Commit
•
b1170d9
1
Parent(s):
37eef19
End of training
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 29.86,
|
| 3 |
+
"eval_accuracy": 0.929384965831435,
|
| 4 |
+
"eval_loss": 0.22778519988059998,
|
| 5 |
+
"eval_runtime": 15.9342,
|
| 6 |
+
"eval_samples_per_second": 27.551,
|
| 7 |
+
"eval_steps_per_second": 6.903,
|
| 8 |
+
"total_flos": 2.1986190125590118e+18,
|
| 9 |
+
"train_loss": 0.3750012589521729,
|
| 10 |
+
"train_runtime": 3633.2931,
|
| 11 |
+
"train_samples_per_second": 14.466,
|
| 12 |
+
"train_steps_per_second": 0.9
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 29.86,
|
| 3 |
+
"eval_accuracy": 0.929384965831435,
|
| 4 |
+
"eval_loss": 0.22778519988059998,
|
| 5 |
+
"eval_runtime": 15.9342,
|
| 6 |
+
"eval_samples_per_second": 27.551,
|
| 7 |
+
"eval_steps_per_second": 6.903
|
| 8 |
+
}
|
runs/Aug19_00-27-58_ip-172-16-95-1.us-west-2.compute.internal/events.out.tfevents.1692480779.ip-172-16-95-1.us-west-2.compute.internal.15426.13
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2905021f4398e94831392e4a9b33f275664e345716430845e12306a7ab9f052d
|
| 3 |
+
size 411
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 29.86,
|
| 3 |
+
"total_flos": 2.1986190125590118e+18,
|
| 4 |
+
"train_loss": 0.3750012589521729,
|
| 5 |
+
"train_runtime": 3633.2931,
|
| 6 |
+
"train_samples_per_second": 14.466,
|
| 7 |
+
"train_steps_per_second": 0.9
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2257 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.929384965831435,
|
| 3 |
+
"best_model_checkpoint": "mobilenet_v2_1.0_224-cxr-view/checkpoint-2518",
|
| 4 |
+
"epoch": 29.863013698630137,
|
| 5 |
+
"global_step": 3270,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.09,
|
| 12 |
+
"learning_rate": 1.5290519877675842e-07,
|
| 13 |
+
"loss": 0.7183,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.18,
|
| 18 |
+
"learning_rate": 3.0581039755351683e-07,
|
| 19 |
+
"loss": 0.6605,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.27,
|
| 24 |
+
"learning_rate": 4.587155963302753e-07,
|
| 25 |
+
"loss": 0.6883,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.37,
|
| 30 |
+
"learning_rate": 6.116207951070337e-07,
|
| 31 |
+
"loss": 0.719,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.46,
|
| 36 |
+
"learning_rate": 7.645259938837921e-07,
|
| 37 |
+
"loss": 0.6913,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.55,
|
| 42 |
+
"learning_rate": 9.174311926605506e-07,
|
| 43 |
+
"loss": 0.6944,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.64,
|
| 48 |
+
"learning_rate": 1.070336391437309e-06,
|
| 49 |
+
"loss": 0.6874,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.73,
|
| 54 |
+
"learning_rate": 1.2232415902140673e-06,
|
| 55 |
+
"loss": 0.7119,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.82,
|
| 60 |
+
"learning_rate": 1.3761467889908258e-06,
|
| 61 |
+
"loss": 0.7093,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.91,
|
| 66 |
+
"learning_rate": 1.5290519877675841e-06,
|
| 67 |
+
"loss": 0.7049,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 1.0,
|
| 72 |
+
"eval_accuracy": 0.744874715261959,
|
| 73 |
+
"eval_loss": 0.6746256351470947,
|
| 74 |
+
"eval_runtime": 15.5928,
|
| 75 |
+
"eval_samples_per_second": 28.154,
|
| 76 |
+
"eval_steps_per_second": 7.055,
|
| 77 |
+
"step": 109
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 1.0,
|
| 81 |
+
"learning_rate": 1.6819571865443426e-06,
|
| 82 |
+
"loss": 0.6907,
|
| 83 |
+
"step": 110
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 1.1,
|
| 87 |
+
"learning_rate": 1.8348623853211011e-06,
|
| 88 |
+
"loss": 0.6767,
|
| 89 |
+
"step": 120
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 1.19,
|
| 93 |
+
"learning_rate": 1.9877675840978594e-06,
|
| 94 |
+
"loss": 0.6906,
|
| 95 |
+
"step": 130
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 1.28,
|
| 99 |
+
"learning_rate": 2.140672782874618e-06,
|
| 100 |
+
"loss": 0.6775,
|
| 101 |
+
"step": 140
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 1.37,
|
| 105 |
+
"learning_rate": 2.2935779816513764e-06,
|
| 106 |
+
"loss": 0.6731,
|
| 107 |
+
"step": 150
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 1.46,
|
| 111 |
+
"learning_rate": 2.4464831804281347e-06,
|
| 112 |
+
"loss": 0.6895,
|
| 113 |
+
"step": 160
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 1.55,
|
| 117 |
+
"learning_rate": 2.599388379204893e-06,
|
| 118 |
+
"loss": 0.6679,
|
| 119 |
+
"step": 170
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.64,
|
| 123 |
+
"learning_rate": 2.7522935779816517e-06,
|
| 124 |
+
"loss": 0.6846,
|
| 125 |
+
"step": 180
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 1.74,
|
| 129 |
+
"learning_rate": 2.90519877675841e-06,
|
| 130 |
+
"loss": 0.6737,
|
| 131 |
+
"step": 190
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 1.83,
|
| 135 |
+
"learning_rate": 3.0581039755351682e-06,
|
| 136 |
+
"loss": 0.6588,
|
| 137 |
+
"step": 200
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.92,
|
| 141 |
+
"learning_rate": 3.211009174311927e-06,
|
| 142 |
+
"loss": 0.6565,
|
| 143 |
+
"step": 210
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 2.0,
|
| 147 |
+
"eval_accuracy": 0.6742596810933941,
|
| 148 |
+
"eval_loss": 0.6498312950134277,
|
| 149 |
+
"eval_runtime": 15.6514,
|
| 150 |
+
"eval_samples_per_second": 28.049,
|
| 151 |
+
"eval_steps_per_second": 7.028,
|
| 152 |
+
"step": 219
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 2.01,
|
| 156 |
+
"learning_rate": 3.3639143730886852e-06,
|
| 157 |
+
"loss": 0.6365,
|
| 158 |
+
"step": 220
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 2.1,
|
| 162 |
+
"learning_rate": 3.5168195718654435e-06,
|
| 163 |
+
"loss": 0.6319,
|
| 164 |
+
"step": 230
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 2.19,
|
| 168 |
+
"learning_rate": 3.6697247706422022e-06,
|
| 169 |
+
"loss": 0.613,
|
| 170 |
+
"step": 240
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 2.28,
|
| 174 |
+
"learning_rate": 3.8226299694189605e-06,
|
| 175 |
+
"loss": 0.6312,
|
| 176 |
+
"step": 250
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 2.37,
|
| 180 |
+
"learning_rate": 3.975535168195719e-06,
|
| 181 |
+
"loss": 0.631,
|
| 182 |
+
"step": 260
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 2.47,
|
| 186 |
+
"learning_rate": 4.128440366972478e-06,
|
| 187 |
+
"loss": 0.6364,
|
| 188 |
+
"step": 270
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 2.56,
|
| 192 |
+
"learning_rate": 4.281345565749236e-06,
|
| 193 |
+
"loss": 0.6092,
|
| 194 |
+
"step": 280
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 2.65,
|
| 198 |
+
"learning_rate": 4.4342507645259945e-06,
|
| 199 |
+
"loss": 0.6348,
|
| 200 |
+
"step": 290
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 2.74,
|
| 204 |
+
"learning_rate": 4.587155963302753e-06,
|
| 205 |
+
"loss": 0.6171,
|
| 206 |
+
"step": 300
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 2.83,
|
| 210 |
+
"learning_rate": 4.740061162079511e-06,
|
| 211 |
+
"loss": 0.5771,
|
| 212 |
+
"step": 310
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 2.92,
|
| 216 |
+
"learning_rate": 4.892966360856269e-06,
|
| 217 |
+
"loss": 0.5699,
|
| 218 |
+
"step": 320
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 3.0,
|
| 222 |
+
"eval_accuracy": 0.7995444191343963,
|
| 223 |
+
"eval_loss": 0.5730286836624146,
|
| 224 |
+
"eval_runtime": 15.9142,
|
| 225 |
+
"eval_samples_per_second": 27.586,
|
| 226 |
+
"eval_steps_per_second": 6.912,
|
| 227 |
+
"step": 328
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 3.01,
|
| 231 |
+
"learning_rate": 4.994903160040775e-06,
|
| 232 |
+
"loss": 0.5787,
|
| 233 |
+
"step": 330
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 3.11,
|
| 237 |
+
"learning_rate": 4.977913693510024e-06,
|
| 238 |
+
"loss": 0.5584,
|
| 239 |
+
"step": 340
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 3.2,
|
| 243 |
+
"learning_rate": 4.960924226979273e-06,
|
| 244 |
+
"loss": 0.5662,
|
| 245 |
+
"step": 350
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 3.29,
|
| 249 |
+
"learning_rate": 4.943934760448522e-06,
|
| 250 |
+
"loss": 0.5853,
|
| 251 |
+
"step": 360
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 3.38,
|
| 255 |
+
"learning_rate": 4.926945293917771e-06,
|
| 256 |
+
"loss": 0.5639,
|
| 257 |
+
"step": 370
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 3.47,
|
| 261 |
+
"learning_rate": 4.909955827387021e-06,
|
| 262 |
+
"loss": 0.5507,
|
| 263 |
+
"step": 380
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 3.56,
|
| 267 |
+
"learning_rate": 4.892966360856269e-06,
|
| 268 |
+
"loss": 0.5051,
|
| 269 |
+
"step": 390
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 3.65,
|
| 273 |
+
"learning_rate": 4.875976894325519e-06,
|
| 274 |
+
"loss": 0.5559,
|
| 275 |
+
"step": 400
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 3.74,
|
| 279 |
+
"learning_rate": 4.858987427794767e-06,
|
| 280 |
+
"loss": 0.5184,
|
| 281 |
+
"step": 410
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 3.84,
|
| 285 |
+
"learning_rate": 4.841997961264017e-06,
|
| 286 |
+
"loss": 0.5453,
|
| 287 |
+
"step": 420
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 3.93,
|
| 291 |
+
"learning_rate": 4.825008494733266e-06,
|
| 292 |
+
"loss": 0.5702,
|
| 293 |
+
"step": 430
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 4.0,
|
| 297 |
+
"eval_accuracy": 0.8086560364464692,
|
| 298 |
+
"eval_loss": 0.5119441747665405,
|
| 299 |
+
"eval_runtime": 15.7264,
|
| 300 |
+
"eval_samples_per_second": 27.915,
|
| 301 |
+
"eval_steps_per_second": 6.995,
|
| 302 |
+
"step": 438
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 4.02,
|
| 306 |
+
"learning_rate": 4.808019028202515e-06,
|
| 307 |
+
"loss": 0.4758,
|
| 308 |
+
"step": 440
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 4.11,
|
| 312 |
+
"learning_rate": 4.7910295616717635e-06,
|
| 313 |
+
"loss": 0.52,
|
| 314 |
+
"step": 450
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 4.2,
|
| 318 |
+
"learning_rate": 4.774040095141013e-06,
|
| 319 |
+
"loss": 0.5039,
|
| 320 |
+
"step": 460
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 4.29,
|
| 324 |
+
"learning_rate": 4.757050628610262e-06,
|
| 325 |
+
"loss": 0.4593,
|
| 326 |
+
"step": 470
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 4.38,
|
| 330 |
+
"learning_rate": 4.740061162079511e-06,
|
| 331 |
+
"loss": 0.5111,
|
| 332 |
+
"step": 480
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 4.47,
|
| 336 |
+
"learning_rate": 4.72307169554876e-06,
|
| 337 |
+
"loss": 0.5149,
|
| 338 |
+
"step": 490
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 4.57,
|
| 342 |
+
"learning_rate": 4.706082229018009e-06,
|
| 343 |
+
"loss": 0.4369,
|
| 344 |
+
"step": 500
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 4.66,
|
| 348 |
+
"learning_rate": 4.689092762487259e-06,
|
| 349 |
+
"loss": 0.5214,
|
| 350 |
+
"step": 510
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 4.75,
|
| 354 |
+
"learning_rate": 4.672103295956507e-06,
|
| 355 |
+
"loss": 0.4627,
|
| 356 |
+
"step": 520
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 4.84,
|
| 360 |
+
"learning_rate": 4.655113829425757e-06,
|
| 361 |
+
"loss": 0.461,
|
| 362 |
+
"step": 530
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 4.93,
|
| 366 |
+
"learning_rate": 4.638124362895005e-06,
|
| 367 |
+
"loss": 0.4849,
|
| 368 |
+
"step": 540
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 5.0,
|
| 372 |
+
"eval_accuracy": 0.8678815489749431,
|
| 373 |
+
"eval_loss": 0.43555235862731934,
|
| 374 |
+
"eval_runtime": 15.6016,
|
| 375 |
+
"eval_samples_per_second": 28.138,
|
| 376 |
+
"eval_steps_per_second": 7.051,
|
| 377 |
+
"step": 547
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 5.02,
|
| 381 |
+
"learning_rate": 4.621134896364255e-06,
|
| 382 |
+
"loss": 0.4744,
|
| 383 |
+
"step": 550
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 5.11,
|
| 387 |
+
"learning_rate": 4.604145429833503e-06,
|
| 388 |
+
"loss": 0.4267,
|
| 389 |
+
"step": 560
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 5.21,
|
| 393 |
+
"learning_rate": 4.587155963302753e-06,
|
| 394 |
+
"loss": 0.4546,
|
| 395 |
+
"step": 570
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 5.3,
|
| 399 |
+
"learning_rate": 4.570166496772001e-06,
|
| 400 |
+
"loss": 0.3955,
|
| 401 |
+
"step": 580
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 5.39,
|
| 405 |
+
"learning_rate": 4.553177030241251e-06,
|
| 406 |
+
"loss": 0.4499,
|
| 407 |
+
"step": 590
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 5.48,
|
| 411 |
+
"learning_rate": 4.5361875637104995e-06,
|
| 412 |
+
"loss": 0.4315,
|
| 413 |
+
"step": 600
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 5.57,
|
| 417 |
+
"learning_rate": 4.519198097179749e-06,
|
| 418 |
+
"loss": 0.4542,
|
| 419 |
+
"step": 610
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 5.66,
|
| 423 |
+
"learning_rate": 4.502208630648998e-06,
|
| 424 |
+
"loss": 0.5297,
|
| 425 |
+
"step": 620
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 5.75,
|
| 429 |
+
"learning_rate": 4.485219164118247e-06,
|
| 430 |
+
"loss": 0.4189,
|
| 431 |
+
"step": 630
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 5.84,
|
| 435 |
+
"learning_rate": 4.4682296975874964e-06,
|
| 436 |
+
"loss": 0.43,
|
| 437 |
+
"step": 640
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 5.94,
|
| 441 |
+
"learning_rate": 4.451240231056745e-06,
|
| 442 |
+
"loss": 0.356,
|
| 443 |
+
"step": 650
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 6.0,
|
| 447 |
+
"eval_accuracy": 0.8086560364464692,
|
| 448 |
+
"eval_loss": 0.4640910029411316,
|
| 449 |
+
"eval_runtime": 15.7608,
|
| 450 |
+
"eval_samples_per_second": 27.854,
|
| 451 |
+
"eval_steps_per_second": 6.979,
|
| 452 |
+
"step": 657
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 6.03,
|
| 456 |
+
"learning_rate": 4.4342507645259945e-06,
|
| 457 |
+
"loss": 0.4923,
|
| 458 |
+
"step": 660
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 6.12,
|
| 462 |
+
"learning_rate": 4.417261297995243e-06,
|
| 463 |
+
"loss": 0.464,
|
| 464 |
+
"step": 670
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 6.21,
|
| 468 |
+
"learning_rate": 4.400271831464492e-06,
|
| 469 |
+
"loss": 0.3965,
|
| 470 |
+
"step": 680
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 6.3,
|
| 474 |
+
"learning_rate": 4.383282364933741e-06,
|
| 475 |
+
"loss": 0.4148,
|
| 476 |
+
"step": 690
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 6.39,
|
| 480 |
+
"learning_rate": 4.366292898402991e-06,
|
| 481 |
+
"loss": 0.3874,
|
| 482 |
+
"step": 700
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 6.48,
|
| 486 |
+
"learning_rate": 4.349303431872239e-06,
|
| 487 |
+
"loss": 0.3694,
|
| 488 |
+
"step": 710
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 6.58,
|
| 492 |
+
"learning_rate": 4.332313965341489e-06,
|
| 493 |
+
"loss": 0.4263,
|
| 494 |
+
"step": 720
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 6.67,
|
| 498 |
+
"learning_rate": 4.315324498810738e-06,
|
| 499 |
+
"loss": 0.3835,
|
| 500 |
+
"step": 730
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 6.76,
|
| 504 |
+
"learning_rate": 4.298335032279987e-06,
|
| 505 |
+
"loss": 0.4637,
|
| 506 |
+
"step": 740
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 6.85,
|
| 510 |
+
"learning_rate": 4.281345565749236e-06,
|
| 511 |
+
"loss": 0.3923,
|
| 512 |
+
"step": 750
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 6.94,
|
| 516 |
+
"learning_rate": 4.264356099218485e-06,
|
| 517 |
+
"loss": 0.3713,
|
| 518 |
+
"step": 760
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 7.0,
|
| 522 |
+
"eval_accuracy": 0.8678815489749431,
|
| 523 |
+
"eval_loss": 0.3406853675842285,
|
| 524 |
+
"eval_runtime": 15.6486,
|
| 525 |
+
"eval_samples_per_second": 28.054,
|
| 526 |
+
"eval_steps_per_second": 7.029,
|
| 527 |
+
"step": 766
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 7.03,
|
| 531 |
+
"learning_rate": 4.247366632687734e-06,
|
| 532 |
+
"loss": 0.5575,
|
| 533 |
+
"step": 770
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 7.12,
|
| 537 |
+
"learning_rate": 4.230377166156983e-06,
|
| 538 |
+
"loss": 0.4175,
|
| 539 |
+
"step": 780
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 7.21,
|
| 543 |
+
"learning_rate": 4.2133876996262315e-06,
|
| 544 |
+
"loss": 0.4288,
|
| 545 |
+
"step": 790
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 7.31,
|
| 549 |
+
"learning_rate": 4.196398233095481e-06,
|
| 550 |
+
"loss": 0.3972,
|
| 551 |
+
"step": 800
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 7.4,
|
| 555 |
+
"learning_rate": 4.17940876656473e-06,
|
| 556 |
+
"loss": 0.354,
|
| 557 |
+
"step": 810
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 7.49,
|
| 561 |
+
"learning_rate": 4.162419300033979e-06,
|
| 562 |
+
"loss": 0.3354,
|
| 563 |
+
"step": 820
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 7.58,
|
| 567 |
+
"learning_rate": 4.1454298335032285e-06,
|
| 568 |
+
"loss": 0.3047,
|
| 569 |
+
"step": 830
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 7.67,
|
| 573 |
+
"learning_rate": 4.128440366972478e-06,
|
| 574 |
+
"loss": 0.4096,
|
| 575 |
+
"step": 840
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 7.76,
|
| 579 |
+
"learning_rate": 4.1114509004417265e-06,
|
| 580 |
+
"loss": 0.4209,
|
| 581 |
+
"step": 850
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 7.85,
|
| 585 |
+
"learning_rate": 4.094461433910976e-06,
|
| 586 |
+
"loss": 0.3988,
|
| 587 |
+
"step": 860
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 7.95,
|
| 591 |
+
"learning_rate": 4.077471967380225e-06,
|
| 592 |
+
"loss": 0.4571,
|
| 593 |
+
"step": 870
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 8.0,
|
| 597 |
+
"eval_accuracy": 0.7813211845102506,
|
| 598 |
+
"eval_loss": 0.48959481716156006,
|
| 599 |
+
"eval_runtime": 15.6588,
|
| 600 |
+
"eval_samples_per_second": 28.035,
|
| 601 |
+
"eval_steps_per_second": 7.025,
|
| 602 |
+
"step": 876
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 8.04,
|
| 606 |
+
"learning_rate": 4.060482500849473e-06,
|
| 607 |
+
"loss": 0.4078,
|
| 608 |
+
"step": 880
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 8.13,
|
| 612 |
+
"learning_rate": 4.043493034318723e-06,
|
| 613 |
+
"loss": 0.3905,
|
| 614 |
+
"step": 890
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 8.22,
|
| 618 |
+
"learning_rate": 4.026503567787971e-06,
|
| 619 |
+
"loss": 0.4258,
|
| 620 |
+
"step": 900
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 8.31,
|
| 624 |
+
"learning_rate": 4.009514101257221e-06,
|
| 625 |
+
"loss": 0.3991,
|
| 626 |
+
"step": 910
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 8.4,
|
| 630 |
+
"learning_rate": 3.99252463472647e-06,
|
| 631 |
+
"loss": 0.3783,
|
| 632 |
+
"step": 920
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 8.49,
|
| 636 |
+
"learning_rate": 3.975535168195719e-06,
|
| 637 |
+
"loss": 0.4471,
|
| 638 |
+
"step": 930
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 8.58,
|
| 642 |
+
"learning_rate": 3.958545701664968e-06,
|
| 643 |
+
"loss": 0.3437,
|
| 644 |
+
"step": 940
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 8.68,
|
| 648 |
+
"learning_rate": 3.941556235134218e-06,
|
| 649 |
+
"loss": 0.3779,
|
| 650 |
+
"step": 950
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 8.77,
|
| 654 |
+
"learning_rate": 3.924566768603466e-06,
|
| 655 |
+
"loss": 0.2581,
|
| 656 |
+
"step": 960
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 8.86,
|
| 660 |
+
"learning_rate": 3.907577302072715e-06,
|
| 661 |
+
"loss": 0.3267,
|
| 662 |
+
"step": 970
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 8.95,
|
| 666 |
+
"learning_rate": 3.890587835541964e-06,
|
| 667 |
+
"loss": 0.3896,
|
| 668 |
+
"step": 980
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 9.0,
|
| 672 |
+
"eval_accuracy": 0.8883826879271071,
|
| 673 |
+
"eval_loss": 0.3124327063560486,
|
| 674 |
+
"eval_runtime": 15.6592,
|
| 675 |
+
"eval_samples_per_second": 28.035,
|
| 676 |
+
"eval_steps_per_second": 7.025,
|
| 677 |
+
"step": 985
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 9.04,
|
| 681 |
+
"learning_rate": 3.873598369011213e-06,
|
| 682 |
+
"loss": 0.3471,
|
| 683 |
+
"step": 990
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 9.13,
|
| 687 |
+
"learning_rate": 3.8566089024804624e-06,
|
| 688 |
+
"loss": 0.396,
|
| 689 |
+
"step": 1000
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 9.22,
|
| 693 |
+
"learning_rate": 3.839619435949711e-06,
|
| 694 |
+
"loss": 0.3502,
|
| 695 |
+
"step": 1010
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 9.32,
|
| 699 |
+
"learning_rate": 3.8226299694189605e-06,
|
| 700 |
+
"loss": 0.3685,
|
| 701 |
+
"step": 1020
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 9.41,
|
| 705 |
+
"learning_rate": 3.80564050288821e-06,
|
| 706 |
+
"loss": 0.3023,
|
| 707 |
+
"step": 1030
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 9.5,
|
| 711 |
+
"learning_rate": 3.7886510363574586e-06,
|
| 712 |
+
"loss": 0.3511,
|
| 713 |
+
"step": 1040
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 9.59,
|
| 717 |
+
"learning_rate": 3.771661569826708e-06,
|
| 718 |
+
"loss": 0.3186,
|
| 719 |
+
"step": 1050
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 9.68,
|
| 723 |
+
"learning_rate": 3.7546721032959566e-06,
|
| 724 |
+
"loss": 0.4063,
|
| 725 |
+
"step": 1060
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 9.77,
|
| 729 |
+
"learning_rate": 3.7376826367652057e-06,
|
| 730 |
+
"loss": 0.3553,
|
| 731 |
+
"step": 1070
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 9.86,
|
| 735 |
+
"learning_rate": 3.720693170234455e-06,
|
| 736 |
+
"loss": 0.3771,
|
| 737 |
+
"step": 1080
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 9.95,
|
| 741 |
+
"learning_rate": 3.7037037037037037e-06,
|
| 742 |
+
"loss": 0.3422,
|
| 743 |
+
"step": 1090
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 10.0,
|
| 747 |
+
"eval_accuracy": 0.9271070615034168,
|
| 748 |
+
"eval_loss": 0.27909907698631287,
|
| 749 |
+
"eval_runtime": 15.6791,
|
| 750 |
+
"eval_samples_per_second": 27.999,
|
| 751 |
+
"eval_steps_per_second": 7.016,
|
| 752 |
+
"step": 1095
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 10.05,
|
| 756 |
+
"learning_rate": 3.686714237172953e-06,
|
| 757 |
+
"loss": 0.3242,
|
| 758 |
+
"step": 1100
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 10.14,
|
| 762 |
+
"learning_rate": 3.6697247706422022e-06,
|
| 763 |
+
"loss": 0.3179,
|
| 764 |
+
"step": 1110
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 10.23,
|
| 768 |
+
"learning_rate": 3.6527353041114513e-06,
|
| 769 |
+
"loss": 0.3971,
|
| 770 |
+
"step": 1120
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 10.32,
|
| 774 |
+
"learning_rate": 3.6357458375807003e-06,
|
| 775 |
+
"loss": 0.3813,
|
| 776 |
+
"step": 1130
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 10.41,
|
| 780 |
+
"learning_rate": 3.6187563710499497e-06,
|
| 781 |
+
"loss": 0.3048,
|
| 782 |
+
"step": 1140
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 10.5,
|
| 786 |
+
"learning_rate": 3.6017669045191984e-06,
|
| 787 |
+
"loss": 0.3852,
|
| 788 |
+
"step": 1150
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 10.59,
|
| 792 |
+
"learning_rate": 3.5847774379884474e-06,
|
| 793 |
+
"loss": 0.3157,
|
| 794 |
+
"step": 1160
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 10.68,
|
| 798 |
+
"learning_rate": 3.5677879714576964e-06,
|
| 799 |
+
"loss": 0.3868,
|
| 800 |
+
"step": 1170
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 10.78,
|
| 804 |
+
"learning_rate": 3.5507985049269455e-06,
|
| 805 |
+
"loss": 0.3309,
|
| 806 |
+
"step": 1180
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 10.87,
|
| 810 |
+
"learning_rate": 3.533809038396195e-06,
|
| 811 |
+
"loss": 0.3372,
|
| 812 |
+
"step": 1190
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 10.96,
|
| 816 |
+
"learning_rate": 3.5168195718654435e-06,
|
| 817 |
+
"loss": 0.3358,
|
| 818 |
+
"step": 1200
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 11.0,
|
| 822 |
+
"eval_accuracy": 0.8246013667425968,
|
| 823 |
+
"eval_loss": 0.39981135725975037,
|
| 824 |
+
"eval_runtime": 15.5765,
|
| 825 |
+
"eval_samples_per_second": 28.183,
|
| 826 |
+
"eval_steps_per_second": 7.062,
|
| 827 |
+
"step": 1204
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 11.05,
|
| 831 |
+
"learning_rate": 3.499830105334693e-06,
|
| 832 |
+
"loss": 0.2943,
|
| 833 |
+
"step": 1210
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 11.14,
|
| 837 |
+
"learning_rate": 3.482840638803942e-06,
|
| 838 |
+
"loss": 0.4242,
|
| 839 |
+
"step": 1220
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 11.23,
|
| 843 |
+
"learning_rate": 3.4658511722731906e-06,
|
| 844 |
+
"loss": 0.2678,
|
| 845 |
+
"step": 1230
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 11.32,
|
| 849 |
+
"learning_rate": 3.44886170574244e-06,
|
| 850 |
+
"loss": 0.3375,
|
| 851 |
+
"step": 1240
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 11.42,
|
| 855 |
+
"learning_rate": 3.4318722392116887e-06,
|
| 856 |
+
"loss": 0.3046,
|
| 857 |
+
"step": 1250
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 11.51,
|
| 861 |
+
"learning_rate": 3.414882772680938e-06,
|
| 862 |
+
"loss": 0.3193,
|
| 863 |
+
"step": 1260
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 11.6,
|
| 867 |
+
"learning_rate": 3.397893306150187e-06,
|
| 868 |
+
"loss": 0.3037,
|
| 869 |
+
"step": 1270
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 11.69,
|
| 873 |
+
"learning_rate": 3.380903839619436e-06,
|
| 874 |
+
"loss": 0.2976,
|
| 875 |
+
"step": 1280
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 11.78,
|
| 879 |
+
"learning_rate": 3.3639143730886852e-06,
|
| 880 |
+
"loss": 0.4232,
|
| 881 |
+
"step": 1290
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 11.87,
|
| 885 |
+
"learning_rate": 3.3469249065579347e-06,
|
| 886 |
+
"loss": 0.3512,
|
| 887 |
+
"step": 1300
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 11.96,
|
| 891 |
+
"learning_rate": 3.3299354400271833e-06,
|
| 892 |
+
"loss": 0.3658,
|
| 893 |
+
"step": 1310
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 12.0,
|
| 897 |
+
"eval_accuracy": 0.9066059225512528,
|
| 898 |
+
"eval_loss": 0.27161121368408203,
|
| 899 |
+
"eval_runtime": 15.5481,
|
| 900 |
+
"eval_samples_per_second": 28.235,
|
| 901 |
+
"eval_steps_per_second": 7.075,
|
| 902 |
+
"step": 1314
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 12.05,
|
| 906 |
+
"learning_rate": 3.3129459734964323e-06,
|
| 907 |
+
"loss": 0.2998,
|
| 908 |
+
"step": 1320
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 12.15,
|
| 912 |
+
"learning_rate": 3.2959565069656818e-06,
|
| 913 |
+
"loss": 0.3566,
|
| 914 |
+
"step": 1330
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 12.24,
|
| 918 |
+
"learning_rate": 3.2789670404349304e-06,
|
| 919 |
+
"loss": 0.39,
|
| 920 |
+
"step": 1340
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 12.33,
|
| 924 |
+
"learning_rate": 3.26197757390418e-06,
|
| 925 |
+
"loss": 0.2815,
|
| 926 |
+
"step": 1350
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 12.42,
|
| 930 |
+
"learning_rate": 3.2449881073734285e-06,
|
| 931 |
+
"loss": 0.2475,
|
| 932 |
+
"step": 1360
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 12.51,
|
| 936 |
+
"learning_rate": 3.227998640842678e-06,
|
| 937 |
+
"loss": 0.386,
|
| 938 |
+
"step": 1370
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 12.6,
|
| 942 |
+
"learning_rate": 3.211009174311927e-06,
|
| 943 |
+
"loss": 0.328,
|
| 944 |
+
"step": 1380
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 12.69,
|
| 948 |
+
"learning_rate": 3.1940197077811756e-06,
|
| 949 |
+
"loss": 0.355,
|
| 950 |
+
"step": 1390
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 12.79,
|
| 954 |
+
"learning_rate": 3.177030241250425e-06,
|
| 955 |
+
"loss": 0.3311,
|
| 956 |
+
"step": 1400
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 12.88,
|
| 960 |
+
"learning_rate": 3.1600407747196745e-06,
|
| 961 |
+
"loss": 0.3144,
|
| 962 |
+
"step": 1410
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 12.97,
|
| 966 |
+
"learning_rate": 3.143051308188923e-06,
|
| 967 |
+
"loss": 0.4547,
|
| 968 |
+
"step": 1420
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 13.0,
|
| 972 |
+
"eval_accuracy": 0.7972665148063781,
|
| 973 |
+
"eval_loss": 0.5827615261077881,
|
| 974 |
+
"eval_runtime": 15.5305,
|
| 975 |
+
"eval_samples_per_second": 28.267,
|
| 976 |
+
"eval_steps_per_second": 7.083,
|
| 977 |
+
"step": 1423
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 13.06,
|
| 981 |
+
"learning_rate": 3.126061841658172e-06,
|
| 982 |
+
"loss": 0.3003,
|
| 983 |
+
"step": 1430
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 13.15,
|
| 987 |
+
"learning_rate": 3.109072375127421e-06,
|
| 988 |
+
"loss": 0.3906,
|
| 989 |
+
"step": 1440
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 13.24,
|
| 993 |
+
"learning_rate": 3.09208290859667e-06,
|
| 994 |
+
"loss": 0.2784,
|
| 995 |
+
"step": 1450
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 13.33,
|
| 999 |
+
"learning_rate": 3.0750934420659196e-06,
|
| 1000 |
+
"loss": 0.2384,
|
| 1001 |
+
"step": 1460
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 13.42,
|
| 1005 |
+
"learning_rate": 3.0581039755351682e-06,
|
| 1006 |
+
"loss": 0.4107,
|
| 1007 |
+
"step": 1470
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 13.52,
|
| 1011 |
+
"learning_rate": 3.0411145090044177e-06,
|
| 1012 |
+
"loss": 0.3306,
|
| 1013 |
+
"step": 1480
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 13.61,
|
| 1017 |
+
"learning_rate": 3.0241250424736667e-06,
|
| 1018 |
+
"loss": 0.3872,
|
| 1019 |
+
"step": 1490
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 13.7,
|
| 1023 |
+
"learning_rate": 3.0071355759429153e-06,
|
| 1024 |
+
"loss": 0.3002,
|
| 1025 |
+
"step": 1500
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 13.79,
|
| 1029 |
+
"learning_rate": 2.990146109412165e-06,
|
| 1030 |
+
"loss": 0.3406,
|
| 1031 |
+
"step": 1510
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 13.88,
|
| 1035 |
+
"learning_rate": 2.973156642881414e-06,
|
| 1036 |
+
"loss": 0.3361,
|
| 1037 |
+
"step": 1520
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 13.97,
|
| 1041 |
+
"learning_rate": 2.956167176350663e-06,
|
| 1042 |
+
"loss": 0.2615,
|
| 1043 |
+
"step": 1530
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 14.0,
|
| 1047 |
+
"eval_accuracy": 0.8542141230068337,
|
| 1048 |
+
"eval_loss": 0.34459054470062256,
|
| 1049 |
+
"eval_runtime": 14.8701,
|
| 1050 |
+
"eval_samples_per_second": 29.522,
|
| 1051 |
+
"eval_steps_per_second": 7.397,
|
| 1052 |
+
"step": 1533
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 14.06,
|
| 1056 |
+
"learning_rate": 2.939177709819912e-06,
|
| 1057 |
+
"loss": 0.2604,
|
| 1058 |
+
"step": 1540
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 14.16,
|
| 1062 |
+
"learning_rate": 2.922188243289161e-06,
|
| 1063 |
+
"loss": 0.3445,
|
| 1064 |
+
"step": 1550
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 14.25,
|
| 1068 |
+
"learning_rate": 2.90519877675841e-06,
|
| 1069 |
+
"loss": 0.3383,
|
| 1070 |
+
"step": 1560
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 14.34,
|
| 1074 |
+
"learning_rate": 2.8882093102276594e-06,
|
| 1075 |
+
"loss": 0.4095,
|
| 1076 |
+
"step": 1570
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 14.43,
|
| 1080 |
+
"learning_rate": 2.871219843696908e-06,
|
| 1081 |
+
"loss": 0.3362,
|
| 1082 |
+
"step": 1580
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 14.52,
|
| 1086 |
+
"learning_rate": 2.854230377166157e-06,
|
| 1087 |
+
"loss": 0.2658,
|
| 1088 |
+
"step": 1590
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 14.61,
|
| 1092 |
+
"learning_rate": 2.8372409106354065e-06,
|
| 1093 |
+
"loss": 0.2915,
|
| 1094 |
+
"step": 1600
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 14.7,
|
| 1098 |
+
"learning_rate": 2.820251444104655e-06,
|
| 1099 |
+
"loss": 0.3663,
|
| 1100 |
+
"step": 1610
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 14.79,
|
| 1104 |
+
"learning_rate": 2.8032619775739046e-06,
|
| 1105 |
+
"loss": 0.3091,
|
| 1106 |
+
"step": 1620
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 14.89,
|
| 1110 |
+
"learning_rate": 2.786272511043153e-06,
|
| 1111 |
+
"loss": 0.3601,
|
| 1112 |
+
"step": 1630
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 14.98,
|
| 1116 |
+
"learning_rate": 2.7692830445124026e-06,
|
| 1117 |
+
"loss": 0.377,
|
| 1118 |
+
"step": 1640
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 15.0,
|
| 1122 |
+
"eval_accuracy": 0.7312072892938497,
|
| 1123 |
+
"eval_loss": 0.6321907043457031,
|
| 1124 |
+
"eval_runtime": 15.6486,
|
| 1125 |
+
"eval_samples_per_second": 28.054,
|
| 1126 |
+
"eval_steps_per_second": 7.029,
|
| 1127 |
+
"step": 1642
|
| 1128 |
+
},
|
| 1129 |
+
{
|
| 1130 |
+
"epoch": 15.07,
|
| 1131 |
+
"learning_rate": 2.7522935779816517e-06,
|
| 1132 |
+
"loss": 0.3173,
|
| 1133 |
+
"step": 1650
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 15.16,
|
| 1137 |
+
"learning_rate": 2.7353041114509003e-06,
|
| 1138 |
+
"loss": 0.4798,
|
| 1139 |
+
"step": 1660
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 15.25,
|
| 1143 |
+
"learning_rate": 2.7183146449201497e-06,
|
| 1144 |
+
"loss": 0.2549,
|
| 1145 |
+
"step": 1670
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 15.34,
|
| 1149 |
+
"learning_rate": 2.701325178389399e-06,
|
| 1150 |
+
"loss": 0.4005,
|
| 1151 |
+
"step": 1680
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 15.43,
|
| 1155 |
+
"learning_rate": 2.684335711858648e-06,
|
| 1156 |
+
"loss": 0.3434,
|
| 1157 |
+
"step": 1690
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 15.53,
|
| 1161 |
+
"learning_rate": 2.667346245327897e-06,
|
| 1162 |
+
"loss": 0.2837,
|
| 1163 |
+
"step": 1700
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 15.62,
|
| 1167 |
+
"learning_rate": 2.6503567787971463e-06,
|
| 1168 |
+
"loss": 0.299,
|
| 1169 |
+
"step": 1710
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 15.71,
|
| 1173 |
+
"learning_rate": 2.633367312266395e-06,
|
| 1174 |
+
"loss": 0.2967,
|
| 1175 |
+
"step": 1720
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 15.8,
|
| 1179 |
+
"learning_rate": 2.6163778457356444e-06,
|
| 1180 |
+
"loss": 0.2949,
|
| 1181 |
+
"step": 1730
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 15.89,
|
| 1185 |
+
"learning_rate": 2.599388379204893e-06,
|
| 1186 |
+
"loss": 0.3071,
|
| 1187 |
+
"step": 1740
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 15.98,
|
| 1191 |
+
"learning_rate": 2.5823989126741424e-06,
|
| 1192 |
+
"loss": 0.2846,
|
| 1193 |
+
"step": 1750
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 16.0,
|
| 1197 |
+
"eval_accuracy": 0.9248291571753986,
|
| 1198 |
+
"eval_loss": 0.26208364963531494,
|
| 1199 |
+
"eval_runtime": 15.5182,
|
| 1200 |
+
"eval_samples_per_second": 28.289,
|
| 1201 |
+
"eval_steps_per_second": 7.088,
|
| 1202 |
+
"step": 1752
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 16.07,
|
| 1206 |
+
"learning_rate": 2.5654094461433915e-06,
|
| 1207 |
+
"loss": 0.3589,
|
| 1208 |
+
"step": 1760
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 16.16,
|
| 1212 |
+
"learning_rate": 2.54841997961264e-06,
|
| 1213 |
+
"loss": 0.3031,
|
| 1214 |
+
"step": 1770
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 16.26,
|
| 1218 |
+
"learning_rate": 2.5314305130818895e-06,
|
| 1219 |
+
"loss": 0.4131,
|
| 1220 |
+
"step": 1780
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 16.35,
|
| 1224 |
+
"learning_rate": 2.5144410465511385e-06,
|
| 1225 |
+
"loss": 0.2396,
|
| 1226 |
+
"step": 1790
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 16.44,
|
| 1230 |
+
"learning_rate": 2.4974515800203876e-06,
|
| 1231 |
+
"loss": 0.31,
|
| 1232 |
+
"step": 1800
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 16.53,
|
| 1236 |
+
"learning_rate": 2.4804621134896366e-06,
|
| 1237 |
+
"loss": 0.3869,
|
| 1238 |
+
"step": 1810
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 16.62,
|
| 1242 |
+
"learning_rate": 2.4634726469588856e-06,
|
| 1243 |
+
"loss": 0.2514,
|
| 1244 |
+
"step": 1820
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 16.71,
|
| 1248 |
+
"learning_rate": 2.4464831804281347e-06,
|
| 1249 |
+
"loss": 0.3087,
|
| 1250 |
+
"step": 1830
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 16.8,
|
| 1254 |
+
"learning_rate": 2.4294937138973837e-06,
|
| 1255 |
+
"loss": 0.2488,
|
| 1256 |
+
"step": 1840
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 16.89,
|
| 1260 |
+
"learning_rate": 2.412504247366633e-06,
|
| 1261 |
+
"loss": 0.3072,
|
| 1262 |
+
"step": 1850
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 16.99,
|
| 1266 |
+
"learning_rate": 2.3955147808358818e-06,
|
| 1267 |
+
"loss": 0.3433,
|
| 1268 |
+
"step": 1860
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 17.0,
|
| 1272 |
+
"eval_accuracy": 0.8382687927107062,
|
| 1273 |
+
"eval_loss": 0.37089645862579346,
|
| 1274 |
+
"eval_runtime": 15.1226,
|
| 1275 |
+
"eval_samples_per_second": 29.029,
|
| 1276 |
+
"eval_steps_per_second": 7.274,
|
| 1277 |
+
"step": 1861
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 17.08,
|
| 1281 |
+
"learning_rate": 2.378525314305131e-06,
|
| 1282 |
+
"loss": 0.2911,
|
| 1283 |
+
"step": 1870
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 17.17,
|
| 1287 |
+
"learning_rate": 2.36153584777438e-06,
|
| 1288 |
+
"loss": 0.2909,
|
| 1289 |
+
"step": 1880
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 17.26,
|
| 1293 |
+
"learning_rate": 2.3445463812436293e-06,
|
| 1294 |
+
"loss": 0.3273,
|
| 1295 |
+
"step": 1890
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 17.35,
|
| 1299 |
+
"learning_rate": 2.3275569147128783e-06,
|
| 1300 |
+
"loss": 0.2219,
|
| 1301 |
+
"step": 1900
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 17.44,
|
| 1305 |
+
"learning_rate": 2.3105674481821274e-06,
|
| 1306 |
+
"loss": 0.4021,
|
| 1307 |
+
"step": 1910
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 17.53,
|
| 1311 |
+
"learning_rate": 2.2935779816513764e-06,
|
| 1312 |
+
"loss": 0.3224,
|
| 1313 |
+
"step": 1920
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 17.63,
|
| 1317 |
+
"learning_rate": 2.2765885151206254e-06,
|
| 1318 |
+
"loss": 0.4674,
|
| 1319 |
+
"step": 1930
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 17.72,
|
| 1323 |
+
"learning_rate": 2.2595990485898745e-06,
|
| 1324 |
+
"loss": 0.382,
|
| 1325 |
+
"step": 1940
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 17.81,
|
| 1329 |
+
"learning_rate": 2.2426095820591235e-06,
|
| 1330 |
+
"loss": 0.2151,
|
| 1331 |
+
"step": 1950
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 17.9,
|
| 1335 |
+
"learning_rate": 2.2256201155283725e-06,
|
| 1336 |
+
"loss": 0.3831,
|
| 1337 |
+
"step": 1960
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 17.99,
|
| 1341 |
+
"learning_rate": 2.2086306489976216e-06,
|
| 1342 |
+
"loss": 0.2851,
|
| 1343 |
+
"step": 1970
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 18.0,
|
| 1347 |
+
"eval_accuracy": 0.7312072892938497,
|
| 1348 |
+
"eval_loss": 0.8134105205535889,
|
| 1349 |
+
"eval_runtime": 15.5297,
|
| 1350 |
+
"eval_samples_per_second": 28.268,
|
| 1351 |
+
"eval_steps_per_second": 7.083,
|
| 1352 |
+
"step": 1971
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 18.08,
|
| 1356 |
+
"learning_rate": 2.1916411824668706e-06,
|
| 1357 |
+
"loss": 0.2192,
|
| 1358 |
+
"step": 1980
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 18.17,
|
| 1362 |
+
"learning_rate": 2.1746517159361196e-06,
|
| 1363 |
+
"loss": 0.2858,
|
| 1364 |
+
"step": 1990
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 18.26,
|
| 1368 |
+
"learning_rate": 2.157662249405369e-06,
|
| 1369 |
+
"loss": 0.3613,
|
| 1370 |
+
"step": 2000
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 18.36,
|
| 1374 |
+
"learning_rate": 2.140672782874618e-06,
|
| 1375 |
+
"loss": 0.2634,
|
| 1376 |
+
"step": 2010
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 18.45,
|
| 1380 |
+
"learning_rate": 2.123683316343867e-06,
|
| 1381 |
+
"loss": 0.2932,
|
| 1382 |
+
"step": 2020
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 18.54,
|
| 1386 |
+
"learning_rate": 2.1066938498131157e-06,
|
| 1387 |
+
"loss": 0.2533,
|
| 1388 |
+
"step": 2030
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 18.63,
|
| 1392 |
+
"learning_rate": 2.089704383282365e-06,
|
| 1393 |
+
"loss": 0.3337,
|
| 1394 |
+
"step": 2040
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 18.72,
|
| 1398 |
+
"learning_rate": 2.0727149167516142e-06,
|
| 1399 |
+
"loss": 0.4353,
|
| 1400 |
+
"step": 2050
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 18.81,
|
| 1404 |
+
"learning_rate": 2.0557254502208633e-06,
|
| 1405 |
+
"loss": 0.3477,
|
| 1406 |
+
"step": 2060
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 18.9,
|
| 1410 |
+
"learning_rate": 2.0387359836901123e-06,
|
| 1411 |
+
"loss": 0.2786,
|
| 1412 |
+
"step": 2070
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 19.0,
|
| 1416 |
+
"learning_rate": 2.0217465171593613e-06,
|
| 1417 |
+
"loss": 0.2298,
|
| 1418 |
+
"step": 2080
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 19.0,
|
| 1422 |
+
"eval_accuracy": 0.8314350797266514,
|
| 1423 |
+
"eval_loss": 0.4324374198913574,
|
| 1424 |
+
"eval_runtime": 15.4736,
|
| 1425 |
+
"eval_samples_per_second": 28.371,
|
| 1426 |
+
"eval_steps_per_second": 7.109,
|
| 1427 |
+
"step": 2080
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 19.09,
|
| 1431 |
+
"learning_rate": 2.0047570506286104e-06,
|
| 1432 |
+
"loss": 0.3369,
|
| 1433 |
+
"step": 2090
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 19.18,
|
| 1437 |
+
"learning_rate": 1.9877675840978594e-06,
|
| 1438 |
+
"loss": 0.3051,
|
| 1439 |
+
"step": 2100
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 19.27,
|
| 1443 |
+
"learning_rate": 1.970778117567109e-06,
|
| 1444 |
+
"loss": 0.2775,
|
| 1445 |
+
"step": 2110
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 19.36,
|
| 1449 |
+
"learning_rate": 1.9537886510363575e-06,
|
| 1450 |
+
"loss": 0.2708,
|
| 1451 |
+
"step": 2120
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 19.45,
|
| 1455 |
+
"learning_rate": 1.9367991845056065e-06,
|
| 1456 |
+
"loss": 0.2424,
|
| 1457 |
+
"step": 2130
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 19.54,
|
| 1461 |
+
"learning_rate": 1.9198097179748555e-06,
|
| 1462 |
+
"loss": 0.3806,
|
| 1463 |
+
"step": 2140
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 19.63,
|
| 1467 |
+
"learning_rate": 1.902820251444105e-06,
|
| 1468 |
+
"loss": 0.3573,
|
| 1469 |
+
"step": 2150
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 19.73,
|
| 1473 |
+
"learning_rate": 1.885830784913354e-06,
|
| 1474 |
+
"loss": 0.34,
|
| 1475 |
+
"step": 2160
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 19.82,
|
| 1479 |
+
"learning_rate": 1.8688413183826028e-06,
|
| 1480 |
+
"loss": 0.3165,
|
| 1481 |
+
"step": 2170
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 19.91,
|
| 1485 |
+
"learning_rate": 1.8518518518518519e-06,
|
| 1486 |
+
"loss": 0.2916,
|
| 1487 |
+
"step": 2180
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 20.0,
|
| 1491 |
+
"learning_rate": 1.8348623853211011e-06,
|
| 1492 |
+
"loss": 0.3916,
|
| 1493 |
+
"step": 2190
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 20.0,
|
| 1497 |
+
"eval_accuracy": 0.835990888382688,
|
| 1498 |
+
"eval_loss": 0.3630986213684082,
|
| 1499 |
+
"eval_runtime": 15.5226,
|
| 1500 |
+
"eval_samples_per_second": 28.281,
|
| 1501 |
+
"eval_steps_per_second": 7.086,
|
| 1502 |
+
"step": 2190
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 20.09,
|
| 1506 |
+
"learning_rate": 1.8178729187903501e-06,
|
| 1507 |
+
"loss": 0.2495,
|
| 1508 |
+
"step": 2200
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 20.18,
|
| 1512 |
+
"learning_rate": 1.8008834522595992e-06,
|
| 1513 |
+
"loss": 0.3902,
|
| 1514 |
+
"step": 2210
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 20.27,
|
| 1518 |
+
"learning_rate": 1.7838939857288482e-06,
|
| 1519 |
+
"loss": 0.3424,
|
| 1520 |
+
"step": 2220
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 20.37,
|
| 1524 |
+
"learning_rate": 1.7669045191980975e-06,
|
| 1525 |
+
"loss": 0.3917,
|
| 1526 |
+
"step": 2230
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 20.46,
|
| 1530 |
+
"learning_rate": 1.7499150526673465e-06,
|
| 1531 |
+
"loss": 0.3536,
|
| 1532 |
+
"step": 2240
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 20.55,
|
| 1536 |
+
"learning_rate": 1.7329255861365953e-06,
|
| 1537 |
+
"loss": 0.4027,
|
| 1538 |
+
"step": 2250
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 20.64,
|
| 1542 |
+
"learning_rate": 1.7159361196058443e-06,
|
| 1543 |
+
"loss": 0.3398,
|
| 1544 |
+
"step": 2260
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 20.73,
|
| 1548 |
+
"learning_rate": 1.6989466530750936e-06,
|
| 1549 |
+
"loss": 0.2889,
|
| 1550 |
+
"step": 2270
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 20.82,
|
| 1554 |
+
"learning_rate": 1.6819571865443426e-06,
|
| 1555 |
+
"loss": 0.2609,
|
| 1556 |
+
"step": 2280
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 20.91,
|
| 1560 |
+
"learning_rate": 1.6649677200135916e-06,
|
| 1561 |
+
"loss": 0.3049,
|
| 1562 |
+
"step": 2290
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 21.0,
|
| 1566 |
+
"eval_accuracy": 0.8633257403189066,
|
| 1567 |
+
"eval_loss": 0.34050509333610535,
|
| 1568 |
+
"eval_runtime": 15.539,
|
| 1569 |
+
"eval_samples_per_second": 28.251,
|
| 1570 |
+
"eval_steps_per_second": 7.079,
|
| 1571 |
+
"step": 2299
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 21.0,
|
| 1575 |
+
"learning_rate": 1.6479782534828409e-06,
|
| 1576 |
+
"loss": 0.3258,
|
| 1577 |
+
"step": 2300
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 21.1,
|
| 1581 |
+
"learning_rate": 1.63098878695209e-06,
|
| 1582 |
+
"loss": 0.3284,
|
| 1583 |
+
"step": 2310
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 21.19,
|
| 1587 |
+
"learning_rate": 1.613999320421339e-06,
|
| 1588 |
+
"loss": 0.2377,
|
| 1589 |
+
"step": 2320
|
| 1590 |
+
},
|
| 1591 |
+
{
|
| 1592 |
+
"epoch": 21.28,
|
| 1593 |
+
"learning_rate": 1.5970098538905878e-06,
|
| 1594 |
+
"loss": 0.2422,
|
| 1595 |
+
"step": 2330
|
| 1596 |
+
},
|
| 1597 |
+
{
|
| 1598 |
+
"epoch": 21.37,
|
| 1599 |
+
"learning_rate": 1.5800203873598372e-06,
|
| 1600 |
+
"loss": 0.3573,
|
| 1601 |
+
"step": 2340
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 21.46,
|
| 1605 |
+
"learning_rate": 1.563030920829086e-06,
|
| 1606 |
+
"loss": 0.328,
|
| 1607 |
+
"step": 2350
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 21.55,
|
| 1611 |
+
"learning_rate": 1.546041454298335e-06,
|
| 1612 |
+
"loss": 0.2921,
|
| 1613 |
+
"step": 2360
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 21.64,
|
| 1617 |
+
"learning_rate": 1.5290519877675841e-06,
|
| 1618 |
+
"loss": 0.3243,
|
| 1619 |
+
"step": 2370
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 21.74,
|
| 1623 |
+
"learning_rate": 1.5120625212368334e-06,
|
| 1624 |
+
"loss": 0.276,
|
| 1625 |
+
"step": 2380
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 21.83,
|
| 1629 |
+
"learning_rate": 1.4950730547060824e-06,
|
| 1630 |
+
"loss": 0.4304,
|
| 1631 |
+
"step": 2390
|
| 1632 |
+
},
|
| 1633 |
+
{
|
| 1634 |
+
"epoch": 21.92,
|
| 1635 |
+
"learning_rate": 1.4780835881753314e-06,
|
| 1636 |
+
"loss": 0.3068,
|
| 1637 |
+
"step": 2400
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 22.0,
|
| 1641 |
+
"eval_accuracy": 0.9020501138952164,
|
| 1642 |
+
"eval_loss": 0.2584730386734009,
|
| 1643 |
+
"eval_runtime": 15.0297,
|
| 1644 |
+
"eval_samples_per_second": 29.209,
|
| 1645 |
+
"eval_steps_per_second": 7.319,
|
| 1646 |
+
"step": 2409
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 22.01,
|
| 1650 |
+
"learning_rate": 1.4610941216445805e-06,
|
| 1651 |
+
"loss": 0.2891,
|
| 1652 |
+
"step": 2410
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 22.1,
|
| 1656 |
+
"learning_rate": 1.4441046551138297e-06,
|
| 1657 |
+
"loss": 0.2476,
|
| 1658 |
+
"step": 2420
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 22.19,
|
| 1662 |
+
"learning_rate": 1.4271151885830785e-06,
|
| 1663 |
+
"loss": 0.4231,
|
| 1664 |
+
"step": 2430
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 22.28,
|
| 1668 |
+
"learning_rate": 1.4101257220523276e-06,
|
| 1669 |
+
"loss": 0.3402,
|
| 1670 |
+
"step": 2440
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 22.37,
|
| 1674 |
+
"learning_rate": 1.3931362555215766e-06,
|
| 1675 |
+
"loss": 0.2521,
|
| 1676 |
+
"step": 2450
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 22.47,
|
| 1680 |
+
"learning_rate": 1.3761467889908258e-06,
|
| 1681 |
+
"loss": 0.2561,
|
| 1682 |
+
"step": 2460
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 22.56,
|
| 1686 |
+
"learning_rate": 1.3591573224600749e-06,
|
| 1687 |
+
"loss": 0.3374,
|
| 1688 |
+
"step": 2470
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 22.65,
|
| 1692 |
+
"learning_rate": 1.342167855929324e-06,
|
| 1693 |
+
"loss": 0.2368,
|
| 1694 |
+
"step": 2480
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 22.74,
|
| 1698 |
+
"learning_rate": 1.3251783893985731e-06,
|
| 1699 |
+
"loss": 0.2519,
|
| 1700 |
+
"step": 2490
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 22.83,
|
| 1704 |
+
"learning_rate": 1.3081889228678222e-06,
|
| 1705 |
+
"loss": 0.3171,
|
| 1706 |
+
"step": 2500
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 22.92,
|
| 1710 |
+
"learning_rate": 1.2911994563370712e-06,
|
| 1711 |
+
"loss": 0.3091,
|
| 1712 |
+
"step": 2510
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 23.0,
|
| 1716 |
+
"eval_accuracy": 0.929384965831435,
|
| 1717 |
+
"eval_loss": 0.22778519988059998,
|
| 1718 |
+
"eval_runtime": 15.3862,
|
| 1719 |
+
"eval_samples_per_second": 28.532,
|
| 1720 |
+
"eval_steps_per_second": 7.149,
|
| 1721 |
+
"step": 2518
|
| 1722 |
+
},
|
| 1723 |
+
{
|
| 1724 |
+
"epoch": 23.01,
|
| 1725 |
+
"learning_rate": 1.27420998980632e-06,
|
| 1726 |
+
"loss": 0.3511,
|
| 1727 |
+
"step": 2520
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 23.11,
|
| 1731 |
+
"learning_rate": 1.2572205232755693e-06,
|
| 1732 |
+
"loss": 0.3207,
|
| 1733 |
+
"step": 2530
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 23.2,
|
| 1737 |
+
"learning_rate": 1.2402310567448183e-06,
|
| 1738 |
+
"loss": 0.2925,
|
| 1739 |
+
"step": 2540
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 23.29,
|
| 1743 |
+
"learning_rate": 1.2232415902140673e-06,
|
| 1744 |
+
"loss": 0.2371,
|
| 1745 |
+
"step": 2550
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 23.38,
|
| 1749 |
+
"learning_rate": 1.2062521236833166e-06,
|
| 1750 |
+
"loss": 0.291,
|
| 1751 |
+
"step": 2560
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 23.47,
|
| 1755 |
+
"learning_rate": 1.1892626571525654e-06,
|
| 1756 |
+
"loss": 0.2599,
|
| 1757 |
+
"step": 2570
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 23.56,
|
| 1761 |
+
"learning_rate": 1.1722731906218146e-06,
|
| 1762 |
+
"loss": 0.3854,
|
| 1763 |
+
"step": 2580
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 23.65,
|
| 1767 |
+
"learning_rate": 1.1552837240910637e-06,
|
| 1768 |
+
"loss": 0.3169,
|
| 1769 |
+
"step": 2590
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 23.74,
|
| 1773 |
+
"learning_rate": 1.1382942575603127e-06,
|
| 1774 |
+
"loss": 0.2974,
|
| 1775 |
+
"step": 2600
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 23.84,
|
| 1779 |
+
"learning_rate": 1.1213047910295617e-06,
|
| 1780 |
+
"loss": 0.3056,
|
| 1781 |
+
"step": 2610
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 23.93,
|
| 1785 |
+
"learning_rate": 1.1043153244988108e-06,
|
| 1786 |
+
"loss": 0.2749,
|
| 1787 |
+
"step": 2620
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 24.0,
|
| 1791 |
+
"eval_accuracy": 0.9043280182232346,
|
| 1792 |
+
"eval_loss": 0.29632171988487244,
|
| 1793 |
+
"eval_runtime": 15.3925,
|
| 1794 |
+
"eval_samples_per_second": 28.52,
|
| 1795 |
+
"eval_steps_per_second": 7.146,
|
| 1796 |
+
"step": 2628
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 24.02,
|
| 1800 |
+
"learning_rate": 1.0873258579680598e-06,
|
| 1801 |
+
"loss": 0.2837,
|
| 1802 |
+
"step": 2630
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 24.11,
|
| 1806 |
+
"learning_rate": 1.070336391437309e-06,
|
| 1807 |
+
"loss": 0.2906,
|
| 1808 |
+
"step": 2640
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 24.2,
|
| 1812 |
+
"learning_rate": 1.0533469249065579e-06,
|
| 1813 |
+
"loss": 0.2831,
|
| 1814 |
+
"step": 2650
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 24.29,
|
| 1818 |
+
"learning_rate": 1.0363574583758071e-06,
|
| 1819 |
+
"loss": 0.2712,
|
| 1820 |
+
"step": 2660
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 24.38,
|
| 1824 |
+
"learning_rate": 1.0193679918450562e-06,
|
| 1825 |
+
"loss": 0.2971,
|
| 1826 |
+
"step": 2670
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 24.47,
|
| 1830 |
+
"learning_rate": 1.0023785253143052e-06,
|
| 1831 |
+
"loss": 0.2782,
|
| 1832 |
+
"step": 2680
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 24.57,
|
| 1836 |
+
"learning_rate": 9.853890587835544e-07,
|
| 1837 |
+
"loss": 0.2694,
|
| 1838 |
+
"step": 2690
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 24.66,
|
| 1842 |
+
"learning_rate": 9.683995922528032e-07,
|
| 1843 |
+
"loss": 0.3956,
|
| 1844 |
+
"step": 2700
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 24.75,
|
| 1848 |
+
"learning_rate": 9.514101257220525e-07,
|
| 1849 |
+
"loss": 0.2838,
|
| 1850 |
+
"step": 2710
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 24.84,
|
| 1854 |
+
"learning_rate": 9.344206591913014e-07,
|
| 1855 |
+
"loss": 0.3524,
|
| 1856 |
+
"step": 2720
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 24.93,
|
| 1860 |
+
"learning_rate": 9.174311926605506e-07,
|
| 1861 |
+
"loss": 0.3543,
|
| 1862 |
+
"step": 2730
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 25.0,
|
| 1866 |
+
"eval_accuracy": 0.89749430523918,
|
| 1867 |
+
"eval_loss": 0.2636982798576355,
|
| 1868 |
+
"eval_runtime": 15.4509,
|
| 1869 |
+
"eval_samples_per_second": 28.413,
|
| 1870 |
+
"eval_steps_per_second": 7.119,
|
| 1871 |
+
"step": 2737
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 25.02,
|
| 1875 |
+
"learning_rate": 9.004417261297996e-07,
|
| 1876 |
+
"loss": 0.2743,
|
| 1877 |
+
"step": 2740
|
| 1878 |
+
},
|
| 1879 |
+
{
|
| 1880 |
+
"epoch": 25.11,
|
| 1881 |
+
"learning_rate": 8.834522595990487e-07,
|
| 1882 |
+
"loss": 0.3054,
|
| 1883 |
+
"step": 2750
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 25.21,
|
| 1887 |
+
"learning_rate": 8.664627930682977e-07,
|
| 1888 |
+
"loss": 0.2331,
|
| 1889 |
+
"step": 2760
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 25.3,
|
| 1893 |
+
"learning_rate": 8.494733265375468e-07,
|
| 1894 |
+
"loss": 0.2292,
|
| 1895 |
+
"step": 2770
|
| 1896 |
+
},
|
| 1897 |
+
{
|
| 1898 |
+
"epoch": 25.39,
|
| 1899 |
+
"learning_rate": 8.324838600067958e-07,
|
| 1900 |
+
"loss": 0.3069,
|
| 1901 |
+
"step": 2780
|
| 1902 |
+
},
|
| 1903 |
+
{
|
| 1904 |
+
"epoch": 25.48,
|
| 1905 |
+
"learning_rate": 8.15494393476045e-07,
|
| 1906 |
+
"loss": 0.3656,
|
| 1907 |
+
"step": 2790
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 25.57,
|
| 1911 |
+
"learning_rate": 7.985049269452939e-07,
|
| 1912 |
+
"loss": 0.3221,
|
| 1913 |
+
"step": 2800
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 25.66,
|
| 1917 |
+
"learning_rate": 7.81515460414543e-07,
|
| 1918 |
+
"loss": 0.2209,
|
| 1919 |
+
"step": 2810
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 25.75,
|
| 1923 |
+
"learning_rate": 7.645259938837921e-07,
|
| 1924 |
+
"loss": 0.3988,
|
| 1925 |
+
"step": 2820
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 25.84,
|
| 1929 |
+
"learning_rate": 7.475365273530412e-07,
|
| 1930 |
+
"loss": 0.2766,
|
| 1931 |
+
"step": 2830
|
| 1932 |
+
},
|
| 1933 |
+
{
|
| 1934 |
+
"epoch": 25.94,
|
| 1935 |
+
"learning_rate": 7.305470608222902e-07,
|
| 1936 |
+
"loss": 0.3024,
|
| 1937 |
+
"step": 2840
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 26.0,
|
| 1941 |
+
"eval_accuracy": 0.8997722095671982,
|
| 1942 |
+
"eval_loss": 0.2966226637363434,
|
| 1943 |
+
"eval_runtime": 15.5278,
|
| 1944 |
+
"eval_samples_per_second": 28.272,
|
| 1945 |
+
"eval_steps_per_second": 7.084,
|
| 1946 |
+
"step": 2847
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 26.03,
|
| 1950 |
+
"learning_rate": 7.135575942915393e-07,
|
| 1951 |
+
"loss": 0.2659,
|
| 1952 |
+
"step": 2850
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 26.12,
|
| 1956 |
+
"learning_rate": 6.965681277607883e-07,
|
| 1957 |
+
"loss": 0.32,
|
| 1958 |
+
"step": 2860
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 26.21,
|
| 1962 |
+
"learning_rate": 6.795786612300374e-07,
|
| 1963 |
+
"loss": 0.3336,
|
| 1964 |
+
"step": 2870
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 26.3,
|
| 1968 |
+
"learning_rate": 6.625891946992866e-07,
|
| 1969 |
+
"loss": 0.2326,
|
| 1970 |
+
"step": 2880
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 26.39,
|
| 1974 |
+
"learning_rate": 6.455997281685356e-07,
|
| 1975 |
+
"loss": 0.2657,
|
| 1976 |
+
"step": 2890
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 26.48,
|
| 1980 |
+
"learning_rate": 6.286102616377846e-07,
|
| 1981 |
+
"loss": 0.3482,
|
| 1982 |
+
"step": 2900
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 26.58,
|
| 1986 |
+
"learning_rate": 6.116207951070337e-07,
|
| 1987 |
+
"loss": 0.3089,
|
| 1988 |
+
"step": 2910
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 26.67,
|
| 1992 |
+
"learning_rate": 5.946313285762827e-07,
|
| 1993 |
+
"loss": 0.2407,
|
| 1994 |
+
"step": 2920
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 26.76,
|
| 1998 |
+
"learning_rate": 5.776418620455318e-07,
|
| 1999 |
+
"loss": 0.2431,
|
| 2000 |
+
"step": 2930
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 26.85,
|
| 2004 |
+
"learning_rate": 5.606523955147809e-07,
|
| 2005 |
+
"loss": 0.3574,
|
| 2006 |
+
"step": 2940
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 26.94,
|
| 2010 |
+
"learning_rate": 5.436629289840299e-07,
|
| 2011 |
+
"loss": 0.2593,
|
| 2012 |
+
"step": 2950
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 27.0,
|
| 2016 |
+
"eval_accuracy": 0.8542141230068337,
|
| 2017 |
+
"eval_loss": 0.38415804505348206,
|
| 2018 |
+
"eval_runtime": 15.5444,
|
| 2019 |
+
"eval_samples_per_second": 28.242,
|
| 2020 |
+
"eval_steps_per_second": 7.077,
|
| 2021 |
+
"step": 2956
|
| 2022 |
+
},
|
| 2023 |
+
{
|
| 2024 |
+
"epoch": 27.03,
|
| 2025 |
+
"learning_rate": 5.266734624532789e-07,
|
| 2026 |
+
"loss": 0.1965,
|
| 2027 |
+
"step": 2960
|
| 2028 |
+
},
|
| 2029 |
+
{
|
| 2030 |
+
"epoch": 27.12,
|
| 2031 |
+
"learning_rate": 5.096839959225281e-07,
|
| 2032 |
+
"loss": 0.3291,
|
| 2033 |
+
"step": 2970
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 27.21,
|
| 2037 |
+
"learning_rate": 4.926945293917772e-07,
|
| 2038 |
+
"loss": 0.3044,
|
| 2039 |
+
"step": 2980
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 27.31,
|
| 2043 |
+
"learning_rate": 4.7570506286102625e-07,
|
| 2044 |
+
"loss": 0.238,
|
| 2045 |
+
"step": 2990
|
| 2046 |
+
},
|
| 2047 |
+
{
|
| 2048 |
+
"epoch": 27.4,
|
| 2049 |
+
"learning_rate": 4.587155963302753e-07,
|
| 2050 |
+
"loss": 0.2551,
|
| 2051 |
+
"step": 3000
|
| 2052 |
+
},
|
| 2053 |
+
{
|
| 2054 |
+
"epoch": 27.49,
|
| 2055 |
+
"learning_rate": 4.4172612979952436e-07,
|
| 2056 |
+
"loss": 0.3306,
|
| 2057 |
+
"step": 3010
|
| 2058 |
+
},
|
| 2059 |
+
{
|
| 2060 |
+
"epoch": 27.58,
|
| 2061 |
+
"learning_rate": 4.247366632687734e-07,
|
| 2062 |
+
"loss": 0.3372,
|
| 2063 |
+
"step": 3020
|
| 2064 |
+
},
|
| 2065 |
+
{
|
| 2066 |
+
"epoch": 27.67,
|
| 2067 |
+
"learning_rate": 4.077471967380225e-07,
|
| 2068 |
+
"loss": 0.2652,
|
| 2069 |
+
"step": 3030
|
| 2070 |
+
},
|
| 2071 |
+
{
|
| 2072 |
+
"epoch": 27.76,
|
| 2073 |
+
"learning_rate": 3.907577302072715e-07,
|
| 2074 |
+
"loss": 0.2628,
|
| 2075 |
+
"step": 3040
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 27.85,
|
| 2079 |
+
"learning_rate": 3.737682636765206e-07,
|
| 2080 |
+
"loss": 0.3093,
|
| 2081 |
+
"step": 3050
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 27.95,
|
| 2085 |
+
"learning_rate": 3.5677879714576963e-07,
|
| 2086 |
+
"loss": 0.1979,
|
| 2087 |
+
"step": 3060
|
| 2088 |
+
},
|
| 2089 |
+
{
|
| 2090 |
+
"epoch": 28.0,
|
| 2091 |
+
"eval_accuracy": 0.8883826879271071,
|
| 2092 |
+
"eval_loss": 0.2710832953453064,
|
| 2093 |
+
"eval_runtime": 15.5551,
|
| 2094 |
+
"eval_samples_per_second": 28.222,
|
| 2095 |
+
"eval_steps_per_second": 7.072,
|
| 2096 |
+
"step": 3066
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 28.04,
|
| 2100 |
+
"learning_rate": 3.397893306150187e-07,
|
| 2101 |
+
"loss": 0.3079,
|
| 2102 |
+
"step": 3070
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 28.13,
|
| 2106 |
+
"learning_rate": 3.227998640842678e-07,
|
| 2107 |
+
"loss": 0.2954,
|
| 2108 |
+
"step": 3080
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 28.22,
|
| 2112 |
+
"learning_rate": 3.0581039755351683e-07,
|
| 2113 |
+
"loss": 0.2696,
|
| 2114 |
+
"step": 3090
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 28.31,
|
| 2118 |
+
"learning_rate": 2.888209310227659e-07,
|
| 2119 |
+
"loss": 0.2861,
|
| 2120 |
+
"step": 3100
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 28.4,
|
| 2124 |
+
"learning_rate": 2.7183146449201495e-07,
|
| 2125 |
+
"loss": 0.3212,
|
| 2126 |
+
"step": 3110
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 28.49,
|
| 2130 |
+
"learning_rate": 2.5484199796126404e-07,
|
| 2131 |
+
"loss": 0.3397,
|
| 2132 |
+
"step": 3120
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 28.58,
|
| 2136 |
+
"learning_rate": 2.3785253143051312e-07,
|
| 2137 |
+
"loss": 0.1939,
|
| 2138 |
+
"step": 3130
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 28.68,
|
| 2142 |
+
"learning_rate": 2.2086306489976218e-07,
|
| 2143 |
+
"loss": 0.258,
|
| 2144 |
+
"step": 3140
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 28.77,
|
| 2148 |
+
"learning_rate": 2.0387359836901124e-07,
|
| 2149 |
+
"loss": 0.2138,
|
| 2150 |
+
"step": 3150
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 28.86,
|
| 2154 |
+
"learning_rate": 1.868841318382603e-07,
|
| 2155 |
+
"loss": 0.2533,
|
| 2156 |
+
"step": 3160
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 28.95,
|
| 2160 |
+
"learning_rate": 1.6989466530750936e-07,
|
| 2161 |
+
"loss": 0.2549,
|
| 2162 |
+
"step": 3170
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 29.0,
|
| 2166 |
+
"eval_accuracy": 0.8633257403189066,
|
| 2167 |
+
"eval_loss": 0.3145107626914978,
|
| 2168 |
+
"eval_runtime": 15.5449,
|
| 2169 |
+
"eval_samples_per_second": 28.241,
|
| 2170 |
+
"eval_steps_per_second": 7.076,
|
| 2171 |
+
"step": 3175
|
| 2172 |
+
},
|
| 2173 |
+
{
|
| 2174 |
+
"epoch": 29.04,
|
| 2175 |
+
"learning_rate": 1.5290519877675842e-07,
|
| 2176 |
+
"loss": 0.3133,
|
| 2177 |
+
"step": 3180
|
| 2178 |
+
},
|
| 2179 |
+
{
|
| 2180 |
+
"epoch": 29.13,
|
| 2181 |
+
"learning_rate": 1.3591573224600748e-07,
|
| 2182 |
+
"loss": 0.2734,
|
| 2183 |
+
"step": 3190
|
| 2184 |
+
},
|
| 2185 |
+
{
|
| 2186 |
+
"epoch": 29.22,
|
| 2187 |
+
"learning_rate": 1.1892626571525656e-07,
|
| 2188 |
+
"loss": 0.2771,
|
| 2189 |
+
"step": 3200
|
| 2190 |
+
},
|
| 2191 |
+
{
|
| 2192 |
+
"epoch": 29.32,
|
| 2193 |
+
"learning_rate": 1.0193679918450562e-07,
|
| 2194 |
+
"loss": 0.292,
|
| 2195 |
+
"step": 3210
|
| 2196 |
+
},
|
| 2197 |
+
{
|
| 2198 |
+
"epoch": 29.41,
|
| 2199 |
+
"learning_rate": 8.494733265375468e-08,
|
| 2200 |
+
"loss": 0.3424,
|
| 2201 |
+
"step": 3220
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 29.5,
|
| 2205 |
+
"learning_rate": 6.795786612300374e-08,
|
| 2206 |
+
"loss": 0.282,
|
| 2207 |
+
"step": 3230
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 29.59,
|
| 2211 |
+
"learning_rate": 5.096839959225281e-08,
|
| 2212 |
+
"loss": 0.2225,
|
| 2213 |
+
"step": 3240
|
| 2214 |
+
},
|
| 2215 |
+
{
|
| 2216 |
+
"epoch": 29.68,
|
| 2217 |
+
"learning_rate": 3.397893306150187e-08,
|
| 2218 |
+
"loss": 0.2768,
|
| 2219 |
+
"step": 3250
|
| 2220 |
+
},
|
| 2221 |
+
{
|
| 2222 |
+
"epoch": 29.77,
|
| 2223 |
+
"learning_rate": 1.6989466530750935e-08,
|
| 2224 |
+
"loss": 0.364,
|
| 2225 |
+
"step": 3260
|
| 2226 |
+
},
|
| 2227 |
+
{
|
| 2228 |
+
"epoch": 29.86,
|
| 2229 |
+
"learning_rate": 0.0,
|
| 2230 |
+
"loss": 0.3216,
|
| 2231 |
+
"step": 3270
|
| 2232 |
+
},
|
| 2233 |
+
{
|
| 2234 |
+
"epoch": 29.86,
|
| 2235 |
+
"eval_accuracy": 0.8154897494305239,
|
| 2236 |
+
"eval_loss": 0.4564669728279114,
|
| 2237 |
+
"eval_runtime": 15.3193,
|
| 2238 |
+
"eval_samples_per_second": 28.657,
|
| 2239 |
+
"eval_steps_per_second": 7.181,
|
| 2240 |
+
"step": 3270
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"epoch": 29.86,
|
| 2244 |
+
"step": 3270,
|
| 2245 |
+
"total_flos": 2.1986190125590118e+18,
|
| 2246 |
+
"train_loss": 0.3750012589521729,
|
| 2247 |
+
"train_runtime": 3633.2931,
|
| 2248 |
+
"train_samples_per_second": 14.466,
|
| 2249 |
+
"train_steps_per_second": 0.9
|
| 2250 |
+
}
|
| 2251 |
+
],
|
| 2252 |
+
"max_steps": 3270,
|
| 2253 |
+
"num_train_epochs": 30,
|
| 2254 |
+
"total_flos": 2.1986190125590118e+18,
|
| 2255 |
+
"trial_name": null,
|
| 2256 |
+
"trial_params": null
|
| 2257 |
+
}
|