End of training
Browse files- all_results.json +13 -0
- eval_results.json +8 -0
- runs/Jul21_14-01-07_4643930eec6f/events.out.tfevents.1689958109.4643930eec6f.1577.1 +3 -0
- train_results.json +8 -0
- trainer_state.json +1771 -0
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 11.95,
|
| 3 |
+
"eval_accuracy": 0.9893267651888341,
|
| 4 |
+
"eval_loss": 0.048675429075956345,
|
| 5 |
+
"eval_runtime": 47.9714,
|
| 6 |
+
"eval_samples_per_second": 25.39,
|
| 7 |
+
"eval_steps_per_second": 1.605,
|
| 8 |
+
"total_flos": 1.195686793821323e+19,
|
| 9 |
+
"train_loss": 0.7763721961201283,
|
| 10 |
+
"train_runtime": 9836.7757,
|
| 11 |
+
"train_samples_per_second": 4.454,
|
| 12 |
+
"train_steps_per_second": 0.139
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 11.95,
|
| 3 |
+
"eval_accuracy": 0.9893267651888341,
|
| 4 |
+
"eval_loss": 0.048675429075956345,
|
| 5 |
+
"eval_runtime": 47.9714,
|
| 6 |
+
"eval_samples_per_second": 25.39,
|
| 7 |
+
"eval_steps_per_second": 1.605
|
| 8 |
+
}
|
runs/Jul21_14-01-07_4643930eec6f/events.out.tfevents.1689958109.4643930eec6f.1577.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1215e0f53047a6522881d152c8bec286b1b56281b956aa2e2061e3cba0e17441
|
| 3 |
+
size 411
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 11.95,
|
| 3 |
+
"total_flos": 1.195686793821323e+19,
|
| 4 |
+
"train_loss": 0.7763721961201283,
|
| 5 |
+
"train_runtime": 9836.7757,
|
| 6 |
+
"train_samples_per_second": 4.454,
|
| 7 |
+
"train_steps_per_second": 0.139
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1771 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9893267651888341,
|
| 3 |
+
"best_model_checkpoint": "beit-large-patch16-224-finetuned-Lesion-Classification-HAM10000-AH-60-20-20-Shuffled/checkpoint-1368",
|
| 4 |
+
"epoch": 11.947598253275109,
|
| 5 |
+
"global_step": 1368,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.04,
|
| 12 |
+
"learning_rate": 2.029220779220779e-08,
|
| 13 |
+
"loss": 2.108,
|
| 14 |
+
"step": 5
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.09,
|
| 18 |
+
"learning_rate": 4.058441558441558e-08,
|
| 19 |
+
"loss": 2.2072,
|
| 20 |
+
"step": 10
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.13,
|
| 24 |
+
"learning_rate": 6.087662337662339e-08,
|
| 25 |
+
"loss": 2.2095,
|
| 26 |
+
"step": 15
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.17,
|
| 30 |
+
"learning_rate": 8.116883116883116e-08,
|
| 31 |
+
"loss": 2.285,
|
| 32 |
+
"step": 20
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.22,
|
| 36 |
+
"learning_rate": 1.0146103896103897e-07,
|
| 37 |
+
"loss": 2.1893,
|
| 38 |
+
"step": 25
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.26,
|
| 42 |
+
"learning_rate": 1.2175324675324677e-07,
|
| 43 |
+
"loss": 2.2163,
|
| 44 |
+
"step": 30
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.31,
|
| 48 |
+
"learning_rate": 1.4204545454545455e-07,
|
| 49 |
+
"loss": 2.2329,
|
| 50 |
+
"step": 35
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.35,
|
| 54 |
+
"learning_rate": 1.6233766233766232e-07,
|
| 55 |
+
"loss": 2.1437,
|
| 56 |
+
"step": 40
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.39,
|
| 60 |
+
"learning_rate": 1.8262987012987013e-07,
|
| 61 |
+
"loss": 2.2994,
|
| 62 |
+
"step": 45
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.44,
|
| 66 |
+
"learning_rate": 2.0292207792207793e-07,
|
| 67 |
+
"loss": 2.1974,
|
| 68 |
+
"step": 50
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.48,
|
| 72 |
+
"learning_rate": 2.2321428571428574e-07,
|
| 73 |
+
"loss": 2.1891,
|
| 74 |
+
"step": 55
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.52,
|
| 78 |
+
"learning_rate": 2.4350649350649354e-07,
|
| 79 |
+
"loss": 2.3163,
|
| 80 |
+
"step": 60
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.57,
|
| 84 |
+
"learning_rate": 2.637987012987013e-07,
|
| 85 |
+
"loss": 2.2002,
|
| 86 |
+
"step": 65
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.61,
|
| 90 |
+
"learning_rate": 2.840909090909091e-07,
|
| 91 |
+
"loss": 2.0097,
|
| 92 |
+
"step": 70
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.66,
|
| 96 |
+
"learning_rate": 3.043831168831169e-07,
|
| 97 |
+
"loss": 2.0603,
|
| 98 |
+
"step": 75
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.7,
|
| 102 |
+
"learning_rate": 3.2467532467532465e-07,
|
| 103 |
+
"loss": 1.9911,
|
| 104 |
+
"step": 80
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.74,
|
| 108 |
+
"learning_rate": 3.4496753246753253e-07,
|
| 109 |
+
"loss": 2.0949,
|
| 110 |
+
"step": 85
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.79,
|
| 114 |
+
"learning_rate": 3.6525974025974026e-07,
|
| 115 |
+
"loss": 2.0385,
|
| 116 |
+
"step": 90
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.83,
|
| 120 |
+
"learning_rate": 3.855519480519481e-07,
|
| 121 |
+
"loss": 2.0068,
|
| 122 |
+
"step": 95
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.87,
|
| 126 |
+
"learning_rate": 4.0584415584415586e-07,
|
| 127 |
+
"loss": 2.1027,
|
| 128 |
+
"step": 100
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.92,
|
| 132 |
+
"learning_rate": 4.2613636363636364e-07,
|
| 133 |
+
"loss": 1.9149,
|
| 134 |
+
"step": 105
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.96,
|
| 138 |
+
"learning_rate": 4.4642857142857147e-07,
|
| 139 |
+
"loss": 2.1055,
|
| 140 |
+
"step": 110
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 1.0,
|
| 144 |
+
"eval_accuracy": 0.16009852216748768,
|
| 145 |
+
"eval_loss": 2.0091114044189453,
|
| 146 |
+
"eval_runtime": 991.3741,
|
| 147 |
+
"eval_samples_per_second": 1.229,
|
| 148 |
+
"eval_steps_per_second": 0.078,
|
| 149 |
+
"step": 114
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 1.0,
|
| 153 |
+
"learning_rate": 4.6672077922077925e-07,
|
| 154 |
+
"loss": 1.89,
|
| 155 |
+
"step": 115
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 1.05,
|
| 159 |
+
"learning_rate": 4.870129870129871e-07,
|
| 160 |
+
"loss": 2.0264,
|
| 161 |
+
"step": 120
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 1.09,
|
| 165 |
+
"learning_rate": 5.073051948051948e-07,
|
| 166 |
+
"loss": 2.0185,
|
| 167 |
+
"step": 125
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 1.14,
|
| 171 |
+
"learning_rate": 5.275974025974026e-07,
|
| 172 |
+
"loss": 1.9248,
|
| 173 |
+
"step": 130
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 1.18,
|
| 177 |
+
"learning_rate": 5.478896103896105e-07,
|
| 178 |
+
"loss": 1.8667,
|
| 179 |
+
"step": 135
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.22,
|
| 183 |
+
"learning_rate": 5.681818181818182e-07,
|
| 184 |
+
"loss": 1.8826,
|
| 185 |
+
"step": 140
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 1.27,
|
| 189 |
+
"learning_rate": 5.88474025974026e-07,
|
| 190 |
+
"loss": 1.9451,
|
| 191 |
+
"step": 145
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 1.31,
|
| 195 |
+
"learning_rate": 6.087662337662339e-07,
|
| 196 |
+
"loss": 1.8617,
|
| 197 |
+
"step": 150
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 1.35,
|
| 201 |
+
"learning_rate": 6.290584415584417e-07,
|
| 202 |
+
"loss": 1.8339,
|
| 203 |
+
"step": 155
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 1.4,
|
| 207 |
+
"learning_rate": 6.493506493506493e-07,
|
| 208 |
+
"loss": 1.8251,
|
| 209 |
+
"step": 160
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 1.44,
|
| 213 |
+
"learning_rate": 6.696428571428571e-07,
|
| 214 |
+
"loss": 1.8671,
|
| 215 |
+
"step": 165
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 1.48,
|
| 219 |
+
"learning_rate": 6.899350649350651e-07,
|
| 220 |
+
"loss": 1.8487,
|
| 221 |
+
"step": 170
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.53,
|
| 225 |
+
"learning_rate": 7.102272727272729e-07,
|
| 226 |
+
"loss": 1.7534,
|
| 227 |
+
"step": 175
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 1.57,
|
| 231 |
+
"learning_rate": 7.305194805194805e-07,
|
| 232 |
+
"loss": 1.7212,
|
| 233 |
+
"step": 180
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 1.62,
|
| 237 |
+
"learning_rate": 7.508116883116883e-07,
|
| 238 |
+
"loss": 1.8024,
|
| 239 |
+
"step": 185
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 1.66,
|
| 243 |
+
"learning_rate": 7.711038961038962e-07,
|
| 244 |
+
"loss": 1.7943,
|
| 245 |
+
"step": 190
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 1.7,
|
| 249 |
+
"learning_rate": 7.913961038961039e-07,
|
| 250 |
+
"loss": 1.6546,
|
| 251 |
+
"step": 195
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 1.75,
|
| 255 |
+
"learning_rate": 8.116883116883117e-07,
|
| 256 |
+
"loss": 1.678,
|
| 257 |
+
"step": 200
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 1.79,
|
| 261 |
+
"learning_rate": 8.319805194805196e-07,
|
| 262 |
+
"loss": 1.7724,
|
| 263 |
+
"step": 205
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.83,
|
| 267 |
+
"learning_rate": 8.522727272727273e-07,
|
| 268 |
+
"loss": 1.7413,
|
| 269 |
+
"step": 210
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 1.88,
|
| 273 |
+
"learning_rate": 8.725649350649351e-07,
|
| 274 |
+
"loss": 1.6184,
|
| 275 |
+
"step": 215
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 1.92,
|
| 279 |
+
"learning_rate": 8.928571428571429e-07,
|
| 280 |
+
"loss": 1.6969,
|
| 281 |
+
"step": 220
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 1.97,
|
| 285 |
+
"learning_rate": 9.131493506493508e-07,
|
| 286 |
+
"loss": 1.6582,
|
| 287 |
+
"step": 225
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 2.0,
|
| 291 |
+
"eval_accuracy": 0.4187192118226601,
|
| 292 |
+
"eval_loss": 1.5953072309494019,
|
| 293 |
+
"eval_runtime": 46.7943,
|
| 294 |
+
"eval_samples_per_second": 26.029,
|
| 295 |
+
"eval_steps_per_second": 1.645,
|
| 296 |
+
"step": 229
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 2.01,
|
| 300 |
+
"learning_rate": 9.334415584415585e-07,
|
| 301 |
+
"loss": 1.6087,
|
| 302 |
+
"step": 230
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 2.05,
|
| 306 |
+
"learning_rate": 9.537337662337662e-07,
|
| 307 |
+
"loss": 1.528,
|
| 308 |
+
"step": 235
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 2.1,
|
| 312 |
+
"learning_rate": 9.740259740259742e-07,
|
| 313 |
+
"loss": 1.4899,
|
| 314 |
+
"step": 240
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.14,
|
| 318 |
+
"learning_rate": 9.943181818181819e-07,
|
| 319 |
+
"loss": 1.5472,
|
| 320 |
+
"step": 245
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 2.18,
|
| 324 |
+
"learning_rate": 1.0146103896103896e-06,
|
| 325 |
+
"loss": 1.5591,
|
| 326 |
+
"step": 250
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 2.23,
|
| 330 |
+
"learning_rate": 1.0349025974025975e-06,
|
| 331 |
+
"loss": 1.4602,
|
| 332 |
+
"step": 255
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 2.27,
|
| 336 |
+
"learning_rate": 1.0551948051948053e-06,
|
| 337 |
+
"loss": 1.4764,
|
| 338 |
+
"step": 260
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 2.31,
|
| 342 |
+
"learning_rate": 1.075487012987013e-06,
|
| 343 |
+
"loss": 1.4756,
|
| 344 |
+
"step": 265
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 2.36,
|
| 348 |
+
"learning_rate": 1.095779220779221e-06,
|
| 349 |
+
"loss": 1.4322,
|
| 350 |
+
"step": 270
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 2.4,
|
| 354 |
+
"learning_rate": 1.1160714285714287e-06,
|
| 355 |
+
"loss": 1.4242,
|
| 356 |
+
"step": 275
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.45,
|
| 360 |
+
"learning_rate": 1.1363636363636364e-06,
|
| 361 |
+
"loss": 1.367,
|
| 362 |
+
"step": 280
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 2.49,
|
| 366 |
+
"learning_rate": 1.1566558441558443e-06,
|
| 367 |
+
"loss": 1.3979,
|
| 368 |
+
"step": 285
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 2.53,
|
| 372 |
+
"learning_rate": 1.176948051948052e-06,
|
| 373 |
+
"loss": 1.3273,
|
| 374 |
+
"step": 290
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 2.58,
|
| 378 |
+
"learning_rate": 1.1972402597402598e-06,
|
| 379 |
+
"loss": 1.2942,
|
| 380 |
+
"step": 295
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 2.62,
|
| 384 |
+
"learning_rate": 1.2175324675324677e-06,
|
| 385 |
+
"loss": 1.2446,
|
| 386 |
+
"step": 300
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 2.66,
|
| 390 |
+
"learning_rate": 1.2378246753246754e-06,
|
| 391 |
+
"loss": 1.2468,
|
| 392 |
+
"step": 305
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 2.71,
|
| 396 |
+
"learning_rate": 1.2581168831168834e-06,
|
| 397 |
+
"loss": 1.2983,
|
| 398 |
+
"step": 310
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 2.75,
|
| 402 |
+
"learning_rate": 1.278409090909091e-06,
|
| 403 |
+
"loss": 1.2012,
|
| 404 |
+
"step": 315
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 2.79,
|
| 408 |
+
"learning_rate": 1.2987012987012986e-06,
|
| 409 |
+
"loss": 1.1967,
|
| 410 |
+
"step": 320
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 2.84,
|
| 414 |
+
"learning_rate": 1.3189935064935067e-06,
|
| 415 |
+
"loss": 1.1737,
|
| 416 |
+
"step": 325
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 2.88,
|
| 420 |
+
"learning_rate": 1.3392857142857143e-06,
|
| 421 |
+
"loss": 1.3117,
|
| 422 |
+
"step": 330
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 2.93,
|
| 426 |
+
"learning_rate": 1.359577922077922e-06,
|
| 427 |
+
"loss": 1.1569,
|
| 428 |
+
"step": 335
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 2.97,
|
| 432 |
+
"learning_rate": 1.3798701298701301e-06,
|
| 433 |
+
"loss": 1.2399,
|
| 434 |
+
"step": 340
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 3.0,
|
| 438 |
+
"eval_accuracy": 0.5977011494252874,
|
| 439 |
+
"eval_loss": 1.1053130626678467,
|
| 440 |
+
"eval_runtime": 46.8918,
|
| 441 |
+
"eval_samples_per_second": 25.975,
|
| 442 |
+
"eval_steps_per_second": 1.642,
|
| 443 |
+
"step": 343
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 3.01,
|
| 447 |
+
"learning_rate": 1.4001623376623376e-06,
|
| 448 |
+
"loss": 1.0869,
|
| 449 |
+
"step": 345
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 3.06,
|
| 453 |
+
"learning_rate": 1.4204545454545458e-06,
|
| 454 |
+
"loss": 1.1094,
|
| 455 |
+
"step": 350
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 3.1,
|
| 459 |
+
"learning_rate": 1.4407467532467533e-06,
|
| 460 |
+
"loss": 1.1013,
|
| 461 |
+
"step": 355
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 3.14,
|
| 465 |
+
"learning_rate": 1.461038961038961e-06,
|
| 466 |
+
"loss": 1.1792,
|
| 467 |
+
"step": 360
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 3.19,
|
| 471 |
+
"learning_rate": 1.481331168831169e-06,
|
| 472 |
+
"loss": 1.0155,
|
| 473 |
+
"step": 365
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 3.23,
|
| 477 |
+
"learning_rate": 1.5016233766233767e-06,
|
| 478 |
+
"loss": 1.0912,
|
| 479 |
+
"step": 370
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 3.28,
|
| 483 |
+
"learning_rate": 1.5219155844155844e-06,
|
| 484 |
+
"loss": 1.0208,
|
| 485 |
+
"step": 375
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 3.32,
|
| 489 |
+
"learning_rate": 1.5422077922077924e-06,
|
| 490 |
+
"loss": 1.0427,
|
| 491 |
+
"step": 380
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 3.36,
|
| 495 |
+
"learning_rate": 1.5625e-06,
|
| 496 |
+
"loss": 1.0104,
|
| 497 |
+
"step": 385
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 3.41,
|
| 501 |
+
"learning_rate": 1.5827922077922078e-06,
|
| 502 |
+
"loss": 0.9838,
|
| 503 |
+
"step": 390
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 3.45,
|
| 507 |
+
"learning_rate": 1.6030844155844157e-06,
|
| 508 |
+
"loss": 0.9334,
|
| 509 |
+
"step": 395
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 3.49,
|
| 513 |
+
"learning_rate": 1.6233766233766235e-06,
|
| 514 |
+
"loss": 1.0505,
|
| 515 |
+
"step": 400
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 3.54,
|
| 519 |
+
"learning_rate": 1.6436688311688312e-06,
|
| 520 |
+
"loss": 0.8974,
|
| 521 |
+
"step": 405
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 3.58,
|
| 525 |
+
"learning_rate": 1.6639610389610391e-06,
|
| 526 |
+
"loss": 0.9257,
|
| 527 |
+
"step": 410
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 3.62,
|
| 531 |
+
"learning_rate": 1.6842532467532468e-06,
|
| 532 |
+
"loss": 0.88,
|
| 533 |
+
"step": 415
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 3.67,
|
| 537 |
+
"learning_rate": 1.7045454545454546e-06,
|
| 538 |
+
"loss": 0.9421,
|
| 539 |
+
"step": 420
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 3.71,
|
| 543 |
+
"learning_rate": 1.7248376623376625e-06,
|
| 544 |
+
"loss": 0.8861,
|
| 545 |
+
"step": 425
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 3.76,
|
| 549 |
+
"learning_rate": 1.7451298701298702e-06,
|
| 550 |
+
"loss": 0.7577,
|
| 551 |
+
"step": 430
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 3.8,
|
| 555 |
+
"learning_rate": 1.7654220779220782e-06,
|
| 556 |
+
"loss": 0.8573,
|
| 557 |
+
"step": 435
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 3.84,
|
| 561 |
+
"learning_rate": 1.7857142857142859e-06,
|
| 562 |
+
"loss": 0.8664,
|
| 563 |
+
"step": 440
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 3.89,
|
| 567 |
+
"learning_rate": 1.8060064935064936e-06,
|
| 568 |
+
"loss": 0.859,
|
| 569 |
+
"step": 445
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 3.93,
|
| 573 |
+
"learning_rate": 1.8262987012987016e-06,
|
| 574 |
+
"loss": 0.8762,
|
| 575 |
+
"step": 450
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 3.97,
|
| 579 |
+
"learning_rate": 1.8465909090909093e-06,
|
| 580 |
+
"loss": 0.8417,
|
| 581 |
+
"step": 455
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 4.0,
|
| 585 |
+
"eval_accuracy": 0.7241379310344828,
|
| 586 |
+
"eval_loss": 0.7601869702339172,
|
| 587 |
+
"eval_runtime": 46.7277,
|
| 588 |
+
"eval_samples_per_second": 26.066,
|
| 589 |
+
"eval_steps_per_second": 1.648,
|
| 590 |
+
"step": 458
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 4.02,
|
| 594 |
+
"learning_rate": 1.866883116883117e-06,
|
| 595 |
+
"loss": 0.8784,
|
| 596 |
+
"step": 460
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 4.06,
|
| 600 |
+
"learning_rate": 1.887175324675325e-06,
|
| 601 |
+
"loss": 0.9804,
|
| 602 |
+
"step": 465
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 4.1,
|
| 606 |
+
"learning_rate": 1.9074675324675324e-06,
|
| 607 |
+
"loss": 0.7837,
|
| 608 |
+
"step": 470
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 4.15,
|
| 612 |
+
"learning_rate": 1.92775974025974e-06,
|
| 613 |
+
"loss": 0.7528,
|
| 614 |
+
"step": 475
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 4.19,
|
| 618 |
+
"learning_rate": 1.9480519480519483e-06,
|
| 619 |
+
"loss": 0.6939,
|
| 620 |
+
"step": 480
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 4.24,
|
| 624 |
+
"learning_rate": 1.968344155844156e-06,
|
| 625 |
+
"loss": 0.8287,
|
| 626 |
+
"step": 485
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 4.28,
|
| 630 |
+
"learning_rate": 1.9886363636363638e-06,
|
| 631 |
+
"loss": 0.8655,
|
| 632 |
+
"step": 490
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 4.32,
|
| 636 |
+
"learning_rate": 2.0089285714285715e-06,
|
| 637 |
+
"loss": 0.7398,
|
| 638 |
+
"step": 495
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 4.37,
|
| 642 |
+
"learning_rate": 2.0292207792207792e-06,
|
| 643 |
+
"loss": 0.6669,
|
| 644 |
+
"step": 500
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 4.41,
|
| 648 |
+
"learning_rate": 2.049512987012987e-06,
|
| 649 |
+
"loss": 0.7623,
|
| 650 |
+
"step": 505
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 4.45,
|
| 654 |
+
"learning_rate": 2.069805194805195e-06,
|
| 655 |
+
"loss": 0.7007,
|
| 656 |
+
"step": 510
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 4.5,
|
| 660 |
+
"learning_rate": 2.090097402597403e-06,
|
| 661 |
+
"loss": 0.7341,
|
| 662 |
+
"step": 515
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 4.54,
|
| 666 |
+
"learning_rate": 2.1103896103896105e-06,
|
| 667 |
+
"loss": 0.6959,
|
| 668 |
+
"step": 520
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 4.59,
|
| 672 |
+
"learning_rate": 2.1306818181818183e-06,
|
| 673 |
+
"loss": 0.7649,
|
| 674 |
+
"step": 525
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 4.63,
|
| 678 |
+
"learning_rate": 2.150974025974026e-06,
|
| 679 |
+
"loss": 0.7243,
|
| 680 |
+
"step": 530
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 4.67,
|
| 684 |
+
"learning_rate": 2.171266233766234e-06,
|
| 685 |
+
"loss": 0.6517,
|
| 686 |
+
"step": 535
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 4.72,
|
| 690 |
+
"learning_rate": 2.191558441558442e-06,
|
| 691 |
+
"loss": 0.735,
|
| 692 |
+
"step": 540
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 4.76,
|
| 696 |
+
"learning_rate": 2.2118506493506496e-06,
|
| 697 |
+
"loss": 0.6925,
|
| 698 |
+
"step": 545
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 4.8,
|
| 702 |
+
"learning_rate": 2.2321428571428573e-06,
|
| 703 |
+
"loss": 0.6783,
|
| 704 |
+
"step": 550
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 4.85,
|
| 708 |
+
"learning_rate": 2.252435064935065e-06,
|
| 709 |
+
"loss": 0.5866,
|
| 710 |
+
"step": 555
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 4.89,
|
| 714 |
+
"learning_rate": 2.2727272727272728e-06,
|
| 715 |
+
"loss": 0.5415,
|
| 716 |
+
"step": 560
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 4.93,
|
| 720 |
+
"learning_rate": 2.293019480519481e-06,
|
| 721 |
+
"loss": 0.5739,
|
| 722 |
+
"step": 565
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 4.98,
|
| 726 |
+
"learning_rate": 2.3133116883116886e-06,
|
| 727 |
+
"loss": 0.5517,
|
| 728 |
+
"step": 570
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 5.0,
|
| 732 |
+
"eval_accuracy": 0.8013136288998358,
|
| 733 |
+
"eval_loss": 0.5651180744171143,
|
| 734 |
+
"eval_runtime": 46.8697,
|
| 735 |
+
"eval_samples_per_second": 25.987,
|
| 736 |
+
"eval_steps_per_second": 1.643,
|
| 737 |
+
"step": 572
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 5.02,
|
| 741 |
+
"learning_rate": 2.3336038961038964e-06,
|
| 742 |
+
"loss": 0.7433,
|
| 743 |
+
"step": 575
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 5.07,
|
| 747 |
+
"learning_rate": 2.353896103896104e-06,
|
| 748 |
+
"loss": 0.5786,
|
| 749 |
+
"step": 580
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 5.11,
|
| 753 |
+
"learning_rate": 2.374188311688312e-06,
|
| 754 |
+
"loss": 0.6458,
|
| 755 |
+
"step": 585
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 5.15,
|
| 759 |
+
"learning_rate": 2.3944805194805195e-06,
|
| 760 |
+
"loss": 0.6321,
|
| 761 |
+
"step": 590
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 5.2,
|
| 765 |
+
"learning_rate": 2.4147727272727277e-06,
|
| 766 |
+
"loss": 0.5119,
|
| 767 |
+
"step": 595
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 5.24,
|
| 771 |
+
"learning_rate": 2.4350649350649354e-06,
|
| 772 |
+
"loss": 0.627,
|
| 773 |
+
"step": 600
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 5.28,
|
| 777 |
+
"learning_rate": 2.455357142857143e-06,
|
| 778 |
+
"loss": 0.6677,
|
| 779 |
+
"step": 605
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 5.33,
|
| 783 |
+
"learning_rate": 2.475649350649351e-06,
|
| 784 |
+
"loss": 0.4946,
|
| 785 |
+
"step": 610
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 5.37,
|
| 789 |
+
"learning_rate": 2.4959415584415586e-06,
|
| 790 |
+
"loss": 0.5979,
|
| 791 |
+
"step": 615
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 5.41,
|
| 795 |
+
"learning_rate": 2.5162337662337667e-06,
|
| 796 |
+
"loss": 0.5592,
|
| 797 |
+
"step": 620
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 5.46,
|
| 801 |
+
"learning_rate": 2.536525974025974e-06,
|
| 802 |
+
"loss": 0.5518,
|
| 803 |
+
"step": 625
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 5.5,
|
| 807 |
+
"learning_rate": 2.556818181818182e-06,
|
| 808 |
+
"loss": 0.6418,
|
| 809 |
+
"step": 630
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 5.55,
|
| 813 |
+
"learning_rate": 2.57711038961039e-06,
|
| 814 |
+
"loss": 0.7065,
|
| 815 |
+
"step": 635
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 5.59,
|
| 819 |
+
"learning_rate": 2.597402597402597e-06,
|
| 820 |
+
"loss": 0.6154,
|
| 821 |
+
"step": 640
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 5.63,
|
| 825 |
+
"learning_rate": 2.6176948051948053e-06,
|
| 826 |
+
"loss": 0.6088,
|
| 827 |
+
"step": 645
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 5.68,
|
| 831 |
+
"learning_rate": 2.6379870129870135e-06,
|
| 832 |
+
"loss": 0.5844,
|
| 833 |
+
"step": 650
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 5.72,
|
| 837 |
+
"learning_rate": 2.658279220779221e-06,
|
| 838 |
+
"loss": 0.5261,
|
| 839 |
+
"step": 655
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 5.76,
|
| 843 |
+
"learning_rate": 2.6785714285714285e-06,
|
| 844 |
+
"loss": 0.5826,
|
| 845 |
+
"step": 660
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 5.81,
|
| 849 |
+
"learning_rate": 2.6988636363636367e-06,
|
| 850 |
+
"loss": 0.4568,
|
| 851 |
+
"step": 665
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 5.85,
|
| 855 |
+
"learning_rate": 2.719155844155844e-06,
|
| 856 |
+
"loss": 0.4887,
|
| 857 |
+
"step": 670
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 5.9,
|
| 861 |
+
"learning_rate": 2.739448051948052e-06,
|
| 862 |
+
"loss": 0.4853,
|
| 863 |
+
"step": 675
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 5.94,
|
| 867 |
+
"learning_rate": 2.7597402597402603e-06,
|
| 868 |
+
"loss": 0.5716,
|
| 869 |
+
"step": 680
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 5.98,
|
| 873 |
+
"learning_rate": 2.780032467532468e-06,
|
| 874 |
+
"loss": 0.5777,
|
| 875 |
+
"step": 685
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 6.0,
|
| 879 |
+
"eval_accuracy": 0.8768472906403941,
|
| 880 |
+
"eval_loss": 0.39802247285842896,
|
| 881 |
+
"eval_runtime": 46.8639,
|
| 882 |
+
"eval_samples_per_second": 25.99,
|
| 883 |
+
"eval_steps_per_second": 1.643,
|
| 884 |
+
"step": 687
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 6.03,
|
| 888 |
+
"learning_rate": 2.8003246753246753e-06,
|
| 889 |
+
"loss": 0.5484,
|
| 890 |
+
"step": 690
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 6.07,
|
| 894 |
+
"learning_rate": 2.8206168831168834e-06,
|
| 895 |
+
"loss": 0.4529,
|
| 896 |
+
"step": 695
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 6.11,
|
| 900 |
+
"learning_rate": 2.8409090909090916e-06,
|
| 901 |
+
"loss": 0.4813,
|
| 902 |
+
"step": 700
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 6.16,
|
| 906 |
+
"learning_rate": 2.861201298701299e-06,
|
| 907 |
+
"loss": 0.5704,
|
| 908 |
+
"step": 705
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 6.2,
|
| 912 |
+
"learning_rate": 2.8814935064935066e-06,
|
| 913 |
+
"loss": 0.5216,
|
| 914 |
+
"step": 710
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 6.24,
|
| 918 |
+
"learning_rate": 2.9017857142857148e-06,
|
| 919 |
+
"loss": 0.4718,
|
| 920 |
+
"step": 715
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 6.29,
|
| 924 |
+
"learning_rate": 2.922077922077922e-06,
|
| 925 |
+
"loss": 0.4797,
|
| 926 |
+
"step": 720
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 6.33,
|
| 930 |
+
"learning_rate": 2.94237012987013e-06,
|
| 931 |
+
"loss": 0.4987,
|
| 932 |
+
"step": 725
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 6.38,
|
| 936 |
+
"learning_rate": 2.962662337662338e-06,
|
| 937 |
+
"loss": 0.5066,
|
| 938 |
+
"step": 730
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 6.42,
|
| 942 |
+
"learning_rate": 2.9829545454545457e-06,
|
| 943 |
+
"loss": 0.4662,
|
| 944 |
+
"step": 735
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 6.46,
|
| 948 |
+
"learning_rate": 3.0032467532467534e-06,
|
| 949 |
+
"loss": 0.4463,
|
| 950 |
+
"step": 740
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 6.51,
|
| 954 |
+
"learning_rate": 3.0235389610389615e-06,
|
| 955 |
+
"loss": 0.3972,
|
| 956 |
+
"step": 745
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 6.55,
|
| 960 |
+
"learning_rate": 3.043831168831169e-06,
|
| 961 |
+
"loss": 0.4549,
|
| 962 |
+
"step": 750
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 6.59,
|
| 966 |
+
"learning_rate": 3.064123376623377e-06,
|
| 967 |
+
"loss": 0.4817,
|
| 968 |
+
"step": 755
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 6.64,
|
| 972 |
+
"learning_rate": 3.0844155844155847e-06,
|
| 973 |
+
"loss": 0.4626,
|
| 974 |
+
"step": 760
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 6.68,
|
| 978 |
+
"learning_rate": 3.1047077922077924e-06,
|
| 979 |
+
"loss": 0.3278,
|
| 980 |
+
"step": 765
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 6.72,
|
| 984 |
+
"learning_rate": 3.125e-06,
|
| 985 |
+
"loss": 0.4929,
|
| 986 |
+
"step": 770
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 6.77,
|
| 990 |
+
"learning_rate": 3.1452922077922083e-06,
|
| 991 |
+
"loss": 0.5325,
|
| 992 |
+
"step": 775
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 6.81,
|
| 996 |
+
"learning_rate": 3.1655844155844156e-06,
|
| 997 |
+
"loss": 0.3779,
|
| 998 |
+
"step": 780
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 6.86,
|
| 1002 |
+
"learning_rate": 3.1858766233766237e-06,
|
| 1003 |
+
"loss": 0.3715,
|
| 1004 |
+
"step": 785
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 6.9,
|
| 1008 |
+
"learning_rate": 3.2061688311688315e-06,
|
| 1009 |
+
"loss": 0.3513,
|
| 1010 |
+
"step": 790
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 6.94,
|
| 1014 |
+
"learning_rate": 3.226461038961039e-06,
|
| 1015 |
+
"loss": 0.3366,
|
| 1016 |
+
"step": 795
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 6.99,
|
| 1020 |
+
"learning_rate": 3.246753246753247e-06,
|
| 1021 |
+
"loss": 0.408,
|
| 1022 |
+
"step": 800
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 7.0,
|
| 1026 |
+
"eval_accuracy": 0.9154351395730707,
|
| 1027 |
+
"eval_loss": 0.2912403345108032,
|
| 1028 |
+
"eval_runtime": 46.6881,
|
| 1029 |
+
"eval_samples_per_second": 26.088,
|
| 1030 |
+
"eval_steps_per_second": 1.649,
|
| 1031 |
+
"step": 801
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 7.03,
|
| 1035 |
+
"learning_rate": 3.267045454545455e-06,
|
| 1036 |
+
"loss": 0.2852,
|
| 1037 |
+
"step": 805
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 7.07,
|
| 1041 |
+
"learning_rate": 3.2873376623376624e-06,
|
| 1042 |
+
"loss": 0.4248,
|
| 1043 |
+
"step": 810
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 7.12,
|
| 1047 |
+
"learning_rate": 3.3076298701298705e-06,
|
| 1048 |
+
"loss": 0.3769,
|
| 1049 |
+
"step": 815
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 7.16,
|
| 1053 |
+
"learning_rate": 3.3279220779220782e-06,
|
| 1054 |
+
"loss": 0.4492,
|
| 1055 |
+
"step": 820
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 7.21,
|
| 1059 |
+
"learning_rate": 3.3482142857142855e-06,
|
| 1060 |
+
"loss": 0.4322,
|
| 1061 |
+
"step": 825
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 7.25,
|
| 1065 |
+
"learning_rate": 3.3685064935064937e-06,
|
| 1066 |
+
"loss": 0.4586,
|
| 1067 |
+
"step": 830
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 7.29,
|
| 1071 |
+
"learning_rate": 3.388798701298702e-06,
|
| 1072 |
+
"loss": 0.3588,
|
| 1073 |
+
"step": 835
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 7.34,
|
| 1077 |
+
"learning_rate": 3.409090909090909e-06,
|
| 1078 |
+
"loss": 0.3841,
|
| 1079 |
+
"step": 840
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 7.38,
|
| 1083 |
+
"learning_rate": 3.4293831168831173e-06,
|
| 1084 |
+
"loss": 0.3817,
|
| 1085 |
+
"step": 845
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 7.42,
|
| 1089 |
+
"learning_rate": 3.449675324675325e-06,
|
| 1090 |
+
"loss": 0.326,
|
| 1091 |
+
"step": 850
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 7.47,
|
| 1095 |
+
"learning_rate": 3.4699675324675323e-06,
|
| 1096 |
+
"loss": 0.3644,
|
| 1097 |
+
"step": 855
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 7.51,
|
| 1101 |
+
"learning_rate": 3.4902597402597405e-06,
|
| 1102 |
+
"loss": 0.3328,
|
| 1103 |
+
"step": 860
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 7.55,
|
| 1107 |
+
"learning_rate": 3.5105519480519486e-06,
|
| 1108 |
+
"loss": 0.4362,
|
| 1109 |
+
"step": 865
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 7.6,
|
| 1113 |
+
"learning_rate": 3.5308441558441563e-06,
|
| 1114 |
+
"loss": 0.3732,
|
| 1115 |
+
"step": 870
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 7.64,
|
| 1119 |
+
"learning_rate": 3.5511363636363636e-06,
|
| 1120 |
+
"loss": 0.3732,
|
| 1121 |
+
"step": 875
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 7.69,
|
| 1125 |
+
"learning_rate": 3.5714285714285718e-06,
|
| 1126 |
+
"loss": 0.3757,
|
| 1127 |
+
"step": 880
|
| 1128 |
+
},
|
| 1129 |
+
{
|
| 1130 |
+
"epoch": 7.73,
|
| 1131 |
+
"learning_rate": 3.59172077922078e-06,
|
| 1132 |
+
"loss": 0.3383,
|
| 1133 |
+
"step": 885
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 7.77,
|
| 1137 |
+
"learning_rate": 3.6120129870129872e-06,
|
| 1138 |
+
"loss": 0.3281,
|
| 1139 |
+
"step": 890
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 7.82,
|
| 1143 |
+
"learning_rate": 3.6323051948051954e-06,
|
| 1144 |
+
"loss": 0.2599,
|
| 1145 |
+
"step": 895
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 7.86,
|
| 1149 |
+
"learning_rate": 3.652597402597403e-06,
|
| 1150 |
+
"loss": 0.3384,
|
| 1151 |
+
"step": 900
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 7.9,
|
| 1155 |
+
"learning_rate": 3.6728896103896104e-06,
|
| 1156 |
+
"loss": 0.3587,
|
| 1157 |
+
"step": 905
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 7.95,
|
| 1161 |
+
"learning_rate": 3.6931818181818186e-06,
|
| 1162 |
+
"loss": 0.3419,
|
| 1163 |
+
"step": 910
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 7.99,
|
| 1167 |
+
"learning_rate": 3.7134740259740267e-06,
|
| 1168 |
+
"loss": 0.2395,
|
| 1169 |
+
"step": 915
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 8.0,
|
| 1173 |
+
"eval_accuracy": 0.9417077175697866,
|
| 1174 |
+
"eval_loss": 0.21853221952915192,
|
| 1175 |
+
"eval_runtime": 46.8025,
|
| 1176 |
+
"eval_samples_per_second": 26.024,
|
| 1177 |
+
"eval_steps_per_second": 1.645,
|
| 1178 |
+
"step": 916
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 8.03,
|
| 1182 |
+
"learning_rate": 3.733766233766234e-06,
|
| 1183 |
+
"loss": 0.3855,
|
| 1184 |
+
"step": 920
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 8.08,
|
| 1188 |
+
"learning_rate": 3.7540584415584417e-06,
|
| 1189 |
+
"loss": 0.2681,
|
| 1190 |
+
"step": 925
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 8.12,
|
| 1194 |
+
"learning_rate": 3.77435064935065e-06,
|
| 1195 |
+
"loss": 0.2996,
|
| 1196 |
+
"step": 930
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 8.17,
|
| 1200 |
+
"learning_rate": 3.794642857142857e-06,
|
| 1201 |
+
"loss": 0.2894,
|
| 1202 |
+
"step": 935
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 8.21,
|
| 1206 |
+
"learning_rate": 3.814935064935065e-06,
|
| 1207 |
+
"loss": 0.2479,
|
| 1208 |
+
"step": 940
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 8.25,
|
| 1212 |
+
"learning_rate": 3.8352272727272735e-06,
|
| 1213 |
+
"loss": 0.3879,
|
| 1214 |
+
"step": 945
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 8.3,
|
| 1218 |
+
"learning_rate": 3.85551948051948e-06,
|
| 1219 |
+
"loss": 0.218,
|
| 1220 |
+
"step": 950
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 8.34,
|
| 1224 |
+
"learning_rate": 3.875811688311689e-06,
|
| 1225 |
+
"loss": 0.283,
|
| 1226 |
+
"step": 955
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 8.38,
|
| 1230 |
+
"learning_rate": 3.896103896103897e-06,
|
| 1231 |
+
"loss": 0.3067,
|
| 1232 |
+
"step": 960
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 8.43,
|
| 1236 |
+
"learning_rate": 3.916396103896104e-06,
|
| 1237 |
+
"loss": 0.2708,
|
| 1238 |
+
"step": 965
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 8.47,
|
| 1242 |
+
"learning_rate": 3.936688311688312e-06,
|
| 1243 |
+
"loss": 0.2923,
|
| 1244 |
+
"step": 970
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 8.52,
|
| 1248 |
+
"learning_rate": 3.95698051948052e-06,
|
| 1249 |
+
"loss": 0.352,
|
| 1250 |
+
"step": 975
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 8.56,
|
| 1254 |
+
"learning_rate": 3.9772727272727275e-06,
|
| 1255 |
+
"loss": 0.3366,
|
| 1256 |
+
"step": 980
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 8.6,
|
| 1260 |
+
"learning_rate": 3.997564935064935e-06,
|
| 1261 |
+
"loss": 0.3538,
|
| 1262 |
+
"step": 985
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 8.65,
|
| 1266 |
+
"learning_rate": 4.017857142857143e-06,
|
| 1267 |
+
"loss": 0.3784,
|
| 1268 |
+
"step": 990
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 8.69,
|
| 1272 |
+
"learning_rate": 4.038149350649351e-06,
|
| 1273 |
+
"loss": 0.2374,
|
| 1274 |
+
"step": 995
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 8.73,
|
| 1278 |
+
"learning_rate": 4.0584415584415584e-06,
|
| 1279 |
+
"loss": 0.2373,
|
| 1280 |
+
"step": 1000
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 8.78,
|
| 1284 |
+
"learning_rate": 4.078733766233767e-06,
|
| 1285 |
+
"loss": 0.2347,
|
| 1286 |
+
"step": 1005
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 8.82,
|
| 1290 |
+
"learning_rate": 4.099025974025974e-06,
|
| 1291 |
+
"loss": 0.2406,
|
| 1292 |
+
"step": 1010
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 8.86,
|
| 1296 |
+
"learning_rate": 4.1193181818181825e-06,
|
| 1297 |
+
"loss": 0.2949,
|
| 1298 |
+
"step": 1015
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 8.91,
|
| 1302 |
+
"learning_rate": 4.13961038961039e-06,
|
| 1303 |
+
"loss": 0.2925,
|
| 1304 |
+
"step": 1020
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 8.95,
|
| 1308 |
+
"learning_rate": 4.159902597402598e-06,
|
| 1309 |
+
"loss": 0.343,
|
| 1310 |
+
"step": 1025
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 9.0,
|
| 1314 |
+
"learning_rate": 4.180194805194806e-06,
|
| 1315 |
+
"loss": 0.3613,
|
| 1316 |
+
"step": 1030
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 9.0,
|
| 1320 |
+
"eval_accuracy": 0.9474548440065681,
|
| 1321 |
+
"eval_loss": 0.17530561983585358,
|
| 1322 |
+
"eval_runtime": 46.7366,
|
| 1323 |
+
"eval_samples_per_second": 26.061,
|
| 1324 |
+
"eval_steps_per_second": 1.648,
|
| 1325 |
+
"step": 1030
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 9.04,
|
| 1329 |
+
"learning_rate": 4.200487012987013e-06,
|
| 1330 |
+
"loss": 0.3092,
|
| 1331 |
+
"step": 1035
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 9.08,
|
| 1335 |
+
"learning_rate": 4.220779220779221e-06,
|
| 1336 |
+
"loss": 0.2786,
|
| 1337 |
+
"step": 1040
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 9.13,
|
| 1341 |
+
"learning_rate": 4.241071428571429e-06,
|
| 1342 |
+
"loss": 0.2372,
|
| 1343 |
+
"step": 1045
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 9.17,
|
| 1347 |
+
"learning_rate": 4.2613636363636365e-06,
|
| 1348 |
+
"loss": 0.2254,
|
| 1349 |
+
"step": 1050
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 9.21,
|
| 1353 |
+
"learning_rate": 4.281655844155844e-06,
|
| 1354 |
+
"loss": 0.3296,
|
| 1355 |
+
"step": 1055
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 9.26,
|
| 1359 |
+
"learning_rate": 4.301948051948052e-06,
|
| 1360 |
+
"loss": 0.2538,
|
| 1361 |
+
"step": 1060
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 9.3,
|
| 1365 |
+
"learning_rate": 4.3222402597402605e-06,
|
| 1366 |
+
"loss": 0.2594,
|
| 1367 |
+
"step": 1065
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 9.34,
|
| 1371 |
+
"learning_rate": 4.342532467532468e-06,
|
| 1372 |
+
"loss": 0.2818,
|
| 1373 |
+
"step": 1070
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 9.39,
|
| 1377 |
+
"learning_rate": 4.362824675324675e-06,
|
| 1378 |
+
"loss": 0.225,
|
| 1379 |
+
"step": 1075
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 9.43,
|
| 1383 |
+
"learning_rate": 4.383116883116884e-06,
|
| 1384 |
+
"loss": 0.1928,
|
| 1385 |
+
"step": 1080
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 9.48,
|
| 1389 |
+
"learning_rate": 4.4034090909090914e-06,
|
| 1390 |
+
"loss": 0.1809,
|
| 1391 |
+
"step": 1085
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 9.52,
|
| 1395 |
+
"learning_rate": 4.423701298701299e-06,
|
| 1396 |
+
"loss": 0.289,
|
| 1397 |
+
"step": 1090
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 9.56,
|
| 1401 |
+
"learning_rate": 4.443993506493507e-06,
|
| 1402 |
+
"loss": 0.2408,
|
| 1403 |
+
"step": 1095
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 9.61,
|
| 1407 |
+
"learning_rate": 4.464285714285715e-06,
|
| 1408 |
+
"loss": 0.2477,
|
| 1409 |
+
"step": 1100
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 9.65,
|
| 1413 |
+
"learning_rate": 4.484577922077922e-06,
|
| 1414 |
+
"loss": 0.2281,
|
| 1415 |
+
"step": 1105
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 9.69,
|
| 1419 |
+
"learning_rate": 4.50487012987013e-06,
|
| 1420 |
+
"loss": 0.2231,
|
| 1421 |
+
"step": 1110
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 9.74,
|
| 1425 |
+
"learning_rate": 4.525162337662339e-06,
|
| 1426 |
+
"loss": 0.1817,
|
| 1427 |
+
"step": 1115
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 9.78,
|
| 1431 |
+
"learning_rate": 4.5454545454545455e-06,
|
| 1432 |
+
"loss": 0.2242,
|
| 1433 |
+
"step": 1120
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 9.83,
|
| 1437 |
+
"learning_rate": 4.565746753246753e-06,
|
| 1438 |
+
"loss": 0.314,
|
| 1439 |
+
"step": 1125
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 9.87,
|
| 1443 |
+
"learning_rate": 4.586038961038962e-06,
|
| 1444 |
+
"loss": 0.1974,
|
| 1445 |
+
"step": 1130
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 9.91,
|
| 1449 |
+
"learning_rate": 4.606331168831169e-06,
|
| 1450 |
+
"loss": 0.2517,
|
| 1451 |
+
"step": 1135
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 9.96,
|
| 1455 |
+
"learning_rate": 4.626623376623377e-06,
|
| 1456 |
+
"loss": 0.1563,
|
| 1457 |
+
"step": 1140
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 10.0,
|
| 1461 |
+
"learning_rate": 4.646915584415585e-06,
|
| 1462 |
+
"loss": 0.2408,
|
| 1463 |
+
"step": 1145
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 10.0,
|
| 1467 |
+
"eval_accuracy": 0.9614121510673235,
|
| 1468 |
+
"eval_loss": 0.13528668880462646,
|
| 1469 |
+
"eval_runtime": 46.7513,
|
| 1470 |
+
"eval_samples_per_second": 26.053,
|
| 1471 |
+
"eval_steps_per_second": 1.647,
|
| 1472 |
+
"step": 1145
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 10.04,
|
| 1476 |
+
"learning_rate": 4.667207792207793e-06,
|
| 1477 |
+
"loss": 0.2764,
|
| 1478 |
+
"step": 1150
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 10.09,
|
| 1482 |
+
"learning_rate": 4.6875000000000004e-06,
|
| 1483 |
+
"loss": 0.2571,
|
| 1484 |
+
"step": 1155
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 10.13,
|
| 1488 |
+
"learning_rate": 4.707792207792208e-06,
|
| 1489 |
+
"loss": 0.153,
|
| 1490 |
+
"step": 1160
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 10.17,
|
| 1494 |
+
"learning_rate": 4.728084415584416e-06,
|
| 1495 |
+
"loss": 0.2705,
|
| 1496 |
+
"step": 1165
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 10.22,
|
| 1500 |
+
"learning_rate": 4.748376623376624e-06,
|
| 1501 |
+
"loss": 0.2665,
|
| 1502 |
+
"step": 1170
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 10.26,
|
| 1506 |
+
"learning_rate": 4.768668831168831e-06,
|
| 1507 |
+
"loss": 0.2902,
|
| 1508 |
+
"step": 1175
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 10.31,
|
| 1512 |
+
"learning_rate": 4.788961038961039e-06,
|
| 1513 |
+
"loss": 0.2662,
|
| 1514 |
+
"step": 1180
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 10.35,
|
| 1518 |
+
"learning_rate": 4.809253246753247e-06,
|
| 1519 |
+
"loss": 0.2087,
|
| 1520 |
+
"step": 1185
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 10.39,
|
| 1524 |
+
"learning_rate": 4.829545454545455e-06,
|
| 1525 |
+
"loss": 0.2877,
|
| 1526 |
+
"step": 1190
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 10.44,
|
| 1530 |
+
"learning_rate": 4.849837662337662e-06,
|
| 1531 |
+
"loss": 0.1659,
|
| 1532 |
+
"step": 1195
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 10.48,
|
| 1536 |
+
"learning_rate": 4.870129870129871e-06,
|
| 1537 |
+
"loss": 0.1714,
|
| 1538 |
+
"step": 1200
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 10.52,
|
| 1542 |
+
"learning_rate": 4.8904220779220785e-06,
|
| 1543 |
+
"loss": 0.2592,
|
| 1544 |
+
"step": 1205
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 10.57,
|
| 1548 |
+
"learning_rate": 4.910714285714286e-06,
|
| 1549 |
+
"loss": 0.2289,
|
| 1550 |
+
"step": 1210
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 10.61,
|
| 1554 |
+
"learning_rate": 4.931006493506494e-06,
|
| 1555 |
+
"loss": 0.2291,
|
| 1556 |
+
"step": 1215
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 10.66,
|
| 1560 |
+
"learning_rate": 4.951298701298702e-06,
|
| 1561 |
+
"loss": 0.1795,
|
| 1562 |
+
"step": 1220
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 10.7,
|
| 1566 |
+
"learning_rate": 4.9715909090909094e-06,
|
| 1567 |
+
"loss": 0.2915,
|
| 1568 |
+
"step": 1225
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 10.74,
|
| 1572 |
+
"learning_rate": 4.991883116883117e-06,
|
| 1573 |
+
"loss": 0.1961,
|
| 1574 |
+
"step": 1230
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 10.79,
|
| 1578 |
+
"learning_rate": 4.889705882352942e-06,
|
| 1579 |
+
"loss": 0.2525,
|
| 1580 |
+
"step": 1235
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 10.83,
|
| 1584 |
+
"learning_rate": 4.705882352941177e-06,
|
| 1585 |
+
"loss": 0.1203,
|
| 1586 |
+
"step": 1240
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 10.87,
|
| 1590 |
+
"learning_rate": 4.522058823529412e-06,
|
| 1591 |
+
"loss": 0.2102,
|
| 1592 |
+
"step": 1245
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 10.92,
|
| 1596 |
+
"learning_rate": 4.3382352941176475e-06,
|
| 1597 |
+
"loss": 0.2728,
|
| 1598 |
+
"step": 1250
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 10.96,
|
| 1602 |
+
"learning_rate": 4.154411764705883e-06,
|
| 1603 |
+
"loss": 0.2777,
|
| 1604 |
+
"step": 1255
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 11.0,
|
| 1608 |
+
"eval_accuracy": 0.9860426929392446,
|
| 1609 |
+
"eval_loss": 0.0699055939912796,
|
| 1610 |
+
"eval_runtime": 46.7733,
|
| 1611 |
+
"eval_samples_per_second": 26.04,
|
| 1612 |
+
"eval_steps_per_second": 1.646,
|
| 1613 |
+
"step": 1259
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 11.0,
|
| 1617 |
+
"learning_rate": 3.970588235294118e-06,
|
| 1618 |
+
"loss": 0.1461,
|
| 1619 |
+
"step": 1260
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 11.05,
|
| 1623 |
+
"learning_rate": 3.786764705882353e-06,
|
| 1624 |
+
"loss": 0.2089,
|
| 1625 |
+
"step": 1265
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 11.09,
|
| 1629 |
+
"learning_rate": 3.6029411764705883e-06,
|
| 1630 |
+
"loss": 0.1504,
|
| 1631 |
+
"step": 1270
|
| 1632 |
+
},
|
| 1633 |
+
{
|
| 1634 |
+
"epoch": 11.14,
|
| 1635 |
+
"learning_rate": 3.419117647058824e-06,
|
| 1636 |
+
"loss": 0.1899,
|
| 1637 |
+
"step": 1275
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 11.18,
|
| 1641 |
+
"learning_rate": 3.2352941176470594e-06,
|
| 1642 |
+
"loss": 0.1597,
|
| 1643 |
+
"step": 1280
|
| 1644 |
+
},
|
| 1645 |
+
{
|
| 1646 |
+
"epoch": 11.22,
|
| 1647 |
+
"learning_rate": 3.0514705882352947e-06,
|
| 1648 |
+
"loss": 0.1332,
|
| 1649 |
+
"step": 1285
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 11.27,
|
| 1653 |
+
"learning_rate": 2.8676470588235296e-06,
|
| 1654 |
+
"loss": 0.1553,
|
| 1655 |
+
"step": 1290
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 11.31,
|
| 1659 |
+
"learning_rate": 2.683823529411765e-06,
|
| 1660 |
+
"loss": 0.1826,
|
| 1661 |
+
"step": 1295
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 11.35,
|
| 1665 |
+
"learning_rate": 2.5e-06,
|
| 1666 |
+
"loss": 0.2113,
|
| 1667 |
+
"step": 1300
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 11.4,
|
| 1671 |
+
"learning_rate": 2.3161764705882355e-06,
|
| 1672 |
+
"loss": 0.1745,
|
| 1673 |
+
"step": 1305
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 11.44,
|
| 1677 |
+
"learning_rate": 2.132352941176471e-06,
|
| 1678 |
+
"loss": 0.2032,
|
| 1679 |
+
"step": 1310
|
| 1680 |
+
},
|
| 1681 |
+
{
|
| 1682 |
+
"epoch": 11.48,
|
| 1683 |
+
"learning_rate": 1.948529411764706e-06,
|
| 1684 |
+
"loss": 0.1668,
|
| 1685 |
+
"step": 1315
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 11.53,
|
| 1689 |
+
"learning_rate": 1.7647058823529414e-06,
|
| 1690 |
+
"loss": 0.0964,
|
| 1691 |
+
"step": 1320
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 11.57,
|
| 1695 |
+
"learning_rate": 1.5808823529411765e-06,
|
| 1696 |
+
"loss": 0.0863,
|
| 1697 |
+
"step": 1325
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 11.62,
|
| 1701 |
+
"learning_rate": 1.3970588235294119e-06,
|
| 1702 |
+
"loss": 0.1895,
|
| 1703 |
+
"step": 1330
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 11.66,
|
| 1707 |
+
"learning_rate": 1.2132352941176472e-06,
|
| 1708 |
+
"loss": 0.0883,
|
| 1709 |
+
"step": 1335
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 11.7,
|
| 1713 |
+
"learning_rate": 1.0294117647058825e-06,
|
| 1714 |
+
"loss": 0.1567,
|
| 1715 |
+
"step": 1340
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 11.75,
|
| 1719 |
+
"learning_rate": 8.455882352941178e-07,
|
| 1720 |
+
"loss": 0.1907,
|
| 1721 |
+
"step": 1345
|
| 1722 |
+
},
|
| 1723 |
+
{
|
| 1724 |
+
"epoch": 11.79,
|
| 1725 |
+
"learning_rate": 6.61764705882353e-07,
|
| 1726 |
+
"loss": 0.0819,
|
| 1727 |
+
"step": 1350
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 11.83,
|
| 1731 |
+
"learning_rate": 4.779411764705882e-07,
|
| 1732 |
+
"loss": 0.16,
|
| 1733 |
+
"step": 1355
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 11.88,
|
| 1737 |
+
"learning_rate": 2.9411764705882356e-07,
|
| 1738 |
+
"loss": 0.1602,
|
| 1739 |
+
"step": 1360
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 11.92,
|
| 1743 |
+
"learning_rate": 1.1029411764705884e-07,
|
| 1744 |
+
"loss": 0.1528,
|
| 1745 |
+
"step": 1365
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 11.95,
|
| 1749 |
+
"eval_accuracy": 0.9893267651888341,
|
| 1750 |
+
"eval_loss": 0.048675429075956345,
|
| 1751 |
+
"eval_runtime": 47.0491,
|
| 1752 |
+
"eval_samples_per_second": 25.888,
|
| 1753 |
+
"eval_steps_per_second": 1.637,
|
| 1754 |
+
"step": 1368
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 11.95,
|
| 1758 |
+
"step": 1368,
|
| 1759 |
+
"total_flos": 1.195686793821323e+19,
|
| 1760 |
+
"train_loss": 0.7763721961201283,
|
| 1761 |
+
"train_runtime": 9836.7757,
|
| 1762 |
+
"train_samples_per_second": 4.454,
|
| 1763 |
+
"train_steps_per_second": 0.139
|
| 1764 |
+
}
|
| 1765 |
+
],
|
| 1766 |
+
"max_steps": 1368,
|
| 1767 |
+
"num_train_epochs": 12,
|
| 1768 |
+
"total_flos": 1.195686793821323e+19,
|
| 1769 |
+
"trial_name": null,
|
| 1770 |
+
"trial_params": null
|
| 1771 |
+
}
|