End of training
Browse files- all_results.json +7 -0
- best_results.json +5 -0
- train_results.json +7 -0
- trainer_state.json +1249 -0
all_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 80.0,
|
| 3 |
+
"train_loss": 0.47253459167480466,
|
| 4 |
+
"train_runtime": 1645.8876,
|
| 5 |
+
"train_samples_per_second": 19.442,
|
| 6 |
+
"train_steps_per_second": 1.215
|
| 7 |
+
}
|
best_results.json
ADDED
|
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_epoch": 57,
|
| 3 |
+
"best_eval_accuracy": 0.75,
|
| 4 |
+
"epoch": 80.0
|
| 5 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 80.0,
|
| 3 |
+
"train_loss": 0.47253459167480466,
|
| 4 |
+
"train_runtime": 1645.8876,
|
| 5 |
+
"train_samples_per_second": 19.442,
|
| 6 |
+
"train_steps_per_second": 1.215
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1249 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 80.0,
|
| 5 |
+
"global_step": 2000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 1.0,
|
| 12 |
+
"eval_accuracy": 0.49,
|
| 13 |
+
"eval_loss": 0.5651425123214722,
|
| 14 |
+
"eval_runtime": 2.9028,
|
| 15 |
+
"eval_samples_per_second": 34.45,
|
| 16 |
+
"eval_steps_per_second": 4.478,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"best_epoch": 0,
|
| 21 |
+
"best_eval_accuracy": 0.49,
|
| 22 |
+
"epoch": 1.0,
|
| 23 |
+
"step": 25
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 2.0,
|
| 27 |
+
"eval_accuracy": 0.6,
|
| 28 |
+
"eval_loss": 0.5092301368713379,
|
| 29 |
+
"eval_runtime": 2.9462,
|
| 30 |
+
"eval_samples_per_second": 33.943,
|
| 31 |
+
"eval_steps_per_second": 4.413,
|
| 32 |
+
"step": 50
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"best_epoch": 1,
|
| 36 |
+
"best_eval_accuracy": 0.6,
|
| 37 |
+
"epoch": 2.0,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 3.0,
|
| 42 |
+
"eval_accuracy": 0.63,
|
| 43 |
+
"eval_loss": 0.5062299370765686,
|
| 44 |
+
"eval_runtime": 2.9899,
|
| 45 |
+
"eval_samples_per_second": 33.446,
|
| 46 |
+
"eval_steps_per_second": 4.348,
|
| 47 |
+
"step": 75
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"best_epoch": 2,
|
| 51 |
+
"best_eval_accuracy": 0.63,
|
| 52 |
+
"epoch": 3.0,
|
| 53 |
+
"step": 75
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 4.0,
|
| 57 |
+
"eval_accuracy": 0.64,
|
| 58 |
+
"eval_loss": 0.4843701422214508,
|
| 59 |
+
"eval_runtime": 3.0147,
|
| 60 |
+
"eval_samples_per_second": 33.171,
|
| 61 |
+
"eval_steps_per_second": 4.312,
|
| 62 |
+
"step": 100
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"best_epoch": 3,
|
| 66 |
+
"best_eval_accuracy": 0.64,
|
| 67 |
+
"epoch": 4.0,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 5.0,
|
| 72 |
+
"eval_accuracy": 0.58,
|
| 73 |
+
"eval_loss": 0.441400945186615,
|
| 74 |
+
"eval_runtime": 3.0319,
|
| 75 |
+
"eval_samples_per_second": 32.982,
|
| 76 |
+
"eval_steps_per_second": 4.288,
|
| 77 |
+
"step": 125
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"best_epoch": 3,
|
| 81 |
+
"best_eval_accuracy": 0.64,
|
| 82 |
+
"epoch": 5.0,
|
| 83 |
+
"step": 125
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 6.0,
|
| 87 |
+
"eval_accuracy": 0.58,
|
| 88 |
+
"eval_loss": 0.4265769124031067,
|
| 89 |
+
"eval_runtime": 3.0555,
|
| 90 |
+
"eval_samples_per_second": 32.728,
|
| 91 |
+
"eval_steps_per_second": 4.255,
|
| 92 |
+
"step": 150
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"best_epoch": 3,
|
| 96 |
+
"best_eval_accuracy": 0.64,
|
| 97 |
+
"epoch": 6.0,
|
| 98 |
+
"step": 150
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 7.0,
|
| 102 |
+
"eval_accuracy": 0.62,
|
| 103 |
+
"eval_loss": 0.4179702401161194,
|
| 104 |
+
"eval_runtime": 3.0661,
|
| 105 |
+
"eval_samples_per_second": 32.615,
|
| 106 |
+
"eval_steps_per_second": 4.24,
|
| 107 |
+
"step": 175
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"best_epoch": 3,
|
| 111 |
+
"best_eval_accuracy": 0.64,
|
| 112 |
+
"epoch": 7.0,
|
| 113 |
+
"step": 175
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 8.0,
|
| 117 |
+
"eval_accuracy": 0.64,
|
| 118 |
+
"eval_loss": 0.46249979734420776,
|
| 119 |
+
"eval_runtime": 3.0777,
|
| 120 |
+
"eval_samples_per_second": 32.492,
|
| 121 |
+
"eval_steps_per_second": 4.224,
|
| 122 |
+
"step": 200
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"best_epoch": 3,
|
| 126 |
+
"best_eval_accuracy": 0.64,
|
| 127 |
+
"epoch": 8.0,
|
| 128 |
+
"step": 200
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 9.0,
|
| 132 |
+
"eval_accuracy": 0.61,
|
| 133 |
+
"eval_loss": 0.41327133774757385,
|
| 134 |
+
"eval_runtime": 3.0835,
|
| 135 |
+
"eval_samples_per_second": 32.43,
|
| 136 |
+
"eval_steps_per_second": 4.216,
|
| 137 |
+
"step": 225
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"best_epoch": 3,
|
| 141 |
+
"best_eval_accuracy": 0.64,
|
| 142 |
+
"epoch": 9.0,
|
| 143 |
+
"step": 225
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 10.0,
|
| 147 |
+
"eval_accuracy": 0.63,
|
| 148 |
+
"eval_loss": 0.42823413014411926,
|
| 149 |
+
"eval_runtime": 3.0882,
|
| 150 |
+
"eval_samples_per_second": 32.382,
|
| 151 |
+
"eval_steps_per_second": 4.21,
|
| 152 |
+
"step": 250
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"best_epoch": 3,
|
| 156 |
+
"best_eval_accuracy": 0.64,
|
| 157 |
+
"epoch": 10.0,
|
| 158 |
+
"step": 250
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 11.0,
|
| 162 |
+
"eval_accuracy": 0.57,
|
| 163 |
+
"eval_loss": 0.41956833004951477,
|
| 164 |
+
"eval_runtime": 3.0926,
|
| 165 |
+
"eval_samples_per_second": 32.335,
|
| 166 |
+
"eval_steps_per_second": 4.204,
|
| 167 |
+
"step": 275
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"best_epoch": 3,
|
| 171 |
+
"best_eval_accuracy": 0.64,
|
| 172 |
+
"epoch": 11.0,
|
| 173 |
+
"step": 275
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 12.0,
|
| 177 |
+
"eval_accuracy": 0.6,
|
| 178 |
+
"eval_loss": 0.40657806396484375,
|
| 179 |
+
"eval_runtime": 3.092,
|
| 180 |
+
"eval_samples_per_second": 32.342,
|
| 181 |
+
"eval_steps_per_second": 4.204,
|
| 182 |
+
"step": 300
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"best_epoch": 3,
|
| 186 |
+
"best_eval_accuracy": 0.64,
|
| 187 |
+
"epoch": 12.0,
|
| 188 |
+
"step": 300
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 13.0,
|
| 192 |
+
"eval_accuracy": 0.62,
|
| 193 |
+
"eval_loss": 0.40092143416404724,
|
| 194 |
+
"eval_runtime": 3.0926,
|
| 195 |
+
"eval_samples_per_second": 32.336,
|
| 196 |
+
"eval_steps_per_second": 4.204,
|
| 197 |
+
"step": 325
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"best_epoch": 3,
|
| 201 |
+
"best_eval_accuracy": 0.64,
|
| 202 |
+
"epoch": 13.0,
|
| 203 |
+
"step": 325
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 14.0,
|
| 207 |
+
"eval_accuracy": 0.63,
|
| 208 |
+
"eval_loss": 0.39529332518577576,
|
| 209 |
+
"eval_runtime": 3.0883,
|
| 210 |
+
"eval_samples_per_second": 32.381,
|
| 211 |
+
"eval_steps_per_second": 4.209,
|
| 212 |
+
"step": 350
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"best_epoch": 3,
|
| 216 |
+
"best_eval_accuracy": 0.64,
|
| 217 |
+
"epoch": 14.0,
|
| 218 |
+
"step": 350
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 15.0,
|
| 222 |
+
"eval_accuracy": 0.61,
|
| 223 |
+
"eval_loss": 0.39530593156814575,
|
| 224 |
+
"eval_runtime": 3.0882,
|
| 225 |
+
"eval_samples_per_second": 32.382,
|
| 226 |
+
"eval_steps_per_second": 4.21,
|
| 227 |
+
"step": 375
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"best_epoch": 3,
|
| 231 |
+
"best_eval_accuracy": 0.64,
|
| 232 |
+
"epoch": 15.0,
|
| 233 |
+
"step": 375
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 16.0,
|
| 237 |
+
"eval_accuracy": 0.64,
|
| 238 |
+
"eval_loss": 0.4115433990955353,
|
| 239 |
+
"eval_runtime": 3.085,
|
| 240 |
+
"eval_samples_per_second": 32.415,
|
| 241 |
+
"eval_steps_per_second": 4.214,
|
| 242 |
+
"step": 400
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"best_epoch": 3,
|
| 246 |
+
"best_eval_accuracy": 0.64,
|
| 247 |
+
"epoch": 16.0,
|
| 248 |
+
"step": 400
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 17.0,
|
| 252 |
+
"eval_accuracy": 0.6,
|
| 253 |
+
"eval_loss": 0.38952964544296265,
|
| 254 |
+
"eval_runtime": 3.0886,
|
| 255 |
+
"eval_samples_per_second": 32.377,
|
| 256 |
+
"eval_steps_per_second": 4.209,
|
| 257 |
+
"step": 425
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"best_epoch": 3,
|
| 261 |
+
"best_eval_accuracy": 0.64,
|
| 262 |
+
"epoch": 17.0,
|
| 263 |
+
"step": 425
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 18.0,
|
| 267 |
+
"eval_accuracy": 0.63,
|
| 268 |
+
"eval_loss": 0.427374929189682,
|
| 269 |
+
"eval_runtime": 3.0847,
|
| 270 |
+
"eval_samples_per_second": 32.418,
|
| 271 |
+
"eval_steps_per_second": 4.214,
|
| 272 |
+
"step": 450
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"best_epoch": 3,
|
| 276 |
+
"best_eval_accuracy": 0.64,
|
| 277 |
+
"epoch": 18.0,
|
| 278 |
+
"step": 450
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 19.0,
|
| 282 |
+
"eval_accuracy": 0.64,
|
| 283 |
+
"eval_loss": 0.39974284172058105,
|
| 284 |
+
"eval_runtime": 3.085,
|
| 285 |
+
"eval_samples_per_second": 32.415,
|
| 286 |
+
"eval_steps_per_second": 4.214,
|
| 287 |
+
"step": 475
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"best_epoch": 3,
|
| 291 |
+
"best_eval_accuracy": 0.64,
|
| 292 |
+
"epoch": 19.0,
|
| 293 |
+
"step": 475
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 20.0,
|
| 297 |
+
"learning_rate": 0.0075,
|
| 298 |
+
"loss": 0.6183,
|
| 299 |
+
"step": 500
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 20.0,
|
| 303 |
+
"eval_accuracy": 0.66,
|
| 304 |
+
"eval_loss": 0.3965441584587097,
|
| 305 |
+
"eval_runtime": 3.085,
|
| 306 |
+
"eval_samples_per_second": 32.415,
|
| 307 |
+
"eval_steps_per_second": 4.214,
|
| 308 |
+
"step": 500
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"best_epoch": 19,
|
| 312 |
+
"best_eval_accuracy": 0.66,
|
| 313 |
+
"epoch": 20.0,
|
| 314 |
+
"step": 500
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 21.0,
|
| 318 |
+
"eval_accuracy": 0.68,
|
| 319 |
+
"eval_loss": 0.43516531586647034,
|
| 320 |
+
"eval_runtime": 3.0853,
|
| 321 |
+
"eval_samples_per_second": 32.412,
|
| 322 |
+
"eval_steps_per_second": 4.214,
|
| 323 |
+
"step": 525
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"best_epoch": 20,
|
| 327 |
+
"best_eval_accuracy": 0.68,
|
| 328 |
+
"epoch": 21.0,
|
| 329 |
+
"step": 525
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 22.0,
|
| 333 |
+
"eval_accuracy": 0.69,
|
| 334 |
+
"eval_loss": 0.4253265857696533,
|
| 335 |
+
"eval_runtime": 3.0836,
|
| 336 |
+
"eval_samples_per_second": 32.43,
|
| 337 |
+
"eval_steps_per_second": 4.216,
|
| 338 |
+
"step": 550
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"best_epoch": 21,
|
| 342 |
+
"best_eval_accuracy": 0.69,
|
| 343 |
+
"epoch": 22.0,
|
| 344 |
+
"step": 550
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 23.0,
|
| 348 |
+
"eval_accuracy": 0.66,
|
| 349 |
+
"eval_loss": 0.3890969753265381,
|
| 350 |
+
"eval_runtime": 3.0862,
|
| 351 |
+
"eval_samples_per_second": 32.402,
|
| 352 |
+
"eval_steps_per_second": 4.212,
|
| 353 |
+
"step": 575
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"best_epoch": 21,
|
| 357 |
+
"best_eval_accuracy": 0.69,
|
| 358 |
+
"epoch": 23.0,
|
| 359 |
+
"step": 575
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 24.0,
|
| 363 |
+
"eval_accuracy": 0.69,
|
| 364 |
+
"eval_loss": 0.4323919415473938,
|
| 365 |
+
"eval_runtime": 3.0876,
|
| 366 |
+
"eval_samples_per_second": 32.387,
|
| 367 |
+
"eval_steps_per_second": 4.21,
|
| 368 |
+
"step": 600
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"best_epoch": 21,
|
| 372 |
+
"best_eval_accuracy": 0.69,
|
| 373 |
+
"epoch": 24.0,
|
| 374 |
+
"step": 600
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 25.0,
|
| 378 |
+
"eval_accuracy": 0.73,
|
| 379 |
+
"eval_loss": 0.4395662248134613,
|
| 380 |
+
"eval_runtime": 3.0865,
|
| 381 |
+
"eval_samples_per_second": 32.399,
|
| 382 |
+
"eval_steps_per_second": 4.212,
|
| 383 |
+
"step": 625
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"best_epoch": 24,
|
| 387 |
+
"best_eval_accuracy": 0.73,
|
| 388 |
+
"epoch": 25.0,
|
| 389 |
+
"step": 625
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 26.0,
|
| 393 |
+
"eval_accuracy": 0.68,
|
| 394 |
+
"eval_loss": 0.43163925409317017,
|
| 395 |
+
"eval_runtime": 3.0856,
|
| 396 |
+
"eval_samples_per_second": 32.408,
|
| 397 |
+
"eval_steps_per_second": 4.213,
|
| 398 |
+
"step": 650
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"best_epoch": 24,
|
| 402 |
+
"best_eval_accuracy": 0.73,
|
| 403 |
+
"epoch": 26.0,
|
| 404 |
+
"step": 650
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 27.0,
|
| 408 |
+
"eval_accuracy": 0.67,
|
| 409 |
+
"eval_loss": 0.39514267444610596,
|
| 410 |
+
"eval_runtime": 3.0856,
|
| 411 |
+
"eval_samples_per_second": 32.408,
|
| 412 |
+
"eval_steps_per_second": 4.213,
|
| 413 |
+
"step": 675
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"best_epoch": 24,
|
| 417 |
+
"best_eval_accuracy": 0.73,
|
| 418 |
+
"epoch": 27.0,
|
| 419 |
+
"step": 675
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 28.0,
|
| 423 |
+
"eval_accuracy": 0.68,
|
| 424 |
+
"eval_loss": 0.4021581709384918,
|
| 425 |
+
"eval_runtime": 3.0862,
|
| 426 |
+
"eval_samples_per_second": 32.402,
|
| 427 |
+
"eval_steps_per_second": 4.212,
|
| 428 |
+
"step": 700
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"best_epoch": 24,
|
| 432 |
+
"best_eval_accuracy": 0.73,
|
| 433 |
+
"epoch": 28.0,
|
| 434 |
+
"step": 700
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 29.0,
|
| 438 |
+
"eval_accuracy": 0.68,
|
| 439 |
+
"eval_loss": 0.42091983556747437,
|
| 440 |
+
"eval_runtime": 3.0882,
|
| 441 |
+
"eval_samples_per_second": 32.382,
|
| 442 |
+
"eval_steps_per_second": 4.21,
|
| 443 |
+
"step": 725
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"best_epoch": 24,
|
| 447 |
+
"best_eval_accuracy": 0.73,
|
| 448 |
+
"epoch": 29.0,
|
| 449 |
+
"step": 725
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 30.0,
|
| 453 |
+
"eval_accuracy": 0.7,
|
| 454 |
+
"eval_loss": 0.4499932527542114,
|
| 455 |
+
"eval_runtime": 3.0879,
|
| 456 |
+
"eval_samples_per_second": 32.385,
|
| 457 |
+
"eval_steps_per_second": 4.21,
|
| 458 |
+
"step": 750
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"best_epoch": 24,
|
| 462 |
+
"best_eval_accuracy": 0.73,
|
| 463 |
+
"epoch": 30.0,
|
| 464 |
+
"step": 750
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 31.0,
|
| 468 |
+
"eval_accuracy": 0.71,
|
| 469 |
+
"eval_loss": 0.4072466194629669,
|
| 470 |
+
"eval_runtime": 3.0878,
|
| 471 |
+
"eval_samples_per_second": 32.385,
|
| 472 |
+
"eval_steps_per_second": 4.21,
|
| 473 |
+
"step": 775
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"best_epoch": 24,
|
| 477 |
+
"best_eval_accuracy": 0.73,
|
| 478 |
+
"epoch": 31.0,
|
| 479 |
+
"step": 775
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 32.0,
|
| 483 |
+
"eval_accuracy": 0.7,
|
| 484 |
+
"eval_loss": 0.40178433060646057,
|
| 485 |
+
"eval_runtime": 3.0877,
|
| 486 |
+
"eval_samples_per_second": 32.387,
|
| 487 |
+
"eval_steps_per_second": 4.21,
|
| 488 |
+
"step": 800
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"best_epoch": 24,
|
| 492 |
+
"best_eval_accuracy": 0.73,
|
| 493 |
+
"epoch": 32.0,
|
| 494 |
+
"step": 800
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 33.0,
|
| 498 |
+
"eval_accuracy": 0.7,
|
| 499 |
+
"eval_loss": 0.41905662417411804,
|
| 500 |
+
"eval_runtime": 3.089,
|
| 501 |
+
"eval_samples_per_second": 32.373,
|
| 502 |
+
"eval_steps_per_second": 4.208,
|
| 503 |
+
"step": 825
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"best_epoch": 24,
|
| 507 |
+
"best_eval_accuracy": 0.73,
|
| 508 |
+
"epoch": 33.0,
|
| 509 |
+
"step": 825
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 34.0,
|
| 513 |
+
"eval_accuracy": 0.71,
|
| 514 |
+
"eval_loss": 0.3970845937728882,
|
| 515 |
+
"eval_runtime": 3.0867,
|
| 516 |
+
"eval_samples_per_second": 32.397,
|
| 517 |
+
"eval_steps_per_second": 4.212,
|
| 518 |
+
"step": 850
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"best_epoch": 24,
|
| 522 |
+
"best_eval_accuracy": 0.73,
|
| 523 |
+
"epoch": 34.0,
|
| 524 |
+
"step": 850
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 35.0,
|
| 528 |
+
"eval_accuracy": 0.7,
|
| 529 |
+
"eval_loss": 0.39993271231651306,
|
| 530 |
+
"eval_runtime": 3.088,
|
| 531 |
+
"eval_samples_per_second": 32.383,
|
| 532 |
+
"eval_steps_per_second": 4.21,
|
| 533 |
+
"step": 875
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"best_epoch": 24,
|
| 537 |
+
"best_eval_accuracy": 0.73,
|
| 538 |
+
"epoch": 35.0,
|
| 539 |
+
"step": 875
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 36.0,
|
| 543 |
+
"eval_accuracy": 0.71,
|
| 544 |
+
"eval_loss": 0.40247154235839844,
|
| 545 |
+
"eval_runtime": 3.0865,
|
| 546 |
+
"eval_samples_per_second": 32.399,
|
| 547 |
+
"eval_steps_per_second": 4.212,
|
| 548 |
+
"step": 900
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"best_epoch": 24,
|
| 552 |
+
"best_eval_accuracy": 0.73,
|
| 553 |
+
"epoch": 36.0,
|
| 554 |
+
"step": 900
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 37.0,
|
| 558 |
+
"eval_accuracy": 0.71,
|
| 559 |
+
"eval_loss": 0.4090607762336731,
|
| 560 |
+
"eval_runtime": 3.0867,
|
| 561 |
+
"eval_samples_per_second": 32.397,
|
| 562 |
+
"eval_steps_per_second": 4.212,
|
| 563 |
+
"step": 925
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"best_epoch": 24,
|
| 567 |
+
"best_eval_accuracy": 0.73,
|
| 568 |
+
"epoch": 37.0,
|
| 569 |
+
"step": 925
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 38.0,
|
| 573 |
+
"eval_accuracy": 0.72,
|
| 574 |
+
"eval_loss": 0.4060271084308624,
|
| 575 |
+
"eval_runtime": 3.0865,
|
| 576 |
+
"eval_samples_per_second": 32.399,
|
| 577 |
+
"eval_steps_per_second": 4.212,
|
| 578 |
+
"step": 950
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"best_epoch": 24,
|
| 582 |
+
"best_eval_accuracy": 0.73,
|
| 583 |
+
"epoch": 38.0,
|
| 584 |
+
"step": 950
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 39.0,
|
| 588 |
+
"eval_accuracy": 0.71,
|
| 589 |
+
"eval_loss": 0.4415541887283325,
|
| 590 |
+
"eval_runtime": 3.089,
|
| 591 |
+
"eval_samples_per_second": 32.373,
|
| 592 |
+
"eval_steps_per_second": 4.208,
|
| 593 |
+
"step": 975
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"best_epoch": 24,
|
| 597 |
+
"best_eval_accuracy": 0.73,
|
| 598 |
+
"epoch": 39.0,
|
| 599 |
+
"step": 975
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 40.0,
|
| 603 |
+
"learning_rate": 0.005,
|
| 604 |
+
"loss": 0.4716,
|
| 605 |
+
"step": 1000
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 40.0,
|
| 609 |
+
"eval_accuracy": 0.71,
|
| 610 |
+
"eval_loss": 0.404102623462677,
|
| 611 |
+
"eval_runtime": 3.0888,
|
| 612 |
+
"eval_samples_per_second": 32.375,
|
| 613 |
+
"eval_steps_per_second": 4.209,
|
| 614 |
+
"step": 1000
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"best_epoch": 24,
|
| 618 |
+
"best_eval_accuracy": 0.73,
|
| 619 |
+
"epoch": 40.0,
|
| 620 |
+
"step": 1000
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 41.0,
|
| 624 |
+
"eval_accuracy": 0.72,
|
| 625 |
+
"eval_loss": 0.4100326895713806,
|
| 626 |
+
"eval_runtime": 3.0875,
|
| 627 |
+
"eval_samples_per_second": 32.389,
|
| 628 |
+
"eval_steps_per_second": 4.211,
|
| 629 |
+
"step": 1025
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"best_epoch": 24,
|
| 633 |
+
"best_eval_accuracy": 0.73,
|
| 634 |
+
"epoch": 41.0,
|
| 635 |
+
"step": 1025
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 42.0,
|
| 639 |
+
"eval_accuracy": 0.73,
|
| 640 |
+
"eval_loss": 0.40424102544784546,
|
| 641 |
+
"eval_runtime": 3.0901,
|
| 642 |
+
"eval_samples_per_second": 32.362,
|
| 643 |
+
"eval_steps_per_second": 4.207,
|
| 644 |
+
"step": 1050
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"best_epoch": 24,
|
| 648 |
+
"best_eval_accuracy": 0.73,
|
| 649 |
+
"epoch": 42.0,
|
| 650 |
+
"step": 1050
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 43.0,
|
| 654 |
+
"eval_accuracy": 0.71,
|
| 655 |
+
"eval_loss": 0.3743560314178467,
|
| 656 |
+
"eval_runtime": 3.0886,
|
| 657 |
+
"eval_samples_per_second": 32.377,
|
| 658 |
+
"eval_steps_per_second": 4.209,
|
| 659 |
+
"step": 1075
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"best_epoch": 24,
|
| 663 |
+
"best_eval_accuracy": 0.73,
|
| 664 |
+
"epoch": 43.0,
|
| 665 |
+
"step": 1075
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 44.0,
|
| 669 |
+
"eval_accuracy": 0.71,
|
| 670 |
+
"eval_loss": 0.3826563358306885,
|
| 671 |
+
"eval_runtime": 3.089,
|
| 672 |
+
"eval_samples_per_second": 32.373,
|
| 673 |
+
"eval_steps_per_second": 4.209,
|
| 674 |
+
"step": 1100
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"best_epoch": 24,
|
| 678 |
+
"best_eval_accuracy": 0.73,
|
| 679 |
+
"epoch": 44.0,
|
| 680 |
+
"step": 1100
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 45.0,
|
| 684 |
+
"eval_accuracy": 0.71,
|
| 685 |
+
"eval_loss": 0.39410629868507385,
|
| 686 |
+
"eval_runtime": 3.0915,
|
| 687 |
+
"eval_samples_per_second": 32.346,
|
| 688 |
+
"eval_steps_per_second": 4.205,
|
| 689 |
+
"step": 1125
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"best_epoch": 24,
|
| 693 |
+
"best_eval_accuracy": 0.73,
|
| 694 |
+
"epoch": 45.0,
|
| 695 |
+
"step": 1125
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 46.0,
|
| 699 |
+
"eval_accuracy": 0.73,
|
| 700 |
+
"eval_loss": 0.430500864982605,
|
| 701 |
+
"eval_runtime": 3.0889,
|
| 702 |
+
"eval_samples_per_second": 32.373,
|
| 703 |
+
"eval_steps_per_second": 4.209,
|
| 704 |
+
"step": 1150
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"best_epoch": 24,
|
| 708 |
+
"best_eval_accuracy": 0.73,
|
| 709 |
+
"epoch": 46.0,
|
| 710 |
+
"step": 1150
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 47.0,
|
| 714 |
+
"eval_accuracy": 0.73,
|
| 715 |
+
"eval_loss": 0.40077340602874756,
|
| 716 |
+
"eval_runtime": 3.0873,
|
| 717 |
+
"eval_samples_per_second": 32.39,
|
| 718 |
+
"eval_steps_per_second": 4.211,
|
| 719 |
+
"step": 1175
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"best_epoch": 24,
|
| 723 |
+
"best_eval_accuracy": 0.73,
|
| 724 |
+
"epoch": 47.0,
|
| 725 |
+
"step": 1175
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 48.0,
|
| 729 |
+
"eval_accuracy": 0.73,
|
| 730 |
+
"eval_loss": 0.40273621678352356,
|
| 731 |
+
"eval_runtime": 3.0892,
|
| 732 |
+
"eval_samples_per_second": 32.371,
|
| 733 |
+
"eval_steps_per_second": 4.208,
|
| 734 |
+
"step": 1200
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"best_epoch": 24,
|
| 738 |
+
"best_eval_accuracy": 0.73,
|
| 739 |
+
"epoch": 48.0,
|
| 740 |
+
"step": 1200
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 49.0,
|
| 744 |
+
"eval_accuracy": 0.72,
|
| 745 |
+
"eval_loss": 0.4023691415786743,
|
| 746 |
+
"eval_runtime": 3.0893,
|
| 747 |
+
"eval_samples_per_second": 32.369,
|
| 748 |
+
"eval_steps_per_second": 4.208,
|
| 749 |
+
"step": 1225
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"best_epoch": 24,
|
| 753 |
+
"best_eval_accuracy": 0.73,
|
| 754 |
+
"epoch": 49.0,
|
| 755 |
+
"step": 1225
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 50.0,
|
| 759 |
+
"eval_accuracy": 0.72,
|
| 760 |
+
"eval_loss": 0.3937583267688751,
|
| 761 |
+
"eval_runtime": 3.0877,
|
| 762 |
+
"eval_samples_per_second": 32.387,
|
| 763 |
+
"eval_steps_per_second": 4.21,
|
| 764 |
+
"step": 1250
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"best_epoch": 24,
|
| 768 |
+
"best_eval_accuracy": 0.73,
|
| 769 |
+
"epoch": 50.0,
|
| 770 |
+
"step": 1250
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 51.0,
|
| 774 |
+
"eval_accuracy": 0.73,
|
| 775 |
+
"eval_loss": 0.38425591588020325,
|
| 776 |
+
"eval_runtime": 3.0873,
|
| 777 |
+
"eval_samples_per_second": 32.391,
|
| 778 |
+
"eval_steps_per_second": 4.211,
|
| 779 |
+
"step": 1275
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"best_epoch": 24,
|
| 783 |
+
"best_eval_accuracy": 0.73,
|
| 784 |
+
"epoch": 51.0,
|
| 785 |
+
"step": 1275
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 52.0,
|
| 789 |
+
"eval_accuracy": 0.73,
|
| 790 |
+
"eval_loss": 0.3911306858062744,
|
| 791 |
+
"eval_runtime": 3.0857,
|
| 792 |
+
"eval_samples_per_second": 32.407,
|
| 793 |
+
"eval_steps_per_second": 4.213,
|
| 794 |
+
"step": 1300
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"best_epoch": 24,
|
| 798 |
+
"best_eval_accuracy": 0.73,
|
| 799 |
+
"epoch": 52.0,
|
| 800 |
+
"step": 1300
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 53.0,
|
| 804 |
+
"eval_accuracy": 0.73,
|
| 805 |
+
"eval_loss": 0.3854801654815674,
|
| 806 |
+
"eval_runtime": 3.0878,
|
| 807 |
+
"eval_samples_per_second": 32.385,
|
| 808 |
+
"eval_steps_per_second": 4.21,
|
| 809 |
+
"step": 1325
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"best_epoch": 24,
|
| 813 |
+
"best_eval_accuracy": 0.73,
|
| 814 |
+
"epoch": 53.0,
|
| 815 |
+
"step": 1325
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 54.0,
|
| 819 |
+
"eval_accuracy": 0.72,
|
| 820 |
+
"eval_loss": 0.3934061527252197,
|
| 821 |
+
"eval_runtime": 3.0865,
|
| 822 |
+
"eval_samples_per_second": 32.399,
|
| 823 |
+
"eval_steps_per_second": 4.212,
|
| 824 |
+
"step": 1350
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"best_epoch": 24,
|
| 828 |
+
"best_eval_accuracy": 0.73,
|
| 829 |
+
"epoch": 54.0,
|
| 830 |
+
"step": 1350
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 55.0,
|
| 834 |
+
"eval_accuracy": 0.73,
|
| 835 |
+
"eval_loss": 0.40291014313697815,
|
| 836 |
+
"eval_runtime": 3.0864,
|
| 837 |
+
"eval_samples_per_second": 32.401,
|
| 838 |
+
"eval_steps_per_second": 4.212,
|
| 839 |
+
"step": 1375
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"best_epoch": 24,
|
| 843 |
+
"best_eval_accuracy": 0.73,
|
| 844 |
+
"epoch": 55.0,
|
| 845 |
+
"step": 1375
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 56.0,
|
| 849 |
+
"eval_accuracy": 0.73,
|
| 850 |
+
"eval_loss": 0.38784506916999817,
|
| 851 |
+
"eval_runtime": 3.0835,
|
| 852 |
+
"eval_samples_per_second": 32.43,
|
| 853 |
+
"eval_steps_per_second": 4.216,
|
| 854 |
+
"step": 1400
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"best_epoch": 24,
|
| 858 |
+
"best_eval_accuracy": 0.73,
|
| 859 |
+
"epoch": 56.0,
|
| 860 |
+
"step": 1400
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 57.0,
|
| 864 |
+
"eval_accuracy": 0.72,
|
| 865 |
+
"eval_loss": 0.3839341104030609,
|
| 866 |
+
"eval_runtime": 3.0855,
|
| 867 |
+
"eval_samples_per_second": 32.41,
|
| 868 |
+
"eval_steps_per_second": 4.213,
|
| 869 |
+
"step": 1425
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"best_epoch": 24,
|
| 873 |
+
"best_eval_accuracy": 0.73,
|
| 874 |
+
"epoch": 57.0,
|
| 875 |
+
"step": 1425
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 58.0,
|
| 879 |
+
"eval_accuracy": 0.75,
|
| 880 |
+
"eval_loss": 0.3942648768424988,
|
| 881 |
+
"eval_runtime": 3.0854,
|
| 882 |
+
"eval_samples_per_second": 32.411,
|
| 883 |
+
"eval_steps_per_second": 4.213,
|
| 884 |
+
"step": 1450
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"best_epoch": 57,
|
| 888 |
+
"best_eval_accuracy": 0.75,
|
| 889 |
+
"epoch": 58.0,
|
| 890 |
+
"step": 1450
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 59.0,
|
| 894 |
+
"eval_accuracy": 0.74,
|
| 895 |
+
"eval_loss": 0.3984474539756775,
|
| 896 |
+
"eval_runtime": 3.0848,
|
| 897 |
+
"eval_samples_per_second": 32.417,
|
| 898 |
+
"eval_steps_per_second": 4.214,
|
| 899 |
+
"step": 1475
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"best_epoch": 57,
|
| 903 |
+
"best_eval_accuracy": 0.75,
|
| 904 |
+
"epoch": 59.0,
|
| 905 |
+
"step": 1475
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 60.0,
|
| 909 |
+
"learning_rate": 0.0025,
|
| 910 |
+
"loss": 0.4121,
|
| 911 |
+
"step": 1500
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 60.0,
|
| 915 |
+
"eval_accuracy": 0.71,
|
| 916 |
+
"eval_loss": 0.4064193069934845,
|
| 917 |
+
"eval_runtime": 3.0855,
|
| 918 |
+
"eval_samples_per_second": 32.41,
|
| 919 |
+
"eval_steps_per_second": 4.213,
|
| 920 |
+
"step": 1500
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"best_epoch": 57,
|
| 924 |
+
"best_eval_accuracy": 0.75,
|
| 925 |
+
"epoch": 60.0,
|
| 926 |
+
"step": 1500
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 61.0,
|
| 930 |
+
"eval_accuracy": 0.72,
|
| 931 |
+
"eval_loss": 0.3871249258518219,
|
| 932 |
+
"eval_runtime": 3.0854,
|
| 933 |
+
"eval_samples_per_second": 32.411,
|
| 934 |
+
"eval_steps_per_second": 4.213,
|
| 935 |
+
"step": 1525
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"best_epoch": 57,
|
| 939 |
+
"best_eval_accuracy": 0.75,
|
| 940 |
+
"epoch": 61.0,
|
| 941 |
+
"step": 1525
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 62.0,
|
| 945 |
+
"eval_accuracy": 0.73,
|
| 946 |
+
"eval_loss": 0.41410523653030396,
|
| 947 |
+
"eval_runtime": 3.0839,
|
| 948 |
+
"eval_samples_per_second": 32.427,
|
| 949 |
+
"eval_steps_per_second": 4.215,
|
| 950 |
+
"step": 1550
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"best_epoch": 57,
|
| 954 |
+
"best_eval_accuracy": 0.75,
|
| 955 |
+
"epoch": 62.0,
|
| 956 |
+
"step": 1550
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 63.0,
|
| 960 |
+
"eval_accuracy": 0.72,
|
| 961 |
+
"eval_loss": 0.38500529527664185,
|
| 962 |
+
"eval_runtime": 3.0874,
|
| 963 |
+
"eval_samples_per_second": 32.39,
|
| 964 |
+
"eval_steps_per_second": 4.211,
|
| 965 |
+
"step": 1575
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"best_epoch": 57,
|
| 969 |
+
"best_eval_accuracy": 0.75,
|
| 970 |
+
"epoch": 63.0,
|
| 971 |
+
"step": 1575
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 64.0,
|
| 975 |
+
"eval_accuracy": 0.73,
|
| 976 |
+
"eval_loss": 0.3933054804801941,
|
| 977 |
+
"eval_runtime": 3.0858,
|
| 978 |
+
"eval_samples_per_second": 32.406,
|
| 979 |
+
"eval_steps_per_second": 4.213,
|
| 980 |
+
"step": 1600
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"best_epoch": 57,
|
| 984 |
+
"best_eval_accuracy": 0.75,
|
| 985 |
+
"epoch": 64.0,
|
| 986 |
+
"step": 1600
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 65.0,
|
| 990 |
+
"eval_accuracy": 0.72,
|
| 991 |
+
"eval_loss": 0.40550005435943604,
|
| 992 |
+
"eval_runtime": 3.0847,
|
| 993 |
+
"eval_samples_per_second": 32.418,
|
| 994 |
+
"eval_steps_per_second": 4.214,
|
| 995 |
+
"step": 1625
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"best_epoch": 57,
|
| 999 |
+
"best_eval_accuracy": 0.75,
|
| 1000 |
+
"epoch": 65.0,
|
| 1001 |
+
"step": 1625
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 66.0,
|
| 1005 |
+
"eval_accuracy": 0.72,
|
| 1006 |
+
"eval_loss": 0.38515716791152954,
|
| 1007 |
+
"eval_runtime": 3.084,
|
| 1008 |
+
"eval_samples_per_second": 32.425,
|
| 1009 |
+
"eval_steps_per_second": 4.215,
|
| 1010 |
+
"step": 1650
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"best_epoch": 57,
|
| 1014 |
+
"best_eval_accuracy": 0.75,
|
| 1015 |
+
"epoch": 66.0,
|
| 1016 |
+
"step": 1650
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 67.0,
|
| 1020 |
+
"eval_accuracy": 0.73,
|
| 1021 |
+
"eval_loss": 0.39516574144363403,
|
| 1022 |
+
"eval_runtime": 3.0876,
|
| 1023 |
+
"eval_samples_per_second": 32.387,
|
| 1024 |
+
"eval_steps_per_second": 4.21,
|
| 1025 |
+
"step": 1675
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"best_epoch": 57,
|
| 1029 |
+
"best_eval_accuracy": 0.75,
|
| 1030 |
+
"epoch": 67.0,
|
| 1031 |
+
"step": 1675
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 68.0,
|
| 1035 |
+
"eval_accuracy": 0.72,
|
| 1036 |
+
"eval_loss": 0.38739094138145447,
|
| 1037 |
+
"eval_runtime": 3.0845,
|
| 1038 |
+
"eval_samples_per_second": 32.42,
|
| 1039 |
+
"eval_steps_per_second": 4.215,
|
| 1040 |
+
"step": 1700
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"best_epoch": 57,
|
| 1044 |
+
"best_eval_accuracy": 0.75,
|
| 1045 |
+
"epoch": 68.0,
|
| 1046 |
+
"step": 1700
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 69.0,
|
| 1050 |
+
"eval_accuracy": 0.72,
|
| 1051 |
+
"eval_loss": 0.3999227285385132,
|
| 1052 |
+
"eval_runtime": 3.0852,
|
| 1053 |
+
"eval_samples_per_second": 32.413,
|
| 1054 |
+
"eval_steps_per_second": 4.214,
|
| 1055 |
+
"step": 1725
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"best_epoch": 57,
|
| 1059 |
+
"best_eval_accuracy": 0.75,
|
| 1060 |
+
"epoch": 69.0,
|
| 1061 |
+
"step": 1725
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 70.0,
|
| 1065 |
+
"eval_accuracy": 0.72,
|
| 1066 |
+
"eval_loss": 0.39561355113983154,
|
| 1067 |
+
"eval_runtime": 3.0845,
|
| 1068 |
+
"eval_samples_per_second": 32.42,
|
| 1069 |
+
"eval_steps_per_second": 4.215,
|
| 1070 |
+
"step": 1750
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"best_epoch": 57,
|
| 1074 |
+
"best_eval_accuracy": 0.75,
|
| 1075 |
+
"epoch": 70.0,
|
| 1076 |
+
"step": 1750
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 71.0,
|
| 1080 |
+
"eval_accuracy": 0.72,
|
| 1081 |
+
"eval_loss": 0.39178183674812317,
|
| 1082 |
+
"eval_runtime": 3.0849,
|
| 1083 |
+
"eval_samples_per_second": 32.416,
|
| 1084 |
+
"eval_steps_per_second": 4.214,
|
| 1085 |
+
"step": 1775
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"best_epoch": 57,
|
| 1089 |
+
"best_eval_accuracy": 0.75,
|
| 1090 |
+
"epoch": 71.0,
|
| 1091 |
+
"step": 1775
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 72.0,
|
| 1095 |
+
"eval_accuracy": 0.72,
|
| 1096 |
+
"eval_loss": 0.38587287068367004,
|
| 1097 |
+
"eval_runtime": 3.0854,
|
| 1098 |
+
"eval_samples_per_second": 32.411,
|
| 1099 |
+
"eval_steps_per_second": 4.213,
|
| 1100 |
+
"step": 1800
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"best_epoch": 57,
|
| 1104 |
+
"best_eval_accuracy": 0.75,
|
| 1105 |
+
"epoch": 72.0,
|
| 1106 |
+
"step": 1800
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 73.0,
|
| 1110 |
+
"eval_accuracy": 0.72,
|
| 1111 |
+
"eval_loss": 0.39257708191871643,
|
| 1112 |
+
"eval_runtime": 3.0824,
|
| 1113 |
+
"eval_samples_per_second": 32.443,
|
| 1114 |
+
"eval_steps_per_second": 4.218,
|
| 1115 |
+
"step": 1825
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"best_epoch": 57,
|
| 1119 |
+
"best_eval_accuracy": 0.75,
|
| 1120 |
+
"epoch": 73.0,
|
| 1121 |
+
"step": 1825
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 74.0,
|
| 1125 |
+
"eval_accuracy": 0.72,
|
| 1126 |
+
"eval_loss": 0.38968443870544434,
|
| 1127 |
+
"eval_runtime": 3.084,
|
| 1128 |
+
"eval_samples_per_second": 32.425,
|
| 1129 |
+
"eval_steps_per_second": 4.215,
|
| 1130 |
+
"step": 1850
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"best_epoch": 57,
|
| 1134 |
+
"best_eval_accuracy": 0.75,
|
| 1135 |
+
"epoch": 74.0,
|
| 1136 |
+
"step": 1850
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 75.0,
|
| 1140 |
+
"eval_accuracy": 0.72,
|
| 1141 |
+
"eval_loss": 0.3859497010707855,
|
| 1142 |
+
"eval_runtime": 3.0819,
|
| 1143 |
+
"eval_samples_per_second": 32.447,
|
| 1144 |
+
"eval_steps_per_second": 4.218,
|
| 1145 |
+
"step": 1875
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"best_epoch": 57,
|
| 1149 |
+
"best_eval_accuracy": 0.75,
|
| 1150 |
+
"epoch": 75.0,
|
| 1151 |
+
"step": 1875
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 76.0,
|
| 1155 |
+
"eval_accuracy": 0.72,
|
| 1156 |
+
"eval_loss": 0.38490238785743713,
|
| 1157 |
+
"eval_runtime": 3.083,
|
| 1158 |
+
"eval_samples_per_second": 32.436,
|
| 1159 |
+
"eval_steps_per_second": 4.217,
|
| 1160 |
+
"step": 1900
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"best_epoch": 57,
|
| 1164 |
+
"best_eval_accuracy": 0.75,
|
| 1165 |
+
"epoch": 76.0,
|
| 1166 |
+
"step": 1900
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 77.0,
|
| 1170 |
+
"eval_accuracy": 0.72,
|
| 1171 |
+
"eval_loss": 0.3855893313884735,
|
| 1172 |
+
"eval_runtime": 3.0824,
|
| 1173 |
+
"eval_samples_per_second": 32.442,
|
| 1174 |
+
"eval_steps_per_second": 4.217,
|
| 1175 |
+
"step": 1925
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"best_epoch": 57,
|
| 1179 |
+
"best_eval_accuracy": 0.75,
|
| 1180 |
+
"epoch": 77.0,
|
| 1181 |
+
"step": 1925
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 78.0,
|
| 1185 |
+
"eval_accuracy": 0.72,
|
| 1186 |
+
"eval_loss": 0.3901687562465668,
|
| 1187 |
+
"eval_runtime": 3.0824,
|
| 1188 |
+
"eval_samples_per_second": 32.442,
|
| 1189 |
+
"eval_steps_per_second": 4.217,
|
| 1190 |
+
"step": 1950
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"best_epoch": 57,
|
| 1194 |
+
"best_eval_accuracy": 0.75,
|
| 1195 |
+
"epoch": 78.0,
|
| 1196 |
+
"step": 1950
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 79.0,
|
| 1200 |
+
"eval_accuracy": 0.72,
|
| 1201 |
+
"eval_loss": 0.3903580904006958,
|
| 1202 |
+
"eval_runtime": 3.0842,
|
| 1203 |
+
"eval_samples_per_second": 32.423,
|
| 1204 |
+
"eval_steps_per_second": 4.215,
|
| 1205 |
+
"step": 1975
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"best_epoch": 57,
|
| 1209 |
+
"best_eval_accuracy": 0.75,
|
| 1210 |
+
"epoch": 79.0,
|
| 1211 |
+
"step": 1975
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 80.0,
|
| 1215 |
+
"learning_rate": 0.0,
|
| 1216 |
+
"loss": 0.3881,
|
| 1217 |
+
"step": 2000
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 80.0,
|
| 1221 |
+
"eval_accuracy": 0.72,
|
| 1222 |
+
"eval_loss": 0.3895449936389923,
|
| 1223 |
+
"eval_runtime": 3.0918,
|
| 1224 |
+
"eval_samples_per_second": 32.344,
|
| 1225 |
+
"eval_steps_per_second": 4.205,
|
| 1226 |
+
"step": 2000
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"best_epoch": 57,
|
| 1230 |
+
"best_eval_accuracy": 0.75,
|
| 1231 |
+
"epoch": 80.0,
|
| 1232 |
+
"step": 2000
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 80.0,
|
| 1236 |
+
"step": 2000,
|
| 1237 |
+
"total_flos": 2.9821702864896e+16,
|
| 1238 |
+
"train_loss": 0.47253459167480466,
|
| 1239 |
+
"train_runtime": 1645.8876,
|
| 1240 |
+
"train_samples_per_second": 19.442,
|
| 1241 |
+
"train_steps_per_second": 1.215
|
| 1242 |
+
}
|
| 1243 |
+
],
|
| 1244 |
+
"max_steps": 2000,
|
| 1245 |
+
"num_train_epochs": 80,
|
| 1246 |
+
"total_flos": 2.9821702864896e+16,
|
| 1247 |
+
"trial_name": null,
|
| 1248 |
+
"trial_params": null
|
| 1249 |
+
}
|