End of training
Browse files- all_results.json +13 -0
- eval_results.json +8 -0
- runs/Jul14_20-42-56_c2d4aa47cd7f/events.out.tfevents.1689378766.c2d4aa47cd7f.307.1 +3 -0
- train_results.json +8 -0
- trainer_state.json +2995 -0
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 12.0,
|
| 3 |
+
"eval_accuracy": 0.9938708156529938,
|
| 4 |
+
"eval_loss": 0.0275154709815979,
|
| 5 |
+
"eval_runtime": 91.4223,
|
| 6 |
+
"eval_samples_per_second": 23.2,
|
| 7 |
+
"eval_steps_per_second": 1.455,
|
| 8 |
+
"total_flos": 2.0923685902916665e+19,
|
| 9 |
+
"train_loss": 0.1486589150943784,
|
| 10 |
+
"train_runtime": 11149.9067,
|
| 11 |
+
"train_samples_per_second": 6.848,
|
| 12 |
+
"train_steps_per_second": 0.214
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 12.0,
|
| 3 |
+
"eval_accuracy": 0.9938708156529938,
|
| 4 |
+
"eval_loss": 0.0275154709815979,
|
| 5 |
+
"eval_runtime": 91.4223,
|
| 6 |
+
"eval_samples_per_second": 23.2,
|
| 7 |
+
"eval_steps_per_second": 1.455
|
| 8 |
+
}
|
runs/Jul14_20-42-56_c2d4aa47cd7f/events.out.tfevents.1689378766.c2d4aa47cd7f.307.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6afcd5fb08261c1b13220e3d66fa563cdb52e837f27cbd0387874e499e46ebec
|
| 3 |
+
size 411
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 12.0,
|
| 3 |
+
"total_flos": 2.0923685902916665e+19,
|
| 4 |
+
"train_loss": 0.1486589150943784,
|
| 5 |
+
"train_runtime": 11149.9067,
|
| 6 |
+
"train_samples_per_second": 6.848,
|
| 7 |
+
"train_steps_per_second": 0.214
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2995 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9938708156529938,
|
| 3 |
+
"best_model_checkpoint": "beit-large-patch16-224-finetuned-BreastCancer-Classification-BreakHis-AH-60-20-20/checkpoint-2388",
|
| 4 |
+
"epoch": 12.0,
|
| 5 |
+
"global_step": 2388,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.03,
|
| 12 |
+
"learning_rate": 1.1627906976744186e-08,
|
| 13 |
+
"loss": 0.7335,
|
| 14 |
+
"step": 5
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.05,
|
| 18 |
+
"learning_rate": 2.3255813953488372e-08,
|
| 19 |
+
"loss": 0.7745,
|
| 20 |
+
"step": 10
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.08,
|
| 24 |
+
"learning_rate": 3.488372093023256e-08,
|
| 25 |
+
"loss": 0.7823,
|
| 26 |
+
"step": 15
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.1,
|
| 30 |
+
"learning_rate": 4.6511627906976744e-08,
|
| 31 |
+
"loss": 0.7606,
|
| 32 |
+
"step": 20
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.13,
|
| 36 |
+
"learning_rate": 5.8139534883720935e-08,
|
| 37 |
+
"loss": 0.7698,
|
| 38 |
+
"step": 25
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.15,
|
| 42 |
+
"learning_rate": 6.976744186046512e-08,
|
| 43 |
+
"loss": 0.7576,
|
| 44 |
+
"step": 30
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.18,
|
| 48 |
+
"learning_rate": 8.139534883720931e-08,
|
| 49 |
+
"loss": 0.7102,
|
| 50 |
+
"step": 35
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.2,
|
| 54 |
+
"learning_rate": 9.302325581395349e-08,
|
| 55 |
+
"loss": 0.7435,
|
| 56 |
+
"step": 40
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.23,
|
| 60 |
+
"learning_rate": 1.0465116279069768e-07,
|
| 61 |
+
"loss": 0.7648,
|
| 62 |
+
"step": 45
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.25,
|
| 66 |
+
"learning_rate": 1.1627906976744187e-07,
|
| 67 |
+
"loss": 0.7704,
|
| 68 |
+
"step": 50
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.28,
|
| 72 |
+
"learning_rate": 1.2790697674418605e-07,
|
| 73 |
+
"loss": 0.688,
|
| 74 |
+
"step": 55
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.3,
|
| 78 |
+
"learning_rate": 1.3953488372093024e-07,
|
| 79 |
+
"loss": 0.7343,
|
| 80 |
+
"step": 60
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.33,
|
| 84 |
+
"learning_rate": 1.5116279069767443e-07,
|
| 85 |
+
"loss": 0.7055,
|
| 86 |
+
"step": 65
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.35,
|
| 90 |
+
"learning_rate": 1.6279069767441862e-07,
|
| 91 |
+
"loss": 0.7237,
|
| 92 |
+
"step": 70
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.38,
|
| 96 |
+
"learning_rate": 1.744186046511628e-07,
|
| 97 |
+
"loss": 0.7374,
|
| 98 |
+
"step": 75
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.4,
|
| 102 |
+
"learning_rate": 1.8604651162790698e-07,
|
| 103 |
+
"loss": 0.653,
|
| 104 |
+
"step": 80
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.43,
|
| 108 |
+
"learning_rate": 1.9767441860465117e-07,
|
| 109 |
+
"loss": 0.6931,
|
| 110 |
+
"step": 85
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.45,
|
| 114 |
+
"learning_rate": 2.0930232558139536e-07,
|
| 115 |
+
"loss": 0.638,
|
| 116 |
+
"step": 90
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.48,
|
| 120 |
+
"learning_rate": 2.2093023255813955e-07,
|
| 121 |
+
"loss": 0.6868,
|
| 122 |
+
"step": 95
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.5,
|
| 126 |
+
"learning_rate": 2.3255813953488374e-07,
|
| 127 |
+
"loss": 0.716,
|
| 128 |
+
"step": 100
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.53,
|
| 132 |
+
"learning_rate": 2.4418604651162793e-07,
|
| 133 |
+
"loss": 0.725,
|
| 134 |
+
"step": 105
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.55,
|
| 138 |
+
"learning_rate": 2.558139534883721e-07,
|
| 139 |
+
"loss": 0.6195,
|
| 140 |
+
"step": 110
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.58,
|
| 144 |
+
"learning_rate": 2.674418604651163e-07,
|
| 145 |
+
"loss": 0.5791,
|
| 146 |
+
"step": 115
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.6,
|
| 150 |
+
"learning_rate": 2.790697674418605e-07,
|
| 151 |
+
"loss": 0.6498,
|
| 152 |
+
"step": 120
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.63,
|
| 156 |
+
"learning_rate": 2.906976744186047e-07,
|
| 157 |
+
"loss": 0.6431,
|
| 158 |
+
"step": 125
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.65,
|
| 162 |
+
"learning_rate": 3.0232558139534886e-07,
|
| 163 |
+
"loss": 0.5811,
|
| 164 |
+
"step": 130
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.68,
|
| 168 |
+
"learning_rate": 3.139534883720931e-07,
|
| 169 |
+
"loss": 0.5803,
|
| 170 |
+
"step": 135
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.7,
|
| 174 |
+
"learning_rate": 3.2558139534883724e-07,
|
| 175 |
+
"loss": 0.5635,
|
| 176 |
+
"step": 140
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.73,
|
| 180 |
+
"learning_rate": 3.3720930232558146e-07,
|
| 181 |
+
"loss": 0.5369,
|
| 182 |
+
"step": 145
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.75,
|
| 186 |
+
"learning_rate": 3.488372093023256e-07,
|
| 187 |
+
"loss": 0.5395,
|
| 188 |
+
"step": 150
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.78,
|
| 192 |
+
"learning_rate": 3.6046511627906984e-07,
|
| 193 |
+
"loss": 0.5645,
|
| 194 |
+
"step": 155
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.8,
|
| 198 |
+
"learning_rate": 3.7209302325581396e-07,
|
| 199 |
+
"loss": 0.5327,
|
| 200 |
+
"step": 160
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.83,
|
| 204 |
+
"learning_rate": 3.837209302325582e-07,
|
| 205 |
+
"loss": 0.5035,
|
| 206 |
+
"step": 165
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.85,
|
| 210 |
+
"learning_rate": 3.9534883720930234e-07,
|
| 211 |
+
"loss": 0.5256,
|
| 212 |
+
"step": 170
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.88,
|
| 216 |
+
"learning_rate": 4.0697674418604655e-07,
|
| 217 |
+
"loss": 0.5225,
|
| 218 |
+
"step": 175
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.9,
|
| 222 |
+
"learning_rate": 4.186046511627907e-07,
|
| 223 |
+
"loss": 0.5197,
|
| 224 |
+
"step": 180
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.93,
|
| 228 |
+
"learning_rate": 4.3023255813953494e-07,
|
| 229 |
+
"loss": 0.5052,
|
| 230 |
+
"step": 185
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.95,
|
| 234 |
+
"learning_rate": 4.418604651162791e-07,
|
| 235 |
+
"loss": 0.4875,
|
| 236 |
+
"step": 190
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.98,
|
| 240 |
+
"learning_rate": 4.534883720930233e-07,
|
| 241 |
+
"loss": 0.46,
|
| 242 |
+
"step": 195
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.0,
|
| 246 |
+
"eval_accuracy": 0.8481848184818482,
|
| 247 |
+
"eval_loss": 0.39499393105506897,
|
| 248 |
+
"eval_runtime": 90.9996,
|
| 249 |
+
"eval_samples_per_second": 23.308,
|
| 250 |
+
"eval_steps_per_second": 1.462,
|
| 251 |
+
"step": 199
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 1.01,
|
| 255 |
+
"learning_rate": 4.651162790697675e-07,
|
| 256 |
+
"loss": 0.5344,
|
| 257 |
+
"step": 200
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 1.03,
|
| 261 |
+
"learning_rate": 4.767441860465117e-07,
|
| 262 |
+
"loss": 0.4448,
|
| 263 |
+
"step": 205
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.06,
|
| 267 |
+
"learning_rate": 4.883720930232559e-07,
|
| 268 |
+
"loss": 0.5011,
|
| 269 |
+
"step": 210
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 1.08,
|
| 273 |
+
"learning_rate": 5.000000000000001e-07,
|
| 274 |
+
"loss": 0.4055,
|
| 275 |
+
"step": 215
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 1.11,
|
| 279 |
+
"learning_rate": 5.116279069767442e-07,
|
| 280 |
+
"loss": 0.4714,
|
| 281 |
+
"step": 220
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 1.13,
|
| 285 |
+
"learning_rate": 5.232558139534884e-07,
|
| 286 |
+
"loss": 0.4132,
|
| 287 |
+
"step": 225
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 1.16,
|
| 291 |
+
"learning_rate": 5.348837209302326e-07,
|
| 292 |
+
"loss": 0.4394,
|
| 293 |
+
"step": 230
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 1.18,
|
| 297 |
+
"learning_rate": 5.465116279069768e-07,
|
| 298 |
+
"loss": 0.3649,
|
| 299 |
+
"step": 235
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 1.21,
|
| 303 |
+
"learning_rate": 5.58139534883721e-07,
|
| 304 |
+
"loss": 0.3948,
|
| 305 |
+
"step": 240
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 1.23,
|
| 309 |
+
"learning_rate": 5.697674418604652e-07,
|
| 310 |
+
"loss": 0.4236,
|
| 311 |
+
"step": 245
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 1.26,
|
| 315 |
+
"learning_rate": 5.813953488372094e-07,
|
| 316 |
+
"loss": 0.4115,
|
| 317 |
+
"step": 250
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 1.28,
|
| 321 |
+
"learning_rate": 5.930232558139536e-07,
|
| 322 |
+
"loss": 0.465,
|
| 323 |
+
"step": 255
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 1.31,
|
| 327 |
+
"learning_rate": 6.046511627906977e-07,
|
| 328 |
+
"loss": 0.3335,
|
| 329 |
+
"step": 260
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 1.33,
|
| 333 |
+
"learning_rate": 6.162790697674419e-07,
|
| 334 |
+
"loss": 0.3746,
|
| 335 |
+
"step": 265
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 1.36,
|
| 339 |
+
"learning_rate": 6.279069767441862e-07,
|
| 340 |
+
"loss": 0.3266,
|
| 341 |
+
"step": 270
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 1.38,
|
| 345 |
+
"learning_rate": 6.395348837209303e-07,
|
| 346 |
+
"loss": 0.3936,
|
| 347 |
+
"step": 275
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 1.41,
|
| 351 |
+
"learning_rate": 6.511627906976745e-07,
|
| 352 |
+
"loss": 0.3193,
|
| 353 |
+
"step": 280
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 1.43,
|
| 357 |
+
"learning_rate": 6.627906976744186e-07,
|
| 358 |
+
"loss": 0.3624,
|
| 359 |
+
"step": 285
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 1.46,
|
| 363 |
+
"learning_rate": 6.744186046511629e-07,
|
| 364 |
+
"loss": 0.303,
|
| 365 |
+
"step": 290
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 1.48,
|
| 369 |
+
"learning_rate": 6.86046511627907e-07,
|
| 370 |
+
"loss": 0.3427,
|
| 371 |
+
"step": 295
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 1.51,
|
| 375 |
+
"learning_rate": 6.976744186046513e-07,
|
| 376 |
+
"loss": 0.2622,
|
| 377 |
+
"step": 300
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 1.53,
|
| 381 |
+
"learning_rate": 7.093023255813954e-07,
|
| 382 |
+
"loss": 0.2665,
|
| 383 |
+
"step": 305
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 1.56,
|
| 387 |
+
"learning_rate": 7.209302325581397e-07,
|
| 388 |
+
"loss": 0.3244,
|
| 389 |
+
"step": 310
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 1.58,
|
| 393 |
+
"learning_rate": 7.325581395348838e-07,
|
| 394 |
+
"loss": 0.2232,
|
| 395 |
+
"step": 315
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 1.61,
|
| 399 |
+
"learning_rate": 7.441860465116279e-07,
|
| 400 |
+
"loss": 0.2798,
|
| 401 |
+
"step": 320
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 1.63,
|
| 405 |
+
"learning_rate": 7.558139534883721e-07,
|
| 406 |
+
"loss": 0.261,
|
| 407 |
+
"step": 325
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 1.66,
|
| 411 |
+
"learning_rate": 7.674418604651165e-07,
|
| 412 |
+
"loss": 0.2758,
|
| 413 |
+
"step": 330
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 1.68,
|
| 417 |
+
"learning_rate": 7.790697674418606e-07,
|
| 418 |
+
"loss": 0.3212,
|
| 419 |
+
"step": 335
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 1.71,
|
| 423 |
+
"learning_rate": 7.906976744186047e-07,
|
| 424 |
+
"loss": 0.2536,
|
| 425 |
+
"step": 340
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 1.73,
|
| 429 |
+
"learning_rate": 8.023255813953489e-07,
|
| 430 |
+
"loss": 0.2719,
|
| 431 |
+
"step": 345
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 1.76,
|
| 435 |
+
"learning_rate": 8.139534883720931e-07,
|
| 436 |
+
"loss": 0.2754,
|
| 437 |
+
"step": 350
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 1.78,
|
| 441 |
+
"learning_rate": 8.255813953488373e-07,
|
| 442 |
+
"loss": 0.187,
|
| 443 |
+
"step": 355
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 1.81,
|
| 447 |
+
"learning_rate": 8.372093023255814e-07,
|
| 448 |
+
"loss": 0.2383,
|
| 449 |
+
"step": 360
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 1.83,
|
| 453 |
+
"learning_rate": 8.488372093023257e-07,
|
| 454 |
+
"loss": 0.3075,
|
| 455 |
+
"step": 365
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 1.86,
|
| 459 |
+
"learning_rate": 8.604651162790699e-07,
|
| 460 |
+
"loss": 0.2802,
|
| 461 |
+
"step": 370
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 1.88,
|
| 465 |
+
"learning_rate": 8.720930232558141e-07,
|
| 466 |
+
"loss": 0.2815,
|
| 467 |
+
"step": 375
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 1.91,
|
| 471 |
+
"learning_rate": 8.837209302325582e-07,
|
| 472 |
+
"loss": 0.2916,
|
| 473 |
+
"step": 380
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 1.93,
|
| 477 |
+
"learning_rate": 8.953488372093023e-07,
|
| 478 |
+
"loss": 0.2769,
|
| 479 |
+
"step": 385
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 1.96,
|
| 483 |
+
"learning_rate": 9.069767441860466e-07,
|
| 484 |
+
"loss": 0.2174,
|
| 485 |
+
"step": 390
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 1.98,
|
| 489 |
+
"learning_rate": 9.186046511627907e-07,
|
| 490 |
+
"loss": 0.2048,
|
| 491 |
+
"step": 395
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 2.0,
|
| 495 |
+
"eval_accuracy": 0.918906176331919,
|
| 496 |
+
"eval_loss": 0.18860779702663422,
|
| 497 |
+
"eval_runtime": 91.0197,
|
| 498 |
+
"eval_samples_per_second": 23.303,
|
| 499 |
+
"eval_steps_per_second": 1.461,
|
| 500 |
+
"step": 398
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 2.01,
|
| 504 |
+
"learning_rate": 9.30232558139535e-07,
|
| 505 |
+
"loss": 0.2442,
|
| 506 |
+
"step": 400
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 2.04,
|
| 510 |
+
"learning_rate": 9.418604651162791e-07,
|
| 511 |
+
"loss": 0.2001,
|
| 512 |
+
"step": 405
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 2.06,
|
| 516 |
+
"learning_rate": 9.534883720930234e-07,
|
| 517 |
+
"loss": 0.2522,
|
| 518 |
+
"step": 410
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 2.09,
|
| 522 |
+
"learning_rate": 9.651162790697675e-07,
|
| 523 |
+
"loss": 0.1624,
|
| 524 |
+
"step": 415
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 2.11,
|
| 528 |
+
"learning_rate": 9.767441860465117e-07,
|
| 529 |
+
"loss": 0.2131,
|
| 530 |
+
"step": 420
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 2.14,
|
| 534 |
+
"learning_rate": 9.88372093023256e-07,
|
| 535 |
+
"loss": 0.1761,
|
| 536 |
+
"step": 425
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 2.16,
|
| 540 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 541 |
+
"loss": 0.2182,
|
| 542 |
+
"step": 430
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 2.19,
|
| 546 |
+
"learning_rate": 1.0116279069767444e-06,
|
| 547 |
+
"loss": 0.2105,
|
| 548 |
+
"step": 435
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 2.21,
|
| 552 |
+
"learning_rate": 1.0232558139534884e-06,
|
| 553 |
+
"loss": 0.2178,
|
| 554 |
+
"step": 440
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 2.24,
|
| 558 |
+
"learning_rate": 1.0348837209302326e-06,
|
| 559 |
+
"loss": 0.1795,
|
| 560 |
+
"step": 445
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 2.26,
|
| 564 |
+
"learning_rate": 1.0465116279069768e-06,
|
| 565 |
+
"loss": 0.1685,
|
| 566 |
+
"step": 450
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 2.29,
|
| 570 |
+
"learning_rate": 1.058139534883721e-06,
|
| 571 |
+
"loss": 0.2405,
|
| 572 |
+
"step": 455
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 2.31,
|
| 576 |
+
"learning_rate": 1.0697674418604653e-06,
|
| 577 |
+
"loss": 0.1603,
|
| 578 |
+
"step": 460
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 2.34,
|
| 582 |
+
"learning_rate": 1.0813953488372093e-06,
|
| 583 |
+
"loss": 0.2294,
|
| 584 |
+
"step": 465
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 2.36,
|
| 588 |
+
"learning_rate": 1.0930232558139537e-06,
|
| 589 |
+
"loss": 0.2107,
|
| 590 |
+
"step": 470
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 2.39,
|
| 594 |
+
"learning_rate": 1.1046511627906977e-06,
|
| 595 |
+
"loss": 0.1622,
|
| 596 |
+
"step": 475
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 2.41,
|
| 600 |
+
"learning_rate": 1.116279069767442e-06,
|
| 601 |
+
"loss": 0.1739,
|
| 602 |
+
"step": 480
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 2.44,
|
| 606 |
+
"learning_rate": 1.1279069767441861e-06,
|
| 607 |
+
"loss": 0.2156,
|
| 608 |
+
"step": 485
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 2.46,
|
| 612 |
+
"learning_rate": 1.1395348837209304e-06,
|
| 613 |
+
"loss": 0.2137,
|
| 614 |
+
"step": 490
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 2.49,
|
| 618 |
+
"learning_rate": 1.1511627906976746e-06,
|
| 619 |
+
"loss": 0.2177,
|
| 620 |
+
"step": 495
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 2.51,
|
| 624 |
+
"learning_rate": 1.1627906976744188e-06,
|
| 625 |
+
"loss": 0.2202,
|
| 626 |
+
"step": 500
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 2.54,
|
| 630 |
+
"learning_rate": 1.1744186046511628e-06,
|
| 631 |
+
"loss": 0.1772,
|
| 632 |
+
"step": 505
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 2.56,
|
| 636 |
+
"learning_rate": 1.1860465116279072e-06,
|
| 637 |
+
"loss": 0.1435,
|
| 638 |
+
"step": 510
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 2.59,
|
| 642 |
+
"learning_rate": 1.1976744186046512e-06,
|
| 643 |
+
"loss": 0.1728,
|
| 644 |
+
"step": 515
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 2.61,
|
| 648 |
+
"learning_rate": 1.2093023255813954e-06,
|
| 649 |
+
"loss": 0.1686,
|
| 650 |
+
"step": 520
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 2.64,
|
| 654 |
+
"learning_rate": 1.2209302325581397e-06,
|
| 655 |
+
"loss": 0.1815,
|
| 656 |
+
"step": 525
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 2.66,
|
| 660 |
+
"learning_rate": 1.2325581395348839e-06,
|
| 661 |
+
"loss": 0.2048,
|
| 662 |
+
"step": 530
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 2.69,
|
| 666 |
+
"learning_rate": 1.244186046511628e-06,
|
| 667 |
+
"loss": 0.1522,
|
| 668 |
+
"step": 535
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 2.71,
|
| 672 |
+
"learning_rate": 1.2558139534883723e-06,
|
| 673 |
+
"loss": 0.1424,
|
| 674 |
+
"step": 540
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 2.74,
|
| 678 |
+
"learning_rate": 1.2674418604651165e-06,
|
| 679 |
+
"loss": 0.1503,
|
| 680 |
+
"step": 545
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 2.76,
|
| 684 |
+
"learning_rate": 1.2790697674418605e-06,
|
| 685 |
+
"loss": 0.2181,
|
| 686 |
+
"step": 550
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 2.79,
|
| 690 |
+
"learning_rate": 1.2906976744186048e-06,
|
| 691 |
+
"loss": 0.1607,
|
| 692 |
+
"step": 555
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 2.81,
|
| 696 |
+
"learning_rate": 1.302325581395349e-06,
|
| 697 |
+
"loss": 0.1129,
|
| 698 |
+
"step": 560
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 2.84,
|
| 702 |
+
"learning_rate": 1.313953488372093e-06,
|
| 703 |
+
"loss": 0.1496,
|
| 704 |
+
"step": 565
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 2.86,
|
| 708 |
+
"learning_rate": 1.3255813953488372e-06,
|
| 709 |
+
"loss": 0.1261,
|
| 710 |
+
"step": 570
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 2.89,
|
| 714 |
+
"learning_rate": 1.3372093023255814e-06,
|
| 715 |
+
"loss": 0.1951,
|
| 716 |
+
"step": 575
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 2.91,
|
| 720 |
+
"learning_rate": 1.3488372093023258e-06,
|
| 721 |
+
"loss": 0.2033,
|
| 722 |
+
"step": 580
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 2.94,
|
| 726 |
+
"learning_rate": 1.36046511627907e-06,
|
| 727 |
+
"loss": 0.2468,
|
| 728 |
+
"step": 585
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 2.96,
|
| 732 |
+
"learning_rate": 1.372093023255814e-06,
|
| 733 |
+
"loss": 0.1552,
|
| 734 |
+
"step": 590
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 2.99,
|
| 738 |
+
"learning_rate": 1.3837209302325583e-06,
|
| 739 |
+
"loss": 0.182,
|
| 740 |
+
"step": 595
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 3.0,
|
| 744 |
+
"eval_accuracy": 0.9481376709099482,
|
| 745 |
+
"eval_loss": 0.1382409781217575,
|
| 746 |
+
"eval_runtime": 91.1965,
|
| 747 |
+
"eval_samples_per_second": 23.257,
|
| 748 |
+
"eval_steps_per_second": 1.458,
|
| 749 |
+
"step": 597
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 3.02,
|
| 753 |
+
"learning_rate": 1.3953488372093025e-06,
|
| 754 |
+
"loss": 0.1781,
|
| 755 |
+
"step": 600
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 3.04,
|
| 759 |
+
"learning_rate": 1.4069767441860465e-06,
|
| 760 |
+
"loss": 0.2034,
|
| 761 |
+
"step": 605
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 3.07,
|
| 765 |
+
"learning_rate": 1.4186046511627907e-06,
|
| 766 |
+
"loss": 0.1188,
|
| 767 |
+
"step": 610
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 3.09,
|
| 771 |
+
"learning_rate": 1.430232558139535e-06,
|
| 772 |
+
"loss": 0.1816,
|
| 773 |
+
"step": 615
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 3.12,
|
| 777 |
+
"learning_rate": 1.4418604651162794e-06,
|
| 778 |
+
"loss": 0.1081,
|
| 779 |
+
"step": 620
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 3.14,
|
| 783 |
+
"learning_rate": 1.4534883720930234e-06,
|
| 784 |
+
"loss": 0.1409,
|
| 785 |
+
"step": 625
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 3.17,
|
| 789 |
+
"learning_rate": 1.4651162790697676e-06,
|
| 790 |
+
"loss": 0.1834,
|
| 791 |
+
"step": 630
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 3.19,
|
| 795 |
+
"learning_rate": 1.4767441860465118e-06,
|
| 796 |
+
"loss": 0.2097,
|
| 797 |
+
"step": 635
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 3.22,
|
| 801 |
+
"learning_rate": 1.4883720930232558e-06,
|
| 802 |
+
"loss": 0.145,
|
| 803 |
+
"step": 640
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 3.24,
|
| 807 |
+
"learning_rate": 1.5e-06,
|
| 808 |
+
"loss": 0.156,
|
| 809 |
+
"step": 645
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 3.27,
|
| 813 |
+
"learning_rate": 1.5116279069767443e-06,
|
| 814 |
+
"loss": 0.1742,
|
| 815 |
+
"step": 650
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 3.29,
|
| 819 |
+
"learning_rate": 1.5232558139534885e-06,
|
| 820 |
+
"loss": 0.1477,
|
| 821 |
+
"step": 655
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 3.32,
|
| 825 |
+
"learning_rate": 1.534883720930233e-06,
|
| 826 |
+
"loss": 0.0985,
|
| 827 |
+
"step": 660
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 3.34,
|
| 831 |
+
"learning_rate": 1.546511627906977e-06,
|
| 832 |
+
"loss": 0.1889,
|
| 833 |
+
"step": 665
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 3.37,
|
| 837 |
+
"learning_rate": 1.5581395348837211e-06,
|
| 838 |
+
"loss": 0.1832,
|
| 839 |
+
"step": 670
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 3.39,
|
| 843 |
+
"learning_rate": 1.5697674418604653e-06,
|
| 844 |
+
"loss": 0.1582,
|
| 845 |
+
"step": 675
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 3.42,
|
| 849 |
+
"learning_rate": 1.5813953488372093e-06,
|
| 850 |
+
"loss": 0.1872,
|
| 851 |
+
"step": 680
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 3.44,
|
| 855 |
+
"learning_rate": 1.5930232558139536e-06,
|
| 856 |
+
"loss": 0.1571,
|
| 857 |
+
"step": 685
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 3.47,
|
| 861 |
+
"learning_rate": 1.6046511627906978e-06,
|
| 862 |
+
"loss": 0.1407,
|
| 863 |
+
"step": 690
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 3.49,
|
| 867 |
+
"learning_rate": 1.6162790697674418e-06,
|
| 868 |
+
"loss": 0.0965,
|
| 869 |
+
"step": 695
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 3.52,
|
| 873 |
+
"learning_rate": 1.6279069767441862e-06,
|
| 874 |
+
"loss": 0.0725,
|
| 875 |
+
"step": 700
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 3.54,
|
| 879 |
+
"learning_rate": 1.6395348837209304e-06,
|
| 880 |
+
"loss": 0.1101,
|
| 881 |
+
"step": 705
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 3.57,
|
| 885 |
+
"learning_rate": 1.6511627906976747e-06,
|
| 886 |
+
"loss": 0.1048,
|
| 887 |
+
"step": 710
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 3.59,
|
| 891 |
+
"learning_rate": 1.6627906976744187e-06,
|
| 892 |
+
"loss": 0.1131,
|
| 893 |
+
"step": 715
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 3.62,
|
| 897 |
+
"learning_rate": 1.6744186046511629e-06,
|
| 898 |
+
"loss": 0.1339,
|
| 899 |
+
"step": 720
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 3.64,
|
| 903 |
+
"learning_rate": 1.686046511627907e-06,
|
| 904 |
+
"loss": 0.1134,
|
| 905 |
+
"step": 725
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 3.67,
|
| 909 |
+
"learning_rate": 1.6976744186046513e-06,
|
| 910 |
+
"loss": 0.1316,
|
| 911 |
+
"step": 730
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 3.69,
|
| 915 |
+
"learning_rate": 1.7093023255813953e-06,
|
| 916 |
+
"loss": 0.1277,
|
| 917 |
+
"step": 735
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 3.72,
|
| 921 |
+
"learning_rate": 1.7209302325581397e-06,
|
| 922 |
+
"loss": 0.1207,
|
| 923 |
+
"step": 740
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 3.74,
|
| 927 |
+
"learning_rate": 1.732558139534884e-06,
|
| 928 |
+
"loss": 0.1694,
|
| 929 |
+
"step": 745
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 3.77,
|
| 933 |
+
"learning_rate": 1.7441860465116282e-06,
|
| 934 |
+
"loss": 0.1486,
|
| 935 |
+
"step": 750
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 3.79,
|
| 939 |
+
"learning_rate": 1.7558139534883722e-06,
|
| 940 |
+
"loss": 0.0856,
|
| 941 |
+
"step": 755
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 3.82,
|
| 945 |
+
"learning_rate": 1.7674418604651164e-06,
|
| 946 |
+
"loss": 0.1068,
|
| 947 |
+
"step": 760
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 3.84,
|
| 951 |
+
"learning_rate": 1.7790697674418606e-06,
|
| 952 |
+
"loss": 0.0985,
|
| 953 |
+
"step": 765
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 3.87,
|
| 957 |
+
"learning_rate": 1.7906976744186046e-06,
|
| 958 |
+
"loss": 0.1445,
|
| 959 |
+
"step": 770
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 3.89,
|
| 963 |
+
"learning_rate": 1.8023255813953488e-06,
|
| 964 |
+
"loss": 0.0848,
|
| 965 |
+
"step": 775
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 3.92,
|
| 969 |
+
"learning_rate": 1.8139534883720933e-06,
|
| 970 |
+
"loss": 0.1124,
|
| 971 |
+
"step": 780
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 3.94,
|
| 975 |
+
"learning_rate": 1.8255813953488375e-06,
|
| 976 |
+
"loss": 0.0806,
|
| 977 |
+
"step": 785
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 3.97,
|
| 981 |
+
"learning_rate": 1.8372093023255815e-06,
|
| 982 |
+
"loss": 0.078,
|
| 983 |
+
"step": 790
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 3.99,
|
| 987 |
+
"learning_rate": 1.8488372093023257e-06,
|
| 988 |
+
"loss": 0.0826,
|
| 989 |
+
"step": 795
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 4.0,
|
| 993 |
+
"eval_accuracy": 0.9693540782649693,
|
| 994 |
+
"eval_loss": 0.0760357454419136,
|
| 995 |
+
"eval_runtime": 90.9354,
|
| 996 |
+
"eval_samples_per_second": 23.324,
|
| 997 |
+
"eval_steps_per_second": 1.463,
|
| 998 |
+
"step": 796
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 4.02,
|
| 1002 |
+
"learning_rate": 1.86046511627907e-06,
|
| 1003 |
+
"loss": 0.0578,
|
| 1004 |
+
"step": 800
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 4.05,
|
| 1008 |
+
"learning_rate": 1.872093023255814e-06,
|
| 1009 |
+
"loss": 0.2264,
|
| 1010 |
+
"step": 805
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 4.07,
|
| 1014 |
+
"learning_rate": 1.8837209302325582e-06,
|
| 1015 |
+
"loss": 0.0851,
|
| 1016 |
+
"step": 810
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 4.1,
|
| 1020 |
+
"learning_rate": 1.8953488372093024e-06,
|
| 1021 |
+
"loss": 0.0803,
|
| 1022 |
+
"step": 815
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 4.12,
|
| 1026 |
+
"learning_rate": 1.9069767441860468e-06,
|
| 1027 |
+
"loss": 0.104,
|
| 1028 |
+
"step": 820
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 4.15,
|
| 1032 |
+
"learning_rate": 1.918604651162791e-06,
|
| 1033 |
+
"loss": 0.1087,
|
| 1034 |
+
"step": 825
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 4.17,
|
| 1038 |
+
"learning_rate": 1.930232558139535e-06,
|
| 1039 |
+
"loss": 0.1494,
|
| 1040 |
+
"step": 830
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 4.2,
|
| 1044 |
+
"learning_rate": 1.941860465116279e-06,
|
| 1045 |
+
"loss": 0.1459,
|
| 1046 |
+
"step": 835
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 4.22,
|
| 1050 |
+
"learning_rate": 1.9534883720930235e-06,
|
| 1051 |
+
"loss": 0.0806,
|
| 1052 |
+
"step": 840
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 4.25,
|
| 1056 |
+
"learning_rate": 1.9651162790697675e-06,
|
| 1057 |
+
"loss": 0.0517,
|
| 1058 |
+
"step": 845
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 4.27,
|
| 1062 |
+
"learning_rate": 1.976744186046512e-06,
|
| 1063 |
+
"loss": 0.1205,
|
| 1064 |
+
"step": 850
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 4.3,
|
| 1068 |
+
"learning_rate": 1.988372093023256e-06,
|
| 1069 |
+
"loss": 0.1168,
|
| 1070 |
+
"step": 855
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 4.32,
|
| 1074 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 1075 |
+
"loss": 0.1178,
|
| 1076 |
+
"step": 860
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 4.35,
|
| 1080 |
+
"learning_rate": 2.0116279069767443e-06,
|
| 1081 |
+
"loss": 0.0972,
|
| 1082 |
+
"step": 865
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 4.37,
|
| 1086 |
+
"learning_rate": 2.0232558139534888e-06,
|
| 1087 |
+
"loss": 0.1218,
|
| 1088 |
+
"step": 870
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 4.4,
|
| 1092 |
+
"learning_rate": 2.0348837209302328e-06,
|
| 1093 |
+
"loss": 0.0609,
|
| 1094 |
+
"step": 875
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 4.42,
|
| 1098 |
+
"learning_rate": 2.0465116279069768e-06,
|
| 1099 |
+
"loss": 0.0926,
|
| 1100 |
+
"step": 880
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 4.45,
|
| 1104 |
+
"learning_rate": 2.058139534883721e-06,
|
| 1105 |
+
"loss": 0.0923,
|
| 1106 |
+
"step": 885
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 4.47,
|
| 1110 |
+
"learning_rate": 2.0697674418604652e-06,
|
| 1111 |
+
"loss": 0.0673,
|
| 1112 |
+
"step": 890
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 4.5,
|
| 1116 |
+
"learning_rate": 2.0813953488372092e-06,
|
| 1117 |
+
"loss": 0.0915,
|
| 1118 |
+
"step": 895
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 4.52,
|
| 1122 |
+
"learning_rate": 2.0930232558139536e-06,
|
| 1123 |
+
"loss": 0.0987,
|
| 1124 |
+
"step": 900
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 4.55,
|
| 1128 |
+
"learning_rate": 2.104651162790698e-06,
|
| 1129 |
+
"loss": 0.1189,
|
| 1130 |
+
"step": 905
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 4.57,
|
| 1134 |
+
"learning_rate": 2.116279069767442e-06,
|
| 1135 |
+
"loss": 0.0615,
|
| 1136 |
+
"step": 910
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 4.6,
|
| 1140 |
+
"learning_rate": 2.127906976744186e-06,
|
| 1141 |
+
"loss": 0.1131,
|
| 1142 |
+
"step": 915
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 4.62,
|
| 1146 |
+
"learning_rate": 2.1395348837209305e-06,
|
| 1147 |
+
"loss": 0.1153,
|
| 1148 |
+
"step": 920
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 4.65,
|
| 1152 |
+
"learning_rate": 2.1511627906976745e-06,
|
| 1153 |
+
"loss": 0.0741,
|
| 1154 |
+
"step": 925
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 4.67,
|
| 1158 |
+
"learning_rate": 2.1627906976744185e-06,
|
| 1159 |
+
"loss": 0.1233,
|
| 1160 |
+
"step": 930
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 4.7,
|
| 1164 |
+
"learning_rate": 2.174418604651163e-06,
|
| 1165 |
+
"loss": 0.1245,
|
| 1166 |
+
"step": 935
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 4.72,
|
| 1170 |
+
"learning_rate": 2.1860465116279074e-06,
|
| 1171 |
+
"loss": 0.058,
|
| 1172 |
+
"step": 940
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 4.75,
|
| 1176 |
+
"learning_rate": 2.1976744186046514e-06,
|
| 1177 |
+
"loss": 0.0409,
|
| 1178 |
+
"step": 945
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 4.77,
|
| 1182 |
+
"learning_rate": 2.2093023255813954e-06,
|
| 1183 |
+
"loss": 0.0828,
|
| 1184 |
+
"step": 950
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 4.8,
|
| 1188 |
+
"learning_rate": 2.22093023255814e-06,
|
| 1189 |
+
"loss": 0.1098,
|
| 1190 |
+
"step": 955
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 4.82,
|
| 1194 |
+
"learning_rate": 2.232558139534884e-06,
|
| 1195 |
+
"loss": 0.0561,
|
| 1196 |
+
"step": 960
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 4.85,
|
| 1200 |
+
"learning_rate": 2.244186046511628e-06,
|
| 1201 |
+
"loss": 0.1561,
|
| 1202 |
+
"step": 965
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 4.87,
|
| 1206 |
+
"learning_rate": 2.2558139534883723e-06,
|
| 1207 |
+
"loss": 0.1465,
|
| 1208 |
+
"step": 970
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 4.9,
|
| 1212 |
+
"learning_rate": 2.2674418604651163e-06,
|
| 1213 |
+
"loss": 0.068,
|
| 1214 |
+
"step": 975
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 4.92,
|
| 1218 |
+
"learning_rate": 2.2790697674418607e-06,
|
| 1219 |
+
"loss": 0.1774,
|
| 1220 |
+
"step": 980
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 4.95,
|
| 1224 |
+
"learning_rate": 2.2906976744186047e-06,
|
| 1225 |
+
"loss": 0.1152,
|
| 1226 |
+
"step": 985
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 4.97,
|
| 1230 |
+
"learning_rate": 2.302325581395349e-06,
|
| 1231 |
+
"loss": 0.0964,
|
| 1232 |
+
"step": 990
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 5.0,
|
| 1236 |
+
"learning_rate": 2.313953488372093e-06,
|
| 1237 |
+
"loss": 0.0886,
|
| 1238 |
+
"step": 995
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 5.0,
|
| 1242 |
+
"eval_accuracy": 0.9787835926449788,
|
| 1243 |
+
"eval_loss": 0.059965480118989944,
|
| 1244 |
+
"eval_runtime": 90.9509,
|
| 1245 |
+
"eval_samples_per_second": 23.32,
|
| 1246 |
+
"eval_steps_per_second": 1.462,
|
| 1247 |
+
"step": 995
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 5.03,
|
| 1251 |
+
"learning_rate": 2.3255813953488376e-06,
|
| 1252 |
+
"loss": 0.0869,
|
| 1253 |
+
"step": 1000
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 5.05,
|
| 1257 |
+
"learning_rate": 2.3372093023255816e-06,
|
| 1258 |
+
"loss": 0.1005,
|
| 1259 |
+
"step": 1005
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 5.08,
|
| 1263 |
+
"learning_rate": 2.3488372093023256e-06,
|
| 1264 |
+
"loss": 0.0859,
|
| 1265 |
+
"step": 1010
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 5.1,
|
| 1269 |
+
"learning_rate": 2.36046511627907e-06,
|
| 1270 |
+
"loss": 0.0807,
|
| 1271 |
+
"step": 1015
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 5.13,
|
| 1275 |
+
"learning_rate": 2.3720930232558144e-06,
|
| 1276 |
+
"loss": 0.0645,
|
| 1277 |
+
"step": 1020
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 5.15,
|
| 1281 |
+
"learning_rate": 2.3837209302325585e-06,
|
| 1282 |
+
"loss": 0.0598,
|
| 1283 |
+
"step": 1025
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 5.18,
|
| 1287 |
+
"learning_rate": 2.3953488372093025e-06,
|
| 1288 |
+
"loss": 0.1259,
|
| 1289 |
+
"step": 1030
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 5.2,
|
| 1293 |
+
"learning_rate": 2.406976744186047e-06,
|
| 1294 |
+
"loss": 0.0914,
|
| 1295 |
+
"step": 1035
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 5.23,
|
| 1299 |
+
"learning_rate": 2.418604651162791e-06,
|
| 1300 |
+
"loss": 0.1101,
|
| 1301 |
+
"step": 1040
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 5.25,
|
| 1305 |
+
"learning_rate": 2.430232558139535e-06,
|
| 1306 |
+
"loss": 0.1553,
|
| 1307 |
+
"step": 1045
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 5.28,
|
| 1311 |
+
"learning_rate": 2.4418604651162793e-06,
|
| 1312 |
+
"loss": 0.0465,
|
| 1313 |
+
"step": 1050
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 5.3,
|
| 1317 |
+
"learning_rate": 2.4534883720930233e-06,
|
| 1318 |
+
"loss": 0.0226,
|
| 1319 |
+
"step": 1055
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 5.33,
|
| 1323 |
+
"learning_rate": 2.4651162790697678e-06,
|
| 1324 |
+
"loss": 0.089,
|
| 1325 |
+
"step": 1060
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 5.35,
|
| 1329 |
+
"learning_rate": 2.4767441860465118e-06,
|
| 1330 |
+
"loss": 0.0578,
|
| 1331 |
+
"step": 1065
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 5.38,
|
| 1335 |
+
"learning_rate": 2.488372093023256e-06,
|
| 1336 |
+
"loss": 0.0772,
|
| 1337 |
+
"step": 1070
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 5.4,
|
| 1341 |
+
"learning_rate": 2.5e-06,
|
| 1342 |
+
"loss": 0.066,
|
| 1343 |
+
"step": 1075
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 5.43,
|
| 1347 |
+
"learning_rate": 2.5116279069767446e-06,
|
| 1348 |
+
"loss": 0.0877,
|
| 1349 |
+
"step": 1080
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 5.45,
|
| 1353 |
+
"learning_rate": 2.5232558139534886e-06,
|
| 1354 |
+
"loss": 0.107,
|
| 1355 |
+
"step": 1085
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 5.48,
|
| 1359 |
+
"learning_rate": 2.534883720930233e-06,
|
| 1360 |
+
"loss": 0.1334,
|
| 1361 |
+
"step": 1090
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 5.5,
|
| 1365 |
+
"learning_rate": 2.5465116279069767e-06,
|
| 1366 |
+
"loss": 0.0577,
|
| 1367 |
+
"step": 1095
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 5.53,
|
| 1371 |
+
"learning_rate": 2.558139534883721e-06,
|
| 1372 |
+
"loss": 0.1463,
|
| 1373 |
+
"step": 1100
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 5.55,
|
| 1377 |
+
"learning_rate": 2.569767441860465e-06,
|
| 1378 |
+
"loss": 0.097,
|
| 1379 |
+
"step": 1105
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 5.58,
|
| 1383 |
+
"learning_rate": 2.5813953488372095e-06,
|
| 1384 |
+
"loss": 0.0964,
|
| 1385 |
+
"step": 1110
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 5.6,
|
| 1389 |
+
"learning_rate": 2.593023255813954e-06,
|
| 1390 |
+
"loss": 0.0306,
|
| 1391 |
+
"step": 1115
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 5.63,
|
| 1395 |
+
"learning_rate": 2.604651162790698e-06,
|
| 1396 |
+
"loss": 0.1012,
|
| 1397 |
+
"step": 1120
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 5.65,
|
| 1401 |
+
"learning_rate": 2.6162790697674424e-06,
|
| 1402 |
+
"loss": 0.1271,
|
| 1403 |
+
"step": 1125
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 5.68,
|
| 1407 |
+
"learning_rate": 2.627906976744186e-06,
|
| 1408 |
+
"loss": 0.0381,
|
| 1409 |
+
"step": 1130
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 5.7,
|
| 1413 |
+
"learning_rate": 2.6395348837209304e-06,
|
| 1414 |
+
"loss": 0.1062,
|
| 1415 |
+
"step": 1135
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 5.73,
|
| 1419 |
+
"learning_rate": 2.6511627906976744e-06,
|
| 1420 |
+
"loss": 0.0798,
|
| 1421 |
+
"step": 1140
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 5.75,
|
| 1425 |
+
"learning_rate": 2.662790697674419e-06,
|
| 1426 |
+
"loss": 0.0583,
|
| 1427 |
+
"step": 1145
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 5.78,
|
| 1431 |
+
"learning_rate": 2.674418604651163e-06,
|
| 1432 |
+
"loss": 0.1107,
|
| 1433 |
+
"step": 1150
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 5.8,
|
| 1437 |
+
"learning_rate": 2.6860465116279073e-06,
|
| 1438 |
+
"loss": 0.0535,
|
| 1439 |
+
"step": 1155
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 5.83,
|
| 1443 |
+
"learning_rate": 2.6976744186046517e-06,
|
| 1444 |
+
"loss": 0.1115,
|
| 1445 |
+
"step": 1160
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 5.85,
|
| 1449 |
+
"learning_rate": 2.7093023255813957e-06,
|
| 1450 |
+
"loss": 0.0569,
|
| 1451 |
+
"step": 1165
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 5.88,
|
| 1455 |
+
"learning_rate": 2.72093023255814e-06,
|
| 1456 |
+
"loss": 0.0927,
|
| 1457 |
+
"step": 1170
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 5.9,
|
| 1461 |
+
"learning_rate": 2.7325581395348837e-06,
|
| 1462 |
+
"loss": 0.0326,
|
| 1463 |
+
"step": 1175
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 5.93,
|
| 1467 |
+
"learning_rate": 2.744186046511628e-06,
|
| 1468 |
+
"loss": 0.1111,
|
| 1469 |
+
"step": 1180
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 5.95,
|
| 1473 |
+
"learning_rate": 2.755813953488372e-06,
|
| 1474 |
+
"loss": 0.0345,
|
| 1475 |
+
"step": 1185
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 5.98,
|
| 1479 |
+
"learning_rate": 2.7674418604651166e-06,
|
| 1480 |
+
"loss": 0.0896,
|
| 1481 |
+
"step": 1190
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 6.0,
|
| 1485 |
+
"eval_accuracy": 0.9801980198019802,
|
| 1486 |
+
"eval_loss": 0.05234839394688606,
|
| 1487 |
+
"eval_runtime": 91.0139,
|
| 1488 |
+
"eval_samples_per_second": 23.304,
|
| 1489 |
+
"eval_steps_per_second": 1.461,
|
| 1490 |
+
"step": 1194
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 6.01,
|
| 1494 |
+
"learning_rate": 2.779069767441861e-06,
|
| 1495 |
+
"loss": 0.0247,
|
| 1496 |
+
"step": 1195
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 6.03,
|
| 1500 |
+
"learning_rate": 2.790697674418605e-06,
|
| 1501 |
+
"loss": 0.06,
|
| 1502 |
+
"step": 1200
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 6.06,
|
| 1506 |
+
"learning_rate": 2.8023255813953494e-06,
|
| 1507 |
+
"loss": 0.0695,
|
| 1508 |
+
"step": 1205
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 6.08,
|
| 1512 |
+
"learning_rate": 2.813953488372093e-06,
|
| 1513 |
+
"loss": 0.0554,
|
| 1514 |
+
"step": 1210
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 6.11,
|
| 1518 |
+
"learning_rate": 2.8255813953488374e-06,
|
| 1519 |
+
"loss": 0.1587,
|
| 1520 |
+
"step": 1215
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 6.13,
|
| 1524 |
+
"learning_rate": 2.8372093023255815e-06,
|
| 1525 |
+
"loss": 0.0628,
|
| 1526 |
+
"step": 1220
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 6.16,
|
| 1530 |
+
"learning_rate": 2.848837209302326e-06,
|
| 1531 |
+
"loss": 0.073,
|
| 1532 |
+
"step": 1225
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 6.18,
|
| 1536 |
+
"learning_rate": 2.86046511627907e-06,
|
| 1537 |
+
"loss": 0.0835,
|
| 1538 |
+
"step": 1230
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 6.21,
|
| 1542 |
+
"learning_rate": 2.8720930232558143e-06,
|
| 1543 |
+
"loss": 0.1044,
|
| 1544 |
+
"step": 1235
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 6.23,
|
| 1548 |
+
"learning_rate": 2.8837209302325587e-06,
|
| 1549 |
+
"loss": 0.0537,
|
| 1550 |
+
"step": 1240
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 6.26,
|
| 1554 |
+
"learning_rate": 2.8953488372093023e-06,
|
| 1555 |
+
"loss": 0.1153,
|
| 1556 |
+
"step": 1245
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 6.28,
|
| 1560 |
+
"learning_rate": 2.9069767441860468e-06,
|
| 1561 |
+
"loss": 0.0685,
|
| 1562 |
+
"step": 1250
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 6.31,
|
| 1566 |
+
"learning_rate": 2.9186046511627908e-06,
|
| 1567 |
+
"loss": 0.0602,
|
| 1568 |
+
"step": 1255
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 6.33,
|
| 1572 |
+
"learning_rate": 2.930232558139535e-06,
|
| 1573 |
+
"loss": 0.0731,
|
| 1574 |
+
"step": 1260
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 6.36,
|
| 1578 |
+
"learning_rate": 2.941860465116279e-06,
|
| 1579 |
+
"loss": 0.0509,
|
| 1580 |
+
"step": 1265
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 6.38,
|
| 1584 |
+
"learning_rate": 2.9534883720930236e-06,
|
| 1585 |
+
"loss": 0.0426,
|
| 1586 |
+
"step": 1270
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 6.41,
|
| 1590 |
+
"learning_rate": 2.965116279069768e-06,
|
| 1591 |
+
"loss": 0.0578,
|
| 1592 |
+
"step": 1275
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 6.43,
|
| 1596 |
+
"learning_rate": 2.9767441860465116e-06,
|
| 1597 |
+
"loss": 0.0487,
|
| 1598 |
+
"step": 1280
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 6.46,
|
| 1602 |
+
"learning_rate": 2.988372093023256e-06,
|
| 1603 |
+
"loss": 0.0266,
|
| 1604 |
+
"step": 1285
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 6.48,
|
| 1608 |
+
"learning_rate": 3e-06,
|
| 1609 |
+
"loss": 0.0566,
|
| 1610 |
+
"step": 1290
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 6.51,
|
| 1614 |
+
"learning_rate": 3.0116279069767445e-06,
|
| 1615 |
+
"loss": 0.0443,
|
| 1616 |
+
"step": 1295
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 6.53,
|
| 1620 |
+
"learning_rate": 3.0232558139534885e-06,
|
| 1621 |
+
"loss": 0.0457,
|
| 1622 |
+
"step": 1300
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 6.56,
|
| 1626 |
+
"learning_rate": 3.034883720930233e-06,
|
| 1627 |
+
"loss": 0.122,
|
| 1628 |
+
"step": 1305
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 6.58,
|
| 1632 |
+
"learning_rate": 3.046511627906977e-06,
|
| 1633 |
+
"loss": 0.1308,
|
| 1634 |
+
"step": 1310
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 6.61,
|
| 1638 |
+
"learning_rate": 3.0581395348837214e-06,
|
| 1639 |
+
"loss": 0.0771,
|
| 1640 |
+
"step": 1315
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 6.63,
|
| 1644 |
+
"learning_rate": 3.069767441860466e-06,
|
| 1645 |
+
"loss": 0.0594,
|
| 1646 |
+
"step": 1320
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 6.66,
|
| 1650 |
+
"learning_rate": 3.0813953488372094e-06,
|
| 1651 |
+
"loss": 0.0398,
|
| 1652 |
+
"step": 1325
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 6.68,
|
| 1656 |
+
"learning_rate": 3.093023255813954e-06,
|
| 1657 |
+
"loss": 0.0973,
|
| 1658 |
+
"step": 1330
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 6.71,
|
| 1662 |
+
"learning_rate": 3.104651162790698e-06,
|
| 1663 |
+
"loss": 0.174,
|
| 1664 |
+
"step": 1335
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 6.73,
|
| 1668 |
+
"learning_rate": 3.1162790697674423e-06,
|
| 1669 |
+
"loss": 0.0669,
|
| 1670 |
+
"step": 1340
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 6.76,
|
| 1674 |
+
"learning_rate": 3.1279069767441863e-06,
|
| 1675 |
+
"loss": 0.0592,
|
| 1676 |
+
"step": 1345
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 6.78,
|
| 1680 |
+
"learning_rate": 3.1395348837209307e-06,
|
| 1681 |
+
"loss": 0.0337,
|
| 1682 |
+
"step": 1350
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 6.81,
|
| 1686 |
+
"learning_rate": 3.151162790697675e-06,
|
| 1687 |
+
"loss": 0.0338,
|
| 1688 |
+
"step": 1355
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 6.83,
|
| 1692 |
+
"learning_rate": 3.1627906976744187e-06,
|
| 1693 |
+
"loss": 0.0956,
|
| 1694 |
+
"step": 1360
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 6.86,
|
| 1698 |
+
"learning_rate": 3.174418604651163e-06,
|
| 1699 |
+
"loss": 0.0236,
|
| 1700 |
+
"step": 1365
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 6.88,
|
| 1704 |
+
"learning_rate": 3.186046511627907e-06,
|
| 1705 |
+
"loss": 0.0381,
|
| 1706 |
+
"step": 1370
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 6.91,
|
| 1710 |
+
"learning_rate": 3.1976744186046516e-06,
|
| 1711 |
+
"loss": 0.0674,
|
| 1712 |
+
"step": 1375
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 6.93,
|
| 1716 |
+
"learning_rate": 3.2093023255813956e-06,
|
| 1717 |
+
"loss": 0.0448,
|
| 1718 |
+
"step": 1380
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 6.96,
|
| 1722 |
+
"learning_rate": 3.22093023255814e-06,
|
| 1723 |
+
"loss": 0.0564,
|
| 1724 |
+
"step": 1385
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 6.98,
|
| 1728 |
+
"learning_rate": 3.2325581395348836e-06,
|
| 1729 |
+
"loss": 0.0774,
|
| 1730 |
+
"step": 1390
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 7.0,
|
| 1734 |
+
"eval_accuracy": 0.9825553983969826,
|
| 1735 |
+
"eval_loss": 0.04816773906350136,
|
| 1736 |
+
"eval_runtime": 90.6447,
|
| 1737 |
+
"eval_samples_per_second": 23.399,
|
| 1738 |
+
"eval_steps_per_second": 1.467,
|
| 1739 |
+
"step": 1393
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 7.01,
|
| 1743 |
+
"learning_rate": 3.244186046511628e-06,
|
| 1744 |
+
"loss": 0.1127,
|
| 1745 |
+
"step": 1395
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 7.04,
|
| 1749 |
+
"learning_rate": 3.2558139534883724e-06,
|
| 1750 |
+
"loss": 0.0509,
|
| 1751 |
+
"step": 1400
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 7.06,
|
| 1755 |
+
"learning_rate": 3.2674418604651164e-06,
|
| 1756 |
+
"loss": 0.1105,
|
| 1757 |
+
"step": 1405
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 7.09,
|
| 1761 |
+
"learning_rate": 3.279069767441861e-06,
|
| 1762 |
+
"loss": 0.0226,
|
| 1763 |
+
"step": 1410
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 7.11,
|
| 1767 |
+
"learning_rate": 3.290697674418605e-06,
|
| 1768 |
+
"loss": 0.0518,
|
| 1769 |
+
"step": 1415
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 7.14,
|
| 1773 |
+
"learning_rate": 3.3023255813953493e-06,
|
| 1774 |
+
"loss": 0.155,
|
| 1775 |
+
"step": 1420
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 7.16,
|
| 1779 |
+
"learning_rate": 3.313953488372093e-06,
|
| 1780 |
+
"loss": 0.0815,
|
| 1781 |
+
"step": 1425
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 7.19,
|
| 1785 |
+
"learning_rate": 3.3255813953488373e-06,
|
| 1786 |
+
"loss": 0.0785,
|
| 1787 |
+
"step": 1430
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 7.21,
|
| 1791 |
+
"learning_rate": 3.3372093023255817e-06,
|
| 1792 |
+
"loss": 0.0339,
|
| 1793 |
+
"step": 1435
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 7.24,
|
| 1797 |
+
"learning_rate": 3.3488372093023258e-06,
|
| 1798 |
+
"loss": 0.0358,
|
| 1799 |
+
"step": 1440
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 7.26,
|
| 1803 |
+
"learning_rate": 3.36046511627907e-06,
|
| 1804 |
+
"loss": 0.0259,
|
| 1805 |
+
"step": 1445
|
| 1806 |
+
},
|
| 1807 |
+
{
|
| 1808 |
+
"epoch": 7.29,
|
| 1809 |
+
"learning_rate": 3.372093023255814e-06,
|
| 1810 |
+
"loss": 0.0391,
|
| 1811 |
+
"step": 1450
|
| 1812 |
+
},
|
| 1813 |
+
{
|
| 1814 |
+
"epoch": 7.31,
|
| 1815 |
+
"learning_rate": 3.3837209302325586e-06,
|
| 1816 |
+
"loss": 0.0233,
|
| 1817 |
+
"step": 1455
|
| 1818 |
+
},
|
| 1819 |
+
{
|
| 1820 |
+
"epoch": 7.34,
|
| 1821 |
+
"learning_rate": 3.3953488372093026e-06,
|
| 1822 |
+
"loss": 0.024,
|
| 1823 |
+
"step": 1460
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 7.36,
|
| 1827 |
+
"learning_rate": 3.406976744186047e-06,
|
| 1828 |
+
"loss": 0.0789,
|
| 1829 |
+
"step": 1465
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 7.39,
|
| 1833 |
+
"learning_rate": 3.4186046511627906e-06,
|
| 1834 |
+
"loss": 0.0406,
|
| 1835 |
+
"step": 1470
|
| 1836 |
+
},
|
| 1837 |
+
{
|
| 1838 |
+
"epoch": 7.41,
|
| 1839 |
+
"learning_rate": 3.430232558139535e-06,
|
| 1840 |
+
"loss": 0.0524,
|
| 1841 |
+
"step": 1475
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 7.44,
|
| 1845 |
+
"learning_rate": 3.4418604651162795e-06,
|
| 1846 |
+
"loss": 0.1186,
|
| 1847 |
+
"step": 1480
|
| 1848 |
+
},
|
| 1849 |
+
{
|
| 1850 |
+
"epoch": 7.46,
|
| 1851 |
+
"learning_rate": 3.4534883720930235e-06,
|
| 1852 |
+
"loss": 0.0434,
|
| 1853 |
+
"step": 1485
|
| 1854 |
+
},
|
| 1855 |
+
{
|
| 1856 |
+
"epoch": 7.49,
|
| 1857 |
+
"learning_rate": 3.465116279069768e-06,
|
| 1858 |
+
"loss": 0.084,
|
| 1859 |
+
"step": 1490
|
| 1860 |
+
},
|
| 1861 |
+
{
|
| 1862 |
+
"epoch": 7.51,
|
| 1863 |
+
"learning_rate": 3.476744186046512e-06,
|
| 1864 |
+
"loss": 0.0475,
|
| 1865 |
+
"step": 1495
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 7.54,
|
| 1869 |
+
"learning_rate": 3.4883720930232564e-06,
|
| 1870 |
+
"loss": 0.1015,
|
| 1871 |
+
"step": 1500
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 7.56,
|
| 1875 |
+
"learning_rate": 3.5e-06,
|
| 1876 |
+
"loss": 0.042,
|
| 1877 |
+
"step": 1505
|
| 1878 |
+
},
|
| 1879 |
+
{
|
| 1880 |
+
"epoch": 7.59,
|
| 1881 |
+
"learning_rate": 3.5116279069767444e-06,
|
| 1882 |
+
"loss": 0.0287,
|
| 1883 |
+
"step": 1510
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 7.61,
|
| 1887 |
+
"learning_rate": 3.523255813953489e-06,
|
| 1888 |
+
"loss": 0.064,
|
| 1889 |
+
"step": 1515
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 7.64,
|
| 1893 |
+
"learning_rate": 3.534883720930233e-06,
|
| 1894 |
+
"loss": 0.095,
|
| 1895 |
+
"step": 1520
|
| 1896 |
+
},
|
| 1897 |
+
{
|
| 1898 |
+
"epoch": 7.66,
|
| 1899 |
+
"learning_rate": 3.5465116279069772e-06,
|
| 1900 |
+
"loss": 0.0677,
|
| 1901 |
+
"step": 1525
|
| 1902 |
+
},
|
| 1903 |
+
{
|
| 1904 |
+
"epoch": 7.69,
|
| 1905 |
+
"learning_rate": 3.5581395348837212e-06,
|
| 1906 |
+
"loss": 0.05,
|
| 1907 |
+
"step": 1530
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 7.71,
|
| 1911 |
+
"learning_rate": 3.5697674418604657e-06,
|
| 1912 |
+
"loss": 0.0435,
|
| 1913 |
+
"step": 1535
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 7.74,
|
| 1917 |
+
"learning_rate": 3.5813953488372093e-06,
|
| 1918 |
+
"loss": 0.1378,
|
| 1919 |
+
"step": 1540
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 7.76,
|
| 1923 |
+
"learning_rate": 3.5930232558139537e-06,
|
| 1924 |
+
"loss": 0.1013,
|
| 1925 |
+
"step": 1545
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 7.79,
|
| 1929 |
+
"learning_rate": 3.6046511627906977e-06,
|
| 1930 |
+
"loss": 0.0508,
|
| 1931 |
+
"step": 1550
|
| 1932 |
+
},
|
| 1933 |
+
{
|
| 1934 |
+
"epoch": 7.81,
|
| 1935 |
+
"learning_rate": 3.616279069767442e-06,
|
| 1936 |
+
"loss": 0.0078,
|
| 1937 |
+
"step": 1555
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 7.84,
|
| 1941 |
+
"learning_rate": 3.6279069767441866e-06,
|
| 1942 |
+
"loss": 0.0256,
|
| 1943 |
+
"step": 1560
|
| 1944 |
+
},
|
| 1945 |
+
{
|
| 1946 |
+
"epoch": 7.86,
|
| 1947 |
+
"learning_rate": 3.6395348837209306e-06,
|
| 1948 |
+
"loss": 0.0381,
|
| 1949 |
+
"step": 1565
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 7.89,
|
| 1953 |
+
"learning_rate": 3.651162790697675e-06,
|
| 1954 |
+
"loss": 0.0318,
|
| 1955 |
+
"step": 1570
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 7.91,
|
| 1959 |
+
"learning_rate": 3.6627906976744186e-06,
|
| 1960 |
+
"loss": 0.1263,
|
| 1961 |
+
"step": 1575
|
| 1962 |
+
},
|
| 1963 |
+
{
|
| 1964 |
+
"epoch": 7.94,
|
| 1965 |
+
"learning_rate": 3.674418604651163e-06,
|
| 1966 |
+
"loss": 0.0309,
|
| 1967 |
+
"step": 1580
|
| 1968 |
+
},
|
| 1969 |
+
{
|
| 1970 |
+
"epoch": 7.96,
|
| 1971 |
+
"learning_rate": 3.686046511627907e-06,
|
| 1972 |
+
"loss": 0.0699,
|
| 1973 |
+
"step": 1585
|
| 1974 |
+
},
|
| 1975 |
+
{
|
| 1976 |
+
"epoch": 7.99,
|
| 1977 |
+
"learning_rate": 3.6976744186046514e-06,
|
| 1978 |
+
"loss": 0.0876,
|
| 1979 |
+
"step": 1590
|
| 1980 |
+
},
|
| 1981 |
+
{
|
| 1982 |
+
"epoch": 8.0,
|
| 1983 |
+
"eval_accuracy": 0.9877416313059877,
|
| 1984 |
+
"eval_loss": 0.0289468877017498,
|
| 1985 |
+
"eval_runtime": 90.5383,
|
| 1986 |
+
"eval_samples_per_second": 23.427,
|
| 1987 |
+
"eval_steps_per_second": 1.469,
|
| 1988 |
+
"step": 1592
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 8.02,
|
| 1992 |
+
"learning_rate": 3.709302325581396e-06,
|
| 1993 |
+
"loss": 0.0537,
|
| 1994 |
+
"step": 1595
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 8.04,
|
| 1998 |
+
"learning_rate": 3.72093023255814e-06,
|
| 1999 |
+
"loss": 0.0633,
|
| 2000 |
+
"step": 1600
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 8.07,
|
| 2004 |
+
"learning_rate": 3.7325581395348843e-06,
|
| 2005 |
+
"loss": 0.0803,
|
| 2006 |
+
"step": 1605
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 8.09,
|
| 2010 |
+
"learning_rate": 3.744186046511628e-06,
|
| 2011 |
+
"loss": 0.0594,
|
| 2012 |
+
"step": 1610
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 8.12,
|
| 2016 |
+
"learning_rate": 3.7558139534883727e-06,
|
| 2017 |
+
"loss": 0.0894,
|
| 2018 |
+
"step": 1615
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 8.14,
|
| 2022 |
+
"learning_rate": 3.7674418604651163e-06,
|
| 2023 |
+
"loss": 0.0336,
|
| 2024 |
+
"step": 1620
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 8.17,
|
| 2028 |
+
"learning_rate": 3.7790697674418607e-06,
|
| 2029 |
+
"loss": 0.0151,
|
| 2030 |
+
"step": 1625
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 8.19,
|
| 2034 |
+
"learning_rate": 3.7906976744186048e-06,
|
| 2035 |
+
"loss": 0.0575,
|
| 2036 |
+
"step": 1630
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"epoch": 8.22,
|
| 2040 |
+
"learning_rate": 3.802325581395349e-06,
|
| 2041 |
+
"loss": 0.0224,
|
| 2042 |
+
"step": 1635
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 8.24,
|
| 2046 |
+
"learning_rate": 3.8139534883720936e-06,
|
| 2047 |
+
"loss": 0.0561,
|
| 2048 |
+
"step": 1640
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 8.27,
|
| 2052 |
+
"learning_rate": 3.825581395348838e-06,
|
| 2053 |
+
"loss": 0.0333,
|
| 2054 |
+
"step": 1645
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 8.29,
|
| 2058 |
+
"learning_rate": 3.837209302325582e-06,
|
| 2059 |
+
"loss": 0.0881,
|
| 2060 |
+
"step": 1650
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 8.32,
|
| 2064 |
+
"learning_rate": 3.848837209302326e-06,
|
| 2065 |
+
"loss": 0.0562,
|
| 2066 |
+
"step": 1655
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 8.34,
|
| 2070 |
+
"learning_rate": 3.86046511627907e-06,
|
| 2071 |
+
"loss": 0.0381,
|
| 2072 |
+
"step": 1660
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 8.37,
|
| 2076 |
+
"learning_rate": 3.872093023255814e-06,
|
| 2077 |
+
"loss": 0.0172,
|
| 2078 |
+
"step": 1665
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 8.39,
|
| 2082 |
+
"learning_rate": 3.883720930232558e-06,
|
| 2083 |
+
"loss": 0.0262,
|
| 2084 |
+
"step": 1670
|
| 2085 |
+
},
|
| 2086 |
+
{
|
| 2087 |
+
"epoch": 8.42,
|
| 2088 |
+
"learning_rate": 3.8953488372093025e-06,
|
| 2089 |
+
"loss": 0.1236,
|
| 2090 |
+
"step": 1675
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 8.44,
|
| 2094 |
+
"learning_rate": 3.906976744186047e-06,
|
| 2095 |
+
"loss": 0.0794,
|
| 2096 |
+
"step": 1680
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 8.47,
|
| 2100 |
+
"learning_rate": 3.918604651162791e-06,
|
| 2101 |
+
"loss": 0.0319,
|
| 2102 |
+
"step": 1685
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 8.49,
|
| 2106 |
+
"learning_rate": 3.930232558139535e-06,
|
| 2107 |
+
"loss": 0.0114,
|
| 2108 |
+
"step": 1690
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 8.52,
|
| 2112 |
+
"learning_rate": 3.941860465116279e-06,
|
| 2113 |
+
"loss": 0.0872,
|
| 2114 |
+
"step": 1695
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 8.54,
|
| 2118 |
+
"learning_rate": 3.953488372093024e-06,
|
| 2119 |
+
"loss": 0.0592,
|
| 2120 |
+
"step": 1700
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 8.57,
|
| 2124 |
+
"learning_rate": 3.965116279069768e-06,
|
| 2125 |
+
"loss": 0.0436,
|
| 2126 |
+
"step": 1705
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 8.59,
|
| 2130 |
+
"learning_rate": 3.976744186046512e-06,
|
| 2131 |
+
"loss": 0.0328,
|
| 2132 |
+
"step": 1710
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 8.62,
|
| 2136 |
+
"learning_rate": 3.988372093023256e-06,
|
| 2137 |
+
"loss": 0.0602,
|
| 2138 |
+
"step": 1715
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 8.64,
|
| 2142 |
+
"learning_rate": 4.000000000000001e-06,
|
| 2143 |
+
"loss": 0.0482,
|
| 2144 |
+
"step": 1720
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 8.67,
|
| 2148 |
+
"learning_rate": 4.011627906976744e-06,
|
| 2149 |
+
"loss": 0.012,
|
| 2150 |
+
"step": 1725
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 8.69,
|
| 2154 |
+
"learning_rate": 4.023255813953489e-06,
|
| 2155 |
+
"loss": 0.0723,
|
| 2156 |
+
"step": 1730
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 8.72,
|
| 2160 |
+
"learning_rate": 4.034883720930233e-06,
|
| 2161 |
+
"loss": 0.0854,
|
| 2162 |
+
"step": 1735
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 8.74,
|
| 2166 |
+
"learning_rate": 4.0465116279069775e-06,
|
| 2167 |
+
"loss": 0.0185,
|
| 2168 |
+
"step": 1740
|
| 2169 |
+
},
|
| 2170 |
+
{
|
| 2171 |
+
"epoch": 8.77,
|
| 2172 |
+
"learning_rate": 4.058139534883721e-06,
|
| 2173 |
+
"loss": 0.0684,
|
| 2174 |
+
"step": 1745
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 8.79,
|
| 2178 |
+
"learning_rate": 4.0697674418604655e-06,
|
| 2179 |
+
"loss": 0.0528,
|
| 2180 |
+
"step": 1750
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 8.82,
|
| 2184 |
+
"learning_rate": 4.08139534883721e-06,
|
| 2185 |
+
"loss": 0.0708,
|
| 2186 |
+
"step": 1755
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 8.84,
|
| 2190 |
+
"learning_rate": 4.0930232558139536e-06,
|
| 2191 |
+
"loss": 0.0283,
|
| 2192 |
+
"step": 1760
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 8.87,
|
| 2196 |
+
"learning_rate": 4.104651162790698e-06,
|
| 2197 |
+
"loss": 0.0126,
|
| 2198 |
+
"step": 1765
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"epoch": 8.89,
|
| 2202 |
+
"learning_rate": 4.116279069767442e-06,
|
| 2203 |
+
"loss": 0.1266,
|
| 2204 |
+
"step": 1770
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 8.92,
|
| 2208 |
+
"learning_rate": 4.127906976744187e-06,
|
| 2209 |
+
"loss": 0.138,
|
| 2210 |
+
"step": 1775
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"epoch": 8.94,
|
| 2214 |
+
"learning_rate": 4.1395348837209304e-06,
|
| 2215 |
+
"loss": 0.1034,
|
| 2216 |
+
"step": 1780
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 8.97,
|
| 2220 |
+
"learning_rate": 4.151162790697675e-06,
|
| 2221 |
+
"loss": 0.0554,
|
| 2222 |
+
"step": 1785
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 8.99,
|
| 2226 |
+
"learning_rate": 4.1627906976744184e-06,
|
| 2227 |
+
"loss": 0.1105,
|
| 2228 |
+
"step": 1790
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 9.0,
|
| 2232 |
+
"eval_accuracy": 0.9820839226779821,
|
| 2233 |
+
"eval_loss": 0.05802316591143608,
|
| 2234 |
+
"eval_runtime": 90.801,
|
| 2235 |
+
"eval_samples_per_second": 23.359,
|
| 2236 |
+
"eval_steps_per_second": 1.465,
|
| 2237 |
+
"step": 1791
|
| 2238 |
+
},
|
| 2239 |
+
{
|
| 2240 |
+
"epoch": 9.02,
|
| 2241 |
+
"learning_rate": 4.174418604651163e-06,
|
| 2242 |
+
"loss": 0.0654,
|
| 2243 |
+
"step": 1795
|
| 2244 |
+
},
|
| 2245 |
+
{
|
| 2246 |
+
"epoch": 9.05,
|
| 2247 |
+
"learning_rate": 4.186046511627907e-06,
|
| 2248 |
+
"loss": 0.0111,
|
| 2249 |
+
"step": 1800
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 9.07,
|
| 2253 |
+
"learning_rate": 4.197674418604652e-06,
|
| 2254 |
+
"loss": 0.0222,
|
| 2255 |
+
"step": 1805
|
| 2256 |
+
},
|
| 2257 |
+
{
|
| 2258 |
+
"epoch": 9.1,
|
| 2259 |
+
"learning_rate": 4.209302325581396e-06,
|
| 2260 |
+
"loss": 0.0517,
|
| 2261 |
+
"step": 1810
|
| 2262 |
+
},
|
| 2263 |
+
{
|
| 2264 |
+
"epoch": 9.12,
|
| 2265 |
+
"learning_rate": 4.22093023255814e-06,
|
| 2266 |
+
"loss": 0.0376,
|
| 2267 |
+
"step": 1815
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 9.15,
|
| 2271 |
+
"learning_rate": 4.232558139534884e-06,
|
| 2272 |
+
"loss": 0.0881,
|
| 2273 |
+
"step": 1820
|
| 2274 |
+
},
|
| 2275 |
+
{
|
| 2276 |
+
"epoch": 9.17,
|
| 2277 |
+
"learning_rate": 4.244186046511628e-06,
|
| 2278 |
+
"loss": 0.0317,
|
| 2279 |
+
"step": 1825
|
| 2280 |
+
},
|
| 2281 |
+
{
|
| 2282 |
+
"epoch": 9.2,
|
| 2283 |
+
"learning_rate": 4.255813953488372e-06,
|
| 2284 |
+
"loss": 0.0075,
|
| 2285 |
+
"step": 1830
|
| 2286 |
+
},
|
| 2287 |
+
{
|
| 2288 |
+
"epoch": 9.22,
|
| 2289 |
+
"learning_rate": 4.267441860465117e-06,
|
| 2290 |
+
"loss": 0.0807,
|
| 2291 |
+
"step": 1835
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 9.25,
|
| 2295 |
+
"learning_rate": 4.279069767441861e-06,
|
| 2296 |
+
"loss": 0.0307,
|
| 2297 |
+
"step": 1840
|
| 2298 |
+
},
|
| 2299 |
+
{
|
| 2300 |
+
"epoch": 9.27,
|
| 2301 |
+
"learning_rate": 4.2906976744186055e-06,
|
| 2302 |
+
"loss": 0.0826,
|
| 2303 |
+
"step": 1845
|
| 2304 |
+
},
|
| 2305 |
+
{
|
| 2306 |
+
"epoch": 9.3,
|
| 2307 |
+
"learning_rate": 4.302325581395349e-06,
|
| 2308 |
+
"loss": 0.0526,
|
| 2309 |
+
"step": 1850
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 9.32,
|
| 2313 |
+
"learning_rate": 4.3139534883720935e-06,
|
| 2314 |
+
"loss": 0.0748,
|
| 2315 |
+
"step": 1855
|
| 2316 |
+
},
|
| 2317 |
+
{
|
| 2318 |
+
"epoch": 9.35,
|
| 2319 |
+
"learning_rate": 4.325581395348837e-06,
|
| 2320 |
+
"loss": 0.02,
|
| 2321 |
+
"step": 1860
|
| 2322 |
+
},
|
| 2323 |
+
{
|
| 2324 |
+
"epoch": 9.37,
|
| 2325 |
+
"learning_rate": 4.3372093023255815e-06,
|
| 2326 |
+
"loss": 0.0865,
|
| 2327 |
+
"step": 1865
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 9.4,
|
| 2331 |
+
"learning_rate": 4.348837209302326e-06,
|
| 2332 |
+
"loss": 0.0962,
|
| 2333 |
+
"step": 1870
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 9.42,
|
| 2337 |
+
"learning_rate": 4.36046511627907e-06,
|
| 2338 |
+
"loss": 0.0448,
|
| 2339 |
+
"step": 1875
|
| 2340 |
+
},
|
| 2341 |
+
{
|
| 2342 |
+
"epoch": 9.45,
|
| 2343 |
+
"learning_rate": 4.372093023255815e-06,
|
| 2344 |
+
"loss": 0.0414,
|
| 2345 |
+
"step": 1880
|
| 2346 |
+
},
|
| 2347 |
+
{
|
| 2348 |
+
"epoch": 9.47,
|
| 2349 |
+
"learning_rate": 4.383720930232558e-06,
|
| 2350 |
+
"loss": 0.0367,
|
| 2351 |
+
"step": 1885
|
| 2352 |
+
},
|
| 2353 |
+
{
|
| 2354 |
+
"epoch": 9.5,
|
| 2355 |
+
"learning_rate": 4.395348837209303e-06,
|
| 2356 |
+
"loss": 0.018,
|
| 2357 |
+
"step": 1890
|
| 2358 |
+
},
|
| 2359 |
+
{
|
| 2360 |
+
"epoch": 9.52,
|
| 2361 |
+
"learning_rate": 4.406976744186046e-06,
|
| 2362 |
+
"loss": 0.0238,
|
| 2363 |
+
"step": 1895
|
| 2364 |
+
},
|
| 2365 |
+
{
|
| 2366 |
+
"epoch": 9.55,
|
| 2367 |
+
"learning_rate": 4.418604651162791e-06,
|
| 2368 |
+
"loss": 0.0449,
|
| 2369 |
+
"step": 1900
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 9.57,
|
| 2373 |
+
"learning_rate": 4.430232558139535e-06,
|
| 2374 |
+
"loss": 0.0461,
|
| 2375 |
+
"step": 1905
|
| 2376 |
+
},
|
| 2377 |
+
{
|
| 2378 |
+
"epoch": 9.6,
|
| 2379 |
+
"learning_rate": 4.44186046511628e-06,
|
| 2380 |
+
"loss": 0.0269,
|
| 2381 |
+
"step": 1910
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 9.62,
|
| 2385 |
+
"learning_rate": 4.453488372093024e-06,
|
| 2386 |
+
"loss": 0.1015,
|
| 2387 |
+
"step": 1915
|
| 2388 |
+
},
|
| 2389 |
+
{
|
| 2390 |
+
"epoch": 9.65,
|
| 2391 |
+
"learning_rate": 4.465116279069768e-06,
|
| 2392 |
+
"loss": 0.0235,
|
| 2393 |
+
"step": 1920
|
| 2394 |
+
},
|
| 2395 |
+
{
|
| 2396 |
+
"epoch": 9.67,
|
| 2397 |
+
"learning_rate": 4.476744186046512e-06,
|
| 2398 |
+
"loss": 0.0445,
|
| 2399 |
+
"step": 1925
|
| 2400 |
+
},
|
| 2401 |
+
{
|
| 2402 |
+
"epoch": 9.7,
|
| 2403 |
+
"learning_rate": 4.488372093023256e-06,
|
| 2404 |
+
"loss": 0.0707,
|
| 2405 |
+
"step": 1930
|
| 2406 |
+
},
|
| 2407 |
+
{
|
| 2408 |
+
"epoch": 9.72,
|
| 2409 |
+
"learning_rate": 4.5e-06,
|
| 2410 |
+
"loss": 0.0562,
|
| 2411 |
+
"step": 1935
|
| 2412 |
+
},
|
| 2413 |
+
{
|
| 2414 |
+
"epoch": 9.75,
|
| 2415 |
+
"learning_rate": 4.5116279069767445e-06,
|
| 2416 |
+
"loss": 0.0444,
|
| 2417 |
+
"step": 1940
|
| 2418 |
+
},
|
| 2419 |
+
{
|
| 2420 |
+
"epoch": 9.77,
|
| 2421 |
+
"learning_rate": 4.523255813953489e-06,
|
| 2422 |
+
"loss": 0.0733,
|
| 2423 |
+
"step": 1945
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 9.8,
|
| 2427 |
+
"learning_rate": 4.5348837209302326e-06,
|
| 2428 |
+
"loss": 0.041,
|
| 2429 |
+
"step": 1950
|
| 2430 |
+
},
|
| 2431 |
+
{
|
| 2432 |
+
"epoch": 9.82,
|
| 2433 |
+
"learning_rate": 4.546511627906977e-06,
|
| 2434 |
+
"loss": 0.0205,
|
| 2435 |
+
"step": 1955
|
| 2436 |
+
},
|
| 2437 |
+
{
|
| 2438 |
+
"epoch": 9.85,
|
| 2439 |
+
"learning_rate": 4.558139534883721e-06,
|
| 2440 |
+
"loss": 0.067,
|
| 2441 |
+
"step": 1960
|
| 2442 |
+
},
|
| 2443 |
+
{
|
| 2444 |
+
"epoch": 9.87,
|
| 2445 |
+
"learning_rate": 4.569767441860465e-06,
|
| 2446 |
+
"loss": 0.0548,
|
| 2447 |
+
"step": 1965
|
| 2448 |
+
},
|
| 2449 |
+
{
|
| 2450 |
+
"epoch": 9.9,
|
| 2451 |
+
"learning_rate": 4.5813953488372094e-06,
|
| 2452 |
+
"loss": 0.0704,
|
| 2453 |
+
"step": 1970
|
| 2454 |
+
},
|
| 2455 |
+
{
|
| 2456 |
+
"epoch": 9.92,
|
| 2457 |
+
"learning_rate": 4.593023255813954e-06,
|
| 2458 |
+
"loss": 0.0203,
|
| 2459 |
+
"step": 1975
|
| 2460 |
+
},
|
| 2461 |
+
{
|
| 2462 |
+
"epoch": 9.95,
|
| 2463 |
+
"learning_rate": 4.604651162790698e-06,
|
| 2464 |
+
"loss": 0.0189,
|
| 2465 |
+
"step": 1980
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 9.97,
|
| 2469 |
+
"learning_rate": 4.616279069767442e-06,
|
| 2470 |
+
"loss": 0.0342,
|
| 2471 |
+
"step": 1985
|
| 2472 |
+
},
|
| 2473 |
+
{
|
| 2474 |
+
"epoch": 10.0,
|
| 2475 |
+
"learning_rate": 4.627906976744186e-06,
|
| 2476 |
+
"loss": 0.0289,
|
| 2477 |
+
"step": 1990
|
| 2478 |
+
},
|
| 2479 |
+
{
|
| 2480 |
+
"epoch": 10.0,
|
| 2481 |
+
"eval_accuracy": 0.9924563884959925,
|
| 2482 |
+
"eval_loss": 0.029388291761279106,
|
| 2483 |
+
"eval_runtime": 90.9871,
|
| 2484 |
+
"eval_samples_per_second": 23.311,
|
| 2485 |
+
"eval_steps_per_second": 1.462,
|
| 2486 |
+
"step": 1990
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 10.03,
|
| 2490 |
+
"learning_rate": 4.639534883720931e-06,
|
| 2491 |
+
"loss": 0.0206,
|
| 2492 |
+
"step": 1995
|
| 2493 |
+
},
|
| 2494 |
+
{
|
| 2495 |
+
"epoch": 10.05,
|
| 2496 |
+
"learning_rate": 4.651162790697675e-06,
|
| 2497 |
+
"loss": 0.0817,
|
| 2498 |
+
"step": 2000
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 10.08,
|
| 2502 |
+
"learning_rate": 4.66279069767442e-06,
|
| 2503 |
+
"loss": 0.0143,
|
| 2504 |
+
"step": 2005
|
| 2505 |
+
},
|
| 2506 |
+
{
|
| 2507 |
+
"epoch": 10.1,
|
| 2508 |
+
"learning_rate": 4.674418604651163e-06,
|
| 2509 |
+
"loss": 0.0036,
|
| 2510 |
+
"step": 2010
|
| 2511 |
+
},
|
| 2512 |
+
{
|
| 2513 |
+
"epoch": 10.13,
|
| 2514 |
+
"learning_rate": 4.686046511627908e-06,
|
| 2515 |
+
"loss": 0.0335,
|
| 2516 |
+
"step": 2015
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 10.15,
|
| 2520 |
+
"learning_rate": 4.697674418604651e-06,
|
| 2521 |
+
"loss": 0.0304,
|
| 2522 |
+
"step": 2020
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 10.18,
|
| 2526 |
+
"learning_rate": 4.709302325581396e-06,
|
| 2527 |
+
"loss": 0.099,
|
| 2528 |
+
"step": 2025
|
| 2529 |
+
},
|
| 2530 |
+
{
|
| 2531 |
+
"epoch": 10.2,
|
| 2532 |
+
"learning_rate": 4.72093023255814e-06,
|
| 2533 |
+
"loss": 0.037,
|
| 2534 |
+
"step": 2030
|
| 2535 |
+
},
|
| 2536 |
+
{
|
| 2537 |
+
"epoch": 10.23,
|
| 2538 |
+
"learning_rate": 4.7325581395348845e-06,
|
| 2539 |
+
"loss": 0.0534,
|
| 2540 |
+
"step": 2035
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 10.25,
|
| 2544 |
+
"learning_rate": 4.744186046511629e-06,
|
| 2545 |
+
"loss": 0.0079,
|
| 2546 |
+
"step": 2040
|
| 2547 |
+
},
|
| 2548 |
+
{
|
| 2549 |
+
"epoch": 10.28,
|
| 2550 |
+
"learning_rate": 4.7558139534883725e-06,
|
| 2551 |
+
"loss": 0.0551,
|
| 2552 |
+
"step": 2045
|
| 2553 |
+
},
|
| 2554 |
+
{
|
| 2555 |
+
"epoch": 10.3,
|
| 2556 |
+
"learning_rate": 4.767441860465117e-06,
|
| 2557 |
+
"loss": 0.0175,
|
| 2558 |
+
"step": 2050
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 10.33,
|
| 2562 |
+
"learning_rate": 4.7790697674418605e-06,
|
| 2563 |
+
"loss": 0.0048,
|
| 2564 |
+
"step": 2055
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 10.35,
|
| 2568 |
+
"learning_rate": 4.790697674418605e-06,
|
| 2569 |
+
"loss": 0.0216,
|
| 2570 |
+
"step": 2060
|
| 2571 |
+
},
|
| 2572 |
+
{
|
| 2573 |
+
"epoch": 10.38,
|
| 2574 |
+
"learning_rate": 4.802325581395349e-06,
|
| 2575 |
+
"loss": 0.0821,
|
| 2576 |
+
"step": 2065
|
| 2577 |
+
},
|
| 2578 |
+
{
|
| 2579 |
+
"epoch": 10.4,
|
| 2580 |
+
"learning_rate": 4.813953488372094e-06,
|
| 2581 |
+
"loss": 0.0045,
|
| 2582 |
+
"step": 2070
|
| 2583 |
+
},
|
| 2584 |
+
{
|
| 2585 |
+
"epoch": 10.43,
|
| 2586 |
+
"learning_rate": 4.825581395348838e-06,
|
| 2587 |
+
"loss": 0.0971,
|
| 2588 |
+
"step": 2075
|
| 2589 |
+
},
|
| 2590 |
+
{
|
| 2591 |
+
"epoch": 10.45,
|
| 2592 |
+
"learning_rate": 4.837209302325582e-06,
|
| 2593 |
+
"loss": 0.0088,
|
| 2594 |
+
"step": 2080
|
| 2595 |
+
},
|
| 2596 |
+
{
|
| 2597 |
+
"epoch": 10.48,
|
| 2598 |
+
"learning_rate": 4.848837209302326e-06,
|
| 2599 |
+
"loss": 0.0364,
|
| 2600 |
+
"step": 2085
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 10.5,
|
| 2604 |
+
"learning_rate": 4.86046511627907e-06,
|
| 2605 |
+
"loss": 0.0391,
|
| 2606 |
+
"step": 2090
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 10.53,
|
| 2610 |
+
"learning_rate": 4.872093023255814e-06,
|
| 2611 |
+
"loss": 0.0233,
|
| 2612 |
+
"step": 2095
|
| 2613 |
+
},
|
| 2614 |
+
{
|
| 2615 |
+
"epoch": 10.55,
|
| 2616 |
+
"learning_rate": 4.883720930232559e-06,
|
| 2617 |
+
"loss": 0.0171,
|
| 2618 |
+
"step": 2100
|
| 2619 |
+
},
|
| 2620 |
+
{
|
| 2621 |
+
"epoch": 10.58,
|
| 2622 |
+
"learning_rate": 4.895348837209303e-06,
|
| 2623 |
+
"loss": 0.0406,
|
| 2624 |
+
"step": 2105
|
| 2625 |
+
},
|
| 2626 |
+
{
|
| 2627 |
+
"epoch": 10.6,
|
| 2628 |
+
"learning_rate": 4.906976744186047e-06,
|
| 2629 |
+
"loss": 0.0124,
|
| 2630 |
+
"step": 2110
|
| 2631 |
+
},
|
| 2632 |
+
{
|
| 2633 |
+
"epoch": 10.63,
|
| 2634 |
+
"learning_rate": 4.918604651162791e-06,
|
| 2635 |
+
"loss": 0.042,
|
| 2636 |
+
"step": 2115
|
| 2637 |
+
},
|
| 2638 |
+
{
|
| 2639 |
+
"epoch": 10.65,
|
| 2640 |
+
"learning_rate": 4.9302325581395355e-06,
|
| 2641 |
+
"loss": 0.0625,
|
| 2642 |
+
"step": 2120
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 10.68,
|
| 2646 |
+
"learning_rate": 4.941860465116279e-06,
|
| 2647 |
+
"loss": 0.0485,
|
| 2648 |
+
"step": 2125
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 10.7,
|
| 2652 |
+
"learning_rate": 4.9534883720930235e-06,
|
| 2653 |
+
"loss": 0.0038,
|
| 2654 |
+
"step": 2130
|
| 2655 |
+
},
|
| 2656 |
+
{
|
| 2657 |
+
"epoch": 10.73,
|
| 2658 |
+
"learning_rate": 4.965116279069768e-06,
|
| 2659 |
+
"loss": 0.0635,
|
| 2660 |
+
"step": 2135
|
| 2661 |
+
},
|
| 2662 |
+
{
|
| 2663 |
+
"epoch": 10.75,
|
| 2664 |
+
"learning_rate": 4.976744186046512e-06,
|
| 2665 |
+
"loss": 0.0032,
|
| 2666 |
+
"step": 2140
|
| 2667 |
+
},
|
| 2668 |
+
{
|
| 2669 |
+
"epoch": 10.78,
|
| 2670 |
+
"learning_rate": 4.988372093023256e-06,
|
| 2671 |
+
"loss": 0.0495,
|
| 2672 |
+
"step": 2145
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 10.8,
|
| 2676 |
+
"learning_rate": 5e-06,
|
| 2677 |
+
"loss": 0.08,
|
| 2678 |
+
"step": 2150
|
| 2679 |
+
},
|
| 2680 |
+
{
|
| 2681 |
+
"epoch": 10.83,
|
| 2682 |
+
"learning_rate": 4.894957983193277e-06,
|
| 2683 |
+
"loss": 0.0515,
|
| 2684 |
+
"step": 2155
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 10.85,
|
| 2688 |
+
"learning_rate": 4.7899159663865555e-06,
|
| 2689 |
+
"loss": 0.0506,
|
| 2690 |
+
"step": 2160
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 10.88,
|
| 2694 |
+
"learning_rate": 4.684873949579832e-06,
|
| 2695 |
+
"loss": 0.0511,
|
| 2696 |
+
"step": 2165
|
| 2697 |
+
},
|
| 2698 |
+
{
|
| 2699 |
+
"epoch": 10.9,
|
| 2700 |
+
"learning_rate": 4.57983193277311e-06,
|
| 2701 |
+
"loss": 0.0343,
|
| 2702 |
+
"step": 2170
|
| 2703 |
+
},
|
| 2704 |
+
{
|
| 2705 |
+
"epoch": 10.93,
|
| 2706 |
+
"learning_rate": 4.474789915966387e-06,
|
| 2707 |
+
"loss": 0.129,
|
| 2708 |
+
"step": 2175
|
| 2709 |
+
},
|
| 2710 |
+
{
|
| 2711 |
+
"epoch": 10.95,
|
| 2712 |
+
"learning_rate": 4.369747899159664e-06,
|
| 2713 |
+
"loss": 0.0301,
|
| 2714 |
+
"step": 2180
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 10.98,
|
| 2718 |
+
"learning_rate": 4.264705882352942e-06,
|
| 2719 |
+
"loss": 0.0594,
|
| 2720 |
+
"step": 2185
|
| 2721 |
+
},
|
| 2722 |
+
{
|
| 2723 |
+
"epoch": 11.0,
|
| 2724 |
+
"eval_accuracy": 0.9905704856199906,
|
| 2725 |
+
"eval_loss": 0.03312527760863304,
|
| 2726 |
+
"eval_runtime": 90.7854,
|
| 2727 |
+
"eval_samples_per_second": 23.363,
|
| 2728 |
+
"eval_steps_per_second": 1.465,
|
| 2729 |
+
"step": 2189
|
| 2730 |
+
},
|
| 2731 |
+
{
|
| 2732 |
+
"epoch": 11.01,
|
| 2733 |
+
"learning_rate": 4.159663865546219e-06,
|
| 2734 |
+
"loss": 0.0181,
|
| 2735 |
+
"step": 2190
|
| 2736 |
+
},
|
| 2737 |
+
{
|
| 2738 |
+
"epoch": 11.03,
|
| 2739 |
+
"learning_rate": 4.054621848739496e-06,
|
| 2740 |
+
"loss": 0.0074,
|
| 2741 |
+
"step": 2195
|
| 2742 |
+
},
|
| 2743 |
+
{
|
| 2744 |
+
"epoch": 11.06,
|
| 2745 |
+
"learning_rate": 3.9495798319327735e-06,
|
| 2746 |
+
"loss": 0.0418,
|
| 2747 |
+
"step": 2200
|
| 2748 |
+
},
|
| 2749 |
+
{
|
| 2750 |
+
"epoch": 11.08,
|
| 2751 |
+
"learning_rate": 3.844537815126051e-06,
|
| 2752 |
+
"loss": 0.0897,
|
| 2753 |
+
"step": 2205
|
| 2754 |
+
},
|
| 2755 |
+
{
|
| 2756 |
+
"epoch": 11.11,
|
| 2757 |
+
"learning_rate": 3.739495798319328e-06,
|
| 2758 |
+
"loss": 0.0387,
|
| 2759 |
+
"step": 2210
|
| 2760 |
+
},
|
| 2761 |
+
{
|
| 2762 |
+
"epoch": 11.13,
|
| 2763 |
+
"learning_rate": 3.6344537815126053e-06,
|
| 2764 |
+
"loss": 0.0034,
|
| 2765 |
+
"step": 2215
|
| 2766 |
+
},
|
| 2767 |
+
{
|
| 2768 |
+
"epoch": 11.16,
|
| 2769 |
+
"learning_rate": 3.529411764705883e-06,
|
| 2770 |
+
"loss": 0.0177,
|
| 2771 |
+
"step": 2220
|
| 2772 |
+
},
|
| 2773 |
+
{
|
| 2774 |
+
"epoch": 11.18,
|
| 2775 |
+
"learning_rate": 3.42436974789916e-06,
|
| 2776 |
+
"loss": 0.0186,
|
| 2777 |
+
"step": 2225
|
| 2778 |
+
},
|
| 2779 |
+
{
|
| 2780 |
+
"epoch": 11.21,
|
| 2781 |
+
"learning_rate": 3.319327731092437e-06,
|
| 2782 |
+
"loss": 0.0361,
|
| 2783 |
+
"step": 2230
|
| 2784 |
+
},
|
| 2785 |
+
{
|
| 2786 |
+
"epoch": 11.23,
|
| 2787 |
+
"learning_rate": 3.2142857142857147e-06,
|
| 2788 |
+
"loss": 0.0547,
|
| 2789 |
+
"step": 2235
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 11.26,
|
| 2793 |
+
"learning_rate": 3.109243697478992e-06,
|
| 2794 |
+
"loss": 0.1177,
|
| 2795 |
+
"step": 2240
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 11.28,
|
| 2799 |
+
"learning_rate": 3.004201680672269e-06,
|
| 2800 |
+
"loss": 0.0343,
|
| 2801 |
+
"step": 2245
|
| 2802 |
+
},
|
| 2803 |
+
{
|
| 2804 |
+
"epoch": 11.31,
|
| 2805 |
+
"learning_rate": 2.8991596638655466e-06,
|
| 2806 |
+
"loss": 0.0807,
|
| 2807 |
+
"step": 2250
|
| 2808 |
+
},
|
| 2809 |
+
{
|
| 2810 |
+
"epoch": 11.33,
|
| 2811 |
+
"learning_rate": 2.7941176470588237e-06,
|
| 2812 |
+
"loss": 0.0048,
|
| 2813 |
+
"step": 2255
|
| 2814 |
+
},
|
| 2815 |
+
{
|
| 2816 |
+
"epoch": 11.36,
|
| 2817 |
+
"learning_rate": 2.689075630252101e-06,
|
| 2818 |
+
"loss": 0.0128,
|
| 2819 |
+
"step": 2260
|
| 2820 |
+
},
|
| 2821 |
+
{
|
| 2822 |
+
"epoch": 11.38,
|
| 2823 |
+
"learning_rate": 2.5840336134453784e-06,
|
| 2824 |
+
"loss": 0.0142,
|
| 2825 |
+
"step": 2265
|
| 2826 |
+
},
|
| 2827 |
+
{
|
| 2828 |
+
"epoch": 11.41,
|
| 2829 |
+
"learning_rate": 2.4789915966386555e-06,
|
| 2830 |
+
"loss": 0.0401,
|
| 2831 |
+
"step": 2270
|
| 2832 |
+
},
|
| 2833 |
+
{
|
| 2834 |
+
"epoch": 11.43,
|
| 2835 |
+
"learning_rate": 2.3739495798319327e-06,
|
| 2836 |
+
"loss": 0.002,
|
| 2837 |
+
"step": 2275
|
| 2838 |
+
},
|
| 2839 |
+
{
|
| 2840 |
+
"epoch": 11.46,
|
| 2841 |
+
"learning_rate": 2.2689075630252102e-06,
|
| 2842 |
+
"loss": 0.0072,
|
| 2843 |
+
"step": 2280
|
| 2844 |
+
},
|
| 2845 |
+
{
|
| 2846 |
+
"epoch": 11.48,
|
| 2847 |
+
"learning_rate": 2.163865546218488e-06,
|
| 2848 |
+
"loss": 0.0341,
|
| 2849 |
+
"step": 2285
|
| 2850 |
+
},
|
| 2851 |
+
{
|
| 2852 |
+
"epoch": 11.51,
|
| 2853 |
+
"learning_rate": 2.058823529411765e-06,
|
| 2854 |
+
"loss": 0.0059,
|
| 2855 |
+
"step": 2290
|
| 2856 |
+
},
|
| 2857 |
+
{
|
| 2858 |
+
"epoch": 11.53,
|
| 2859 |
+
"learning_rate": 1.953781512605042e-06,
|
| 2860 |
+
"loss": 0.0588,
|
| 2861 |
+
"step": 2295
|
| 2862 |
+
},
|
| 2863 |
+
{
|
| 2864 |
+
"epoch": 11.56,
|
| 2865 |
+
"learning_rate": 1.8487394957983196e-06,
|
| 2866 |
+
"loss": 0.0601,
|
| 2867 |
+
"step": 2300
|
| 2868 |
+
},
|
| 2869 |
+
{
|
| 2870 |
+
"epoch": 11.58,
|
| 2871 |
+
"learning_rate": 1.7436974789915968e-06,
|
| 2872 |
+
"loss": 0.0146,
|
| 2873 |
+
"step": 2305
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 11.61,
|
| 2877 |
+
"learning_rate": 1.6386554621848741e-06,
|
| 2878 |
+
"loss": 0.0052,
|
| 2879 |
+
"step": 2310
|
| 2880 |
+
},
|
| 2881 |
+
{
|
| 2882 |
+
"epoch": 11.63,
|
| 2883 |
+
"learning_rate": 1.5336134453781515e-06,
|
| 2884 |
+
"loss": 0.0437,
|
| 2885 |
+
"step": 2315
|
| 2886 |
+
},
|
| 2887 |
+
{
|
| 2888 |
+
"epoch": 11.66,
|
| 2889 |
+
"learning_rate": 1.4285714285714286e-06,
|
| 2890 |
+
"loss": 0.0075,
|
| 2891 |
+
"step": 2320
|
| 2892 |
+
},
|
| 2893 |
+
{
|
| 2894 |
+
"epoch": 11.68,
|
| 2895 |
+
"learning_rate": 1.323529411764706e-06,
|
| 2896 |
+
"loss": 0.0561,
|
| 2897 |
+
"step": 2325
|
| 2898 |
+
},
|
| 2899 |
+
{
|
| 2900 |
+
"epoch": 11.71,
|
| 2901 |
+
"learning_rate": 1.2184873949579833e-06,
|
| 2902 |
+
"loss": 0.0718,
|
| 2903 |
+
"step": 2330
|
| 2904 |
+
},
|
| 2905 |
+
{
|
| 2906 |
+
"epoch": 11.73,
|
| 2907 |
+
"learning_rate": 1.1134453781512607e-06,
|
| 2908 |
+
"loss": 0.0006,
|
| 2909 |
+
"step": 2335
|
| 2910 |
+
},
|
| 2911 |
+
{
|
| 2912 |
+
"epoch": 11.76,
|
| 2913 |
+
"learning_rate": 1.0084033613445378e-06,
|
| 2914 |
+
"loss": 0.0184,
|
| 2915 |
+
"step": 2340
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 11.78,
|
| 2919 |
+
"learning_rate": 9.033613445378152e-07,
|
| 2920 |
+
"loss": 0.0341,
|
| 2921 |
+
"step": 2345
|
| 2922 |
+
},
|
| 2923 |
+
{
|
| 2924 |
+
"epoch": 11.81,
|
| 2925 |
+
"learning_rate": 7.983193277310924e-07,
|
| 2926 |
+
"loss": 0.0058,
|
| 2927 |
+
"step": 2350
|
| 2928 |
+
},
|
| 2929 |
+
{
|
| 2930 |
+
"epoch": 11.83,
|
| 2931 |
+
"learning_rate": 6.932773109243699e-07,
|
| 2932 |
+
"loss": 0.0086,
|
| 2933 |
+
"step": 2355
|
| 2934 |
+
},
|
| 2935 |
+
{
|
| 2936 |
+
"epoch": 11.86,
|
| 2937 |
+
"learning_rate": 5.882352941176471e-07,
|
| 2938 |
+
"loss": 0.0563,
|
| 2939 |
+
"step": 2360
|
| 2940 |
+
},
|
| 2941 |
+
{
|
| 2942 |
+
"epoch": 11.88,
|
| 2943 |
+
"learning_rate": 4.831932773109245e-07,
|
| 2944 |
+
"loss": 0.0036,
|
| 2945 |
+
"step": 2365
|
| 2946 |
+
},
|
| 2947 |
+
{
|
| 2948 |
+
"epoch": 11.91,
|
| 2949 |
+
"learning_rate": 3.781512605042017e-07,
|
| 2950 |
+
"loss": 0.018,
|
| 2951 |
+
"step": 2370
|
| 2952 |
+
},
|
| 2953 |
+
{
|
| 2954 |
+
"epoch": 11.93,
|
| 2955 |
+
"learning_rate": 2.73109243697479e-07,
|
| 2956 |
+
"loss": 0.0058,
|
| 2957 |
+
"step": 2375
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 11.96,
|
| 2961 |
+
"learning_rate": 1.680672268907563e-07,
|
| 2962 |
+
"loss": 0.0008,
|
| 2963 |
+
"step": 2380
|
| 2964 |
+
},
|
| 2965 |
+
{
|
| 2966 |
+
"epoch": 11.98,
|
| 2967 |
+
"learning_rate": 6.302521008403361e-08,
|
| 2968 |
+
"loss": 0.0011,
|
| 2969 |
+
"step": 2385
|
| 2970 |
+
},
|
| 2971 |
+
{
|
| 2972 |
+
"epoch": 12.0,
|
| 2973 |
+
"eval_accuracy": 0.9938708156529938,
|
| 2974 |
+
"eval_loss": 0.0275154709815979,
|
| 2975 |
+
"eval_runtime": 90.8878,
|
| 2976 |
+
"eval_samples_per_second": 23.336,
|
| 2977 |
+
"eval_steps_per_second": 1.463,
|
| 2978 |
+
"step": 2388
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 12.0,
|
| 2982 |
+
"step": 2388,
|
| 2983 |
+
"total_flos": 2.0923685902916665e+19,
|
| 2984 |
+
"train_loss": 0.1486589150943784,
|
| 2985 |
+
"train_runtime": 11149.9067,
|
| 2986 |
+
"train_samples_per_second": 6.848,
|
| 2987 |
+
"train_steps_per_second": 0.214
|
| 2988 |
+
}
|
| 2989 |
+
],
|
| 2990 |
+
"max_steps": 2388,
|
| 2991 |
+
"num_train_epochs": 12,
|
| 2992 |
+
"total_flos": 2.0923685902916665e+19,
|
| 2993 |
+
"trial_name": null,
|
| 2994 |
+
"trial_params": null
|
| 2995 |
+
}
|