ahmedALM1221
commited on
Commit
•
4b5f698
1
Parent(s):
acc38c5
Training in progress, epoch 1
Browse files
all_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 12.0,
|
3 |
+
"total_flos": 1.175377212318081e+18,
|
4 |
+
"train_loss": 1.2422215939382386,
|
5 |
+
"train_runtime": 2576.2267,
|
6 |
+
"train_samples_per_second": 18.124,
|
7 |
+
"train_steps_per_second": 0.568
|
8 |
+
}
|
runs/Jul06_13-48-03_86ee152215b1/events.out.tfevents.1688654933.86ee152215b1.2511.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:538fef20e86300d33443fa1b6072523bf657f2edfd448876fd579cdd49657637
|
3 |
+
size 8437
|
train_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 12.0,
|
3 |
+
"total_flos": 1.175377212318081e+18,
|
4 |
+
"train_loss": 1.2422215939382386,
|
5 |
+
"train_runtime": 2576.2267,
|
6 |
+
"train_samples_per_second": 18.124,
|
7 |
+
"train_steps_per_second": 0.568
|
8 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,1885 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 0.8273381294964028,
|
3 |
+
"best_model_checkpoint": "convnextv2-tiny-22k-224-finetuned-eurosat-50/checkpoint-1464",
|
4 |
+
"epoch": 12.0,
|
5 |
+
"global_step": 1464,
|
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": 1.8968133535660093e-08,
|
13 |
+
"loss": 2.0068,
|
14 |
+
"step": 5
|
15 |
+
},
|
16 |
+
{
|
17 |
+
"epoch": 0.08,
|
18 |
+
"learning_rate": 3.7936267071320186e-08,
|
19 |
+
"loss": 2.0511,
|
20 |
+
"step": 10
|
21 |
+
},
|
22 |
+
{
|
23 |
+
"epoch": 0.12,
|
24 |
+
"learning_rate": 5.6904400606980276e-08,
|
25 |
+
"loss": 1.995,
|
26 |
+
"step": 15
|
27 |
+
},
|
28 |
+
{
|
29 |
+
"epoch": 0.16,
|
30 |
+
"learning_rate": 7.587253414264037e-08,
|
31 |
+
"loss": 2.0273,
|
32 |
+
"step": 20
|
33 |
+
},
|
34 |
+
{
|
35 |
+
"epoch": 0.2,
|
36 |
+
"learning_rate": 9.484066767830046e-08,
|
37 |
+
"loss": 2.0036,
|
38 |
+
"step": 25
|
39 |
+
},
|
40 |
+
{
|
41 |
+
"epoch": 0.25,
|
42 |
+
"learning_rate": 1.1380880121396055e-07,
|
43 |
+
"loss": 2.0435,
|
44 |
+
"step": 30
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.29,
|
48 |
+
"learning_rate": 1.3277693474962064e-07,
|
49 |
+
"loss": 2.0563,
|
50 |
+
"step": 35
|
51 |
+
},
|
52 |
+
{
|
53 |
+
"epoch": 0.33,
|
54 |
+
"learning_rate": 1.5174506828528074e-07,
|
55 |
+
"loss": 2.0733,
|
56 |
+
"step": 40
|
57 |
+
},
|
58 |
+
{
|
59 |
+
"epoch": 0.37,
|
60 |
+
"learning_rate": 1.7071320182094083e-07,
|
61 |
+
"loss": 2.0279,
|
62 |
+
"step": 45
|
63 |
+
},
|
64 |
+
{
|
65 |
+
"epoch": 0.41,
|
66 |
+
"learning_rate": 1.896813353566009e-07,
|
67 |
+
"loss": 2.0142,
|
68 |
+
"step": 50
|
69 |
+
},
|
70 |
+
{
|
71 |
+
"epoch": 0.45,
|
72 |
+
"learning_rate": 2.0864946889226102e-07,
|
73 |
+
"loss": 2.0612,
|
74 |
+
"step": 55
|
75 |
+
},
|
76 |
+
{
|
77 |
+
"epoch": 0.49,
|
78 |
+
"learning_rate": 2.276176024279211e-07,
|
79 |
+
"loss": 2.0662,
|
80 |
+
"step": 60
|
81 |
+
},
|
82 |
+
{
|
83 |
+
"epoch": 0.53,
|
84 |
+
"learning_rate": 2.465857359635812e-07,
|
85 |
+
"loss": 2.0601,
|
86 |
+
"step": 65
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.57,
|
90 |
+
"learning_rate": 2.6555386949924127e-07,
|
91 |
+
"loss": 1.9407,
|
92 |
+
"step": 70
|
93 |
+
},
|
94 |
+
{
|
95 |
+
"epoch": 0.61,
|
96 |
+
"learning_rate": 2.845220030349014e-07,
|
97 |
+
"loss": 2.0653,
|
98 |
+
"step": 75
|
99 |
+
},
|
100 |
+
{
|
101 |
+
"epoch": 0.66,
|
102 |
+
"learning_rate": 3.034901365705615e-07,
|
103 |
+
"loss": 2.0545,
|
104 |
+
"step": 80
|
105 |
+
},
|
106 |
+
{
|
107 |
+
"epoch": 0.7,
|
108 |
+
"learning_rate": 3.2245827010622155e-07,
|
109 |
+
"loss": 2.0208,
|
110 |
+
"step": 85
|
111 |
+
},
|
112 |
+
{
|
113 |
+
"epoch": 0.74,
|
114 |
+
"learning_rate": 3.4142640364188166e-07,
|
115 |
+
"loss": 2.0267,
|
116 |
+
"step": 90
|
117 |
+
},
|
118 |
+
{
|
119 |
+
"epoch": 0.78,
|
120 |
+
"learning_rate": 3.6039453717754177e-07,
|
121 |
+
"loss": 2.0156,
|
122 |
+
"step": 95
|
123 |
+
},
|
124 |
+
{
|
125 |
+
"epoch": 0.82,
|
126 |
+
"learning_rate": 3.793626707132018e-07,
|
127 |
+
"loss": 2.0217,
|
128 |
+
"step": 100
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.86,
|
132 |
+
"learning_rate": 3.9833080424886193e-07,
|
133 |
+
"loss": 2.0383,
|
134 |
+
"step": 105
|
135 |
+
},
|
136 |
+
{
|
137 |
+
"epoch": 0.9,
|
138 |
+
"learning_rate": 4.1729893778452204e-07,
|
139 |
+
"loss": 1.9975,
|
140 |
+
"step": 110
|
141 |
+
},
|
142 |
+
{
|
143 |
+
"epoch": 0.94,
|
144 |
+
"learning_rate": 4.3626707132018215e-07,
|
145 |
+
"loss": 1.9487,
|
146 |
+
"step": 115
|
147 |
+
},
|
148 |
+
{
|
149 |
+
"epoch": 0.98,
|
150 |
+
"learning_rate": 4.552352048558422e-07,
|
151 |
+
"loss": 2.0127,
|
152 |
+
"step": 120
|
153 |
+
},
|
154 |
+
{
|
155 |
+
"epoch": 1.0,
|
156 |
+
"eval_accuracy": 0.19424460431654678,
|
157 |
+
"eval_loss": 1.9837570190429688,
|
158 |
+
"eval_runtime": 379.9947,
|
159 |
+
"eval_samples_per_second": 2.561,
|
160 |
+
"eval_steps_per_second": 0.161,
|
161 |
+
"step": 122
|
162 |
+
},
|
163 |
+
{
|
164 |
+
"epoch": 1.02,
|
165 |
+
"learning_rate": 4.742033383915023e-07,
|
166 |
+
"loss": 1.9781,
|
167 |
+
"step": 125
|
168 |
+
},
|
169 |
+
{
|
170 |
+
"epoch": 1.07,
|
171 |
+
"learning_rate": 4.931714719271624e-07,
|
172 |
+
"loss": 1.9947,
|
173 |
+
"step": 130
|
174 |
+
},
|
175 |
+
{
|
176 |
+
"epoch": 1.11,
|
177 |
+
"learning_rate": 5.121396054628225e-07,
|
178 |
+
"loss": 1.9917,
|
179 |
+
"step": 135
|
180 |
+
},
|
181 |
+
{
|
182 |
+
"epoch": 1.15,
|
183 |
+
"learning_rate": 5.311077389984825e-07,
|
184 |
+
"loss": 1.9269,
|
185 |
+
"step": 140
|
186 |
+
},
|
187 |
+
{
|
188 |
+
"epoch": 1.19,
|
189 |
+
"learning_rate": 5.500758725341426e-07,
|
190 |
+
"loss": 2.0058,
|
191 |
+
"step": 145
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 1.23,
|
195 |
+
"learning_rate": 5.690440060698028e-07,
|
196 |
+
"loss": 1.9766,
|
197 |
+
"step": 150
|
198 |
+
},
|
199 |
+
{
|
200 |
+
"epoch": 1.27,
|
201 |
+
"learning_rate": 5.880121396054629e-07,
|
202 |
+
"loss": 2.0333,
|
203 |
+
"step": 155
|
204 |
+
},
|
205 |
+
{
|
206 |
+
"epoch": 1.31,
|
207 |
+
"learning_rate": 6.06980273141123e-07,
|
208 |
+
"loss": 1.948,
|
209 |
+
"step": 160
|
210 |
+
},
|
211 |
+
{
|
212 |
+
"epoch": 1.35,
|
213 |
+
"learning_rate": 6.259484066767831e-07,
|
214 |
+
"loss": 1.9442,
|
215 |
+
"step": 165
|
216 |
+
},
|
217 |
+
{
|
218 |
+
"epoch": 1.39,
|
219 |
+
"learning_rate": 6.449165402124431e-07,
|
220 |
+
"loss": 1.9335,
|
221 |
+
"step": 170
|
222 |
+
},
|
223 |
+
{
|
224 |
+
"epoch": 1.43,
|
225 |
+
"learning_rate": 6.638846737481032e-07,
|
226 |
+
"loss": 1.9478,
|
227 |
+
"step": 175
|
228 |
+
},
|
229 |
+
{
|
230 |
+
"epoch": 1.48,
|
231 |
+
"learning_rate": 6.828528072837633e-07,
|
232 |
+
"loss": 1.9518,
|
233 |
+
"step": 180
|
234 |
+
},
|
235 |
+
{
|
236 |
+
"epoch": 1.52,
|
237 |
+
"learning_rate": 7.018209408194235e-07,
|
238 |
+
"loss": 1.9652,
|
239 |
+
"step": 185
|
240 |
+
},
|
241 |
+
{
|
242 |
+
"epoch": 1.56,
|
243 |
+
"learning_rate": 7.207890743550835e-07,
|
244 |
+
"loss": 1.9323,
|
245 |
+
"step": 190
|
246 |
+
},
|
247 |
+
{
|
248 |
+
"epoch": 1.6,
|
249 |
+
"learning_rate": 7.397572078907436e-07,
|
250 |
+
"loss": 1.8917,
|
251 |
+
"step": 195
|
252 |
+
},
|
253 |
+
{
|
254 |
+
"epoch": 1.64,
|
255 |
+
"learning_rate": 7.587253414264036e-07,
|
256 |
+
"loss": 1.8973,
|
257 |
+
"step": 200
|
258 |
+
},
|
259 |
+
{
|
260 |
+
"epoch": 1.68,
|
261 |
+
"learning_rate": 7.776934749620638e-07,
|
262 |
+
"loss": 1.8717,
|
263 |
+
"step": 205
|
264 |
+
},
|
265 |
+
{
|
266 |
+
"epoch": 1.72,
|
267 |
+
"learning_rate": 7.966616084977239e-07,
|
268 |
+
"loss": 1.9361,
|
269 |
+
"step": 210
|
270 |
+
},
|
271 |
+
{
|
272 |
+
"epoch": 1.76,
|
273 |
+
"learning_rate": 8.156297420333841e-07,
|
274 |
+
"loss": 1.8694,
|
275 |
+
"step": 215
|
276 |
+
},
|
277 |
+
{
|
278 |
+
"epoch": 1.8,
|
279 |
+
"learning_rate": 8.345978755690441e-07,
|
280 |
+
"loss": 1.9716,
|
281 |
+
"step": 220
|
282 |
+
},
|
283 |
+
{
|
284 |
+
"epoch": 1.84,
|
285 |
+
"learning_rate": 8.535660091047042e-07,
|
286 |
+
"loss": 1.9019,
|
287 |
+
"step": 225
|
288 |
+
},
|
289 |
+
{
|
290 |
+
"epoch": 1.89,
|
291 |
+
"learning_rate": 8.725341426403643e-07,
|
292 |
+
"loss": 1.9313,
|
293 |
+
"step": 230
|
294 |
+
},
|
295 |
+
{
|
296 |
+
"epoch": 1.93,
|
297 |
+
"learning_rate": 8.915022761760243e-07,
|
298 |
+
"loss": 1.887,
|
299 |
+
"step": 235
|
300 |
+
},
|
301 |
+
{
|
302 |
+
"epoch": 1.97,
|
303 |
+
"learning_rate": 9.104704097116844e-07,
|
304 |
+
"loss": 1.8499,
|
305 |
+
"step": 240
|
306 |
+
},
|
307 |
+
{
|
308 |
+
"epoch": 2.0,
|
309 |
+
"eval_accuracy": 0.24563206577595068,
|
310 |
+
"eval_loss": 1.8666632175445557,
|
311 |
+
"eval_runtime": 5.443,
|
312 |
+
"eval_samples_per_second": 178.763,
|
313 |
+
"eval_steps_per_second": 11.207,
|
314 |
+
"step": 244
|
315 |
+
},
|
316 |
+
{
|
317 |
+
"epoch": 2.01,
|
318 |
+
"learning_rate": 9.294385432473445e-07,
|
319 |
+
"loss": 1.8892,
|
320 |
+
"step": 245
|
321 |
+
},
|
322 |
+
{
|
323 |
+
"epoch": 2.05,
|
324 |
+
"learning_rate": 9.484066767830046e-07,
|
325 |
+
"loss": 1.8887,
|
326 |
+
"step": 250
|
327 |
+
},
|
328 |
+
{
|
329 |
+
"epoch": 2.09,
|
330 |
+
"learning_rate": 9.673748103186647e-07,
|
331 |
+
"loss": 1.8473,
|
332 |
+
"step": 255
|
333 |
+
},
|
334 |
+
{
|
335 |
+
"epoch": 2.13,
|
336 |
+
"learning_rate": 9.863429438543249e-07,
|
337 |
+
"loss": 1.8769,
|
338 |
+
"step": 260
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 2.17,
|
342 |
+
"learning_rate": 1.005311077389985e-06,
|
343 |
+
"loss": 1.8652,
|
344 |
+
"step": 265
|
345 |
+
},
|
346 |
+
{
|
347 |
+
"epoch": 2.21,
|
348 |
+
"learning_rate": 1.024279210925645e-06,
|
349 |
+
"loss": 1.798,
|
350 |
+
"step": 270
|
351 |
+
},
|
352 |
+
{
|
353 |
+
"epoch": 2.25,
|
354 |
+
"learning_rate": 1.043247344461305e-06,
|
355 |
+
"loss": 1.8467,
|
356 |
+
"step": 275
|
357 |
+
},
|
358 |
+
{
|
359 |
+
"epoch": 2.3,
|
360 |
+
"learning_rate": 1.062215477996965e-06,
|
361 |
+
"loss": 1.7636,
|
362 |
+
"step": 280
|
363 |
+
},
|
364 |
+
{
|
365 |
+
"epoch": 2.34,
|
366 |
+
"learning_rate": 1.0811836115326254e-06,
|
367 |
+
"loss": 1.8451,
|
368 |
+
"step": 285
|
369 |
+
},
|
370 |
+
{
|
371 |
+
"epoch": 2.38,
|
372 |
+
"learning_rate": 1.1001517450682853e-06,
|
373 |
+
"loss": 1.8375,
|
374 |
+
"step": 290
|
375 |
+
},
|
376 |
+
{
|
377 |
+
"epoch": 2.42,
|
378 |
+
"learning_rate": 1.1191198786039454e-06,
|
379 |
+
"loss": 1.8289,
|
380 |
+
"step": 295
|
381 |
+
},
|
382 |
+
{
|
383 |
+
"epoch": 2.46,
|
384 |
+
"learning_rate": 1.1380880121396055e-06,
|
385 |
+
"loss": 1.756,
|
386 |
+
"step": 300
|
387 |
+
},
|
388 |
+
{
|
389 |
+
"epoch": 2.5,
|
390 |
+
"learning_rate": 1.1570561456752656e-06,
|
391 |
+
"loss": 1.8203,
|
392 |
+
"step": 305
|
393 |
+
},
|
394 |
+
{
|
395 |
+
"epoch": 2.54,
|
396 |
+
"learning_rate": 1.1760242792109257e-06,
|
397 |
+
"loss": 1.8125,
|
398 |
+
"step": 310
|
399 |
+
},
|
400 |
+
{
|
401 |
+
"epoch": 2.58,
|
402 |
+
"learning_rate": 1.1949924127465858e-06,
|
403 |
+
"loss": 1.8137,
|
404 |
+
"step": 315
|
405 |
+
},
|
406 |
+
{
|
407 |
+
"epoch": 2.62,
|
408 |
+
"learning_rate": 1.213960546282246e-06,
|
409 |
+
"loss": 1.7922,
|
410 |
+
"step": 320
|
411 |
+
},
|
412 |
+
{
|
413 |
+
"epoch": 2.66,
|
414 |
+
"learning_rate": 1.232928679817906e-06,
|
415 |
+
"loss": 1.8181,
|
416 |
+
"step": 325
|
417 |
+
},
|
418 |
+
{
|
419 |
+
"epoch": 2.7,
|
420 |
+
"learning_rate": 1.2518968133535662e-06,
|
421 |
+
"loss": 1.7922,
|
422 |
+
"step": 330
|
423 |
+
},
|
424 |
+
{
|
425 |
+
"epoch": 2.75,
|
426 |
+
"learning_rate": 1.2708649468892263e-06,
|
427 |
+
"loss": 1.7696,
|
428 |
+
"step": 335
|
429 |
+
},
|
430 |
+
{
|
431 |
+
"epoch": 2.79,
|
432 |
+
"learning_rate": 1.2898330804248862e-06,
|
433 |
+
"loss": 1.7614,
|
434 |
+
"step": 340
|
435 |
+
},
|
436 |
+
{
|
437 |
+
"epoch": 2.83,
|
438 |
+
"learning_rate": 1.3088012139605463e-06,
|
439 |
+
"loss": 1.7979,
|
440 |
+
"step": 345
|
441 |
+
},
|
442 |
+
{
|
443 |
+
"epoch": 2.87,
|
444 |
+
"learning_rate": 1.3277693474962064e-06,
|
445 |
+
"loss": 1.7467,
|
446 |
+
"step": 350
|
447 |
+
},
|
448 |
+
{
|
449 |
+
"epoch": 2.91,
|
450 |
+
"learning_rate": 1.3467374810318665e-06,
|
451 |
+
"loss": 1.7974,
|
452 |
+
"step": 355
|
453 |
+
},
|
454 |
+
{
|
455 |
+
"epoch": 2.95,
|
456 |
+
"learning_rate": 1.3657056145675266e-06,
|
457 |
+
"loss": 1.7483,
|
458 |
+
"step": 360
|
459 |
+
},
|
460 |
+
{
|
461 |
+
"epoch": 2.99,
|
462 |
+
"learning_rate": 1.384673748103187e-06,
|
463 |
+
"loss": 1.8024,
|
464 |
+
"step": 365
|
465 |
+
},
|
466 |
+
{
|
467 |
+
"epoch": 3.0,
|
468 |
+
"eval_accuracy": 0.37923946557040084,
|
469 |
+
"eval_loss": 1.7246524095535278,
|
470 |
+
"eval_runtime": 5.4824,
|
471 |
+
"eval_samples_per_second": 177.476,
|
472 |
+
"eval_steps_per_second": 11.126,
|
473 |
+
"step": 366
|
474 |
+
},
|
475 |
+
{
|
476 |
+
"epoch": 3.03,
|
477 |
+
"learning_rate": 1.403641881638847e-06,
|
478 |
+
"loss": 1.6331,
|
479 |
+
"step": 370
|
480 |
+
},
|
481 |
+
{
|
482 |
+
"epoch": 3.07,
|
483 |
+
"learning_rate": 1.422610015174507e-06,
|
484 |
+
"loss": 1.699,
|
485 |
+
"step": 375
|
486 |
+
},
|
487 |
+
{
|
488 |
+
"epoch": 3.11,
|
489 |
+
"learning_rate": 1.441578148710167e-06,
|
490 |
+
"loss": 1.7516,
|
491 |
+
"step": 380
|
492 |
+
},
|
493 |
+
{
|
494 |
+
"epoch": 3.16,
|
495 |
+
"learning_rate": 1.4605462822458272e-06,
|
496 |
+
"loss": 1.7463,
|
497 |
+
"step": 385
|
498 |
+
},
|
499 |
+
{
|
500 |
+
"epoch": 3.2,
|
501 |
+
"learning_rate": 1.4795144157814873e-06,
|
502 |
+
"loss": 1.7216,
|
503 |
+
"step": 390
|
504 |
+
},
|
505 |
+
{
|
506 |
+
"epoch": 3.24,
|
507 |
+
"learning_rate": 1.4984825493171474e-06,
|
508 |
+
"loss": 1.7106,
|
509 |
+
"step": 395
|
510 |
+
},
|
511 |
+
{
|
512 |
+
"epoch": 3.28,
|
513 |
+
"learning_rate": 1.5174506828528073e-06,
|
514 |
+
"loss": 1.6734,
|
515 |
+
"step": 400
|
516 |
+
},
|
517 |
+
{
|
518 |
+
"epoch": 3.32,
|
519 |
+
"learning_rate": 1.5364188163884674e-06,
|
520 |
+
"loss": 1.6436,
|
521 |
+
"step": 405
|
522 |
+
},
|
523 |
+
{
|
524 |
+
"epoch": 3.36,
|
525 |
+
"learning_rate": 1.5553869499241275e-06,
|
526 |
+
"loss": 1.6588,
|
527 |
+
"step": 410
|
528 |
+
},
|
529 |
+
{
|
530 |
+
"epoch": 3.4,
|
531 |
+
"learning_rate": 1.5743550834597876e-06,
|
532 |
+
"loss": 1.6423,
|
533 |
+
"step": 415
|
534 |
+
},
|
535 |
+
{
|
536 |
+
"epoch": 3.44,
|
537 |
+
"learning_rate": 1.5933232169954477e-06,
|
538 |
+
"loss": 1.5818,
|
539 |
+
"step": 420
|
540 |
+
},
|
541 |
+
{
|
542 |
+
"epoch": 3.48,
|
543 |
+
"learning_rate": 1.6122913505311078e-06,
|
544 |
+
"loss": 1.6783,
|
545 |
+
"step": 425
|
546 |
+
},
|
547 |
+
{
|
548 |
+
"epoch": 3.52,
|
549 |
+
"learning_rate": 1.6312594840667682e-06,
|
550 |
+
"loss": 1.6432,
|
551 |
+
"step": 430
|
552 |
+
},
|
553 |
+
{
|
554 |
+
"epoch": 3.57,
|
555 |
+
"learning_rate": 1.6502276176024283e-06,
|
556 |
+
"loss": 1.6706,
|
557 |
+
"step": 435
|
558 |
+
},
|
559 |
+
{
|
560 |
+
"epoch": 3.61,
|
561 |
+
"learning_rate": 1.6691957511380882e-06,
|
562 |
+
"loss": 1.6687,
|
563 |
+
"step": 440
|
564 |
+
},
|
565 |
+
{
|
566 |
+
"epoch": 3.65,
|
567 |
+
"learning_rate": 1.6881638846737483e-06,
|
568 |
+
"loss": 1.6264,
|
569 |
+
"step": 445
|
570 |
+
},
|
571 |
+
{
|
572 |
+
"epoch": 3.69,
|
573 |
+
"learning_rate": 1.7071320182094084e-06,
|
574 |
+
"loss": 1.6226,
|
575 |
+
"step": 450
|
576 |
+
},
|
577 |
+
{
|
578 |
+
"epoch": 3.73,
|
579 |
+
"learning_rate": 1.7261001517450685e-06,
|
580 |
+
"loss": 1.7115,
|
581 |
+
"step": 455
|
582 |
+
},
|
583 |
+
{
|
584 |
+
"epoch": 3.77,
|
585 |
+
"learning_rate": 1.7450682852807286e-06,
|
586 |
+
"loss": 1.6301,
|
587 |
+
"step": 460
|
588 |
+
},
|
589 |
+
{
|
590 |
+
"epoch": 3.81,
|
591 |
+
"learning_rate": 1.7640364188163885e-06,
|
592 |
+
"loss": 1.6291,
|
593 |
+
"step": 465
|
594 |
+
},
|
595 |
+
{
|
596 |
+
"epoch": 3.85,
|
597 |
+
"learning_rate": 1.7830045523520486e-06,
|
598 |
+
"loss": 1.6008,
|
599 |
+
"step": 470
|
600 |
+
},
|
601 |
+
{
|
602 |
+
"epoch": 3.89,
|
603 |
+
"learning_rate": 1.8019726858877087e-06,
|
604 |
+
"loss": 1.5817,
|
605 |
+
"step": 475
|
606 |
+
},
|
607 |
+
{
|
608 |
+
"epoch": 3.93,
|
609 |
+
"learning_rate": 1.8209408194233688e-06,
|
610 |
+
"loss": 1.6208,
|
611 |
+
"step": 480
|
612 |
+
},
|
613 |
+
{
|
614 |
+
"epoch": 3.98,
|
615 |
+
"learning_rate": 1.839908952959029e-06,
|
616 |
+
"loss": 1.5952,
|
617 |
+
"step": 485
|
618 |
+
},
|
619 |
+
{
|
620 |
+
"epoch": 4.0,
|
621 |
+
"eval_accuracy": 0.48612538540596095,
|
622 |
+
"eval_loss": 1.5539824962615967,
|
623 |
+
"eval_runtime": 5.5954,
|
624 |
+
"eval_samples_per_second": 173.892,
|
625 |
+
"eval_steps_per_second": 10.902,
|
626 |
+
"step": 488
|
627 |
+
},
|
628 |
+
{
|
629 |
+
"epoch": 4.02,
|
630 |
+
"learning_rate": 1.858877086494689e-06,
|
631 |
+
"loss": 1.5884,
|
632 |
+
"step": 490
|
633 |
+
},
|
634 |
+
{
|
635 |
+
"epoch": 4.06,
|
636 |
+
"learning_rate": 1.8778452200303494e-06,
|
637 |
+
"loss": 1.5637,
|
638 |
+
"step": 495
|
639 |
+
},
|
640 |
+
{
|
641 |
+
"epoch": 4.1,
|
642 |
+
"learning_rate": 1.8968133535660093e-06,
|
643 |
+
"loss": 1.5989,
|
644 |
+
"step": 500
|
645 |
+
},
|
646 |
+
{
|
647 |
+
"epoch": 4.14,
|
648 |
+
"learning_rate": 1.9157814871016696e-06,
|
649 |
+
"loss": 1.5478,
|
650 |
+
"step": 505
|
651 |
+
},
|
652 |
+
{
|
653 |
+
"epoch": 4.18,
|
654 |
+
"learning_rate": 1.9347496206373295e-06,
|
655 |
+
"loss": 1.5079,
|
656 |
+
"step": 510
|
657 |
+
},
|
658 |
+
{
|
659 |
+
"epoch": 4.22,
|
660 |
+
"learning_rate": 1.9537177541729894e-06,
|
661 |
+
"loss": 1.5757,
|
662 |
+
"step": 515
|
663 |
+
},
|
664 |
+
{
|
665 |
+
"epoch": 4.26,
|
666 |
+
"learning_rate": 1.9726858877086497e-06,
|
667 |
+
"loss": 1.495,
|
668 |
+
"step": 520
|
669 |
+
},
|
670 |
+
{
|
671 |
+
"epoch": 4.3,
|
672 |
+
"learning_rate": 1.9916540212443096e-06,
|
673 |
+
"loss": 1.5159,
|
674 |
+
"step": 525
|
675 |
+
},
|
676 |
+
{
|
677 |
+
"epoch": 4.34,
|
678 |
+
"learning_rate": 2.01062215477997e-06,
|
679 |
+
"loss": 1.5065,
|
680 |
+
"step": 530
|
681 |
+
},
|
682 |
+
{
|
683 |
+
"epoch": 4.39,
|
684 |
+
"learning_rate": 2.02959028831563e-06,
|
685 |
+
"loss": 1.4111,
|
686 |
+
"step": 535
|
687 |
+
},
|
688 |
+
{
|
689 |
+
"epoch": 4.43,
|
690 |
+
"learning_rate": 2.04855842185129e-06,
|
691 |
+
"loss": 1.5118,
|
692 |
+
"step": 540
|
693 |
+
},
|
694 |
+
{
|
695 |
+
"epoch": 4.47,
|
696 |
+
"learning_rate": 2.06752655538695e-06,
|
697 |
+
"loss": 1.4989,
|
698 |
+
"step": 545
|
699 |
+
},
|
700 |
+
{
|
701 |
+
"epoch": 4.51,
|
702 |
+
"learning_rate": 2.08649468892261e-06,
|
703 |
+
"loss": 1.4438,
|
704 |
+
"step": 550
|
705 |
+
},
|
706 |
+
{
|
707 |
+
"epoch": 4.55,
|
708 |
+
"learning_rate": 2.1054628224582703e-06,
|
709 |
+
"loss": 1.469,
|
710 |
+
"step": 555
|
711 |
+
},
|
712 |
+
{
|
713 |
+
"epoch": 4.59,
|
714 |
+
"learning_rate": 2.12443095599393e-06,
|
715 |
+
"loss": 1.4363,
|
716 |
+
"step": 560
|
717 |
+
},
|
718 |
+
{
|
719 |
+
"epoch": 4.63,
|
720 |
+
"learning_rate": 2.1433990895295905e-06,
|
721 |
+
"loss": 1.3541,
|
722 |
+
"step": 565
|
723 |
+
},
|
724 |
+
{
|
725 |
+
"epoch": 4.67,
|
726 |
+
"learning_rate": 2.162367223065251e-06,
|
727 |
+
"loss": 1.4832,
|
728 |
+
"step": 570
|
729 |
+
},
|
730 |
+
{
|
731 |
+
"epoch": 4.71,
|
732 |
+
"learning_rate": 2.1813353566009107e-06,
|
733 |
+
"loss": 1.5036,
|
734 |
+
"step": 575
|
735 |
+
},
|
736 |
+
{
|
737 |
+
"epoch": 4.75,
|
738 |
+
"learning_rate": 2.2003034901365706e-06,
|
739 |
+
"loss": 1.3954,
|
740 |
+
"step": 580
|
741 |
+
},
|
742 |
+
{
|
743 |
+
"epoch": 4.8,
|
744 |
+
"learning_rate": 2.219271623672231e-06,
|
745 |
+
"loss": 1.4314,
|
746 |
+
"step": 585
|
747 |
+
},
|
748 |
+
{
|
749 |
+
"epoch": 4.84,
|
750 |
+
"learning_rate": 2.238239757207891e-06,
|
751 |
+
"loss": 1.3898,
|
752 |
+
"step": 590
|
753 |
+
},
|
754 |
+
{
|
755 |
+
"epoch": 4.88,
|
756 |
+
"learning_rate": 2.257207890743551e-06,
|
757 |
+
"loss": 1.3733,
|
758 |
+
"step": 595
|
759 |
+
},
|
760 |
+
{
|
761 |
+
"epoch": 4.92,
|
762 |
+
"learning_rate": 2.276176024279211e-06,
|
763 |
+
"loss": 1.4053,
|
764 |
+
"step": 600
|
765 |
+
},
|
766 |
+
{
|
767 |
+
"epoch": 4.96,
|
768 |
+
"learning_rate": 2.2951441578148714e-06,
|
769 |
+
"loss": 1.3945,
|
770 |
+
"step": 605
|
771 |
+
},
|
772 |
+
{
|
773 |
+
"epoch": 5.0,
|
774 |
+
"learning_rate": 2.3141122913505313e-06,
|
775 |
+
"loss": 1.3867,
|
776 |
+
"step": 610
|
777 |
+
},
|
778 |
+
{
|
779 |
+
"epoch": 5.0,
|
780 |
+
"eval_accuracy": 0.5549845837615622,
|
781 |
+
"eval_loss": 1.3568464517593384,
|
782 |
+
"eval_runtime": 5.4729,
|
783 |
+
"eval_samples_per_second": 177.784,
|
784 |
+
"eval_steps_per_second": 11.146,
|
785 |
+
"step": 610
|
786 |
+
},
|
787 |
+
{
|
788 |
+
"epoch": 5.04,
|
789 |
+
"learning_rate": 2.333080424886191e-06,
|
790 |
+
"loss": 1.3388,
|
791 |
+
"step": 615
|
792 |
+
},
|
793 |
+
{
|
794 |
+
"epoch": 5.08,
|
795 |
+
"learning_rate": 2.3520485584218515e-06,
|
796 |
+
"loss": 1.3141,
|
797 |
+
"step": 620
|
798 |
+
},
|
799 |
+
{
|
800 |
+
"epoch": 5.12,
|
801 |
+
"learning_rate": 2.3710166919575114e-06,
|
802 |
+
"loss": 1.337,
|
803 |
+
"step": 625
|
804 |
+
},
|
805 |
+
{
|
806 |
+
"epoch": 5.16,
|
807 |
+
"learning_rate": 2.3899848254931717e-06,
|
808 |
+
"loss": 1.3374,
|
809 |
+
"step": 630
|
810 |
+
},
|
811 |
+
{
|
812 |
+
"epoch": 5.2,
|
813 |
+
"learning_rate": 2.408952959028832e-06,
|
814 |
+
"loss": 1.3687,
|
815 |
+
"step": 635
|
816 |
+
},
|
817 |
+
{
|
818 |
+
"epoch": 5.25,
|
819 |
+
"learning_rate": 2.427921092564492e-06,
|
820 |
+
"loss": 1.3385,
|
821 |
+
"step": 640
|
822 |
+
},
|
823 |
+
{
|
824 |
+
"epoch": 5.29,
|
825 |
+
"learning_rate": 2.446889226100152e-06,
|
826 |
+
"loss": 1.2406,
|
827 |
+
"step": 645
|
828 |
+
},
|
829 |
+
{
|
830 |
+
"epoch": 5.33,
|
831 |
+
"learning_rate": 2.465857359635812e-06,
|
832 |
+
"loss": 1.2842,
|
833 |
+
"step": 650
|
834 |
+
},
|
835 |
+
{
|
836 |
+
"epoch": 5.37,
|
837 |
+
"learning_rate": 2.484825493171472e-06,
|
838 |
+
"loss": 1.2734,
|
839 |
+
"step": 655
|
840 |
+
},
|
841 |
+
{
|
842 |
+
"epoch": 5.41,
|
843 |
+
"learning_rate": 2.5037936267071324e-06,
|
844 |
+
"loss": 1.3318,
|
845 |
+
"step": 660
|
846 |
+
},
|
847 |
+
{
|
848 |
+
"epoch": 5.45,
|
849 |
+
"learning_rate": 2.5227617602427923e-06,
|
850 |
+
"loss": 1.2897,
|
851 |
+
"step": 665
|
852 |
+
},
|
853 |
+
{
|
854 |
+
"epoch": 5.49,
|
855 |
+
"learning_rate": 2.5417298937784526e-06,
|
856 |
+
"loss": 1.1894,
|
857 |
+
"step": 670
|
858 |
+
},
|
859 |
+
{
|
860 |
+
"epoch": 5.53,
|
861 |
+
"learning_rate": 2.5606980273141125e-06,
|
862 |
+
"loss": 1.2152,
|
863 |
+
"step": 675
|
864 |
+
},
|
865 |
+
{
|
866 |
+
"epoch": 5.57,
|
867 |
+
"learning_rate": 2.5796661608497724e-06,
|
868 |
+
"loss": 1.3102,
|
869 |
+
"step": 680
|
870 |
+
},
|
871 |
+
{
|
872 |
+
"epoch": 5.61,
|
873 |
+
"learning_rate": 2.5986342943854327e-06,
|
874 |
+
"loss": 1.2221,
|
875 |
+
"step": 685
|
876 |
+
},
|
877 |
+
{
|
878 |
+
"epoch": 5.66,
|
879 |
+
"learning_rate": 2.6176024279210926e-06,
|
880 |
+
"loss": 1.2936,
|
881 |
+
"step": 690
|
882 |
+
},
|
883 |
+
{
|
884 |
+
"epoch": 5.7,
|
885 |
+
"learning_rate": 2.636570561456753e-06,
|
886 |
+
"loss": 1.2431,
|
887 |
+
"step": 695
|
888 |
+
},
|
889 |
+
{
|
890 |
+
"epoch": 5.74,
|
891 |
+
"learning_rate": 2.655538694992413e-06,
|
892 |
+
"loss": 1.1876,
|
893 |
+
"step": 700
|
894 |
+
},
|
895 |
+
{
|
896 |
+
"epoch": 5.78,
|
897 |
+
"learning_rate": 2.6745068285280727e-06,
|
898 |
+
"loss": 1.1703,
|
899 |
+
"step": 705
|
900 |
+
},
|
901 |
+
{
|
902 |
+
"epoch": 5.82,
|
903 |
+
"learning_rate": 2.693474962063733e-06,
|
904 |
+
"loss": 1.1677,
|
905 |
+
"step": 710
|
906 |
+
},
|
907 |
+
{
|
908 |
+
"epoch": 5.86,
|
909 |
+
"learning_rate": 2.712443095599393e-06,
|
910 |
+
"loss": 1.215,
|
911 |
+
"step": 715
|
912 |
+
},
|
913 |
+
{
|
914 |
+
"epoch": 5.9,
|
915 |
+
"learning_rate": 2.7314112291350532e-06,
|
916 |
+
"loss": 1.2071,
|
917 |
+
"step": 720
|
918 |
+
},
|
919 |
+
{
|
920 |
+
"epoch": 5.94,
|
921 |
+
"learning_rate": 2.7503793626707136e-06,
|
922 |
+
"loss": 1.2239,
|
923 |
+
"step": 725
|
924 |
+
},
|
925 |
+
{
|
926 |
+
"epoch": 5.98,
|
927 |
+
"learning_rate": 2.769347496206374e-06,
|
928 |
+
"loss": 1.1846,
|
929 |
+
"step": 730
|
930 |
+
},
|
931 |
+
{
|
932 |
+
"epoch": 6.0,
|
933 |
+
"eval_accuracy": 0.6372045220966084,
|
934 |
+
"eval_loss": 1.1521209478378296,
|
935 |
+
"eval_runtime": 5.5004,
|
936 |
+
"eval_samples_per_second": 176.897,
|
937 |
+
"eval_steps_per_second": 11.09,
|
938 |
+
"step": 732
|
939 |
+
},
|
940 |
+
{
|
941 |
+
"epoch": 6.02,
|
942 |
+
"learning_rate": 2.7883156297420338e-06,
|
943 |
+
"loss": 1.2198,
|
944 |
+
"step": 735
|
945 |
+
},
|
946 |
+
{
|
947 |
+
"epoch": 6.07,
|
948 |
+
"learning_rate": 2.807283763277694e-06,
|
949 |
+
"loss": 1.1424,
|
950 |
+
"step": 740
|
951 |
+
},
|
952 |
+
{
|
953 |
+
"epoch": 6.11,
|
954 |
+
"learning_rate": 2.826251896813354e-06,
|
955 |
+
"loss": 1.1408,
|
956 |
+
"step": 745
|
957 |
+
},
|
958 |
+
{
|
959 |
+
"epoch": 6.15,
|
960 |
+
"learning_rate": 2.845220030349014e-06,
|
961 |
+
"loss": 1.1362,
|
962 |
+
"step": 750
|
963 |
+
},
|
964 |
+
{
|
965 |
+
"epoch": 6.19,
|
966 |
+
"learning_rate": 2.8641881638846742e-06,
|
967 |
+
"loss": 1.1493,
|
968 |
+
"step": 755
|
969 |
+
},
|
970 |
+
{
|
971 |
+
"epoch": 6.23,
|
972 |
+
"learning_rate": 2.883156297420334e-06,
|
973 |
+
"loss": 1.0286,
|
974 |
+
"step": 760
|
975 |
+
},
|
976 |
+
{
|
977 |
+
"epoch": 6.27,
|
978 |
+
"learning_rate": 2.9021244309559944e-06,
|
979 |
+
"loss": 1.1429,
|
980 |
+
"step": 765
|
981 |
+
},
|
982 |
+
{
|
983 |
+
"epoch": 6.31,
|
984 |
+
"learning_rate": 2.9210925644916543e-06,
|
985 |
+
"loss": 1.1113,
|
986 |
+
"step": 770
|
987 |
+
},
|
988 |
+
{
|
989 |
+
"epoch": 6.35,
|
990 |
+
"learning_rate": 2.9400606980273142e-06,
|
991 |
+
"loss": 1.1336,
|
992 |
+
"step": 775
|
993 |
+
},
|
994 |
+
{
|
995 |
+
"epoch": 6.39,
|
996 |
+
"learning_rate": 2.9590288315629746e-06,
|
997 |
+
"loss": 1.1171,
|
998 |
+
"step": 780
|
999 |
+
},
|
1000 |
+
{
|
1001 |
+
"epoch": 6.43,
|
1002 |
+
"learning_rate": 2.9779969650986345e-06,
|
1003 |
+
"loss": 1.2444,
|
1004 |
+
"step": 785
|
1005 |
+
},
|
1006 |
+
{
|
1007 |
+
"epoch": 6.48,
|
1008 |
+
"learning_rate": 2.9969650986342948e-06,
|
1009 |
+
"loss": 1.0057,
|
1010 |
+
"step": 790
|
1011 |
+
},
|
1012 |
+
{
|
1013 |
+
"epoch": 6.52,
|
1014 |
+
"learning_rate": 3.0159332321699547e-06,
|
1015 |
+
"loss": 1.0284,
|
1016 |
+
"step": 795
|
1017 |
+
},
|
1018 |
+
{
|
1019 |
+
"epoch": 6.56,
|
1020 |
+
"learning_rate": 3.0349013657056146e-06,
|
1021 |
+
"loss": 1.0093,
|
1022 |
+
"step": 800
|
1023 |
+
},
|
1024 |
+
{
|
1025 |
+
"epoch": 6.6,
|
1026 |
+
"learning_rate": 3.053869499241275e-06,
|
1027 |
+
"loss": 1.1139,
|
1028 |
+
"step": 805
|
1029 |
+
},
|
1030 |
+
{
|
1031 |
+
"epoch": 6.64,
|
1032 |
+
"learning_rate": 3.072837632776935e-06,
|
1033 |
+
"loss": 1.0426,
|
1034 |
+
"step": 810
|
1035 |
+
},
|
1036 |
+
{
|
1037 |
+
"epoch": 6.68,
|
1038 |
+
"learning_rate": 3.091805766312595e-06,
|
1039 |
+
"loss": 1.0411,
|
1040 |
+
"step": 815
|
1041 |
+
},
|
1042 |
+
{
|
1043 |
+
"epoch": 6.72,
|
1044 |
+
"learning_rate": 3.110773899848255e-06,
|
1045 |
+
"loss": 0.9774,
|
1046 |
+
"step": 820
|
1047 |
+
},
|
1048 |
+
{
|
1049 |
+
"epoch": 6.76,
|
1050 |
+
"learning_rate": 3.1297420333839153e-06,
|
1051 |
+
"loss": 1.0463,
|
1052 |
+
"step": 825
|
1053 |
+
},
|
1054 |
+
{
|
1055 |
+
"epoch": 6.8,
|
1056 |
+
"learning_rate": 3.1487101669195752e-06,
|
1057 |
+
"loss": 1.0936,
|
1058 |
+
"step": 830
|
1059 |
+
},
|
1060 |
+
{
|
1061 |
+
"epoch": 6.84,
|
1062 |
+
"learning_rate": 3.167678300455235e-06,
|
1063 |
+
"loss": 1.0528,
|
1064 |
+
"step": 835
|
1065 |
+
},
|
1066 |
+
{
|
1067 |
+
"epoch": 6.89,
|
1068 |
+
"learning_rate": 3.1866464339908955e-06,
|
1069 |
+
"loss": 0.9179,
|
1070 |
+
"step": 840
|
1071 |
+
},
|
1072 |
+
{
|
1073 |
+
"epoch": 6.93,
|
1074 |
+
"learning_rate": 3.2056145675265554e-06,
|
1075 |
+
"loss": 1.0432,
|
1076 |
+
"step": 845
|
1077 |
+
},
|
1078 |
+
{
|
1079 |
+
"epoch": 6.97,
|
1080 |
+
"learning_rate": 3.2245827010622157e-06,
|
1081 |
+
"loss": 1.0063,
|
1082 |
+
"step": 850
|
1083 |
+
},
|
1084 |
+
{
|
1085 |
+
"epoch": 7.0,
|
1086 |
+
"eval_accuracy": 0.6824254881808839,
|
1087 |
+
"eval_loss": 0.9649463891983032,
|
1088 |
+
"eval_runtime": 5.5867,
|
1089 |
+
"eval_samples_per_second": 174.163,
|
1090 |
+
"eval_steps_per_second": 10.919,
|
1091 |
+
"step": 854
|
1092 |
+
},
|
1093 |
+
{
|
1094 |
+
"epoch": 7.01,
|
1095 |
+
"learning_rate": 3.2435508345978756e-06,
|
1096 |
+
"loss": 0.9927,
|
1097 |
+
"step": 855
|
1098 |
+
},
|
1099 |
+
{
|
1100 |
+
"epoch": 7.05,
|
1101 |
+
"learning_rate": 3.2625189681335363e-06,
|
1102 |
+
"loss": 0.8729,
|
1103 |
+
"step": 860
|
1104 |
+
},
|
1105 |
+
{
|
1106 |
+
"epoch": 7.09,
|
1107 |
+
"learning_rate": 3.2814871016691962e-06,
|
1108 |
+
"loss": 0.938,
|
1109 |
+
"step": 865
|
1110 |
+
},
|
1111 |
+
{
|
1112 |
+
"epoch": 7.13,
|
1113 |
+
"learning_rate": 3.3004552352048565e-06,
|
1114 |
+
"loss": 0.9457,
|
1115 |
+
"step": 870
|
1116 |
+
},
|
1117 |
+
{
|
1118 |
+
"epoch": 7.17,
|
1119 |
+
"learning_rate": 3.3194233687405164e-06,
|
1120 |
+
"loss": 0.9752,
|
1121 |
+
"step": 875
|
1122 |
+
},
|
1123 |
+
{
|
1124 |
+
"epoch": 7.21,
|
1125 |
+
"learning_rate": 3.3383915022761763e-06,
|
1126 |
+
"loss": 0.979,
|
1127 |
+
"step": 880
|
1128 |
+
},
|
1129 |
+
{
|
1130 |
+
"epoch": 7.25,
|
1131 |
+
"learning_rate": 3.3573596358118367e-06,
|
1132 |
+
"loss": 0.9645,
|
1133 |
+
"step": 885
|
1134 |
+
},
|
1135 |
+
{
|
1136 |
+
"epoch": 7.3,
|
1137 |
+
"learning_rate": 3.3763277693474965e-06,
|
1138 |
+
"loss": 0.9079,
|
1139 |
+
"step": 890
|
1140 |
+
},
|
1141 |
+
{
|
1142 |
+
"epoch": 7.34,
|
1143 |
+
"learning_rate": 3.395295902883157e-06,
|
1144 |
+
"loss": 0.9343,
|
1145 |
+
"step": 895
|
1146 |
+
},
|
1147 |
+
{
|
1148 |
+
"epoch": 7.38,
|
1149 |
+
"learning_rate": 3.4142640364188168e-06,
|
1150 |
+
"loss": 0.9246,
|
1151 |
+
"step": 900
|
1152 |
+
},
|
1153 |
+
{
|
1154 |
+
"epoch": 7.42,
|
1155 |
+
"learning_rate": 3.4332321699544767e-06,
|
1156 |
+
"loss": 0.9111,
|
1157 |
+
"step": 905
|
1158 |
+
},
|
1159 |
+
{
|
1160 |
+
"epoch": 7.46,
|
1161 |
+
"learning_rate": 3.452200303490137e-06,
|
1162 |
+
"loss": 1.0186,
|
1163 |
+
"step": 910
|
1164 |
+
},
|
1165 |
+
{
|
1166 |
+
"epoch": 7.5,
|
1167 |
+
"learning_rate": 3.471168437025797e-06,
|
1168 |
+
"loss": 0.8782,
|
1169 |
+
"step": 915
|
1170 |
+
},
|
1171 |
+
{
|
1172 |
+
"epoch": 7.54,
|
1173 |
+
"learning_rate": 3.490136570561457e-06,
|
1174 |
+
"loss": 0.9733,
|
1175 |
+
"step": 920
|
1176 |
+
},
|
1177 |
+
{
|
1178 |
+
"epoch": 7.58,
|
1179 |
+
"learning_rate": 3.509104704097117e-06,
|
1180 |
+
"loss": 0.9269,
|
1181 |
+
"step": 925
|
1182 |
+
},
|
1183 |
+
{
|
1184 |
+
"epoch": 7.62,
|
1185 |
+
"learning_rate": 3.528072837632777e-06,
|
1186 |
+
"loss": 0.9199,
|
1187 |
+
"step": 930
|
1188 |
+
},
|
1189 |
+
{
|
1190 |
+
"epoch": 7.66,
|
1191 |
+
"learning_rate": 3.5470409711684373e-06,
|
1192 |
+
"loss": 0.8964,
|
1193 |
+
"step": 935
|
1194 |
+
},
|
1195 |
+
{
|
1196 |
+
"epoch": 7.7,
|
1197 |
+
"learning_rate": 3.5660091047040972e-06,
|
1198 |
+
"loss": 0.849,
|
1199 |
+
"step": 940
|
1200 |
+
},
|
1201 |
+
{
|
1202 |
+
"epoch": 7.75,
|
1203 |
+
"learning_rate": 3.5849772382397575e-06,
|
1204 |
+
"loss": 0.9597,
|
1205 |
+
"step": 945
|
1206 |
+
},
|
1207 |
+
{
|
1208 |
+
"epoch": 7.79,
|
1209 |
+
"learning_rate": 3.6039453717754174e-06,
|
1210 |
+
"loss": 0.8816,
|
1211 |
+
"step": 950
|
1212 |
+
},
|
1213 |
+
{
|
1214 |
+
"epoch": 7.83,
|
1215 |
+
"learning_rate": 3.6229135053110773e-06,
|
1216 |
+
"loss": 0.8838,
|
1217 |
+
"step": 955
|
1218 |
+
},
|
1219 |
+
{
|
1220 |
+
"epoch": 7.87,
|
1221 |
+
"learning_rate": 3.6418816388467377e-06,
|
1222 |
+
"loss": 0.8685,
|
1223 |
+
"step": 960
|
1224 |
+
},
|
1225 |
+
{
|
1226 |
+
"epoch": 7.91,
|
1227 |
+
"learning_rate": 3.6608497723823976e-06,
|
1228 |
+
"loss": 0.8977,
|
1229 |
+
"step": 965
|
1230 |
+
},
|
1231 |
+
{
|
1232 |
+
"epoch": 7.95,
|
1233 |
+
"learning_rate": 3.679817905918058e-06,
|
1234 |
+
"loss": 0.8598,
|
1235 |
+
"step": 970
|
1236 |
+
},
|
1237 |
+
{
|
1238 |
+
"epoch": 7.99,
|
1239 |
+
"learning_rate": 3.6987860394537178e-06,
|
1240 |
+
"loss": 0.8932,
|
1241 |
+
"step": 975
|
1242 |
+
},
|
1243 |
+
{
|
1244 |
+
"epoch": 8.0,
|
1245 |
+
"eval_accuracy": 0.7307297019527236,
|
1246 |
+
"eval_loss": 0.8284448981285095,
|
1247 |
+
"eval_runtime": 5.579,
|
1248 |
+
"eval_samples_per_second": 174.404,
|
1249 |
+
"eval_steps_per_second": 10.934,
|
1250 |
+
"step": 976
|
1251 |
+
},
|
1252 |
+
{
|
1253 |
+
"epoch": 8.03,
|
1254 |
+
"learning_rate": 3.717754172989378e-06,
|
1255 |
+
"loss": 0.8395,
|
1256 |
+
"step": 980
|
1257 |
+
},
|
1258 |
+
{
|
1259 |
+
"epoch": 8.07,
|
1260 |
+
"learning_rate": 3.736722306525038e-06,
|
1261 |
+
"loss": 0.827,
|
1262 |
+
"step": 985
|
1263 |
+
},
|
1264 |
+
{
|
1265 |
+
"epoch": 8.11,
|
1266 |
+
"learning_rate": 3.7556904400606987e-06,
|
1267 |
+
"loss": 0.8556,
|
1268 |
+
"step": 990
|
1269 |
+
},
|
1270 |
+
{
|
1271 |
+
"epoch": 8.16,
|
1272 |
+
"learning_rate": 3.7746585735963586e-06,
|
1273 |
+
"loss": 0.7637,
|
1274 |
+
"step": 995
|
1275 |
+
},
|
1276 |
+
{
|
1277 |
+
"epoch": 8.2,
|
1278 |
+
"learning_rate": 3.7936267071320185e-06,
|
1279 |
+
"loss": 0.8405,
|
1280 |
+
"step": 1000
|
1281 |
+
},
|
1282 |
+
{
|
1283 |
+
"epoch": 8.24,
|
1284 |
+
"learning_rate": 3.812594840667679e-06,
|
1285 |
+
"loss": 0.8749,
|
1286 |
+
"step": 1005
|
1287 |
+
},
|
1288 |
+
{
|
1289 |
+
"epoch": 8.28,
|
1290 |
+
"learning_rate": 3.831562974203339e-06,
|
1291 |
+
"loss": 0.7677,
|
1292 |
+
"step": 1010
|
1293 |
+
},
|
1294 |
+
{
|
1295 |
+
"epoch": 8.32,
|
1296 |
+
"learning_rate": 3.850531107738999e-06,
|
1297 |
+
"loss": 0.8192,
|
1298 |
+
"step": 1015
|
1299 |
+
},
|
1300 |
+
{
|
1301 |
+
"epoch": 8.36,
|
1302 |
+
"learning_rate": 3.869499241274659e-06,
|
1303 |
+
"loss": 0.8431,
|
1304 |
+
"step": 1020
|
1305 |
+
},
|
1306 |
+
{
|
1307 |
+
"epoch": 8.4,
|
1308 |
+
"learning_rate": 3.888467374810319e-06,
|
1309 |
+
"loss": 0.7641,
|
1310 |
+
"step": 1025
|
1311 |
+
},
|
1312 |
+
{
|
1313 |
+
"epoch": 8.44,
|
1314 |
+
"learning_rate": 3.907435508345979e-06,
|
1315 |
+
"loss": 0.8932,
|
1316 |
+
"step": 1030
|
1317 |
+
},
|
1318 |
+
{
|
1319 |
+
"epoch": 8.48,
|
1320 |
+
"learning_rate": 3.9264036418816395e-06,
|
1321 |
+
"loss": 0.8155,
|
1322 |
+
"step": 1035
|
1323 |
+
},
|
1324 |
+
{
|
1325 |
+
"epoch": 8.52,
|
1326 |
+
"learning_rate": 3.945371775417299e-06,
|
1327 |
+
"loss": 0.789,
|
1328 |
+
"step": 1040
|
1329 |
+
},
|
1330 |
+
{
|
1331 |
+
"epoch": 8.57,
|
1332 |
+
"learning_rate": 3.964339908952959e-06,
|
1333 |
+
"loss": 0.8227,
|
1334 |
+
"step": 1045
|
1335 |
+
},
|
1336 |
+
{
|
1337 |
+
"epoch": 8.61,
|
1338 |
+
"learning_rate": 3.983308042488619e-06,
|
1339 |
+
"loss": 0.788,
|
1340 |
+
"step": 1050
|
1341 |
+
},
|
1342 |
+
{
|
1343 |
+
"epoch": 8.65,
|
1344 |
+
"learning_rate": 4.00227617602428e-06,
|
1345 |
+
"loss": 0.8094,
|
1346 |
+
"step": 1055
|
1347 |
+
},
|
1348 |
+
{
|
1349 |
+
"epoch": 8.69,
|
1350 |
+
"learning_rate": 4.02124430955994e-06,
|
1351 |
+
"loss": 0.7638,
|
1352 |
+
"step": 1060
|
1353 |
+
},
|
1354 |
+
{
|
1355 |
+
"epoch": 8.73,
|
1356 |
+
"learning_rate": 4.0402124430956e-06,
|
1357 |
+
"loss": 0.8325,
|
1358 |
+
"step": 1065
|
1359 |
+
},
|
1360 |
+
{
|
1361 |
+
"epoch": 8.77,
|
1362 |
+
"learning_rate": 4.05918057663126e-06,
|
1363 |
+
"loss": 0.7966,
|
1364 |
+
"step": 1070
|
1365 |
+
},
|
1366 |
+
{
|
1367 |
+
"epoch": 8.81,
|
1368 |
+
"learning_rate": 4.0781487101669195e-06,
|
1369 |
+
"loss": 0.7768,
|
1370 |
+
"step": 1075
|
1371 |
+
},
|
1372 |
+
{
|
1373 |
+
"epoch": 8.85,
|
1374 |
+
"learning_rate": 4.09711684370258e-06,
|
1375 |
+
"loss": 0.7694,
|
1376 |
+
"step": 1080
|
1377 |
+
},
|
1378 |
+
{
|
1379 |
+
"epoch": 8.89,
|
1380 |
+
"learning_rate": 4.11608497723824e-06,
|
1381 |
+
"loss": 0.8056,
|
1382 |
+
"step": 1085
|
1383 |
+
},
|
1384 |
+
{
|
1385 |
+
"epoch": 8.93,
|
1386 |
+
"learning_rate": 4.1350531107739e-06,
|
1387 |
+
"loss": 0.9008,
|
1388 |
+
"step": 1090
|
1389 |
+
},
|
1390 |
+
{
|
1391 |
+
"epoch": 8.98,
|
1392 |
+
"learning_rate": 4.15402124430956e-06,
|
1393 |
+
"loss": 0.7576,
|
1394 |
+
"step": 1095
|
1395 |
+
},
|
1396 |
+
{
|
1397 |
+
"epoch": 9.0,
|
1398 |
+
"eval_accuracy": 0.7780061664953751,
|
1399 |
+
"eval_loss": 0.7217260003089905,
|
1400 |
+
"eval_runtime": 5.5246,
|
1401 |
+
"eval_samples_per_second": 176.122,
|
1402 |
+
"eval_steps_per_second": 11.042,
|
1403 |
+
"step": 1098
|
1404 |
+
},
|
1405 |
+
{
|
1406 |
+
"epoch": 9.02,
|
1407 |
+
"learning_rate": 4.17298937784522e-06,
|
1408 |
+
"loss": 0.8358,
|
1409 |
+
"step": 1100
|
1410 |
+
},
|
1411 |
+
{
|
1412 |
+
"epoch": 9.06,
|
1413 |
+
"learning_rate": 4.191957511380881e-06,
|
1414 |
+
"loss": 0.8025,
|
1415 |
+
"step": 1105
|
1416 |
+
},
|
1417 |
+
{
|
1418 |
+
"epoch": 9.1,
|
1419 |
+
"learning_rate": 4.2109256449165405e-06,
|
1420 |
+
"loss": 0.7243,
|
1421 |
+
"step": 1110
|
1422 |
+
},
|
1423 |
+
{
|
1424 |
+
"epoch": 9.14,
|
1425 |
+
"learning_rate": 4.2298937784522004e-06,
|
1426 |
+
"loss": 0.7422,
|
1427 |
+
"step": 1115
|
1428 |
+
},
|
1429 |
+
{
|
1430 |
+
"epoch": 9.18,
|
1431 |
+
"learning_rate": 4.24886191198786e-06,
|
1432 |
+
"loss": 0.6503,
|
1433 |
+
"step": 1120
|
1434 |
+
},
|
1435 |
+
{
|
1436 |
+
"epoch": 9.22,
|
1437 |
+
"learning_rate": 4.267830045523521e-06,
|
1438 |
+
"loss": 0.8027,
|
1439 |
+
"step": 1125
|
1440 |
+
},
|
1441 |
+
{
|
1442 |
+
"epoch": 9.26,
|
1443 |
+
"learning_rate": 4.286798179059181e-06,
|
1444 |
+
"loss": 0.7721,
|
1445 |
+
"step": 1130
|
1446 |
+
},
|
1447 |
+
{
|
1448 |
+
"epoch": 9.3,
|
1449 |
+
"learning_rate": 4.305766312594841e-06,
|
1450 |
+
"loss": 0.7889,
|
1451 |
+
"step": 1135
|
1452 |
+
},
|
1453 |
+
{
|
1454 |
+
"epoch": 9.34,
|
1455 |
+
"learning_rate": 4.324734446130502e-06,
|
1456 |
+
"loss": 0.6659,
|
1457 |
+
"step": 1140
|
1458 |
+
},
|
1459 |
+
{
|
1460 |
+
"epoch": 9.39,
|
1461 |
+
"learning_rate": 4.3437025796661615e-06,
|
1462 |
+
"loss": 0.7961,
|
1463 |
+
"step": 1145
|
1464 |
+
},
|
1465 |
+
{
|
1466 |
+
"epoch": 9.43,
|
1467 |
+
"learning_rate": 4.362670713201821e-06,
|
1468 |
+
"loss": 0.7102,
|
1469 |
+
"step": 1150
|
1470 |
+
},
|
1471 |
+
{
|
1472 |
+
"epoch": 9.47,
|
1473 |
+
"learning_rate": 4.381638846737481e-06,
|
1474 |
+
"loss": 0.7001,
|
1475 |
+
"step": 1155
|
1476 |
+
},
|
1477 |
+
{
|
1478 |
+
"epoch": 9.51,
|
1479 |
+
"learning_rate": 4.400606980273141e-06,
|
1480 |
+
"loss": 0.7438,
|
1481 |
+
"step": 1160
|
1482 |
+
},
|
1483 |
+
{
|
1484 |
+
"epoch": 9.55,
|
1485 |
+
"learning_rate": 4.419575113808802e-06,
|
1486 |
+
"loss": 0.8474,
|
1487 |
+
"step": 1165
|
1488 |
+
},
|
1489 |
+
{
|
1490 |
+
"epoch": 9.59,
|
1491 |
+
"learning_rate": 4.438543247344462e-06,
|
1492 |
+
"loss": 0.6325,
|
1493 |
+
"step": 1170
|
1494 |
+
},
|
1495 |
+
{
|
1496 |
+
"epoch": 9.63,
|
1497 |
+
"learning_rate": 4.457511380880122e-06,
|
1498 |
+
"loss": 0.6162,
|
1499 |
+
"step": 1175
|
1500 |
+
},
|
1501 |
+
{
|
1502 |
+
"epoch": 9.67,
|
1503 |
+
"learning_rate": 4.476479514415782e-06,
|
1504 |
+
"loss": 0.7161,
|
1505 |
+
"step": 1180
|
1506 |
+
},
|
1507 |
+
{
|
1508 |
+
"epoch": 9.71,
|
1509 |
+
"learning_rate": 4.4954476479514415e-06,
|
1510 |
+
"loss": 0.6548,
|
1511 |
+
"step": 1185
|
1512 |
+
},
|
1513 |
+
{
|
1514 |
+
"epoch": 9.75,
|
1515 |
+
"learning_rate": 4.514415781487102e-06,
|
1516 |
+
"loss": 0.7685,
|
1517 |
+
"step": 1190
|
1518 |
+
},
|
1519 |
+
{
|
1520 |
+
"epoch": 9.8,
|
1521 |
+
"learning_rate": 4.533383915022762e-06,
|
1522 |
+
"loss": 0.7216,
|
1523 |
+
"step": 1195
|
1524 |
+
},
|
1525 |
+
{
|
1526 |
+
"epoch": 9.84,
|
1527 |
+
"learning_rate": 4.552352048558422e-06,
|
1528 |
+
"loss": 0.7184,
|
1529 |
+
"step": 1200
|
1530 |
+
},
|
1531 |
+
{
|
1532 |
+
"epoch": 9.88,
|
1533 |
+
"learning_rate": 4.571320182094082e-06,
|
1534 |
+
"loss": 0.7909,
|
1535 |
+
"step": 1205
|
1536 |
+
},
|
1537 |
+
{
|
1538 |
+
"epoch": 9.92,
|
1539 |
+
"learning_rate": 4.590288315629743e-06,
|
1540 |
+
"loss": 0.6041,
|
1541 |
+
"step": 1210
|
1542 |
+
},
|
1543 |
+
{
|
1544 |
+
"epoch": 9.96,
|
1545 |
+
"learning_rate": 4.609256449165403e-06,
|
1546 |
+
"loss": 0.6572,
|
1547 |
+
"step": 1215
|
1548 |
+
},
|
1549 |
+
{
|
1550 |
+
"epoch": 10.0,
|
1551 |
+
"learning_rate": 4.6282245827010625e-06,
|
1552 |
+
"loss": 0.6742,
|
1553 |
+
"step": 1220
|
1554 |
+
},
|
1555 |
+
{
|
1556 |
+
"epoch": 10.0,
|
1557 |
+
"eval_accuracy": 0.7923946557040082,
|
1558 |
+
"eval_loss": 0.6411532759666443,
|
1559 |
+
"eval_runtime": 5.5151,
|
1560 |
+
"eval_samples_per_second": 176.423,
|
1561 |
+
"eval_steps_per_second": 11.06,
|
1562 |
+
"step": 1220
|
1563 |
+
},
|
1564 |
+
{
|
1565 |
+
"epoch": 10.04,
|
1566 |
+
"learning_rate": 4.647192716236722e-06,
|
1567 |
+
"loss": 0.6705,
|
1568 |
+
"step": 1225
|
1569 |
+
},
|
1570 |
+
{
|
1571 |
+
"epoch": 10.08,
|
1572 |
+
"learning_rate": 4.666160849772382e-06,
|
1573 |
+
"loss": 0.8139,
|
1574 |
+
"step": 1230
|
1575 |
+
},
|
1576 |
+
{
|
1577 |
+
"epoch": 10.12,
|
1578 |
+
"learning_rate": 4.685128983308043e-06,
|
1579 |
+
"loss": 0.7255,
|
1580 |
+
"step": 1235
|
1581 |
+
},
|
1582 |
+
{
|
1583 |
+
"epoch": 10.16,
|
1584 |
+
"learning_rate": 4.704097116843703e-06,
|
1585 |
+
"loss": 0.6437,
|
1586 |
+
"step": 1240
|
1587 |
+
},
|
1588 |
+
{
|
1589 |
+
"epoch": 10.2,
|
1590 |
+
"learning_rate": 4.723065250379363e-06,
|
1591 |
+
"loss": 0.6594,
|
1592 |
+
"step": 1245
|
1593 |
+
},
|
1594 |
+
{
|
1595 |
+
"epoch": 10.25,
|
1596 |
+
"learning_rate": 4.742033383915023e-06,
|
1597 |
+
"loss": 0.7319,
|
1598 |
+
"step": 1250
|
1599 |
+
},
|
1600 |
+
{
|
1601 |
+
"epoch": 10.29,
|
1602 |
+
"learning_rate": 4.7610015174506835e-06,
|
1603 |
+
"loss": 0.6408,
|
1604 |
+
"step": 1255
|
1605 |
+
},
|
1606 |
+
{
|
1607 |
+
"epoch": 10.33,
|
1608 |
+
"learning_rate": 4.779969650986343e-06,
|
1609 |
+
"loss": 0.6922,
|
1610 |
+
"step": 1260
|
1611 |
+
},
|
1612 |
+
{
|
1613 |
+
"epoch": 10.37,
|
1614 |
+
"learning_rate": 4.798937784522003e-06,
|
1615 |
+
"loss": 0.5613,
|
1616 |
+
"step": 1265
|
1617 |
+
},
|
1618 |
+
{
|
1619 |
+
"epoch": 10.41,
|
1620 |
+
"learning_rate": 4.817905918057664e-06,
|
1621 |
+
"loss": 0.6178,
|
1622 |
+
"step": 1270
|
1623 |
+
},
|
1624 |
+
{
|
1625 |
+
"epoch": 10.45,
|
1626 |
+
"learning_rate": 4.836874051593324e-06,
|
1627 |
+
"loss": 0.651,
|
1628 |
+
"step": 1275
|
1629 |
+
},
|
1630 |
+
{
|
1631 |
+
"epoch": 10.49,
|
1632 |
+
"learning_rate": 4.855842185128984e-06,
|
1633 |
+
"loss": 0.6139,
|
1634 |
+
"step": 1280
|
1635 |
+
},
|
1636 |
+
{
|
1637 |
+
"epoch": 10.53,
|
1638 |
+
"learning_rate": 4.874810318664644e-06,
|
1639 |
+
"loss": 0.5995,
|
1640 |
+
"step": 1285
|
1641 |
+
},
|
1642 |
+
{
|
1643 |
+
"epoch": 10.57,
|
1644 |
+
"learning_rate": 4.893778452200304e-06,
|
1645 |
+
"loss": 0.6759,
|
1646 |
+
"step": 1290
|
1647 |
+
},
|
1648 |
+
{
|
1649 |
+
"epoch": 10.61,
|
1650 |
+
"learning_rate": 4.912746585735964e-06,
|
1651 |
+
"loss": 0.5631,
|
1652 |
+
"step": 1295
|
1653 |
+
},
|
1654 |
+
{
|
1655 |
+
"epoch": 10.66,
|
1656 |
+
"learning_rate": 4.931714719271624e-06,
|
1657 |
+
"loss": 0.6437,
|
1658 |
+
"step": 1300
|
1659 |
+
},
|
1660 |
+
{
|
1661 |
+
"epoch": 10.7,
|
1662 |
+
"learning_rate": 4.950682852807284e-06,
|
1663 |
+
"loss": 0.6328,
|
1664 |
+
"step": 1305
|
1665 |
+
},
|
1666 |
+
{
|
1667 |
+
"epoch": 10.74,
|
1668 |
+
"learning_rate": 4.969650986342944e-06,
|
1669 |
+
"loss": 0.5672,
|
1670 |
+
"step": 1310
|
1671 |
+
},
|
1672 |
+
{
|
1673 |
+
"epoch": 10.78,
|
1674 |
+
"learning_rate": 4.988619119878604e-06,
|
1675 |
+
"loss": 0.6716,
|
1676 |
+
"step": 1315
|
1677 |
+
},
|
1678 |
+
{
|
1679 |
+
"epoch": 10.82,
|
1680 |
+
"learning_rate": 4.931506849315069e-06,
|
1681 |
+
"loss": 0.6037,
|
1682 |
+
"step": 1320
|
1683 |
+
},
|
1684 |
+
{
|
1685 |
+
"epoch": 10.86,
|
1686 |
+
"learning_rate": 4.76027397260274e-06,
|
1687 |
+
"loss": 0.6432,
|
1688 |
+
"step": 1325
|
1689 |
+
},
|
1690 |
+
{
|
1691 |
+
"epoch": 10.9,
|
1692 |
+
"learning_rate": 4.589041095890411e-06,
|
1693 |
+
"loss": 0.7613,
|
1694 |
+
"step": 1330
|
1695 |
+
},
|
1696 |
+
{
|
1697 |
+
"epoch": 10.94,
|
1698 |
+
"learning_rate": 4.4178082191780825e-06,
|
1699 |
+
"loss": 0.5567,
|
1700 |
+
"step": 1335
|
1701 |
+
},
|
1702 |
+
{
|
1703 |
+
"epoch": 10.98,
|
1704 |
+
"learning_rate": 4.246575342465754e-06,
|
1705 |
+
"loss": 0.6398,
|
1706 |
+
"step": 1340
|
1707 |
+
},
|
1708 |
+
{
|
1709 |
+
"epoch": 11.0,
|
1710 |
+
"eval_accuracy": 0.816032887975334,
|
1711 |
+
"eval_loss": 0.56792151927948,
|
1712 |
+
"eval_runtime": 5.5954,
|
1713 |
+
"eval_samples_per_second": 173.893,
|
1714 |
+
"eval_steps_per_second": 10.902,
|
1715 |
+
"step": 1342
|
1716 |
+
},
|
1717 |
+
{
|
1718 |
+
"epoch": 11.02,
|
1719 |
+
"learning_rate": 4.075342465753426e-06,
|
1720 |
+
"loss": 0.6378,
|
1721 |
+
"step": 1345
|
1722 |
+
},
|
1723 |
+
{
|
1724 |
+
"epoch": 11.07,
|
1725 |
+
"learning_rate": 3.904109589041096e-06,
|
1726 |
+
"loss": 0.5403,
|
1727 |
+
"step": 1350
|
1728 |
+
},
|
1729 |
+
{
|
1730 |
+
"epoch": 11.11,
|
1731 |
+
"learning_rate": 3.7328767123287675e-06,
|
1732 |
+
"loss": 0.5622,
|
1733 |
+
"step": 1355
|
1734 |
+
},
|
1735 |
+
{
|
1736 |
+
"epoch": 11.15,
|
1737 |
+
"learning_rate": 3.5616438356164386e-06,
|
1738 |
+
"loss": 0.599,
|
1739 |
+
"step": 1360
|
1740 |
+
},
|
1741 |
+
{
|
1742 |
+
"epoch": 11.19,
|
1743 |
+
"learning_rate": 3.39041095890411e-06,
|
1744 |
+
"loss": 0.5112,
|
1745 |
+
"step": 1365
|
1746 |
+
},
|
1747 |
+
{
|
1748 |
+
"epoch": 11.23,
|
1749 |
+
"learning_rate": 3.2191780821917813e-06,
|
1750 |
+
"loss": 0.6468,
|
1751 |
+
"step": 1370
|
1752 |
+
},
|
1753 |
+
{
|
1754 |
+
"epoch": 11.27,
|
1755 |
+
"learning_rate": 3.0479452054794525e-06,
|
1756 |
+
"loss": 0.5667,
|
1757 |
+
"step": 1375
|
1758 |
+
},
|
1759 |
+
{
|
1760 |
+
"epoch": 11.31,
|
1761 |
+
"learning_rate": 2.876712328767123e-06,
|
1762 |
+
"loss": 0.5676,
|
1763 |
+
"step": 1380
|
1764 |
+
},
|
1765 |
+
{
|
1766 |
+
"epoch": 11.35,
|
1767 |
+
"learning_rate": 2.705479452054795e-06,
|
1768 |
+
"loss": 0.5675,
|
1769 |
+
"step": 1385
|
1770 |
+
},
|
1771 |
+
{
|
1772 |
+
"epoch": 11.39,
|
1773 |
+
"learning_rate": 2.534246575342466e-06,
|
1774 |
+
"loss": 0.6044,
|
1775 |
+
"step": 1390
|
1776 |
+
},
|
1777 |
+
{
|
1778 |
+
"epoch": 11.43,
|
1779 |
+
"learning_rate": 2.363013698630137e-06,
|
1780 |
+
"loss": 0.5977,
|
1781 |
+
"step": 1395
|
1782 |
+
},
|
1783 |
+
{
|
1784 |
+
"epoch": 11.48,
|
1785 |
+
"learning_rate": 2.191780821917808e-06,
|
1786 |
+
"loss": 0.5705,
|
1787 |
+
"step": 1400
|
1788 |
+
},
|
1789 |
+
{
|
1790 |
+
"epoch": 11.52,
|
1791 |
+
"learning_rate": 2.0205479452054797e-06,
|
1792 |
+
"loss": 0.614,
|
1793 |
+
"step": 1405
|
1794 |
+
},
|
1795 |
+
{
|
1796 |
+
"epoch": 11.56,
|
1797 |
+
"learning_rate": 1.8493150684931507e-06,
|
1798 |
+
"loss": 0.5503,
|
1799 |
+
"step": 1410
|
1800 |
+
},
|
1801 |
+
{
|
1802 |
+
"epoch": 11.6,
|
1803 |
+
"learning_rate": 1.678082191780822e-06,
|
1804 |
+
"loss": 0.6357,
|
1805 |
+
"step": 1415
|
1806 |
+
},
|
1807 |
+
{
|
1808 |
+
"epoch": 11.64,
|
1809 |
+
"learning_rate": 1.5068493150684932e-06,
|
1810 |
+
"loss": 0.6212,
|
1811 |
+
"step": 1420
|
1812 |
+
},
|
1813 |
+
{
|
1814 |
+
"epoch": 11.68,
|
1815 |
+
"learning_rate": 1.3356164383561645e-06,
|
1816 |
+
"loss": 0.4834,
|
1817 |
+
"step": 1425
|
1818 |
+
},
|
1819 |
+
{
|
1820 |
+
"epoch": 11.72,
|
1821 |
+
"learning_rate": 1.1643835616438357e-06,
|
1822 |
+
"loss": 0.5066,
|
1823 |
+
"step": 1430
|
1824 |
+
},
|
1825 |
+
{
|
1826 |
+
"epoch": 11.76,
|
1827 |
+
"learning_rate": 9.931506849315068e-07,
|
1828 |
+
"loss": 0.5996,
|
1829 |
+
"step": 1435
|
1830 |
+
},
|
1831 |
+
{
|
1832 |
+
"epoch": 11.8,
|
1833 |
+
"learning_rate": 8.219178082191781e-07,
|
1834 |
+
"loss": 0.5948,
|
1835 |
+
"step": 1440
|
1836 |
+
},
|
1837 |
+
{
|
1838 |
+
"epoch": 11.84,
|
1839 |
+
"learning_rate": 6.506849315068493e-07,
|
1840 |
+
"loss": 0.4941,
|
1841 |
+
"step": 1445
|
1842 |
+
},
|
1843 |
+
{
|
1844 |
+
"epoch": 11.89,
|
1845 |
+
"learning_rate": 4.794520547945206e-07,
|
1846 |
+
"loss": 0.6554,
|
1847 |
+
"step": 1450
|
1848 |
+
},
|
1849 |
+
{
|
1850 |
+
"epoch": 11.93,
|
1851 |
+
"learning_rate": 3.082191780821918e-07,
|
1852 |
+
"loss": 0.4939,
|
1853 |
+
"step": 1455
|
1854 |
+
},
|
1855 |
+
{
|
1856 |
+
"epoch": 11.97,
|
1857 |
+
"learning_rate": 1.36986301369863e-07,
|
1858 |
+
"loss": 0.6243,
|
1859 |
+
"step": 1460
|
1860 |
+
},
|
1861 |
+
{
|
1862 |
+
"epoch": 12.0,
|
1863 |
+
"eval_accuracy": 0.8273381294964028,
|
1864 |
+
"eval_loss": 0.5332313776016235,
|
1865 |
+
"eval_runtime": 5.5171,
|
1866 |
+
"eval_samples_per_second": 176.361,
|
1867 |
+
"eval_steps_per_second": 11.057,
|
1868 |
+
"step": 1464
|
1869 |
+
},
|
1870 |
+
{
|
1871 |
+
"epoch": 12.0,
|
1872 |
+
"step": 1464,
|
1873 |
+
"total_flos": 1.175377212318081e+18,
|
1874 |
+
"train_loss": 1.2422215939382386,
|
1875 |
+
"train_runtime": 2576.2267,
|
1876 |
+
"train_samples_per_second": 18.124,
|
1877 |
+
"train_steps_per_second": 0.568
|
1878 |
+
}
|
1879 |
+
],
|
1880 |
+
"max_steps": 1464,
|
1881 |
+
"num_train_epochs": 12,
|
1882 |
+
"total_flos": 1.175377212318081e+18,
|
1883 |
+
"trial_name": null,
|
1884 |
+
"trial_params": null
|
1885 |
+
}
|