Commit
•
1bd1272
1
Parent(s):
e5cca3e
End of training
Browse files- all_results.json +11 -0
- eval_results.json +7 -0
- runs/Mar13_16-24-48_d08e23154428/events.out.tfevents.1647190790.d08e23154428.4470.2 +3 -0
- train_results.json +7 -0
- trainer_state.json +2505 -0
all_results.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 100.0,
|
| 3 |
+
"eval_loss": 0.2674923837184906,
|
| 4 |
+
"eval_runtime": 2.6371,
|
| 5 |
+
"eval_samples_per_second": 117.554,
|
| 6 |
+
"eval_steps_per_second": 1.896,
|
| 7 |
+
"train_loss": 0.1087875749383654,
|
| 8 |
+
"train_runtime": 1716.2917,
|
| 9 |
+
"train_samples_per_second": 102.022,
|
| 10 |
+
"train_steps_per_second": 1.631
|
| 11 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 100.0,
|
| 3 |
+
"eval_loss": 0.2674923837184906,
|
| 4 |
+
"eval_runtime": 2.6371,
|
| 5 |
+
"eval_samples_per_second": 117.554,
|
| 6 |
+
"eval_steps_per_second": 1.896
|
| 7 |
+
}
|
runs/Mar13_16-24-48_d08e23154428/events.out.tfevents.1647190790.d08e23154428.4470.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c5b5ceb152e23ae0cda10b05e10ba38c8aeb66fc3d5c1c871173eff378331230
|
| 3 |
+
size 311
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 100.0,
|
| 3 |
+
"train_loss": 0.1087875749383654,
|
| 4 |
+
"train_runtime": 1716.2917,
|
| 5 |
+
"train_samples_per_second": 102.022,
|
| 6 |
+
"train_steps_per_second": 1.631
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2505 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.26857295632362366,
|
| 3 |
+
"best_model_checkpoint": "./vit-mae-flysheet/checkpoint-2716",
|
| 4 |
+
"epoch": 100.0,
|
| 5 |
+
"global_step": 2800,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.36,
|
| 12 |
+
"learning_rate": 2.3437499999999998e-07,
|
| 13 |
+
"loss": 2.2918,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.71,
|
| 18 |
+
"learning_rate": 5.691964285714285e-07,
|
| 19 |
+
"loss": 2.284,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 1.0,
|
| 24 |
+
"eval_loss": 2.281158924102783,
|
| 25 |
+
"eval_runtime": 2.6825,
|
| 26 |
+
"eval_samples_per_second": 115.564,
|
| 27 |
+
"eval_steps_per_second": 1.864,
|
| 28 |
+
"step": 28
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"epoch": 1.07,
|
| 32 |
+
"learning_rate": 9.040178571428571e-07,
|
| 33 |
+
"loss": 2.29,
|
| 34 |
+
"step": 30
|
| 35 |
+
},
|
| 36 |
+
{
|
| 37 |
+
"epoch": 1.43,
|
| 38 |
+
"learning_rate": 1.2388392857142857e-06,
|
| 39 |
+
"loss": 2.1474,
|
| 40 |
+
"step": 40
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"epoch": 1.79,
|
| 44 |
+
"learning_rate": 1.573660714285714e-06,
|
| 45 |
+
"loss": 2.137,
|
| 46 |
+
"step": 50
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 2.0,
|
| 50 |
+
"eval_loss": 2.0287907123565674,
|
| 51 |
+
"eval_runtime": 2.9845,
|
| 52 |
+
"eval_samples_per_second": 103.869,
|
| 53 |
+
"eval_steps_per_second": 1.675,
|
| 54 |
+
"step": 56
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 2.14,
|
| 58 |
+
"learning_rate": 1.9084821428571425e-06,
|
| 59 |
+
"loss": 2.0038,
|
| 60 |
+
"step": 60
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 2.5,
|
| 64 |
+
"learning_rate": 2.2433035714285713e-06,
|
| 65 |
+
"loss": 1.8154,
|
| 66 |
+
"step": 70
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 2.86,
|
| 70 |
+
"learning_rate": 2.578125e-06,
|
| 71 |
+
"loss": 1.6016,
|
| 72 |
+
"step": 80
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 3.0,
|
| 76 |
+
"eval_loss": 1.243688702583313,
|
| 77 |
+
"eval_runtime": 3.1026,
|
| 78 |
+
"eval_samples_per_second": 99.916,
|
| 79 |
+
"eval_steps_per_second": 1.612,
|
| 80 |
+
"step": 84
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 3.21,
|
| 84 |
+
"learning_rate": 2.9129464285714283e-06,
|
| 85 |
+
"loss": 1.2624,
|
| 86 |
+
"step": 90
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 3.57,
|
| 90 |
+
"learning_rate": 3.2477678571428566e-06,
|
| 91 |
+
"loss": 0.9693,
|
| 92 |
+
"step": 100
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 3.93,
|
| 96 |
+
"learning_rate": 3.5825892857142853e-06,
|
| 97 |
+
"loss": 0.8055,
|
| 98 |
+
"step": 110
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 4.0,
|
| 102 |
+
"eval_loss": 0.7418646812438965,
|
| 103 |
+
"eval_runtime": 3.0666,
|
| 104 |
+
"eval_samples_per_second": 101.089,
|
| 105 |
+
"eval_steps_per_second": 1.63,
|
| 106 |
+
"step": 112
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 4.29,
|
| 110 |
+
"learning_rate": 3.917410714285714e-06,
|
| 111 |
+
"loss": 0.7145,
|
| 112 |
+
"step": 120
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 4.64,
|
| 116 |
+
"learning_rate": 4.252232142857143e-06,
|
| 117 |
+
"loss": 0.6144,
|
| 118 |
+
"step": 130
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 5.0,
|
| 122 |
+
"learning_rate": 4.587053571428571e-06,
|
| 123 |
+
"loss": 0.5304,
|
| 124 |
+
"step": 140
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 5.0,
|
| 128 |
+
"eval_loss": 0.5150813460350037,
|
| 129 |
+
"eval_runtime": 2.9639,
|
| 130 |
+
"eval_samples_per_second": 104.591,
|
| 131 |
+
"eval_steps_per_second": 1.687,
|
| 132 |
+
"step": 140
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 5.36,
|
| 136 |
+
"learning_rate": 4.921874999999999e-06,
|
| 137 |
+
"loss": 0.4939,
|
| 138 |
+
"step": 150
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 5.71,
|
| 142 |
+
"learning_rate": 5.256696428571428e-06,
|
| 143 |
+
"loss": 0.4873,
|
| 144 |
+
"step": 160
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 6.0,
|
| 148 |
+
"eval_loss": 0.4884074628353119,
|
| 149 |
+
"eval_runtime": 3.1297,
|
| 150 |
+
"eval_samples_per_second": 99.052,
|
| 151 |
+
"eval_steps_per_second": 1.598,
|
| 152 |
+
"step": 168
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 6.07,
|
| 156 |
+
"learning_rate": 5.591517857142857e-06,
|
| 157 |
+
"loss": 0.465,
|
| 158 |
+
"step": 170
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 6.43,
|
| 162 |
+
"learning_rate": 5.926339285714285e-06,
|
| 163 |
+
"loss": 0.4548,
|
| 164 |
+
"step": 180
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 6.79,
|
| 168 |
+
"learning_rate": 6.261160714285713e-06,
|
| 169 |
+
"loss": 0.442,
|
| 170 |
+
"step": 190
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 7.0,
|
| 174 |
+
"eval_loss": 0.44406256079673767,
|
| 175 |
+
"eval_runtime": 3.1182,
|
| 176 |
+
"eval_samples_per_second": 99.418,
|
| 177 |
+
"eval_steps_per_second": 1.604,
|
| 178 |
+
"step": 196
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 7.14,
|
| 182 |
+
"learning_rate": 6.595982142857143e-06,
|
| 183 |
+
"loss": 0.4262,
|
| 184 |
+
"step": 200
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 7.5,
|
| 188 |
+
"learning_rate": 6.930803571428571e-06,
|
| 189 |
+
"loss": 0.4275,
|
| 190 |
+
"step": 210
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 7.86,
|
| 194 |
+
"learning_rate": 7.2656249999999996e-06,
|
| 195 |
+
"loss": 0.4039,
|
| 196 |
+
"step": 220
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 8.0,
|
| 200 |
+
"eval_loss": 0.41594529151916504,
|
| 201 |
+
"eval_runtime": 3.0271,
|
| 202 |
+
"eval_samples_per_second": 102.41,
|
| 203 |
+
"eval_steps_per_second": 1.652,
|
| 204 |
+
"step": 224
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 8.21,
|
| 208 |
+
"learning_rate": 7.600446428571428e-06,
|
| 209 |
+
"loss": 0.3925,
|
| 210 |
+
"step": 230
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 8.57,
|
| 214 |
+
"learning_rate": 7.935267857142856e-06,
|
| 215 |
+
"loss": 0.4092,
|
| 216 |
+
"step": 240
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 8.93,
|
| 220 |
+
"learning_rate": 8.270089285714285e-06,
|
| 221 |
+
"loss": 0.3866,
|
| 222 |
+
"step": 250
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 9.0,
|
| 226 |
+
"eval_loss": 0.39747074246406555,
|
| 227 |
+
"eval_runtime": 3.1219,
|
| 228 |
+
"eval_samples_per_second": 99.3,
|
| 229 |
+
"eval_steps_per_second": 1.602,
|
| 230 |
+
"step": 252
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 9.29,
|
| 234 |
+
"learning_rate": 8.604910714285714e-06,
|
| 235 |
+
"loss": 0.3675,
|
| 236 |
+
"step": 260
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 9.64,
|
| 240 |
+
"learning_rate": 8.939732142857142e-06,
|
| 241 |
+
"loss": 0.3827,
|
| 242 |
+
"step": 270
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 10.0,
|
| 246 |
+
"learning_rate": 9.274553571428571e-06,
|
| 247 |
+
"loss": 0.391,
|
| 248 |
+
"step": 280
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 10.0,
|
| 252 |
+
"eval_loss": 0.38686203956604004,
|
| 253 |
+
"eval_runtime": 2.6794,
|
| 254 |
+
"eval_samples_per_second": 115.697,
|
| 255 |
+
"eval_steps_per_second": 1.866,
|
| 256 |
+
"step": 280
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 10.36,
|
| 260 |
+
"learning_rate": 9.609374999999998e-06,
|
| 261 |
+
"loss": 0.3701,
|
| 262 |
+
"step": 290
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 10.71,
|
| 266 |
+
"learning_rate": 9.944196428571429e-06,
|
| 267 |
+
"loss": 0.3549,
|
| 268 |
+
"step": 300
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 11.0,
|
| 272 |
+
"eval_loss": 0.38007286190986633,
|
| 273 |
+
"eval_runtime": 2.6889,
|
| 274 |
+
"eval_samples_per_second": 115.289,
|
| 275 |
+
"eval_steps_per_second": 1.859,
|
| 276 |
+
"step": 308
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 11.07,
|
| 280 |
+
"learning_rate": 1.0279017857142857e-05,
|
| 281 |
+
"loss": 0.3618,
|
| 282 |
+
"step": 310
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 11.43,
|
| 286 |
+
"learning_rate": 1.0613839285714284e-05,
|
| 287 |
+
"loss": 0.357,
|
| 288 |
+
"step": 320
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 11.79,
|
| 292 |
+
"learning_rate": 1.0948660714285715e-05,
|
| 293 |
+
"loss": 0.3462,
|
| 294 |
+
"step": 330
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 12.0,
|
| 298 |
+
"eval_loss": 0.35770875215530396,
|
| 299 |
+
"eval_runtime": 2.6881,
|
| 300 |
+
"eval_samples_per_second": 115.324,
|
| 301 |
+
"eval_steps_per_second": 1.86,
|
| 302 |
+
"step": 336
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 12.14,
|
| 306 |
+
"learning_rate": 1.1283482142857142e-05,
|
| 307 |
+
"loss": 0.3413,
|
| 308 |
+
"step": 340
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 12.5,
|
| 312 |
+
"learning_rate": 1.161830357142857e-05,
|
| 313 |
+
"loss": 0.341,
|
| 314 |
+
"step": 350
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 12.86,
|
| 318 |
+
"learning_rate": 1.1953124999999997e-05,
|
| 319 |
+
"loss": 0.3402,
|
| 320 |
+
"step": 360
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 13.0,
|
| 324 |
+
"eval_loss": 0.3519320487976074,
|
| 325 |
+
"eval_runtime": 2.6575,
|
| 326 |
+
"eval_samples_per_second": 116.65,
|
| 327 |
+
"eval_steps_per_second": 1.881,
|
| 328 |
+
"step": 364
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 13.21,
|
| 332 |
+
"learning_rate": 1.2287946428571428e-05,
|
| 333 |
+
"loss": 0.3281,
|
| 334 |
+
"step": 370
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 13.57,
|
| 338 |
+
"learning_rate": 1.2622767857142857e-05,
|
| 339 |
+
"loss": 0.3423,
|
| 340 |
+
"step": 380
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 13.93,
|
| 344 |
+
"learning_rate": 1.2957589285714284e-05,
|
| 345 |
+
"loss": 0.3357,
|
| 346 |
+
"step": 390
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 14.0,
|
| 350 |
+
"eval_loss": 0.344655305147171,
|
| 351 |
+
"eval_runtime": 2.6732,
|
| 352 |
+
"eval_samples_per_second": 115.968,
|
| 353 |
+
"eval_steps_per_second": 1.87,
|
| 354 |
+
"step": 392
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 14.29,
|
| 358 |
+
"learning_rate": 1.3292410714285714e-05,
|
| 359 |
+
"loss": 0.3254,
|
| 360 |
+
"step": 400
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 14.64,
|
| 364 |
+
"learning_rate": 1.3627232142857141e-05,
|
| 365 |
+
"loss": 0.3231,
|
| 366 |
+
"step": 410
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 15.0,
|
| 370 |
+
"learning_rate": 1.396205357142857e-05,
|
| 371 |
+
"loss": 0.3474,
|
| 372 |
+
"step": 420
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 15.0,
|
| 376 |
+
"eval_loss": 0.33689987659454346,
|
| 377 |
+
"eval_runtime": 2.6785,
|
| 378 |
+
"eval_samples_per_second": 115.736,
|
| 379 |
+
"eval_steps_per_second": 1.867,
|
| 380 |
+
"step": 420
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 15.36,
|
| 384 |
+
"learning_rate": 1.4296874999999999e-05,
|
| 385 |
+
"loss": 0.3117,
|
| 386 |
+
"step": 430
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 15.71,
|
| 390 |
+
"learning_rate": 1.4631696428571427e-05,
|
| 391 |
+
"loss": 0.3254,
|
| 392 |
+
"step": 440
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 16.0,
|
| 396 |
+
"eval_loss": 0.33861759305000305,
|
| 397 |
+
"eval_runtime": 2.6231,
|
| 398 |
+
"eval_samples_per_second": 118.183,
|
| 399 |
+
"eval_steps_per_second": 1.906,
|
| 400 |
+
"step": 448
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 16.07,
|
| 404 |
+
"learning_rate": 1.4966517857142856e-05,
|
| 405 |
+
"loss": 0.3221,
|
| 406 |
+
"step": 450
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 16.43,
|
| 410 |
+
"learning_rate": 1.5301339285714285e-05,
|
| 411 |
+
"loss": 0.3347,
|
| 412 |
+
"step": 460
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 16.79,
|
| 416 |
+
"learning_rate": 1.5636160714285715e-05,
|
| 417 |
+
"loss": 0.3033,
|
| 418 |
+
"step": 470
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 17.0,
|
| 422 |
+
"eval_loss": 0.3294062316417694,
|
| 423 |
+
"eval_runtime": 2.6872,
|
| 424 |
+
"eval_samples_per_second": 115.362,
|
| 425 |
+
"eval_steps_per_second": 1.861,
|
| 426 |
+
"step": 476
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 17.14,
|
| 430 |
+
"learning_rate": 1.5970982142857142e-05,
|
| 431 |
+
"loss": 0.3135,
|
| 432 |
+
"step": 480
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 17.5,
|
| 436 |
+
"learning_rate": 1.630580357142857e-05,
|
| 437 |
+
"loss": 0.3104,
|
| 438 |
+
"step": 490
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 17.86,
|
| 442 |
+
"learning_rate": 1.6640624999999996e-05,
|
| 443 |
+
"loss": 0.3047,
|
| 444 |
+
"step": 500
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 18.0,
|
| 448 |
+
"eval_loss": 0.32739999890327454,
|
| 449 |
+
"eval_runtime": 2.6527,
|
| 450 |
+
"eval_samples_per_second": 116.864,
|
| 451 |
+
"eval_steps_per_second": 1.885,
|
| 452 |
+
"step": 504
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 18.21,
|
| 456 |
+
"learning_rate": 1.6975446428571427e-05,
|
| 457 |
+
"loss": 0.3088,
|
| 458 |
+
"step": 510
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 18.57,
|
| 462 |
+
"learning_rate": 1.7310267857142857e-05,
|
| 463 |
+
"loss": 0.3079,
|
| 464 |
+
"step": 520
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 18.93,
|
| 468 |
+
"learning_rate": 1.7645089285714284e-05,
|
| 469 |
+
"loss": 0.3103,
|
| 470 |
+
"step": 530
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 19.0,
|
| 474 |
+
"eval_loss": 0.3209254741668701,
|
| 475 |
+
"eval_runtime": 2.6351,
|
| 476 |
+
"eval_samples_per_second": 117.643,
|
| 477 |
+
"eval_steps_per_second": 1.897,
|
| 478 |
+
"step": 532
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 19.29,
|
| 482 |
+
"learning_rate": 1.7979910714285714e-05,
|
| 483 |
+
"loss": 0.3004,
|
| 484 |
+
"step": 540
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 19.64,
|
| 488 |
+
"learning_rate": 1.831473214285714e-05,
|
| 489 |
+
"loss": 0.3037,
|
| 490 |
+
"step": 550
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 20.0,
|
| 494 |
+
"learning_rate": 1.864955357142857e-05,
|
| 495 |
+
"loss": 0.3067,
|
| 496 |
+
"step": 560
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 20.0,
|
| 500 |
+
"eval_loss": 0.3185921013355255,
|
| 501 |
+
"eval_runtime": 2.6798,
|
| 502 |
+
"eval_samples_per_second": 115.68,
|
| 503 |
+
"eval_steps_per_second": 1.866,
|
| 504 |
+
"step": 560
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 20.36,
|
| 508 |
+
"learning_rate": 1.8984375e-05,
|
| 509 |
+
"loss": 0.3007,
|
| 510 |
+
"step": 570
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 20.71,
|
| 514 |
+
"learning_rate": 1.9319196428571426e-05,
|
| 515 |
+
"loss": 0.2959,
|
| 516 |
+
"step": 580
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 21.0,
|
| 520 |
+
"eval_loss": 0.31901559233665466,
|
| 521 |
+
"eval_runtime": 2.6847,
|
| 522 |
+
"eval_samples_per_second": 115.468,
|
| 523 |
+
"eval_steps_per_second": 1.862,
|
| 524 |
+
"step": 588
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 21.07,
|
| 528 |
+
"learning_rate": 1.9654017857142856e-05,
|
| 529 |
+
"loss": 0.3031,
|
| 530 |
+
"step": 590
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 21.43,
|
| 534 |
+
"learning_rate": 1.9988839285714283e-05,
|
| 535 |
+
"loss": 0.3107,
|
| 536 |
+
"step": 600
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 21.79,
|
| 540 |
+
"learning_rate": 2.032366071428571e-05,
|
| 541 |
+
"loss": 0.2899,
|
| 542 |
+
"step": 610
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 22.0,
|
| 546 |
+
"eval_loss": 0.3146977126598358,
|
| 547 |
+
"eval_runtime": 2.6772,
|
| 548 |
+
"eval_samples_per_second": 115.792,
|
| 549 |
+
"eval_steps_per_second": 1.868,
|
| 550 |
+
"step": 616
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 22.14,
|
| 554 |
+
"learning_rate": 2.0658482142857144e-05,
|
| 555 |
+
"loss": 0.2868,
|
| 556 |
+
"step": 620
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 22.5,
|
| 560 |
+
"learning_rate": 2.099330357142857e-05,
|
| 561 |
+
"loss": 0.3018,
|
| 562 |
+
"step": 630
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 22.86,
|
| 566 |
+
"learning_rate": 2.1328125e-05,
|
| 567 |
+
"loss": 0.2872,
|
| 568 |
+
"step": 640
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 23.0,
|
| 572 |
+
"eval_loss": 0.30817940831184387,
|
| 573 |
+
"eval_runtime": 2.6553,
|
| 574 |
+
"eval_samples_per_second": 116.748,
|
| 575 |
+
"eval_steps_per_second": 1.883,
|
| 576 |
+
"step": 644
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 23.21,
|
| 580 |
+
"learning_rate": 2.1662946428571425e-05,
|
| 581 |
+
"loss": 0.3033,
|
| 582 |
+
"step": 650
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 23.57,
|
| 586 |
+
"learning_rate": 2.1997767857142856e-05,
|
| 587 |
+
"loss": 0.2946,
|
| 588 |
+
"step": 660
|
| 589 |
+
},
|
| 590 |
+
{
|
| 591 |
+
"epoch": 23.93,
|
| 592 |
+
"learning_rate": 2.2332589285714283e-05,
|
| 593 |
+
"loss": 0.2956,
|
| 594 |
+
"step": 670
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 24.0,
|
| 598 |
+
"eval_loss": 0.3069721758365631,
|
| 599 |
+
"eval_runtime": 2.6278,
|
| 600 |
+
"eval_samples_per_second": 117.971,
|
| 601 |
+
"eval_steps_per_second": 1.903,
|
| 602 |
+
"step": 672
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 24.29,
|
| 606 |
+
"learning_rate": 2.266741071428571e-05,
|
| 607 |
+
"loss": 0.2754,
|
| 608 |
+
"step": 680
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 24.64,
|
| 612 |
+
"learning_rate": 2.3002232142857144e-05,
|
| 613 |
+
"loss": 0.3039,
|
| 614 |
+
"step": 690
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 25.0,
|
| 618 |
+
"learning_rate": 2.333705357142857e-05,
|
| 619 |
+
"loss": 0.2865,
|
| 620 |
+
"step": 700
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 25.0,
|
| 624 |
+
"eval_loss": 0.30717733502388,
|
| 625 |
+
"eval_runtime": 2.6953,
|
| 626 |
+
"eval_samples_per_second": 115.015,
|
| 627 |
+
"eval_steps_per_second": 1.855,
|
| 628 |
+
"step": 700
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"epoch": 25.36,
|
| 632 |
+
"learning_rate": 2.3671874999999998e-05,
|
| 633 |
+
"loss": 0.2865,
|
| 634 |
+
"step": 710
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 25.71,
|
| 638 |
+
"learning_rate": 2.4006696428571425e-05,
|
| 639 |
+
"loss": 0.2947,
|
| 640 |
+
"step": 720
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 26.0,
|
| 644 |
+
"eval_loss": 0.3071895241737366,
|
| 645 |
+
"eval_runtime": 2.6411,
|
| 646 |
+
"eval_samples_per_second": 117.376,
|
| 647 |
+
"eval_steps_per_second": 1.893,
|
| 648 |
+
"step": 728
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 26.07,
|
| 652 |
+
"learning_rate": 2.4341517857142855e-05,
|
| 653 |
+
"loss": 0.2831,
|
| 654 |
+
"step": 730
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 26.43,
|
| 658 |
+
"learning_rate": 2.4676339285714282e-05,
|
| 659 |
+
"loss": 0.3008,
|
| 660 |
+
"step": 740
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 26.79,
|
| 664 |
+
"learning_rate": 2.501116071428571e-05,
|
| 665 |
+
"loss": 0.2811,
|
| 666 |
+
"step": 750
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 27.0,
|
| 670 |
+
"eval_loss": 0.3130524158477783,
|
| 671 |
+
"eval_runtime": 2.683,
|
| 672 |
+
"eval_samples_per_second": 115.544,
|
| 673 |
+
"eval_steps_per_second": 1.864,
|
| 674 |
+
"step": 756
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 27.14,
|
| 678 |
+
"learning_rate": 2.5345982142857143e-05,
|
| 679 |
+
"loss": 0.2814,
|
| 680 |
+
"step": 760
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 27.5,
|
| 684 |
+
"learning_rate": 2.568080357142857e-05,
|
| 685 |
+
"loss": 0.2872,
|
| 686 |
+
"step": 770
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 27.86,
|
| 690 |
+
"learning_rate": 2.6015624999999997e-05,
|
| 691 |
+
"loss": 0.2935,
|
| 692 |
+
"step": 780
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 28.0,
|
| 696 |
+
"eval_loss": 0.3069266080856323,
|
| 697 |
+
"eval_runtime": 2.6568,
|
| 698 |
+
"eval_samples_per_second": 116.681,
|
| 699 |
+
"eval_steps_per_second": 1.882,
|
| 700 |
+
"step": 784
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 28.21,
|
| 704 |
+
"learning_rate": 2.6350446428571424e-05,
|
| 705 |
+
"loss": 0.2968,
|
| 706 |
+
"step": 790
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 28.57,
|
| 710 |
+
"learning_rate": 2.6685267857142855e-05,
|
| 711 |
+
"loss": 0.284,
|
| 712 |
+
"step": 800
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 28.93,
|
| 716 |
+
"learning_rate": 2.702008928571428e-05,
|
| 717 |
+
"loss": 0.2814,
|
| 718 |
+
"step": 810
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 29.0,
|
| 722 |
+
"eval_loss": 0.30426502227783203,
|
| 723 |
+
"eval_runtime": 2.6601,
|
| 724 |
+
"eval_samples_per_second": 116.537,
|
| 725 |
+
"eval_steps_per_second": 1.88,
|
| 726 |
+
"step": 812
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 29.29,
|
| 730 |
+
"learning_rate": 2.7354910714285712e-05,
|
| 731 |
+
"loss": 0.2963,
|
| 732 |
+
"step": 820
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 29.64,
|
| 736 |
+
"learning_rate": 2.7689732142857142e-05,
|
| 737 |
+
"loss": 0.2891,
|
| 738 |
+
"step": 830
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 30.0,
|
| 742 |
+
"learning_rate": 2.802455357142857e-05,
|
| 743 |
+
"loss": 0.2753,
|
| 744 |
+
"step": 840
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 30.0,
|
| 748 |
+
"eval_loss": 0.2983907461166382,
|
| 749 |
+
"eval_runtime": 2.7173,
|
| 750 |
+
"eval_samples_per_second": 114.083,
|
| 751 |
+
"eval_steps_per_second": 1.84,
|
| 752 |
+
"step": 840
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 30.36,
|
| 756 |
+
"learning_rate": 2.8359374999999996e-05,
|
| 757 |
+
"loss": 0.2894,
|
| 758 |
+
"step": 850
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 30.71,
|
| 762 |
+
"learning_rate": 2.8694196428571423e-05,
|
| 763 |
+
"loss": 0.2823,
|
| 764 |
+
"step": 860
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 31.0,
|
| 768 |
+
"eval_loss": 0.2994972765445709,
|
| 769 |
+
"eval_runtime": 2.6498,
|
| 770 |
+
"eval_samples_per_second": 116.99,
|
| 771 |
+
"eval_steps_per_second": 1.887,
|
| 772 |
+
"step": 868
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 31.07,
|
| 776 |
+
"learning_rate": 2.9029017857142854e-05,
|
| 777 |
+
"loss": 0.2792,
|
| 778 |
+
"step": 870
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 31.43,
|
| 782 |
+
"learning_rate": 2.9363839285714284e-05,
|
| 783 |
+
"loss": 0.2755,
|
| 784 |
+
"step": 880
|
| 785 |
+
},
|
| 786 |
+
{
|
| 787 |
+
"epoch": 31.79,
|
| 788 |
+
"learning_rate": 2.969866071428571e-05,
|
| 789 |
+
"loss": 0.2962,
|
| 790 |
+
"step": 890
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 32.0,
|
| 794 |
+
"eval_loss": 0.30121392011642456,
|
| 795 |
+
"eval_runtime": 2.6677,
|
| 796 |
+
"eval_samples_per_second": 116.206,
|
| 797 |
+
"eval_steps_per_second": 1.874,
|
| 798 |
+
"step": 896
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 32.14,
|
| 802 |
+
"learning_rate": 3.0033482142857142e-05,
|
| 803 |
+
"loss": 0.2835,
|
| 804 |
+
"step": 900
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 32.5,
|
| 808 |
+
"learning_rate": 3.036830357142857e-05,
|
| 809 |
+
"loss": 0.2724,
|
| 810 |
+
"step": 910
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 32.86,
|
| 814 |
+
"learning_rate": 3.0703124999999996e-05,
|
| 815 |
+
"loss": 0.2869,
|
| 816 |
+
"step": 920
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 33.0,
|
| 820 |
+
"eval_loss": 0.30497127771377563,
|
| 821 |
+
"eval_runtime": 2.6483,
|
| 822 |
+
"eval_samples_per_second": 117.058,
|
| 823 |
+
"eval_steps_per_second": 1.888,
|
| 824 |
+
"step": 924
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 33.21,
|
| 828 |
+
"learning_rate": 3.1037946428571426e-05,
|
| 829 |
+
"loss": 0.2812,
|
| 830 |
+
"step": 930
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 33.57,
|
| 834 |
+
"learning_rate": 3.137276785714286e-05,
|
| 835 |
+
"loss": 0.2855,
|
| 836 |
+
"step": 940
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 33.93,
|
| 840 |
+
"learning_rate": 3.170758928571428e-05,
|
| 841 |
+
"loss": 0.2833,
|
| 842 |
+
"step": 950
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 34.0,
|
| 846 |
+
"eval_loss": 0.29598191380500793,
|
| 847 |
+
"eval_runtime": 2.6636,
|
| 848 |
+
"eval_samples_per_second": 116.384,
|
| 849 |
+
"eval_steps_per_second": 1.877,
|
| 850 |
+
"step": 952
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 34.29,
|
| 854 |
+
"learning_rate": 3.204241071428571e-05,
|
| 855 |
+
"loss": 0.2771,
|
| 856 |
+
"step": 960
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 34.64,
|
| 860 |
+
"learning_rate": 3.237723214285714e-05,
|
| 861 |
+
"loss": 0.2753,
|
| 862 |
+
"step": 970
|
| 863 |
+
},
|
| 864 |
+
{
|
| 865 |
+
"epoch": 35.0,
|
| 866 |
+
"learning_rate": 3.271205357142857e-05,
|
| 867 |
+
"loss": 0.2892,
|
| 868 |
+
"step": 980
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 35.0,
|
| 872 |
+
"eval_loss": 0.3039288818836212,
|
| 873 |
+
"eval_runtime": 2.6706,
|
| 874 |
+
"eval_samples_per_second": 116.078,
|
| 875 |
+
"eval_steps_per_second": 1.872,
|
| 876 |
+
"step": 980
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 35.36,
|
| 880 |
+
"learning_rate": 3.3046874999999995e-05,
|
| 881 |
+
"loss": 0.2915,
|
| 882 |
+
"step": 990
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 35.71,
|
| 886 |
+
"learning_rate": 3.3381696428571426e-05,
|
| 887 |
+
"loss": 0.2764,
|
| 888 |
+
"step": 1000
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 36.0,
|
| 892 |
+
"eval_loss": 0.301042377948761,
|
| 893 |
+
"eval_runtime": 2.6356,
|
| 894 |
+
"eval_samples_per_second": 117.622,
|
| 895 |
+
"eval_steps_per_second": 1.897,
|
| 896 |
+
"step": 1008
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 36.07,
|
| 900 |
+
"learning_rate": 3.3716517857142856e-05,
|
| 901 |
+
"loss": 0.2854,
|
| 902 |
+
"step": 1010
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 36.43,
|
| 906 |
+
"learning_rate": 3.405133928571428e-05,
|
| 907 |
+
"loss": 0.2874,
|
| 908 |
+
"step": 1020
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 36.79,
|
| 912 |
+
"learning_rate": 3.438616071428571e-05,
|
| 913 |
+
"loss": 0.2807,
|
| 914 |
+
"step": 1030
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 37.0,
|
| 918 |
+
"eval_loss": 0.2998359799385071,
|
| 919 |
+
"eval_runtime": 2.6115,
|
| 920 |
+
"eval_samples_per_second": 118.706,
|
| 921 |
+
"eval_steps_per_second": 1.915,
|
| 922 |
+
"step": 1036
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"epoch": 37.14,
|
| 926 |
+
"learning_rate": 3.472098214285714e-05,
|
| 927 |
+
"loss": 0.2629,
|
| 928 |
+
"step": 1040
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 37.5,
|
| 932 |
+
"learning_rate": 3.505580357142857e-05,
|
| 933 |
+
"loss": 0.2759,
|
| 934 |
+
"step": 1050
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 37.86,
|
| 938 |
+
"learning_rate": 3.5390624999999995e-05,
|
| 939 |
+
"loss": 0.2843,
|
| 940 |
+
"step": 1060
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 38.0,
|
| 944 |
+
"eval_loss": 0.2989421784877777,
|
| 945 |
+
"eval_runtime": 2.6315,
|
| 946 |
+
"eval_samples_per_second": 117.805,
|
| 947 |
+
"eval_steps_per_second": 1.9,
|
| 948 |
+
"step": 1064
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 38.21,
|
| 952 |
+
"learning_rate": 3.5725446428571425e-05,
|
| 953 |
+
"loss": 0.284,
|
| 954 |
+
"step": 1070
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 38.57,
|
| 958 |
+
"learning_rate": 3.6060267857142855e-05,
|
| 959 |
+
"loss": 0.2786,
|
| 960 |
+
"step": 1080
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 38.93,
|
| 964 |
+
"learning_rate": 3.639508928571428e-05,
|
| 965 |
+
"loss": 0.2808,
|
| 966 |
+
"step": 1090
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 39.0,
|
| 970 |
+
"eval_loss": 0.2969764173030853,
|
| 971 |
+
"eval_runtime": 2.6345,
|
| 972 |
+
"eval_samples_per_second": 117.671,
|
| 973 |
+
"eval_steps_per_second": 1.898,
|
| 974 |
+
"step": 1092
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 39.29,
|
| 978 |
+
"learning_rate": 3.672991071428571e-05,
|
| 979 |
+
"loss": 0.2748,
|
| 980 |
+
"step": 1100
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 39.64,
|
| 984 |
+
"learning_rate": 3.706473214285714e-05,
|
| 985 |
+
"loss": 0.2751,
|
| 986 |
+
"step": 1110
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 40.0,
|
| 990 |
+
"learning_rate": 3.739955357142857e-05,
|
| 991 |
+
"loss": 0.2862,
|
| 992 |
+
"step": 1120
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 40.0,
|
| 996 |
+
"eval_loss": 0.29402607679367065,
|
| 997 |
+
"eval_runtime": 2.6346,
|
| 998 |
+
"eval_samples_per_second": 117.666,
|
| 999 |
+
"eval_steps_per_second": 1.898,
|
| 1000 |
+
"step": 1120
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 40.36,
|
| 1004 |
+
"learning_rate": 3.7499989987931416e-05,
|
| 1005 |
+
"loss": 0.2951,
|
| 1006 |
+
"step": 1130
|
| 1007 |
+
},
|
| 1008 |
+
{
|
| 1009 |
+
"epoch": 40.71,
|
| 1010 |
+
"learning_rate": 3.7499940949253886e-05,
|
| 1011 |
+
"loss": 0.2601,
|
| 1012 |
+
"step": 1140
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 41.0,
|
| 1016 |
+
"eval_loss": 0.2952007055282593,
|
| 1017 |
+
"eval_runtime": 2.6361,
|
| 1018 |
+
"eval_samples_per_second": 117.597,
|
| 1019 |
+
"eval_steps_per_second": 1.897,
|
| 1020 |
+
"step": 1148
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 41.07,
|
| 1024 |
+
"learning_rate": 3.749985104512278e-05,
|
| 1025 |
+
"loss": 0.2796,
|
| 1026 |
+
"step": 1150
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 41.43,
|
| 1030 |
+
"learning_rate": 3.7499720275734064e-05,
|
| 1031 |
+
"loss": 0.2759,
|
| 1032 |
+
"step": 1160
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 41.79,
|
| 1036 |
+
"learning_rate": 3.7499548641372725e-05,
|
| 1037 |
+
"loss": 0.2742,
|
| 1038 |
+
"step": 1170
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 42.0,
|
| 1042 |
+
"eval_loss": 0.2939775884151459,
|
| 1043 |
+
"eval_runtime": 2.6378,
|
| 1044 |
+
"eval_samples_per_second": 117.522,
|
| 1045 |
+
"eval_steps_per_second": 1.896,
|
| 1046 |
+
"step": 1176
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 42.14,
|
| 1050 |
+
"learning_rate": 3.7499336142412856e-05,
|
| 1051 |
+
"loss": 0.2757,
|
| 1052 |
+
"step": 1180
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 42.5,
|
| 1056 |
+
"learning_rate": 3.749908277931758e-05,
|
| 1057 |
+
"loss": 0.2801,
|
| 1058 |
+
"step": 1190
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 42.86,
|
| 1062 |
+
"learning_rate": 3.749878855263912e-05,
|
| 1063 |
+
"loss": 0.2791,
|
| 1064 |
+
"step": 1200
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 43.0,
|
| 1068 |
+
"eval_loss": 0.29970163106918335,
|
| 1069 |
+
"eval_runtime": 2.6144,
|
| 1070 |
+
"eval_samples_per_second": 118.572,
|
| 1071 |
+
"eval_steps_per_second": 1.912,
|
| 1072 |
+
"step": 1204
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 43.21,
|
| 1076 |
+
"learning_rate": 3.7498453463018734e-05,
|
| 1077 |
+
"loss": 0.2744,
|
| 1078 |
+
"step": 1210
|
| 1079 |
+
},
|
| 1080 |
+
{
|
| 1081 |
+
"epoch": 43.57,
|
| 1082 |
+
"learning_rate": 3.749807751118675e-05,
|
| 1083 |
+
"loss": 0.2793,
|
| 1084 |
+
"step": 1220
|
| 1085 |
+
},
|
| 1086 |
+
{
|
| 1087 |
+
"epoch": 43.93,
|
| 1088 |
+
"learning_rate": 3.749766069796256e-05,
|
| 1089 |
+
"loss": 0.2759,
|
| 1090 |
+
"step": 1230
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 44.0,
|
| 1094 |
+
"eval_loss": 0.2951151728630066,
|
| 1095 |
+
"eval_runtime": 2.6705,
|
| 1096 |
+
"eval_samples_per_second": 116.083,
|
| 1097 |
+
"eval_steps_per_second": 1.872,
|
| 1098 |
+
"step": 1232
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 44.29,
|
| 1102 |
+
"learning_rate": 3.749720302425459e-05,
|
| 1103 |
+
"loss": 0.2802,
|
| 1104 |
+
"step": 1240
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 44.64,
|
| 1108 |
+
"learning_rate": 3.749670449106036e-05,
|
| 1109 |
+
"loss": 0.268,
|
| 1110 |
+
"step": 1250
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 45.0,
|
| 1114 |
+
"learning_rate": 3.749616509946641e-05,
|
| 1115 |
+
"loss": 0.2819,
|
| 1116 |
+
"step": 1260
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 45.0,
|
| 1120 |
+
"eval_loss": 0.2895747125148773,
|
| 1121 |
+
"eval_runtime": 2.6558,
|
| 1122 |
+
"eval_samples_per_second": 116.726,
|
| 1123 |
+
"eval_steps_per_second": 1.883,
|
| 1124 |
+
"step": 1260
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 45.36,
|
| 1128 |
+
"learning_rate": 3.7495584850648345e-05,
|
| 1129 |
+
"loss": 0.2697,
|
| 1130 |
+
"step": 1270
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 45.71,
|
| 1134 |
+
"learning_rate": 3.749496374587082e-05,
|
| 1135 |
+
"loss": 0.287,
|
| 1136 |
+
"step": 1280
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 46.0,
|
| 1140 |
+
"eval_loss": 0.2938202917575836,
|
| 1141 |
+
"eval_runtime": 2.6182,
|
| 1142 |
+
"eval_samples_per_second": 118.403,
|
| 1143 |
+
"eval_steps_per_second": 1.91,
|
| 1144 |
+
"step": 1288
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 46.07,
|
| 1148 |
+
"learning_rate": 3.749430178648752e-05,
|
| 1149 |
+
"loss": 0.2736,
|
| 1150 |
+
"step": 1290
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 46.43,
|
| 1154 |
+
"learning_rate": 3.7493598973941194e-05,
|
| 1155 |
+
"loss": 0.2817,
|
| 1156 |
+
"step": 1300
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 46.79,
|
| 1160 |
+
"learning_rate": 3.749285530976362e-05,
|
| 1161 |
+
"loss": 0.2711,
|
| 1162 |
+
"step": 1310
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 47.0,
|
| 1166 |
+
"eval_loss": 0.2973059117794037,
|
| 1167 |
+
"eval_runtime": 2.6436,
|
| 1168 |
+
"eval_samples_per_second": 117.264,
|
| 1169 |
+
"eval_steps_per_second": 1.891,
|
| 1170 |
+
"step": 1316
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 47.14,
|
| 1174 |
+
"learning_rate": 3.749207079557561e-05,
|
| 1175 |
+
"loss": 0.2761,
|
| 1176 |
+
"step": 1320
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 47.5,
|
| 1180 |
+
"learning_rate": 3.749124543308701e-05,
|
| 1181 |
+
"loss": 0.2732,
|
| 1182 |
+
"step": 1330
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"epoch": 47.86,
|
| 1186 |
+
"learning_rate": 3.74903792240967e-05,
|
| 1187 |
+
"loss": 0.2782,
|
| 1188 |
+
"step": 1340
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"epoch": 48.0,
|
| 1192 |
+
"eval_loss": 0.2946121096611023,
|
| 1193 |
+
"eval_runtime": 2.6239,
|
| 1194 |
+
"eval_samples_per_second": 118.144,
|
| 1195 |
+
"eval_steps_per_second": 1.906,
|
| 1196 |
+
"step": 1344
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 48.21,
|
| 1200 |
+
"learning_rate": 3.748947217049258e-05,
|
| 1201 |
+
"loss": 0.2682,
|
| 1202 |
+
"step": 1350
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 48.57,
|
| 1206 |
+
"learning_rate": 3.7488524274251566e-05,
|
| 1207 |
+
"loss": 0.2765,
|
| 1208 |
+
"step": 1360
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 48.93,
|
| 1212 |
+
"learning_rate": 3.748753553743959e-05,
|
| 1213 |
+
"loss": 0.2674,
|
| 1214 |
+
"step": 1370
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 49.0,
|
| 1218 |
+
"eval_loss": 0.29130813479423523,
|
| 1219 |
+
"eval_runtime": 2.6539,
|
| 1220 |
+
"eval_samples_per_second": 116.81,
|
| 1221 |
+
"eval_steps_per_second": 1.884,
|
| 1222 |
+
"step": 1372
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 49.29,
|
| 1226 |
+
"learning_rate": 3.748650596221162e-05,
|
| 1227 |
+
"loss": 0.2833,
|
| 1228 |
+
"step": 1380
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 49.64,
|
| 1232 |
+
"learning_rate": 3.74854355508116e-05,
|
| 1233 |
+
"loss": 0.2701,
|
| 1234 |
+
"step": 1390
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 50.0,
|
| 1238 |
+
"learning_rate": 3.748432430557249e-05,
|
| 1239 |
+
"loss": 0.268,
|
| 1240 |
+
"step": 1400
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"epoch": 50.0,
|
| 1244 |
+
"eval_loss": 0.29444822669029236,
|
| 1245 |
+
"eval_runtime": 2.6587,
|
| 1246 |
+
"eval_samples_per_second": 116.597,
|
| 1247 |
+
"eval_steps_per_second": 1.881,
|
| 1248 |
+
"step": 1400
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 50.36,
|
| 1252 |
+
"learning_rate": 3.748317222891625e-05,
|
| 1253 |
+
"loss": 0.2743,
|
| 1254 |
+
"step": 1410
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 50.71,
|
| 1258 |
+
"learning_rate": 3.748197932335383e-05,
|
| 1259 |
+
"loss": 0.2624,
|
| 1260 |
+
"step": 1420
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"epoch": 51.0,
|
| 1264 |
+
"eval_loss": 0.29401645064353943,
|
| 1265 |
+
"eval_runtime": 2.6105,
|
| 1266 |
+
"eval_samples_per_second": 118.751,
|
| 1267 |
+
"eval_steps_per_second": 1.915,
|
| 1268 |
+
"step": 1428
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 51.07,
|
| 1272 |
+
"learning_rate": 3.748074559148516e-05,
|
| 1273 |
+
"loss": 0.2865,
|
| 1274 |
+
"step": 1430
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 51.43,
|
| 1278 |
+
"learning_rate": 3.747947103599916e-05,
|
| 1279 |
+
"loss": 0.2717,
|
| 1280 |
+
"step": 1440
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 51.79,
|
| 1284 |
+
"learning_rate": 3.747815565967371e-05,
|
| 1285 |
+
"loss": 0.2842,
|
| 1286 |
+
"step": 1450
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 52.0,
|
| 1290 |
+
"eval_loss": 0.29776862263679504,
|
| 1291 |
+
"eval_runtime": 2.6164,
|
| 1292 |
+
"eval_samples_per_second": 118.483,
|
| 1293 |
+
"eval_steps_per_second": 1.911,
|
| 1294 |
+
"step": 1456
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"epoch": 52.14,
|
| 1298 |
+
"learning_rate": 3.747679946537569e-05,
|
| 1299 |
+
"loss": 0.2716,
|
| 1300 |
+
"step": 1460
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 52.5,
|
| 1304 |
+
"learning_rate": 3.7475402456060905e-05,
|
| 1305 |
+
"loss": 0.2776,
|
| 1306 |
+
"step": 1470
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 52.86,
|
| 1310 |
+
"learning_rate": 3.7473964634774137e-05,
|
| 1311 |
+
"loss": 0.2753,
|
| 1312 |
+
"step": 1480
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 53.0,
|
| 1316 |
+
"eval_loss": 0.29514825344085693,
|
| 1317 |
+
"eval_runtime": 2.6208,
|
| 1318 |
+
"eval_samples_per_second": 118.283,
|
| 1319 |
+
"eval_steps_per_second": 1.908,
|
| 1320 |
+
"step": 1484
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 53.21,
|
| 1324 |
+
"learning_rate": 3.747248600464912e-05,
|
| 1325 |
+
"loss": 0.268,
|
| 1326 |
+
"step": 1490
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 53.57,
|
| 1330 |
+
"learning_rate": 3.747096656890852e-05,
|
| 1331 |
+
"loss": 0.2734,
|
| 1332 |
+
"step": 1500
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 53.93,
|
| 1336 |
+
"learning_rate": 3.746940633086394e-05,
|
| 1337 |
+
"loss": 0.2733,
|
| 1338 |
+
"step": 1510
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 54.0,
|
| 1342 |
+
"eval_loss": 0.28796395659446716,
|
| 1343 |
+
"eval_runtime": 2.6647,
|
| 1344 |
+
"eval_samples_per_second": 116.337,
|
| 1345 |
+
"eval_steps_per_second": 1.876,
|
| 1346 |
+
"step": 1512
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 54.29,
|
| 1350 |
+
"learning_rate": 3.746780529391593e-05,
|
| 1351 |
+
"loss": 0.2638,
|
| 1352 |
+
"step": 1520
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 54.64,
|
| 1356 |
+
"learning_rate": 3.7466163461553934e-05,
|
| 1357 |
+
"loss": 0.274,
|
| 1358 |
+
"step": 1530
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 55.0,
|
| 1362 |
+
"learning_rate": 3.746448083735632e-05,
|
| 1363 |
+
"loss": 0.2782,
|
| 1364 |
+
"step": 1540
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 55.0,
|
| 1368 |
+
"eval_loss": 0.29691949486732483,
|
| 1369 |
+
"eval_runtime": 2.6296,
|
| 1370 |
+
"eval_samples_per_second": 117.889,
|
| 1371 |
+
"eval_steps_per_second": 1.901,
|
| 1372 |
+
"step": 1540
|
| 1373 |
+
},
|
| 1374 |
+
{
|
| 1375 |
+
"epoch": 55.36,
|
| 1376 |
+
"learning_rate": 3.746275742499037e-05,
|
| 1377 |
+
"loss": 0.2749,
|
| 1378 |
+
"step": 1550
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 55.71,
|
| 1382 |
+
"learning_rate": 3.746099322821226e-05,
|
| 1383 |
+
"loss": 0.2789,
|
| 1384 |
+
"step": 1560
|
| 1385 |
+
},
|
| 1386 |
+
{
|
| 1387 |
+
"epoch": 56.0,
|
| 1388 |
+
"eval_loss": 0.29188984632492065,
|
| 1389 |
+
"eval_runtime": 2.6147,
|
| 1390 |
+
"eval_samples_per_second": 118.559,
|
| 1391 |
+
"eval_steps_per_second": 1.912,
|
| 1392 |
+
"step": 1568
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"epoch": 56.07,
|
| 1396 |
+
"learning_rate": 3.7459188250867045e-05,
|
| 1397 |
+
"loss": 0.2537,
|
| 1398 |
+
"step": 1570
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"epoch": 56.43,
|
| 1402 |
+
"learning_rate": 3.7457342496888676e-05,
|
| 1403 |
+
"loss": 0.2718,
|
| 1404 |
+
"step": 1580
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 56.79,
|
| 1408 |
+
"learning_rate": 3.745545597029996e-05,
|
| 1409 |
+
"loss": 0.2815,
|
| 1410 |
+
"step": 1590
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 57.0,
|
| 1414 |
+
"eval_loss": 0.291569322347641,
|
| 1415 |
+
"eval_runtime": 2.6148,
|
| 1416 |
+
"eval_samples_per_second": 118.555,
|
| 1417 |
+
"eval_steps_per_second": 1.912,
|
| 1418 |
+
"step": 1596
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 57.14,
|
| 1422 |
+
"learning_rate": 3.7453528675212585e-05,
|
| 1423 |
+
"loss": 0.2663,
|
| 1424 |
+
"step": 1600
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 57.5,
|
| 1428 |
+
"learning_rate": 3.745156061582709e-05,
|
| 1429 |
+
"loss": 0.2785,
|
| 1430 |
+
"step": 1610
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 57.86,
|
| 1434 |
+
"learning_rate": 3.744955179643285e-05,
|
| 1435 |
+
"loss": 0.2629,
|
| 1436 |
+
"step": 1620
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 58.0,
|
| 1440 |
+
"eval_loss": 0.29474544525146484,
|
| 1441 |
+
"eval_runtime": 2.6396,
|
| 1442 |
+
"eval_samples_per_second": 117.441,
|
| 1443 |
+
"eval_steps_per_second": 1.894,
|
| 1444 |
+
"step": 1624
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 58.21,
|
| 1448 |
+
"learning_rate": 3.744750222140808e-05,
|
| 1449 |
+
"loss": 0.2682,
|
| 1450 |
+
"step": 1630
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 58.57,
|
| 1454 |
+
"learning_rate": 3.744541189521983e-05,
|
| 1455 |
+
"loss": 0.2666,
|
| 1456 |
+
"step": 1640
|
| 1457 |
+
},
|
| 1458 |
+
{
|
| 1459 |
+
"epoch": 58.93,
|
| 1460 |
+
"learning_rate": 3.7443280822423956e-05,
|
| 1461 |
+
"loss": 0.2716,
|
| 1462 |
+
"step": 1650
|
| 1463 |
+
},
|
| 1464 |
+
{
|
| 1465 |
+
"epoch": 59.0,
|
| 1466 |
+
"eval_loss": 0.28275418281555176,
|
| 1467 |
+
"eval_runtime": 2.6524,
|
| 1468 |
+
"eval_samples_per_second": 116.876,
|
| 1469 |
+
"eval_steps_per_second": 1.885,
|
| 1470 |
+
"step": 1652
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"epoch": 59.29,
|
| 1474 |
+
"learning_rate": 3.744110900766512e-05,
|
| 1475 |
+
"loss": 0.2775,
|
| 1476 |
+
"step": 1660
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 59.64,
|
| 1480 |
+
"learning_rate": 3.7438896455676804e-05,
|
| 1481 |
+
"loss": 0.2767,
|
| 1482 |
+
"step": 1670
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 60.0,
|
| 1486 |
+
"learning_rate": 3.7436643171281254e-05,
|
| 1487 |
+
"loss": 0.2623,
|
| 1488 |
+
"step": 1680
|
| 1489 |
+
},
|
| 1490 |
+
{
|
| 1491 |
+
"epoch": 60.0,
|
| 1492 |
+
"eval_loss": 0.29241943359375,
|
| 1493 |
+
"eval_runtime": 2.6403,
|
| 1494 |
+
"eval_samples_per_second": 117.412,
|
| 1495 |
+
"eval_steps_per_second": 1.894,
|
| 1496 |
+
"step": 1680
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 60.36,
|
| 1500 |
+
"learning_rate": 1.3997036682248361e-05,
|
| 1501 |
+
"loss": 0.2725,
|
| 1502 |
+
"step": 1690
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 60.71,
|
| 1506 |
+
"learning_rate": 1.3783159273580512e-05,
|
| 1507 |
+
"loss": 0.2773,
|
| 1508 |
+
"step": 1700
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 61.0,
|
| 1512 |
+
"eval_loss": 0.27645179629325867,
|
| 1513 |
+
"eval_runtime": 2.6338,
|
| 1514 |
+
"eval_samples_per_second": 117.701,
|
| 1515 |
+
"eval_steps_per_second": 1.898,
|
| 1516 |
+
"step": 1708
|
| 1517 |
+
},
|
| 1518 |
+
{
|
| 1519 |
+
"epoch": 61.07,
|
| 1520 |
+
"learning_rate": 1.3569974670940111e-05,
|
| 1521 |
+
"loss": 0.2674,
|
| 1522 |
+
"step": 1710
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 61.43,
|
| 1526 |
+
"learning_rate": 1.3357512610649675e-05,
|
| 1527 |
+
"loss": 0.2761,
|
| 1528 |
+
"step": 1720
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 61.79,
|
| 1532 |
+
"learning_rate": 1.3145802728246982e-05,
|
| 1533 |
+
"loss": 0.268,
|
| 1534 |
+
"step": 1730
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 62.0,
|
| 1538 |
+
"eval_loss": 0.2753865420818329,
|
| 1539 |
+
"eval_runtime": 3.0543,
|
| 1540 |
+
"eval_samples_per_second": 101.496,
|
| 1541 |
+
"eval_steps_per_second": 1.637,
|
| 1542 |
+
"step": 1736
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 62.14,
|
| 1546 |
+
"learning_rate": 1.2934874554351332e-05,
|
| 1547 |
+
"loss": 0.2724,
|
| 1548 |
+
"step": 1740
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 62.5,
|
| 1552 |
+
"learning_rate": 1.2724757510544413e-05,
|
| 1553 |
+
"loss": 0.2694,
|
| 1554 |
+
"step": 1750
|
| 1555 |
+
},
|
| 1556 |
+
{
|
| 1557 |
+
"epoch": 62.86,
|
| 1558 |
+
"learning_rate": 1.2515480905266435e-05,
|
| 1559 |
+
"loss": 0.2839,
|
| 1560 |
+
"step": 1760
|
| 1561 |
+
},
|
| 1562 |
+
{
|
| 1563 |
+
"epoch": 63.0,
|
| 1564 |
+
"eval_loss": 0.2744416296482086,
|
| 1565 |
+
"eval_runtime": 2.9673,
|
| 1566 |
+
"eval_samples_per_second": 104.471,
|
| 1567 |
+
"eval_steps_per_second": 1.685,
|
| 1568 |
+
"step": 1764
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 63.21,
|
| 1572 |
+
"learning_rate": 1.2307073929727975e-05,
|
| 1573 |
+
"loss": 0.2731,
|
| 1574 |
+
"step": 1770
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 63.57,
|
| 1578 |
+
"learning_rate": 1.2099565653838216e-05,
|
| 1579 |
+
"loss": 0.277,
|
| 1580 |
+
"step": 1780
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 63.93,
|
| 1584 |
+
"learning_rate": 1.1892985022150073e-05,
|
| 1585 |
+
"loss": 0.2684,
|
| 1586 |
+
"step": 1790
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 64.0,
|
| 1590 |
+
"eval_loss": 0.2744133472442627,
|
| 1591 |
+
"eval_runtime": 3.1325,
|
| 1592 |
+
"eval_samples_per_second": 98.963,
|
| 1593 |
+
"eval_steps_per_second": 1.596,
|
| 1594 |
+
"step": 1792
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 64.29,
|
| 1598 |
+
"learning_rate": 1.1687360849822864e-05,
|
| 1599 |
+
"loss": 0.2694,
|
| 1600 |
+
"step": 1800
|
| 1601 |
+
},
|
| 1602 |
+
{
|
| 1603 |
+
"epoch": 64.64,
|
| 1604 |
+
"learning_rate": 1.1482721818602949e-05,
|
| 1605 |
+
"loss": 0.2639,
|
| 1606 |
+
"step": 1810
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 65.0,
|
| 1610 |
+
"learning_rate": 1.1279096472823067e-05,
|
| 1611 |
+
"loss": 0.2865,
|
| 1612 |
+
"step": 1820
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 65.0,
|
| 1616 |
+
"eval_loss": 0.2716241776943207,
|
| 1617 |
+
"eval_runtime": 2.9389,
|
| 1618 |
+
"eval_samples_per_second": 105.482,
|
| 1619 |
+
"eval_steps_per_second": 1.701,
|
| 1620 |
+
"step": 1820
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 65.36,
|
| 1624 |
+
"learning_rate": 1.1076513215420738e-05,
|
| 1625 |
+
"loss": 0.2651,
|
| 1626 |
+
"step": 1830
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 65.71,
|
| 1630 |
+
"learning_rate": 1.0875000303976493e-05,
|
| 1631 |
+
"loss": 0.2845,
|
| 1632 |
+
"step": 1840
|
| 1633 |
+
},
|
| 1634 |
+
{
|
| 1635 |
+
"epoch": 66.0,
|
| 1636 |
+
"eval_loss": 0.2768835127353668,
|
| 1637 |
+
"eval_runtime": 2.9046,
|
| 1638 |
+
"eval_samples_per_second": 106.726,
|
| 1639 |
+
"eval_steps_per_second": 1.721,
|
| 1640 |
+
"step": 1848
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 66.07,
|
| 1644 |
+
"learning_rate": 1.0674585846772314e-05,
|
| 1645 |
+
"loss": 0.2768,
|
| 1646 |
+
"step": 1850
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 66.43,
|
| 1650 |
+
"learning_rate": 1.04752977988709e-05,
|
| 1651 |
+
"loss": 0.2785,
|
| 1652 |
+
"step": 1860
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 66.79,
|
| 1656 |
+
"learning_rate": 1.0277163958216333e-05,
|
| 1657 |
+
"loss": 0.2663,
|
| 1658 |
+
"step": 1870
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 67.0,
|
| 1662 |
+
"eval_loss": 0.2754151523113251,
|
| 1663 |
+
"eval_runtime": 3.1034,
|
| 1664 |
+
"eval_samples_per_second": 99.891,
|
| 1665 |
+
"eval_steps_per_second": 1.611,
|
| 1666 |
+
"step": 1876
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 67.14,
|
| 1670 |
+
"learning_rate": 1.0080211961756649e-05,
|
| 1671 |
+
"loss": 0.2632,
|
| 1672 |
+
"step": 1880
|
| 1673 |
+
},
|
| 1674 |
+
{
|
| 1675 |
+
"epoch": 67.5,
|
| 1676 |
+
"learning_rate": 9.884469281588848e-06,
|
| 1677 |
+
"loss": 0.2832,
|
| 1678 |
+
"step": 1890
|
| 1679 |
+
},
|
| 1680 |
+
{
|
| 1681 |
+
"epoch": 67.86,
|
| 1682 |
+
"learning_rate": 9.68996322112695e-06,
|
| 1683 |
+
"loss": 0.269,
|
| 1684 |
+
"step": 1900
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"epoch": 68.0,
|
| 1688 |
+
"eval_loss": 0.2737092077732086,
|
| 1689 |
+
"eval_runtime": 3.0377,
|
| 1690 |
+
"eval_samples_per_second": 102.051,
|
| 1691 |
+
"eval_steps_per_second": 1.646,
|
| 1692 |
+
"step": 1904
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 68.21,
|
| 1696 |
+
"learning_rate": 9.496720911293496e-06,
|
| 1697 |
+
"loss": 0.2753,
|
| 1698 |
+
"step": 1910
|
| 1699 |
+
},
|
| 1700 |
+
{
|
| 1701 |
+
"epoch": 68.57,
|
| 1702 |
+
"learning_rate": 9.304769306735192e-06,
|
| 1703 |
+
"loss": 0.2793,
|
| 1704 |
+
"step": 1920
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 68.93,
|
| 1708 |
+
"learning_rate": 9.114135182063088e-06,
|
| 1709 |
+
"loss": 0.2681,
|
| 1710 |
+
"step": 1930
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 69.0,
|
| 1714 |
+
"eval_loss": 0.269718736410141,
|
| 1715 |
+
"eval_runtime": 3.08,
|
| 1716 |
+
"eval_samples_per_second": 100.648,
|
| 1717 |
+
"eval_steps_per_second": 1.623,
|
| 1718 |
+
"step": 1932
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 69.29,
|
| 1722 |
+
"learning_rate": 8.924845128117907e-06,
|
| 1723 |
+
"loss": 0.2763,
|
| 1724 |
+
"step": 1940
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 69.64,
|
| 1728 |
+
"learning_rate": 8.736925548260963e-06,
|
| 1729 |
+
"loss": 0.2679,
|
| 1730 |
+
"step": 1950
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 70.0,
|
| 1734 |
+
"learning_rate": 8.550402654691271e-06,
|
| 1735 |
+
"loss": 0.2748,
|
| 1736 |
+
"step": 1960
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 70.0,
|
| 1740 |
+
"eval_loss": 0.2778591513633728,
|
| 1741 |
+
"eval_runtime": 2.6486,
|
| 1742 |
+
"eval_samples_per_second": 117.044,
|
| 1743 |
+
"eval_steps_per_second": 1.888,
|
| 1744 |
+
"step": 1960
|
| 1745 |
+
},
|
| 1746 |
+
{
|
| 1747 |
+
"epoch": 70.36,
|
| 1748 |
+
"learning_rate": 8.365302464789306e-06,
|
| 1749 |
+
"loss": 0.2758,
|
| 1750 |
+
"step": 1970
|
| 1751 |
+
},
|
| 1752 |
+
{
|
| 1753 |
+
"epoch": 70.71,
|
| 1754 |
+
"learning_rate": 8.181650797487918e-06,
|
| 1755 |
+
"loss": 0.2769,
|
| 1756 |
+
"step": 1980
|
| 1757 |
+
},
|
| 1758 |
+
{
|
| 1759 |
+
"epoch": 71.0,
|
| 1760 |
+
"eval_loss": 0.272790789604187,
|
| 1761 |
+
"eval_runtime": 2.6447,
|
| 1762 |
+
"eval_samples_per_second": 117.214,
|
| 1763 |
+
"eval_steps_per_second": 1.891,
|
| 1764 |
+
"step": 1988
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 71.07,
|
| 1768 |
+
"learning_rate": 7.999473269670997e-06,
|
| 1769 |
+
"loss": 0.2662,
|
| 1770 |
+
"step": 1990
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 71.43,
|
| 1774 |
+
"learning_rate": 7.818795292600176e-06,
|
| 1775 |
+
"loss": 0.2549,
|
| 1776 |
+
"step": 2000
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 71.79,
|
| 1780 |
+
"learning_rate": 7.639642068370416e-06,
|
| 1781 |
+
"loss": 0.2805,
|
| 1782 |
+
"step": 2010
|
| 1783 |
+
},
|
| 1784 |
+
{
|
| 1785 |
+
"epoch": 72.0,
|
| 1786 |
+
"eval_loss": 0.27293434739112854,
|
| 1787 |
+
"eval_runtime": 2.6471,
|
| 1788 |
+
"eval_samples_per_second": 117.111,
|
| 1789 |
+
"eval_steps_per_second": 1.889,
|
| 1790 |
+
"step": 2016
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 72.14,
|
| 1794 |
+
"learning_rate": 7.462038586394576e-06,
|
| 1795 |
+
"loss": 0.2852,
|
| 1796 |
+
"step": 2020
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 72.5,
|
| 1800 |
+
"learning_rate": 7.286009619917803e-06,
|
| 1801 |
+
"loss": 0.2597,
|
| 1802 |
+
"step": 2030
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 72.86,
|
| 1806 |
+
"learning_rate": 7.1115797225619155e-06,
|
| 1807 |
+
"loss": 0.2771,
|
| 1808 |
+
"step": 2040
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 73.0,
|
| 1812 |
+
"eval_loss": 0.272775262594223,
|
| 1813 |
+
"eval_runtime": 2.6711,
|
| 1814 |
+
"eval_samples_per_second": 116.058,
|
| 1815 |
+
"eval_steps_per_second": 1.872,
|
| 1816 |
+
"step": 2044
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 73.21,
|
| 1820 |
+
"learning_rate": 6.938773224900619e-06,
|
| 1821 |
+
"loss": 0.2752,
|
| 1822 |
+
"step": 2050
|
| 1823 |
+
},
|
| 1824 |
+
{
|
| 1825 |
+
"epoch": 73.57,
|
| 1826 |
+
"learning_rate": 6.7676142310656305e-06,
|
| 1827 |
+
"loss": 0.2646,
|
| 1828 |
+
"step": 2060
|
| 1829 |
+
},
|
| 1830 |
+
{
|
| 1831 |
+
"epoch": 73.93,
|
| 1832 |
+
"learning_rate": 6.598126615384559e-06,
|
| 1833 |
+
"loss": 0.2717,
|
| 1834 |
+
"step": 2070
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 74.0,
|
| 1838 |
+
"eval_loss": 0.27492260932922363,
|
| 1839 |
+
"eval_runtime": 2.6449,
|
| 1840 |
+
"eval_samples_per_second": 117.205,
|
| 1841 |
+
"eval_steps_per_second": 1.89,
|
| 1842 |
+
"step": 2072
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 74.29,
|
| 1846 |
+
"learning_rate": 6.430334019050682e-06,
|
| 1847 |
+
"loss": 0.2793,
|
| 1848 |
+
"step": 2080
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 74.64,
|
| 1852 |
+
"learning_rate": 6.264259846825417e-06,
|
| 1853 |
+
"loss": 0.2744,
|
| 1854 |
+
"step": 2090
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 75.0,
|
| 1858 |
+
"learning_rate": 6.099927263773606e-06,
|
| 1859 |
+
"loss": 0.267,
|
| 1860 |
+
"step": 2100
|
| 1861 |
+
},
|
| 1862 |
+
{
|
| 1863 |
+
"epoch": 75.0,
|
| 1864 |
+
"eval_loss": 0.273212730884552,
|
| 1865 |
+
"eval_runtime": 2.6667,
|
| 1866 |
+
"eval_samples_per_second": 116.249,
|
| 1867 |
+
"eval_steps_per_second": 1.875,
|
| 1868 |
+
"step": 2100
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 75.36,
|
| 1872 |
+
"learning_rate": 5.9373591920323426e-06,
|
| 1873 |
+
"loss": 0.2673,
|
| 1874 |
+
"step": 2110
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 75.71,
|
| 1878 |
+
"learning_rate": 5.776578307613639e-06,
|
| 1879 |
+
"loss": 0.2812,
|
| 1880 |
+
"step": 2120
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 76.0,
|
| 1884 |
+
"eval_loss": 0.27429237961769104,
|
| 1885 |
+
"eval_runtime": 2.66,
|
| 1886 |
+
"eval_samples_per_second": 116.541,
|
| 1887 |
+
"eval_steps_per_second": 1.88,
|
| 1888 |
+
"step": 2128
|
| 1889 |
+
},
|
| 1890 |
+
{
|
| 1891 |
+
"epoch": 76.07,
|
| 1892 |
+
"learning_rate": 5.6176070372414275e-06,
|
| 1893 |
+
"loss": 0.2674,
|
| 1894 |
+
"step": 2130
|
| 1895 |
+
},
|
| 1896 |
+
{
|
| 1897 |
+
"epoch": 76.43,
|
| 1898 |
+
"learning_rate": 5.460467555223369e-06,
|
| 1899 |
+
"loss": 0.2696,
|
| 1900 |
+
"step": 2140
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 76.79,
|
| 1904 |
+
"learning_rate": 5.305181780357802e-06,
|
| 1905 |
+
"loss": 0.2749,
|
| 1906 |
+
"step": 2150
|
| 1907 |
+
},
|
| 1908 |
+
{
|
| 1909 |
+
"epoch": 77.0,
|
| 1910 |
+
"eval_loss": 0.2738768756389618,
|
| 1911 |
+
"eval_runtime": 2.6486,
|
| 1912 |
+
"eval_samples_per_second": 117.043,
|
| 1913 |
+
"eval_steps_per_second": 1.888,
|
| 1914 |
+
"step": 2156
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 77.14,
|
| 1918 |
+
"learning_rate": 5.1517713728764e-06,
|
| 1919 |
+
"loss": 0.2837,
|
| 1920 |
+
"step": 2160
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 77.5,
|
| 1924 |
+
"learning_rate": 5.000257731422843e-06,
|
| 1925 |
+
"loss": 0.2606,
|
| 1926 |
+
"step": 2170
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 77.86,
|
| 1930 |
+
"learning_rate": 4.850661990068047e-06,
|
| 1931 |
+
"loss": 0.2746,
|
| 1932 |
+
"step": 2180
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 78.0,
|
| 1936 |
+
"eval_loss": 0.27298134565353394,
|
| 1937 |
+
"eval_runtime": 2.6537,
|
| 1938 |
+
"eval_samples_per_second": 116.817,
|
| 1939 |
+
"eval_steps_per_second": 1.884,
|
| 1940 |
+
"step": 2184
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 78.21,
|
| 1944 |
+
"learning_rate": 4.703005015362185e-06,
|
| 1945 |
+
"loss": 0.2828,
|
| 1946 |
+
"step": 2190
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 78.57,
|
| 1950 |
+
"learning_rate": 4.557307403424147e-06,
|
| 1951 |
+
"loss": 0.2687,
|
| 1952 |
+
"step": 2200
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 78.93,
|
| 1956 |
+
"learning_rate": 4.41358947706863e-06,
|
| 1957 |
+
"loss": 0.2707,
|
| 1958 |
+
"step": 2210
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 79.0,
|
| 1962 |
+
"eval_loss": 0.274308443069458,
|
| 1963 |
+
"eval_runtime": 2.6512,
|
| 1964 |
+
"eval_samples_per_second": 116.929,
|
| 1965 |
+
"eval_steps_per_second": 1.886,
|
| 1966 |
+
"step": 2212
|
| 1967 |
+
},
|
| 1968 |
+
{
|
| 1969 |
+
"epoch": 79.29,
|
| 1970 |
+
"learning_rate": 4.271871282971398e-06,
|
| 1971 |
+
"loss": 0.2788,
|
| 1972 |
+
"step": 2220
|
| 1973 |
+
},
|
| 1974 |
+
{
|
| 1975 |
+
"epoch": 79.64,
|
| 1976 |
+
"learning_rate": 4.1321725888730675e-06,
|
| 1977 |
+
"loss": 0.2692,
|
| 1978 |
+
"step": 2230
|
| 1979 |
+
},
|
| 1980 |
+
{
|
| 1981 |
+
"epoch": 80.0,
|
| 1982 |
+
"learning_rate": 3.994512880821713e-06,
|
| 1983 |
+
"loss": 0.2644,
|
| 1984 |
+
"step": 2240
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"epoch": 80.0,
|
| 1988 |
+
"eval_loss": 0.2740114629268646,
|
| 1989 |
+
"eval_runtime": 2.6658,
|
| 1990 |
+
"eval_samples_per_second": 116.29,
|
| 1991 |
+
"eval_steps_per_second": 1.876,
|
| 1992 |
+
"step": 2240
|
| 1993 |
+
},
|
| 1994 |
+
{
|
| 1995 |
+
"epoch": 80.36,
|
| 1996 |
+
"learning_rate": 3.858911360454918e-06,
|
| 1997 |
+
"loss": 0.2685,
|
| 1998 |
+
"step": 2250
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 80.71,
|
| 2002 |
+
"learning_rate": 3.72538694232134e-06,
|
| 2003 |
+
"loss": 0.2691,
|
| 2004 |
+
"step": 2260
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 81.0,
|
| 2008 |
+
"eval_loss": 0.2726958394050598,
|
| 2009 |
+
"eval_runtime": 2.6341,
|
| 2010 |
+
"eval_samples_per_second": 117.686,
|
| 2011 |
+
"eval_steps_per_second": 1.898,
|
| 2012 |
+
"step": 2268
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 81.07,
|
| 2016 |
+
"learning_rate": 3.593958251242444e-06,
|
| 2017 |
+
"loss": 0.2748,
|
| 2018 |
+
"step": 2270
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 81.43,
|
| 2022 |
+
"learning_rate": 3.4646436197145335e-06,
|
| 2023 |
+
"loss": 0.2689,
|
| 2024 |
+
"step": 2280
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 81.79,
|
| 2028 |
+
"learning_rate": 3.3374610853516906e-06,
|
| 2029 |
+
"loss": 0.2679,
|
| 2030 |
+
"step": 2290
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 82.0,
|
| 2034 |
+
"eval_loss": 0.27709755301475525,
|
| 2035 |
+
"eval_runtime": 2.6545,
|
| 2036 |
+
"eval_samples_per_second": 116.784,
|
| 2037 |
+
"eval_steps_per_second": 1.884,
|
| 2038 |
+
"step": 2296
|
| 2039 |
+
},
|
| 2040 |
+
{
|
| 2041 |
+
"epoch": 82.14,
|
| 2042 |
+
"learning_rate": 3.212428388369718e-06,
|
| 2043 |
+
"loss": 0.2699,
|
| 2044 |
+
"step": 2300
|
| 2045 |
+
},
|
| 2046 |
+
{
|
| 2047 |
+
"epoch": 82.5,
|
| 2048 |
+
"learning_rate": 3.0895629691116723e-06,
|
| 2049 |
+
"loss": 0.2768,
|
| 2050 |
+
"step": 2310
|
| 2051 |
+
},
|
| 2052 |
+
{
|
| 2053 |
+
"epoch": 82.86,
|
| 2054 |
+
"learning_rate": 2.9688819656151164e-06,
|
| 2055 |
+
"loss": 0.2748,
|
| 2056 |
+
"step": 2320
|
| 2057 |
+
},
|
| 2058 |
+
{
|
| 2059 |
+
"epoch": 83.0,
|
| 2060 |
+
"eval_loss": 0.2743559777736664,
|
| 2061 |
+
"eval_runtime": 2.6236,
|
| 2062 |
+
"eval_samples_per_second": 118.158,
|
| 2063 |
+
"eval_steps_per_second": 1.906,
|
| 2064 |
+
"step": 2324
|
| 2065 |
+
},
|
| 2066 |
+
{
|
| 2067 |
+
"epoch": 83.21,
|
| 2068 |
+
"learning_rate": 2.8504022112216642e-06,
|
| 2069 |
+
"loss": 0.2635,
|
| 2070 |
+
"step": 2330
|
| 2071 |
+
},
|
| 2072 |
+
{
|
| 2073 |
+
"epoch": 83.57,
|
| 2074 |
+
"learning_rate": 2.734140232228908e-06,
|
| 2075 |
+
"loss": 0.2667,
|
| 2076 |
+
"step": 2340
|
| 2077 |
+
},
|
| 2078 |
+
{
|
| 2079 |
+
"epoch": 83.93,
|
| 2080 |
+
"learning_rate": 2.6201122455852582e-06,
|
| 2081 |
+
"loss": 0.2744,
|
| 2082 |
+
"step": 2350
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 84.0,
|
| 2086 |
+
"eval_loss": 0.2703389823436737,
|
| 2087 |
+
"eval_runtime": 2.6131,
|
| 2088 |
+
"eval_samples_per_second": 118.631,
|
| 2089 |
+
"eval_steps_per_second": 1.913,
|
| 2090 |
+
"step": 2352
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 84.29,
|
| 2094 |
+
"learning_rate": 2.5083341566278907e-06,
|
| 2095 |
+
"loss": 0.2761,
|
| 2096 |
+
"step": 2360
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 84.64,
|
| 2100 |
+
"learning_rate": 2.398821556864163e-06,
|
| 2101 |
+
"loss": 0.2699,
|
| 2102 |
+
"step": 2370
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 85.0,
|
| 2106 |
+
"learning_rate": 2.291589721796841e-06,
|
| 2107 |
+
"loss": 0.2715,
|
| 2108 |
+
"step": 2380
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 85.0,
|
| 2112 |
+
"eval_loss": 0.27332332730293274,
|
| 2113 |
+
"eval_runtime": 2.6685,
|
| 2114 |
+
"eval_samples_per_second": 116.168,
|
| 2115 |
+
"eval_steps_per_second": 1.874,
|
| 2116 |
+
"step": 2380
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 85.36,
|
| 2120 |
+
"learning_rate": 2.1866536087933395e-06,
|
| 2121 |
+
"loss": 0.263,
|
| 2122 |
+
"step": 2390
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"epoch": 85.71,
|
| 2126 |
+
"learning_rate": 2.084027854999404e-06,
|
| 2127 |
+
"loss": 0.2682,
|
| 2128 |
+
"step": 2400
|
| 2129 |
+
},
|
| 2130 |
+
{
|
| 2131 |
+
"epoch": 86.0,
|
| 2132 |
+
"eval_loss": 0.271467000246048,
|
| 2133 |
+
"eval_runtime": 2.6604,
|
| 2134 |
+
"eval_samples_per_second": 116.522,
|
| 2135 |
+
"eval_steps_per_second": 1.879,
|
| 2136 |
+
"step": 2408
|
| 2137 |
+
},
|
| 2138 |
+
{
|
| 2139 |
+
"epoch": 86.07,
|
| 2140 |
+
"learning_rate": 1.9837267752974077e-06,
|
| 2141 |
+
"loss": 0.2783,
|
| 2142 |
+
"step": 2410
|
| 2143 |
+
},
|
| 2144 |
+
{
|
| 2145 |
+
"epoch": 86.43,
|
| 2146 |
+
"learning_rate": 1.8857643603096489e-06,
|
| 2147 |
+
"loss": 0.2776,
|
| 2148 |
+
"step": 2420
|
| 2149 |
+
},
|
| 2150 |
+
{
|
| 2151 |
+
"epoch": 86.79,
|
| 2152 |
+
"learning_rate": 1.7901542744468242e-06,
|
| 2153 |
+
"loss": 0.2641,
|
| 2154 |
+
"step": 2430
|
| 2155 |
+
},
|
| 2156 |
+
{
|
| 2157 |
+
"epoch": 87.0,
|
| 2158 |
+
"eval_loss": 0.27218252420425415,
|
| 2159 |
+
"eval_runtime": 2.6297,
|
| 2160 |
+
"eval_samples_per_second": 117.884,
|
| 2161 |
+
"eval_steps_per_second": 1.901,
|
| 2162 |
+
"step": 2436
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 87.14,
|
| 2166 |
+
"learning_rate": 1.6969098540020453e-06,
|
| 2167 |
+
"loss": 0.2704,
|
| 2168 |
+
"step": 2440
|
| 2169 |
+
},
|
| 2170 |
+
{
|
| 2171 |
+
"epoch": 87.5,
|
| 2172 |
+
"learning_rate": 1.6060441052906117e-06,
|
| 2173 |
+
"loss": 0.2769,
|
| 2174 |
+
"step": 2450
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 87.86,
|
| 2178 |
+
"learning_rate": 1.517569702835792e-06,
|
| 2179 |
+
"loss": 0.274,
|
| 2180 |
+
"step": 2460
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 88.0,
|
| 2184 |
+
"eval_loss": 0.27482396364212036,
|
| 2185 |
+
"eval_runtime": 2.6311,
|
| 2186 |
+
"eval_samples_per_second": 117.823,
|
| 2187 |
+
"eval_steps_per_second": 1.9,
|
| 2188 |
+
"step": 2464
|
| 2189 |
+
},
|
| 2190 |
+
{
|
| 2191 |
+
"epoch": 88.21,
|
| 2192 |
+
"learning_rate": 1.4314989876009292e-06,
|
| 2193 |
+
"loss": 0.2561,
|
| 2194 |
+
"step": 2470
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"epoch": 88.57,
|
| 2198 |
+
"learning_rate": 1.3478439652680215e-06,
|
| 2199 |
+
"loss": 0.281,
|
| 2200 |
+
"step": 2480
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 88.93,
|
| 2204 |
+
"learning_rate": 1.2666163045631007e-06,
|
| 2205 |
+
"loss": 0.2669,
|
| 2206 |
+
"step": 2490
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"epoch": 89.0,
|
| 2210 |
+
"eval_loss": 0.27534210681915283,
|
| 2211 |
+
"eval_runtime": 2.6393,
|
| 2212 |
+
"eval_samples_per_second": 117.455,
|
| 2213 |
+
"eval_steps_per_second": 1.894,
|
| 2214 |
+
"step": 2492
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 89.29,
|
| 2218 |
+
"learning_rate": 1.1878273356286133e-06,
|
| 2219 |
+
"loss": 0.2765,
|
| 2220 |
+
"step": 2500
|
| 2221 |
+
},
|
| 2222 |
+
{
|
| 2223 |
+
"epoch": 89.64,
|
| 2224 |
+
"learning_rate": 1.1114880484430204e-06,
|
| 2225 |
+
"loss": 0.2635,
|
| 2226 |
+
"step": 2510
|
| 2227 |
+
},
|
| 2228 |
+
{
|
| 2229 |
+
"epoch": 90.0,
|
| 2230 |
+
"learning_rate": 1.0376090912878263e-06,
|
| 2231 |
+
"loss": 0.2707,
|
| 2232 |
+
"step": 2520
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 90.0,
|
| 2236 |
+
"eval_loss": 0.2724357843399048,
|
| 2237 |
+
"eval_runtime": 2.6596,
|
| 2238 |
+
"eval_samples_per_second": 116.558,
|
| 2239 |
+
"eval_steps_per_second": 1.88,
|
| 2240 |
+
"step": 2520
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"epoch": 90.36,
|
| 2244 |
+
"learning_rate": 9.662007692623237e-07,
|
| 2245 |
+
"loss": 0.2667,
|
| 2246 |
+
"step": 2530
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 90.71,
|
| 2250 |
+
"learning_rate": 8.972730428461452e-07,
|
| 2251 |
+
"loss": 0.2755,
|
| 2252 |
+
"step": 2540
|
| 2253 |
+
},
|
| 2254 |
+
{
|
| 2255 |
+
"epoch": 91.0,
|
| 2256 |
+
"eval_loss": 0.27034902572631836,
|
| 2257 |
+
"eval_runtime": 2.6286,
|
| 2258 |
+
"eval_samples_per_second": 117.934,
|
| 2259 |
+
"eval_steps_per_second": 1.902,
|
| 2260 |
+
"step": 2548
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 91.07,
|
| 2264 |
+
"learning_rate": 8.308355265099334e-07,
|
| 2265 |
+
"loss": 0.2692,
|
| 2266 |
+
"step": 2550
|
| 2267 |
+
},
|
| 2268 |
+
{
|
| 2269 |
+
"epoch": 91.43,
|
| 2270 |
+
"learning_rate": 7.668974873742282e-07,
|
| 2271 |
+
"loss": 0.263,
|
| 2272 |
+
"step": 2560
|
| 2273 |
+
},
|
| 2274 |
+
{
|
| 2275 |
+
"epoch": 91.79,
|
| 2276 |
+
"learning_rate": 7.054678439168689e-07,
|
| 2277 |
+
"loss": 0.2769,
|
| 2278 |
+
"step": 2570
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 92.0,
|
| 2282 |
+
"eval_loss": 0.27368903160095215,
|
| 2283 |
+
"eval_runtime": 2.6502,
|
| 2284 |
+
"eval_samples_per_second": 116.973,
|
| 2285 |
+
"eval_steps_per_second": 1.887,
|
| 2286 |
+
"step": 2576
|
| 2287 |
+
},
|
| 2288 |
+
{
|
| 2289 |
+
"epoch": 92.14,
|
| 2290 |
+
"learning_rate": 6.465551647289506e-07,
|
| 2291 |
+
"loss": 0.2722,
|
| 2292 |
+
"step": 2580
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 92.5,
|
| 2296 |
+
"learning_rate": 5.901676673196507e-07,
|
| 2297 |
+
"loss": 0.2828,
|
| 2298 |
+
"step": 2590
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 92.86,
|
| 2302 |
+
"learning_rate": 5.36313216969983e-07,
|
| 2303 |
+
"loss": 0.2659,
|
| 2304 |
+
"step": 2600
|
| 2305 |
+
},
|
| 2306 |
+
{
|
| 2307 |
+
"epoch": 93.0,
|
| 2308 |
+
"eval_loss": 0.2720969319343567,
|
| 2309 |
+
"eval_runtime": 2.6064,
|
| 2310 |
+
"eval_samples_per_second": 118.938,
|
| 2311 |
+
"eval_steps_per_second": 1.918,
|
| 2312 |
+
"step": 2604
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 93.21,
|
| 2316 |
+
"learning_rate": 4.849993256357071e-07,
|
| 2317 |
+
"loss": 0.2608,
|
| 2318 |
+
"step": 2610
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 93.57,
|
| 2322 |
+
"learning_rate": 4.362331508995204e-07,
|
| 2323 |
+
"loss": 0.2731,
|
| 2324 |
+
"step": 2620
|
| 2325 |
+
},
|
| 2326 |
+
{
|
| 2327 |
+
"epoch": 93.93,
|
| 2328 |
+
"learning_rate": 3.9002149497265255e-07,
|
| 2329 |
+
"loss": 0.2674,
|
| 2330 |
+
"step": 2630
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 94.0,
|
| 2334 |
+
"eval_loss": 0.2763327658176422,
|
| 2335 |
+
"eval_runtime": 2.6289,
|
| 2336 |
+
"eval_samples_per_second": 117.922,
|
| 2337 |
+
"eval_steps_per_second": 1.902,
|
| 2338 |
+
"step": 2632
|
| 2339 |
+
},
|
| 2340 |
+
{
|
| 2341 |
+
"epoch": 94.29,
|
| 2342 |
+
"learning_rate": 3.463708037460683e-07,
|
| 2343 |
+
"loss": 0.277,
|
| 2344 |
+
"step": 2640
|
| 2345 |
+
},
|
| 2346 |
+
{
|
| 2347 |
+
"epoch": 94.64,
|
| 2348 |
+
"learning_rate": 3.052871658913533e-07,
|
| 2349 |
+
"loss": 0.2746,
|
| 2350 |
+
"step": 2650
|
| 2351 |
+
},
|
| 2352 |
+
{
|
| 2353 |
+
"epoch": 95.0,
|
| 2354 |
+
"learning_rate": 2.6677631201141856e-07,
|
| 2355 |
+
"loss": 0.2723,
|
| 2356 |
+
"step": 2660
|
| 2357 |
+
},
|
| 2358 |
+
{
|
| 2359 |
+
"epoch": 95.0,
|
| 2360 |
+
"eval_loss": 0.27226099371910095,
|
| 2361 |
+
"eval_runtime": 2.6569,
|
| 2362 |
+
"eval_samples_per_second": 116.678,
|
| 2363 |
+
"eval_steps_per_second": 1.882,
|
| 2364 |
+
"step": 2660
|
| 2365 |
+
},
|
| 2366 |
+
{
|
| 2367 |
+
"epoch": 95.36,
|
| 2368 |
+
"learning_rate": 2.3084361384116886e-07,
|
| 2369 |
+
"loss": 0.261,
|
| 2370 |
+
"step": 2670
|
| 2371 |
+
},
|
| 2372 |
+
{
|
| 2373 |
+
"epoch": 95.71,
|
| 2374 |
+
"learning_rate": 1.9749408349821265e-07,
|
| 2375 |
+
"loss": 0.2723,
|
| 2376 |
+
"step": 2680
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 96.0,
|
| 2380 |
+
"eval_loss": 0.2743915319442749,
|
| 2381 |
+
"eval_runtime": 2.6421,
|
| 2382 |
+
"eval_samples_per_second": 117.329,
|
| 2383 |
+
"eval_steps_per_second": 1.892,
|
| 2384 |
+
"step": 2688
|
| 2385 |
+
},
|
| 2386 |
+
{
|
| 2387 |
+
"epoch": 96.07,
|
| 2388 |
+
"learning_rate": 1.6673237278375036e-07,
|
| 2389 |
+
"loss": 0.2775,
|
| 2390 |
+
"step": 2690
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 96.43,
|
| 2394 |
+
"learning_rate": 1.3856277253369782e-07,
|
| 2395 |
+
"loss": 0.2813,
|
| 2396 |
+
"step": 2700
|
| 2397 |
+
},
|
| 2398 |
+
{
|
| 2399 |
+
"epoch": 96.79,
|
| 2400 |
+
"learning_rate": 1.1298921202018765e-07,
|
| 2401 |
+
"loss": 0.272,
|
| 2402 |
+
"step": 2710
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 97.0,
|
| 2406 |
+
"eval_loss": 0.26857295632362366,
|
| 2407 |
+
"eval_runtime": 2.6133,
|
| 2408 |
+
"eval_samples_per_second": 118.624,
|
| 2409 |
+
"eval_steps_per_second": 1.913,
|
| 2410 |
+
"step": 2716
|
| 2411 |
+
},
|
| 2412 |
+
{
|
| 2413 |
+
"epoch": 97.14,
|
| 2414 |
+
"learning_rate": 9.001525840348495e-08,
|
| 2415 |
+
"loss": 0.2601,
|
| 2416 |
+
"step": 2720
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 97.5,
|
| 2420 |
+
"learning_rate": 6.964411623440816e-08,
|
| 2421 |
+
"loss": 0.2756,
|
| 2422 |
+
"step": 2730
|
| 2423 |
+
},
|
| 2424 |
+
{
|
| 2425 |
+
"epoch": 97.86,
|
| 2426 |
+
"learning_rate": 5.187862700735332e-08,
|
| 2427 |
+
"loss": 0.27,
|
| 2428 |
+
"step": 2740
|
| 2429 |
+
},
|
| 2430 |
+
{
|
| 2431 |
+
"epoch": 98.0,
|
| 2432 |
+
"eval_loss": 0.27275460958480835,
|
| 2433 |
+
"eval_runtime": 2.6408,
|
| 2434 |
+
"eval_samples_per_second": 117.389,
|
| 2435 |
+
"eval_steps_per_second": 1.893,
|
| 2436 |
+
"step": 2744
|
| 2437 |
+
},
|
| 2438 |
+
{
|
| 2439 |
+
"epoch": 98.21,
|
| 2440 |
+
"learning_rate": 3.672126876393816e-08,
|
| 2441 |
+
"loss": 0.2757,
|
| 2442 |
+
"step": 2750
|
| 2443 |
+
},
|
| 2444 |
+
{
|
| 2445 |
+
"epoch": 98.57,
|
| 2446 |
+
"learning_rate": 2.4174155747345297e-08,
|
| 2447 |
+
"loss": 0.2662,
|
| 2448 |
+
"step": 2760
|
| 2449 |
+
},
|
| 2450 |
+
{
|
| 2451 |
+
"epoch": 98.93,
|
| 2452 |
+
"learning_rate": 1.4239038107416445e-08,
|
| 2453 |
+
"loss": 0.2721,
|
| 2454 |
+
"step": 2770
|
| 2455 |
+
},
|
| 2456 |
+
{
|
| 2457 |
+
"epoch": 99.0,
|
| 2458 |
+
"eval_loss": 0.27428731322288513,
|
| 2459 |
+
"eval_runtime": 2.6271,
|
| 2460 |
+
"eval_samples_per_second": 118.001,
|
| 2461 |
+
"eval_steps_per_second": 1.903,
|
| 2462 |
+
"step": 2772
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 99.29,
|
| 2466 |
+
"learning_rate": 6.917301656537283e-09,
|
| 2467 |
+
"loss": 0.2792,
|
| 2468 |
+
"step": 2780
|
| 2469 |
+
},
|
| 2470 |
+
{
|
| 2471 |
+
"epoch": 99.64,
|
| 2472 |
+
"learning_rate": 2.2099676763295825e-09,
|
| 2473 |
+
"loss": 0.2751,
|
| 2474 |
+
"step": 2790
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 100.0,
|
| 2478 |
+
"learning_rate": 1.176927751922463e-10,
|
| 2479 |
+
"loss": 0.2692,
|
| 2480 |
+
"step": 2800
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"epoch": 100.0,
|
| 2484 |
+
"eval_loss": 0.2747902274131775,
|
| 2485 |
+
"eval_runtime": 2.6706,
|
| 2486 |
+
"eval_samples_per_second": 116.08,
|
| 2487 |
+
"eval_steps_per_second": 1.872,
|
| 2488 |
+
"step": 2800
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"epoch": 100.0,
|
| 2492 |
+
"step": 2800,
|
| 2493 |
+
"total_flos": 1.7697633664499712e+19,
|
| 2494 |
+
"train_loss": 0.1087875749383654,
|
| 2495 |
+
"train_runtime": 1716.2917,
|
| 2496 |
+
"train_samples_per_second": 102.022,
|
| 2497 |
+
"train_steps_per_second": 1.631
|
| 2498 |
+
}
|
| 2499 |
+
],
|
| 2500 |
+
"max_steps": 2800,
|
| 2501 |
+
"num_train_epochs": 100,
|
| 2502 |
+
"total_flos": 1.7697633664499712e+19,
|
| 2503 |
+
"trial_name": null,
|
| 2504 |
+
"trial_params": null
|
| 2505 |
+
}
|