": 0.2228,
|
| 675 |
+
"step": 540
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 2.91,
|
| 679 |
+
"learning_rate": 0.00041603053435114505,
|
| 680 |
+
"loss": 0.1618,
|
| 681 |
+
"step": 545
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 2.94,
|
| 685 |
+
"learning_rate": 0.0004198473282442748,
|
| 686 |
+
"loss": 0.1953,
|
| 687 |
+
"step": 550
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 2.96,
|
| 691 |
+
"learning_rate": 0.00042366412213740455,
|
| 692 |
+
"loss": 0.4014,
|
| 693 |
+
"step": 555
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 2.99,
|
| 697 |
+
"learning_rate": 0.00042748091603053435,
|
| 698 |
+
"loss": 0.4962,
|
| 699 |
+
"step": 560
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 3.0,
|
| 703 |
+
"eval_accuracy": 0.9607973421926911,
|
| 704 |
+
"eval_loss": 0.09668433666229248,
|
| 705 |
+
"eval_runtime": 193.2648,
|
| 706 |
+
"eval_samples_per_second": 23.362,
|
| 707 |
+
"eval_steps_per_second": 2.923,
|
| 708 |
+
"step": 561
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 3.02,
|
| 712 |
+
"learning_rate": 0.00043129770992366415,
|
| 713 |
+
"loss": 0.1603,
|
| 714 |
+
"step": 565
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 3.04,
|
| 718 |
+
"learning_rate": 0.0004351145038167939,
|
| 719 |
+
"loss": 0.1989,
|
| 720 |
+
"step": 570
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 3.07,
|
| 724 |
+
"learning_rate": 0.0004389312977099237,
|
| 725 |
+
"loss": 0.126,
|
| 726 |
+
"step": 575
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 3.1,
|
| 730 |
+
"learning_rate": 0.00044274809160305345,
|
| 731 |
+
"loss": 0.4359,
|
| 732 |
+
"step": 580
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 3.12,
|
| 736 |
+
"learning_rate": 0.00044656488549618326,
|
| 737 |
+
"loss": 0.2178,
|
| 738 |
+
"step": 585
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 3.15,
|
| 742 |
+
"learning_rate": 0.00045038167938931295,
|
| 743 |
+
"loss": 0.2256,
|
| 744 |
+
"step": 590
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 3.18,
|
| 748 |
+
"learning_rate": 0.00045419847328244275,
|
| 749 |
+
"loss": 0.2226,
|
| 750 |
+
"step": 595
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 3.2,
|
| 754 |
+
"learning_rate": 0.0004580152671755725,
|
| 755 |
+
"loss": 0.2424,
|
| 756 |
+
"step": 600
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 3.23,
|
| 760 |
+
"learning_rate": 0.0004618320610687023,
|
| 761 |
+
"loss": 0.2644,
|
| 762 |
+
"step": 605
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 3.26,
|
| 766 |
+
"learning_rate": 0.00046564885496183206,
|
| 767 |
+
"loss": 0.2048,
|
| 768 |
+
"step": 610
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 3.28,
|
| 772 |
+
"learning_rate": 0.00046946564885496186,
|
| 773 |
+
"loss": 0.2766,
|
| 774 |
+
"step": 615
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 3.31,
|
| 778 |
+
"learning_rate": 0.0004732824427480916,
|
| 779 |
+
"loss": 0.2327,
|
| 780 |
+
"step": 620
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 3.34,
|
| 784 |
+
"learning_rate": 0.0004770992366412214,
|
| 785 |
+
"loss": 0.2662,
|
| 786 |
+
"step": 625
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 3.36,
|
| 790 |
+
"learning_rate": 0.00048091603053435116,
|
| 791 |
+
"loss": 0.1448,
|
| 792 |
+
"step": 630
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 3.39,
|
| 796 |
+
"learning_rate": 0.0004847328244274809,
|
| 797 |
+
"loss": 0.2287,
|
| 798 |
+
"step": 635
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 3.42,
|
| 802 |
+
"learning_rate": 0.0004885496183206107,
|
| 803 |
+
"loss": 0.1811,
|
| 804 |
+
"step": 640
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 3.44,
|
| 808 |
+
"learning_rate": 0.0004923664122137404,
|
| 809 |
+
"loss": 0.2301,
|
| 810 |
+
"step": 645
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 3.47,
|
| 814 |
+
"learning_rate": 0.0004961832061068703,
|
| 815 |
+
"loss": 0.2068,
|
| 816 |
+
"step": 650
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 3.5,
|
| 820 |
+
"learning_rate": 0.0005,
|
| 821 |
+
"loss": 0.1134,
|
| 822 |
+
"step": 655
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 3.52,
|
| 826 |
+
"learning_rate": 0.000496177370030581,
|
| 827 |
+
"loss": 0.215,
|
| 828 |
+
"step": 660
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 3.55,
|
| 832 |
+
"learning_rate": 0.0004923547400611621,
|
| 833 |
+
"loss": 0.2762,
|
| 834 |
+
"step": 665
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 3.58,
|
| 838 |
+
"learning_rate": 0.0004885321100917432,
|
| 839 |
+
"loss": 0.2555,
|
| 840 |
+
"step": 670
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 3.6,
|
| 844 |
+
"learning_rate": 0.00048470948012232416,
|
| 845 |
+
"loss": 0.3345,
|
| 846 |
+
"step": 675
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 3.63,
|
| 850 |
+
"learning_rate": 0.00048088685015290524,
|
| 851 |
+
"loss": 0.1561,
|
| 852 |
+
"step": 680
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 3.66,
|
| 856 |
+
"learning_rate": 0.00047706422018348627,
|
| 857 |
+
"loss": 0.1449,
|
| 858 |
+
"step": 685
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 3.68,
|
| 862 |
+
"learning_rate": 0.00047324159021406724,
|
| 863 |
+
"loss": 0.3557,
|
| 864 |
+
"step": 690
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 3.71,
|
| 868 |
+
"learning_rate": 0.0004694189602446483,
|
| 869 |
+
"loss": 0.5036,
|
| 870 |
+
"step": 695
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 3.74,
|
| 874 |
+
"learning_rate": 0.00046559633027522934,
|
| 875 |
+
"loss": 0.1832,
|
| 876 |
+
"step": 700
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 3.77,
|
| 880 |
+
"learning_rate": 0.00046177370030581037,
|
| 881 |
+
"loss": 0.4736,
|
| 882 |
+
"step": 705
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 3.79,
|
| 886 |
+
"learning_rate": 0.00045795107033639145,
|
| 887 |
+
"loss": 0.2116,
|
| 888 |
+
"step": 710
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 3.82,
|
| 892 |
+
"learning_rate": 0.0004541284403669725,
|
| 893 |
+
"loss": 0.1522,
|
| 894 |
+
"step": 715
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 3.85,
|
| 898 |
+
"learning_rate": 0.00045030581039755355,
|
| 899 |
+
"loss": 0.1374,
|
| 900 |
+
"step": 720
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 3.87,
|
| 904 |
+
"learning_rate": 0.0004464831804281346,
|
| 905 |
+
"loss": 0.2018,
|
| 906 |
+
"step": 725
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 3.9,
|
| 910 |
+
"learning_rate": 0.0004426605504587156,
|
| 911 |
+
"loss": 0.1472,
|
| 912 |
+
"step": 730
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 3.93,
|
| 916 |
+
"learning_rate": 0.0004388379204892967,
|
| 917 |
+
"loss": 0.2752,
|
| 918 |
+
"step": 735
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 3.95,
|
| 922 |
+
"learning_rate": 0.0004350152905198777,
|
| 923 |
+
"loss": 0.2655,
|
| 924 |
+
"step": 740
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 3.98,
|
| 928 |
+
"learning_rate": 0.00043119266055045873,
|
| 929 |
+
"loss": 0.181,
|
| 930 |
+
"step": 745
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 4.0,
|
| 934 |
+
"eval_accuracy": 0.9295681063122924,
|
| 935 |
+
"eval_loss": 0.16242608428001404,
|
| 936 |
+
"eval_runtime": 193.174,
|
| 937 |
+
"eval_samples_per_second": 23.373,
|
| 938 |
+
"eval_steps_per_second": 2.925,
|
| 939 |
+
"step": 749
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 4.01,
|
| 943 |
+
"learning_rate": 0.00042737003058103976,
|
| 944 |
+
"loss": 0.1413,
|
| 945 |
+
"step": 750
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 4.03,
|
| 949 |
+
"learning_rate": 0.0004235474006116208,
|
| 950 |
+
"loss": 0.1415,
|
| 951 |
+
"step": 755
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 4.06,
|
| 955 |
+
"learning_rate": 0.0004197247706422018,
|
| 956 |
+
"loss": 0.1483,
|
| 957 |
+
"step": 760
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 4.09,
|
| 961 |
+
"learning_rate": 0.0004159021406727829,
|
| 962 |
+
"loss": 0.2537,
|
| 963 |
+
"step": 765
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 4.11,
|
| 967 |
+
"learning_rate": 0.0004120795107033639,
|
| 968 |
+
"loss": 0.2672,
|
| 969 |
+
"step": 770
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 4.14,
|
| 973 |
+
"learning_rate": 0.00040825688073394494,
|
| 974 |
+
"loss": 0.1963,
|
| 975 |
+
"step": 775
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 4.17,
|
| 979 |
+
"learning_rate": 0.000404434250764526,
|
| 980 |
+
"loss": 0.1984,
|
| 981 |
+
"step": 780
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 4.19,
|
| 985 |
+
"learning_rate": 0.00040061162079510704,
|
| 986 |
+
"loss": 0.171,
|
| 987 |
+
"step": 785
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 4.22,
|
| 991 |
+
"learning_rate": 0.00039678899082568807,
|
| 992 |
+
"loss": 0.2535,
|
| 993 |
+
"step": 790
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 4.25,
|
| 997 |
+
"learning_rate": 0.00039296636085626915,
|
| 998 |
+
"loss": 0.2084,
|
| 999 |
+
"step": 795
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 4.27,
|
| 1003 |
+
"learning_rate": 0.00038914373088685017,
|
| 1004 |
+
"loss": 0.122,
|
| 1005 |
+
"step": 800
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 4.3,
|
| 1009 |
+
"learning_rate": 0.0003853211009174312,
|
| 1010 |
+
"loss": 0.1591,
|
| 1011 |
+
"step": 805
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 4.33,
|
| 1015 |
+
"learning_rate": 0.0003814984709480123,
|
| 1016 |
+
"loss": 0.2064,
|
| 1017 |
+
"step": 810
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 4.35,
|
| 1021 |
+
"learning_rate": 0.00037767584097859325,
|
| 1022 |
+
"loss": 0.1932,
|
| 1023 |
+
"step": 815
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 4.38,
|
| 1027 |
+
"learning_rate": 0.00037385321100917427,
|
| 1028 |
+
"loss": 0.1794,
|
| 1029 |
+
"step": 820
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 4.41,
|
| 1033 |
+
"learning_rate": 0.00037003058103975535,
|
| 1034 |
+
"loss": 0.1727,
|
| 1035 |
+
"step": 825
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 4.43,
|
| 1039 |
+
"learning_rate": 0.0003662079510703364,
|
| 1040 |
+
"loss": 0.0987,
|
| 1041 |
+
"step": 830
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 4.46,
|
| 1045 |
+
"learning_rate": 0.00036238532110091746,
|
| 1046 |
+
"loss": 0.1894,
|
| 1047 |
+
"step": 835
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 4.49,
|
| 1051 |
+
"learning_rate": 0.0003585626911314985,
|
| 1052 |
+
"loss": 0.1725,
|
| 1053 |
+
"step": 840
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 4.51,
|
| 1057 |
+
"learning_rate": 0.0003547400611620795,
|
| 1058 |
+
"loss": 0.1334,
|
| 1059 |
+
"step": 845
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 4.54,
|
| 1063 |
+
"learning_rate": 0.0003509174311926606,
|
| 1064 |
+
"loss": 0.1386,
|
| 1065 |
+
"step": 850
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 4.57,
|
| 1069 |
+
"learning_rate": 0.0003470948012232416,
|
| 1070 |
+
"loss": 0.2012,
|
| 1071 |
+
"step": 855
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 4.59,
|
| 1075 |
+
"learning_rate": 0.00034327217125382264,
|
| 1076 |
+
"loss": 0.4956,
|
| 1077 |
+
"step": 860
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 4.62,
|
| 1081 |
+
"learning_rate": 0.0003394495412844037,
|
| 1082 |
+
"loss": 0.16,
|
| 1083 |
+
"step": 865
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 4.65,
|
| 1087 |
+
"learning_rate": 0.00033562691131498474,
|
| 1088 |
+
"loss": 0.2497,
|
| 1089 |
+
"step": 870
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 4.67,
|
| 1093 |
+
"learning_rate": 0.00033180428134556576,
|
| 1094 |
+
"loss": 0.2201,
|
| 1095 |
+
"step": 875
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 4.7,
|
| 1099 |
+
"learning_rate": 0.0003279816513761468,
|
| 1100 |
+
"loss": 0.1424,
|
| 1101 |
+
"step": 880
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 4.73,
|
| 1105 |
+
"learning_rate": 0.0003241590214067278,
|
| 1106 |
+
"loss": 0.1505,
|
| 1107 |
+
"step": 885
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 4.75,
|
| 1111 |
+
"learning_rate": 0.00032033639143730884,
|
| 1112 |
+
"loss": 0.1896,
|
| 1113 |
+
"step": 890
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 4.78,
|
| 1117 |
+
"learning_rate": 0.0003165137614678899,
|
| 1118 |
+
"loss": 0.1165,
|
| 1119 |
+
"step": 895
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 4.81,
|
| 1123 |
+
"learning_rate": 0.00031269113149847094,
|
| 1124 |
+
"loss": 0.0779,
|
| 1125 |
+
"step": 900
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 4.83,
|
| 1129 |
+
"learning_rate": 0.00030886850152905197,
|
| 1130 |
+
"loss": 0.1667,
|
| 1131 |
+
"step": 905
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 4.86,
|
| 1135 |
+
"learning_rate": 0.00030504587155963305,
|
| 1136 |
+
"loss": 0.103,
|
| 1137 |
+
"step": 910
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 4.89,
|
| 1141 |
+
"learning_rate": 0.0003012232415902141,
|
| 1142 |
+
"loss": 0.2255,
|
| 1143 |
+
"step": 915
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 4.91,
|
| 1147 |
+
"learning_rate": 0.0002974006116207951,
|
| 1148 |
+
"loss": 0.2617,
|
| 1149 |
+
"step": 920
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 4.94,
|
| 1153 |
+
"learning_rate": 0.0002935779816513762,
|
| 1154 |
+
"loss": 0.2884,
|
| 1155 |
+
"step": 925
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 4.97,
|
| 1159 |
+
"learning_rate": 0.0002897553516819572,
|
| 1160 |
+
"loss": 0.1863,
|
| 1161 |
+
"step": 930
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 4.99,
|
| 1165 |
+
"learning_rate": 0.0002859327217125383,
|
| 1166 |
+
"loss": 0.2266,
|
| 1167 |
+
"step": 935
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 5.0,
|
| 1171 |
+
"eval_accuracy": 0.9614617940199336,
|
| 1172 |
+
"eval_loss": 0.09752781689167023,
|
| 1173 |
+
"eval_runtime": 193.4672,
|
| 1174 |
+
"eval_samples_per_second": 23.337,
|
| 1175 |
+
"eval_steps_per_second": 2.92,
|
| 1176 |
+
"step": 936
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 5.02,
|
| 1180 |
+
"learning_rate": 0.00028211009174311925,
|
| 1181 |
+
"loss": 0.1735,
|
| 1182 |
+
"step": 940
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"epoch": 5.05,
|
| 1186 |
+
"learning_rate": 0.0002782874617737003,
|
| 1187 |
+
"loss": 0.1106,
|
| 1188 |
+
"step": 945
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"epoch": 5.07,
|
| 1192 |
+
"learning_rate": 0.00027446483180428136,
|
| 1193 |
+
"loss": 0.1575,
|
| 1194 |
+
"step": 950
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 5.1,
|
| 1198 |
+
"learning_rate": 0.0002706422018348624,
|
| 1199 |
+
"loss": 0.1547,
|
| 1200 |
+
"step": 955
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 5.13,
|
| 1204 |
+
"learning_rate": 0.0002668195718654434,
|
| 1205 |
+
"loss": 0.1507,
|
| 1206 |
+
"step": 960
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 5.15,
|
| 1210 |
+
"learning_rate": 0.0002629969418960245,
|
| 1211 |
+
"loss": 0.1887,
|
| 1212 |
+
"step": 965
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 5.18,
|
| 1216 |
+
"learning_rate": 0.0002591743119266055,
|
| 1217 |
+
"loss": 0.1534,
|
| 1218 |
+
"step": 970
|
| 1219 |
+
},
|
| 1220 |
+
{
|
| 1221 |
+
"epoch": 5.21,
|
| 1222 |
+
"learning_rate": 0.00025535168195718654,
|
| 1223 |
+
"loss": 0.0758,
|
| 1224 |
+
"step": 975
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"epoch": 5.23,
|
| 1228 |
+
"learning_rate": 0.0002515290519877676,
|
| 1229 |
+
"loss": 0.0655,
|
| 1230 |
+
"step": 980
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 5.26,
|
| 1234 |
+
"learning_rate": 0.00024770642201834864,
|
| 1235 |
+
"loss": 0.1125,
|
| 1236 |
+
"step": 985
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 5.29,
|
| 1240 |
+
"learning_rate": 0.00024388379204892967,
|
| 1241 |
+
"loss": 0.1932,
|
| 1242 |
+
"step": 990
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 5.31,
|
| 1246 |
+
"learning_rate": 0.0002400611620795107,
|
| 1247 |
+
"loss": 0.1113,
|
| 1248 |
+
"step": 995
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 5.34,
|
| 1252 |
+
"learning_rate": 0.00023623853211009174,
|
| 1253 |
+
"loss": 0.0912,
|
| 1254 |
+
"step": 1000
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 5.37,
|
| 1258 |
+
"learning_rate": 0.0002324159021406728,
|
| 1259 |
+
"loss": 0.1605,
|
| 1260 |
+
"step": 1005
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"epoch": 5.39,
|
| 1264 |
+
"learning_rate": 0.00022859327217125385,
|
| 1265 |
+
"loss": 0.1793,
|
| 1266 |
+
"step": 1010
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 5.42,
|
| 1270 |
+
"learning_rate": 0.00022477064220183487,
|
| 1271 |
+
"loss": 0.0705,
|
| 1272 |
+
"step": 1015
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 5.45,
|
| 1276 |
+
"learning_rate": 0.0002209480122324159,
|
| 1277 |
+
"loss": 0.0778,
|
| 1278 |
+
"step": 1020
|
| 1279 |
+
},
|
| 1280 |
+
{
|
| 1281 |
+
"epoch": 5.47,
|
| 1282 |
+
"learning_rate": 0.00021712538226299695,
|
| 1283 |
+
"loss": 0.1484,
|
| 1284 |
+
"step": 1025
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 5.5,
|
| 1288 |
+
"learning_rate": 0.00021330275229357798,
|
| 1289 |
+
"loss": 0.0914,
|
| 1290 |
+
"step": 1030
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 5.53,
|
| 1294 |
+
"learning_rate": 0.00020948012232415903,
|
| 1295 |
+
"loss": 0.0886,
|
| 1296 |
+
"step": 1035
|
| 1297 |
+
},
|
| 1298 |
+
{
|
| 1299 |
+
"epoch": 5.55,
|
| 1300 |
+
"learning_rate": 0.00020565749235474008,
|
| 1301 |
+
"loss": 0.1647,
|
| 1302 |
+
"step": 1040
|
| 1303 |
+
},
|
| 1304 |
+
{
|
| 1305 |
+
"epoch": 5.58,
|
| 1306 |
+
"learning_rate": 0.0002018348623853211,
|
| 1307 |
+
"loss": 0.0911,
|
| 1308 |
+
"step": 1045
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 5.61,
|
| 1312 |
+
"learning_rate": 0.00019801223241590213,
|
| 1313 |
+
"loss": 0.0603,
|
| 1314 |
+
"step": 1050
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 5.63,
|
| 1318 |
+
"learning_rate": 0.00019418960244648318,
|
| 1319 |
+
"loss": 0.097,
|
| 1320 |
+
"step": 1055
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 5.66,
|
| 1324 |
+
"learning_rate": 0.0001903669724770642,
|
| 1325 |
+
"loss": 0.1378,
|
| 1326 |
+
"step": 1060
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 5.69,
|
| 1330 |
+
"learning_rate": 0.00018654434250764526,
|
| 1331 |
+
"loss": 0.1187,
|
| 1332 |
+
"step": 1065
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 5.71,
|
| 1336 |
+
"learning_rate": 0.0001827217125382263,
|
| 1337 |
+
"loss": 0.1202,
|
| 1338 |
+
"step": 1070
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 5.74,
|
| 1342 |
+
"learning_rate": 0.00017889908256880736,
|
| 1343 |
+
"loss": 0.1565,
|
| 1344 |
+
"step": 1075
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 5.77,
|
| 1348 |
+
"learning_rate": 0.00017507645259938836,
|
| 1349 |
+
"loss": 0.0996,
|
| 1350 |
+
"step": 1080
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 5.79,
|
| 1354 |
+
"learning_rate": 0.00017125382262996941,
|
| 1355 |
+
"loss": 0.0821,
|
| 1356 |
+
"step": 1085
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 5.82,
|
| 1360 |
+
"learning_rate": 0.00016743119266055047,
|
| 1361 |
+
"loss": 0.084,
|
| 1362 |
+
"step": 1090
|
| 1363 |
+
},
|
| 1364 |
+
{
|
| 1365 |
+
"epoch": 5.85,
|
| 1366 |
+
"learning_rate": 0.0001636085626911315,
|
| 1367 |
+
"loss": 0.1211,
|
| 1368 |
+
"step": 1095
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 5.87,
|
| 1372 |
+
"learning_rate": 0.00015978593272171254,
|
| 1373 |
+
"loss": 0.1592,
|
| 1374 |
+
"step": 1100
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 5.9,
|
| 1378 |
+
"learning_rate": 0.0001559633027522936,
|
| 1379 |
+
"loss": 0.1067,
|
| 1380 |
+
"step": 1105
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 5.93,
|
| 1384 |
+
"learning_rate": 0.00015214067278287462,
|
| 1385 |
+
"loss": 0.111,
|
| 1386 |
+
"step": 1110
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 5.95,
|
| 1390 |
+
"learning_rate": 0.00014831804281345565,
|
| 1391 |
+
"loss": 0.0776,
|
| 1392 |
+
"step": 1115
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"epoch": 5.98,
|
| 1396 |
+
"learning_rate": 0.0001444954128440367,
|
| 1397 |
+
"loss": 0.0981,
|
| 1398 |
+
"step": 1120
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"epoch": 6.0,
|
| 1402 |
+
"eval_accuracy": 0.9782945736434109,
|
| 1403 |
+
"eval_loss": 0.05765698477625847,
|
| 1404 |
+
"eval_runtime": 193.5485,
|
| 1405 |
+
"eval_samples_per_second": 23.327,
|
| 1406 |
+
"eval_steps_per_second": 2.919,
|
| 1407 |
+
"step": 1123
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 6.01,
|
| 1411 |
+
"learning_rate": 0.00014067278287461775,
|
| 1412 |
+
"loss": 0.1255,
|
| 1413 |
+
"step": 1125
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 6.03,
|
| 1417 |
+
"learning_rate": 0.00013685015290519878,
|
| 1418 |
+
"loss": 0.093,
|
| 1419 |
+
"step": 1130
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 6.06,
|
| 1423 |
+
"learning_rate": 0.00013302752293577983,
|
| 1424 |
+
"loss": 0.1009,
|
| 1425 |
+
"step": 1135
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 6.09,
|
| 1429 |
+
"learning_rate": 0.00012920489296636088,
|
| 1430 |
+
"loss": 0.0498,
|
| 1431 |
+
"step": 1140
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 6.11,
|
| 1435 |
+
"learning_rate": 0.00012538226299694188,
|
| 1436 |
+
"loss": 0.0594,
|
| 1437 |
+
"step": 1145
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 6.14,
|
| 1441 |
+
"learning_rate": 0.00012155963302752294,
|
| 1442 |
+
"loss": 0.0882,
|
| 1443 |
+
"step": 1150
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 6.17,
|
| 1447 |
+
"learning_rate": 0.00011773700305810397,
|
| 1448 |
+
"loss": 0.3622,
|
| 1449 |
+
"step": 1155
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 6.19,
|
| 1453 |
+
"learning_rate": 0.00011391437308868502,
|
| 1454 |
+
"loss": 0.0536,
|
| 1455 |
+
"step": 1160
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 6.22,
|
| 1459 |
+
"learning_rate": 0.00011009174311926606,
|
| 1460 |
+
"loss": 0.0691,
|
| 1461 |
+
"step": 1165
|
| 1462 |
+
},
|
| 1463 |
+
{
|
| 1464 |
+
"epoch": 6.25,
|
| 1465 |
+
"learning_rate": 0.0001062691131498471,
|
| 1466 |
+
"loss": 0.0965,
|
| 1467 |
+
"step": 1170
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 6.28,
|
| 1471 |
+
"learning_rate": 0.00010244648318042814,
|
| 1472 |
+
"loss": 0.0614,
|
| 1473 |
+
"step": 1175
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 6.3,
|
| 1477 |
+
"learning_rate": 9.862385321100918e-05,
|
| 1478 |
+
"loss": 0.041,
|
| 1479 |
+
"step": 1180
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 6.33,
|
| 1483 |
+
"learning_rate": 9.480122324159021e-05,
|
| 1484 |
+
"loss": 0.0135,
|
| 1485 |
+
"step": 1185
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 6.36,
|
| 1489 |
+
"learning_rate": 9.097859327217125e-05,
|
| 1490 |
+
"loss": 0.0938,
|
| 1491 |
+
"step": 1190
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"epoch": 6.38,
|
| 1495 |
+
"learning_rate": 8.71559633027523e-05,
|
| 1496 |
+
"loss": 0.1184,
|
| 1497 |
+
"step": 1195
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 6.41,
|
| 1501 |
+
"learning_rate": 8.333333333333333e-05,
|
| 1502 |
+
"loss": 0.1004,
|
| 1503 |
+
"step": 1200
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 6.44,
|
| 1507 |
+
"learning_rate": 7.951070336391437e-05,
|
| 1508 |
+
"loss": 0.0319,
|
| 1509 |
+
"step": 1205
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 6.46,
|
| 1513 |
+
"learning_rate": 7.568807339449542e-05,
|
| 1514 |
+
"loss": 0.1098,
|
| 1515 |
+
"step": 1210
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 6.49,
|
| 1519 |
+
"learning_rate": 7.186544342507645e-05,
|
| 1520 |
+
"loss": 0.1152,
|
| 1521 |
+
"step": 1215
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 6.52,
|
| 1525 |
+
"learning_rate": 6.80428134556575e-05,
|
| 1526 |
+
"loss": 0.077,
|
| 1527 |
+
"step": 1220
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 6.54,
|
| 1531 |
+
"learning_rate": 6.422018348623854e-05,
|
| 1532 |
+
"loss": 0.0487,
|
| 1533 |
+
"step": 1225
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 6.57,
|
| 1537 |
+
"learning_rate": 6.0397553516819576e-05,
|
| 1538 |
+
"loss": 0.0873,
|
| 1539 |
+
"step": 1230
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 6.6,
|
| 1543 |
+
"learning_rate": 5.6574923547400615e-05,
|
| 1544 |
+
"loss": 0.0838,
|
| 1545 |
+
"step": 1235
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 6.62,
|
| 1549 |
+
"learning_rate": 5.275229357798165e-05,
|
| 1550 |
+
"loss": 0.0458,
|
| 1551 |
+
"step": 1240
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 6.65,
|
| 1555 |
+
"learning_rate": 4.892966360856269e-05,
|
| 1556 |
+
"loss": 0.1171,
|
| 1557 |
+
"step": 1245
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 6.68,
|
| 1561 |
+
"learning_rate": 4.510703363914373e-05,
|
| 1562 |
+
"loss": 0.0412,
|
| 1563 |
+
"step": 1250
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 6.7,
|
| 1567 |
+
"learning_rate": 4.1284403669724776e-05,
|
| 1568 |
+
"loss": 0.0484,
|
| 1569 |
+
"step": 1255
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 6.73,
|
| 1573 |
+
"learning_rate": 3.746177370030581e-05,
|
| 1574 |
+
"loss": 0.1255,
|
| 1575 |
+
"step": 1260
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 6.76,
|
| 1579 |
+
"learning_rate": 3.3639143730886846e-05,
|
| 1580 |
+
"loss": 0.1212,
|
| 1581 |
+
"step": 1265
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 6.78,
|
| 1585 |
+
"learning_rate": 2.9816513761467892e-05,
|
| 1586 |
+
"loss": 0.0265,
|
| 1587 |
+
"step": 1270
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 6.81,
|
| 1591 |
+
"learning_rate": 2.599388379204893e-05,
|
| 1592 |
+
"loss": 0.0676,
|
| 1593 |
+
"step": 1275
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 6.84,
|
| 1597 |
+
"learning_rate": 2.217125382262997e-05,
|
| 1598 |
+
"loss": 0.0216,
|
| 1599 |
+
"step": 1280
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 6.86,
|
| 1603 |
+
"learning_rate": 1.834862385321101e-05,
|
| 1604 |
+
"loss": 0.0113,
|
| 1605 |
+
"step": 1285
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 6.89,
|
| 1609 |
+
"learning_rate": 1.452599388379205e-05,
|
| 1610 |
+
"loss": 0.0254,
|
| 1611 |
+
"step": 1290
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 6.92,
|
| 1615 |
+
"learning_rate": 1.0703363914373088e-05,
|
| 1616 |
+
"loss": 0.0666,
|
| 1617 |
+
"step": 1295
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 6.94,
|
| 1621 |
+
"learning_rate": 6.880733944954129e-06,
|
| 1622 |
+
"loss": 0.0762,
|
| 1623 |
+
"step": 1300
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 6.97,
|
| 1627 |
+
"learning_rate": 3.0581039755351682e-06,
|
| 1628 |
+
"loss": 0.0629,
|
| 1629 |
+
"step": 1305
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 6.99,
|
| 1633 |
+
"eval_accuracy": 0.992469545957918,
|
| 1634 |
+
"eval_loss": 0.0190800242125988,
|
| 1635 |
+
"eval_runtime": 193.283,
|
| 1636 |
+
"eval_samples_per_second": 23.36,
|
| 1637 |
+
"eval_steps_per_second": 2.923,
|
| 1638 |
+
"step": 1309
|
| 1639 |
+
},
|
| 1640 |
+
{
|
| 1641 |
+
"epoch": 6.99,
|
| 1642 |
+
"step": 1309,
|
| 1643 |
+
"total_flos": 9.636137349860819e+18,
|
| 1644 |
+
"train_loss": 0.2139335812344817,
|
| 1645 |
+
"train_runtime": 11717.3663,
|
| 1646 |
+
"train_samples_per_second": 3.575,
|
| 1647 |
+
"train_steps_per_second": 0.112
|
| 1648 |
+
}
|
| 1649 |
+
],
|
| 1650 |
+
"logging_steps": 5,
|
| 1651 |
+
"max_steps": 1309,
|
| 1652 |
+
"num_train_epochs": 7,
|
| 1653 |
+
"save_steps": 500,
|
| 1654 |
+
"total_flos": 9.636137349860819e+18,
|
| 1655 |
+
"trial_name": null,
|
| 1656 |
+
"trial_params": null
|
| 1657 |
+
}
|