PedroSampaio commited on
Commit
5a97984
1 Parent(s): 69b3d5b

End of training

Browse files
Files changed (4) hide show
  1. all_results.json +13 -0
  2. eval_results.json +8 -0
  3. train_results.json +8 -0
  4. trainer_state.json +4525 -0
all_results.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 7.0,
3
+ "eval_accuracy": 0.9997796191819464,
4
+ "eval_loss": 0.0009974736021831632,
5
+ "eval_runtime": 139.7212,
6
+ "eval_samples_per_second": 162.381,
7
+ "eval_steps_per_second": 10.149,
8
+ "total_flos": 3.714527647408435e+19,
9
+ "train_loss": 0.14141770749387125,
10
+ "train_runtime": 7823.1917,
11
+ "train_samples_per_second": 60.567,
12
+ "train_steps_per_second": 0.946
13
+ }
eval_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 7.0,
3
+ "eval_accuracy": 0.9997796191819464,
4
+ "eval_loss": 0.0009974736021831632,
5
+ "eval_runtime": 139.7212,
6
+ "eval_samples_per_second": 162.381,
7
+ "eval_steps_per_second": 10.149
8
+ }
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 7.0,
3
+ "total_flos": 3.714527647408435e+19,
4
+ "train_loss": 0.14141770749387125,
5
+ "train_runtime": 7823.1917,
6
+ "train_samples_per_second": 60.567,
7
+ "train_steps_per_second": 0.946
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,4525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.9997796191819464,
3
+ "best_model_checkpoint": "swin-base-patch4-window7-224-fruits-360-16-7/checkpoint-7399",
4
+ "epoch": 6.995036634365398,
5
+ "eval_steps": 500,
6
+ "global_step": 7399,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.01,
13
+ "learning_rate": 6.756756756756758e-07,
14
+ "loss": 4.7447,
15
+ "step": 10
16
+ },
17
+ {
18
+ "epoch": 0.02,
19
+ "learning_rate": 1.3513513513513515e-06,
20
+ "loss": 4.7447,
21
+ "step": 20
22
+ },
23
+ {
24
+ "epoch": 0.03,
25
+ "learning_rate": 2.0270270270270273e-06,
26
+ "loss": 4.7306,
27
+ "step": 30
28
+ },
29
+ {
30
+ "epoch": 0.04,
31
+ "learning_rate": 2.702702702702703e-06,
32
+ "loss": 4.7,
33
+ "step": 40
34
+ },
35
+ {
36
+ "epoch": 0.05,
37
+ "learning_rate": 3.3783783783783788e-06,
38
+ "loss": 4.6896,
39
+ "step": 50
40
+ },
41
+ {
42
+ "epoch": 0.06,
43
+ "learning_rate": 4.0540540540540545e-06,
44
+ "loss": 4.6375,
45
+ "step": 60
46
+ },
47
+ {
48
+ "epoch": 0.07,
49
+ "learning_rate": 4.72972972972973e-06,
50
+ "loss": 4.5767,
51
+ "step": 70
52
+ },
53
+ {
54
+ "epoch": 0.08,
55
+ "learning_rate": 5.405405405405406e-06,
56
+ "loss": 4.5196,
57
+ "step": 80
58
+ },
59
+ {
60
+ "epoch": 0.09,
61
+ "learning_rate": 6.081081081081082e-06,
62
+ "loss": 4.4603,
63
+ "step": 90
64
+ },
65
+ {
66
+ "epoch": 0.09,
67
+ "learning_rate": 6.7567567567567575e-06,
68
+ "loss": 4.3467,
69
+ "step": 100
70
+ },
71
+ {
72
+ "epoch": 0.1,
73
+ "learning_rate": 7.432432432432433e-06,
74
+ "loss": 4.2597,
75
+ "step": 110
76
+ },
77
+ {
78
+ "epoch": 0.11,
79
+ "learning_rate": 8.108108108108109e-06,
80
+ "loss": 4.085,
81
+ "step": 120
82
+ },
83
+ {
84
+ "epoch": 0.12,
85
+ "learning_rate": 8.783783783783785e-06,
86
+ "loss": 3.9328,
87
+ "step": 130
88
+ },
89
+ {
90
+ "epoch": 0.13,
91
+ "learning_rate": 9.45945945945946e-06,
92
+ "loss": 3.6433,
93
+ "step": 140
94
+ },
95
+ {
96
+ "epoch": 0.14,
97
+ "learning_rate": 1.0135135135135136e-05,
98
+ "loss": 3.3766,
99
+ "step": 150
100
+ },
101
+ {
102
+ "epoch": 0.15,
103
+ "learning_rate": 1.0810810810810812e-05,
104
+ "loss": 3.1055,
105
+ "step": 160
106
+ },
107
+ {
108
+ "epoch": 0.16,
109
+ "learning_rate": 1.1486486486486488e-05,
110
+ "loss": 2.7427,
111
+ "step": 170
112
+ },
113
+ {
114
+ "epoch": 0.17,
115
+ "learning_rate": 1.2162162162162164e-05,
116
+ "loss": 2.4803,
117
+ "step": 180
118
+ },
119
+ {
120
+ "epoch": 0.18,
121
+ "learning_rate": 1.2837837837837838e-05,
122
+ "loss": 2.1213,
123
+ "step": 190
124
+ },
125
+ {
126
+ "epoch": 0.19,
127
+ "learning_rate": 1.3513513513513515e-05,
128
+ "loss": 1.888,
129
+ "step": 200
130
+ },
131
+ {
132
+ "epoch": 0.2,
133
+ "learning_rate": 1.4189189189189189e-05,
134
+ "loss": 1.513,
135
+ "step": 210
136
+ },
137
+ {
138
+ "epoch": 0.21,
139
+ "learning_rate": 1.4864864864864867e-05,
140
+ "loss": 1.2981,
141
+ "step": 220
142
+ },
143
+ {
144
+ "epoch": 0.22,
145
+ "learning_rate": 1.554054054054054e-05,
146
+ "loss": 1.1383,
147
+ "step": 230
148
+ },
149
+ {
150
+ "epoch": 0.23,
151
+ "learning_rate": 1.6216216216216218e-05,
152
+ "loss": 0.8838,
153
+ "step": 240
154
+ },
155
+ {
156
+ "epoch": 0.24,
157
+ "learning_rate": 1.6891891891891892e-05,
158
+ "loss": 0.7741,
159
+ "step": 250
160
+ },
161
+ {
162
+ "epoch": 0.25,
163
+ "learning_rate": 1.756756756756757e-05,
164
+ "loss": 0.6182,
165
+ "step": 260
166
+ },
167
+ {
168
+ "epoch": 0.26,
169
+ "learning_rate": 1.8243243243243244e-05,
170
+ "loss": 0.5563,
171
+ "step": 270
172
+ },
173
+ {
174
+ "epoch": 0.26,
175
+ "learning_rate": 1.891891891891892e-05,
176
+ "loss": 0.4464,
177
+ "step": 280
178
+ },
179
+ {
180
+ "epoch": 0.27,
181
+ "learning_rate": 1.9594594594594595e-05,
182
+ "loss": 0.4189,
183
+ "step": 290
184
+ },
185
+ {
186
+ "epoch": 0.28,
187
+ "learning_rate": 2.0270270270270273e-05,
188
+ "loss": 0.3484,
189
+ "step": 300
190
+ },
191
+ {
192
+ "epoch": 0.29,
193
+ "learning_rate": 2.0945945945945947e-05,
194
+ "loss": 0.3228,
195
+ "step": 310
196
+ },
197
+ {
198
+ "epoch": 0.3,
199
+ "learning_rate": 2.1621621621621624e-05,
200
+ "loss": 0.2785,
201
+ "step": 320
202
+ },
203
+ {
204
+ "epoch": 0.31,
205
+ "learning_rate": 2.2297297297297298e-05,
206
+ "loss": 0.28,
207
+ "step": 330
208
+ },
209
+ {
210
+ "epoch": 0.32,
211
+ "learning_rate": 2.2972972972972976e-05,
212
+ "loss": 0.1871,
213
+ "step": 340
214
+ },
215
+ {
216
+ "epoch": 0.33,
217
+ "learning_rate": 2.364864864864865e-05,
218
+ "loss": 0.1764,
219
+ "step": 350
220
+ },
221
+ {
222
+ "epoch": 0.34,
223
+ "learning_rate": 2.4324324324324327e-05,
224
+ "loss": 0.1863,
225
+ "step": 360
226
+ },
227
+ {
228
+ "epoch": 0.35,
229
+ "learning_rate": 2.5e-05,
230
+ "loss": 0.1897,
231
+ "step": 370
232
+ },
233
+ {
234
+ "epoch": 0.36,
235
+ "learning_rate": 2.5675675675675675e-05,
236
+ "loss": 0.2142,
237
+ "step": 380
238
+ },
239
+ {
240
+ "epoch": 0.37,
241
+ "learning_rate": 2.635135135135135e-05,
242
+ "loss": 0.1761,
243
+ "step": 390
244
+ },
245
+ {
246
+ "epoch": 0.38,
247
+ "learning_rate": 2.702702702702703e-05,
248
+ "loss": 0.1449,
249
+ "step": 400
250
+ },
251
+ {
252
+ "epoch": 0.39,
253
+ "learning_rate": 2.7702702702702704e-05,
254
+ "loss": 0.1604,
255
+ "step": 410
256
+ },
257
+ {
258
+ "epoch": 0.4,
259
+ "learning_rate": 2.8378378378378378e-05,
260
+ "loss": 0.1487,
261
+ "step": 420
262
+ },
263
+ {
264
+ "epoch": 0.41,
265
+ "learning_rate": 2.9054054054054052e-05,
266
+ "loss": 0.1439,
267
+ "step": 430
268
+ },
269
+ {
270
+ "epoch": 0.42,
271
+ "learning_rate": 2.9729729729729733e-05,
272
+ "loss": 0.1159,
273
+ "step": 440
274
+ },
275
+ {
276
+ "epoch": 0.43,
277
+ "learning_rate": 3.0405405405405407e-05,
278
+ "loss": 0.1141,
279
+ "step": 450
280
+ },
281
+ {
282
+ "epoch": 0.43,
283
+ "learning_rate": 3.108108108108108e-05,
284
+ "loss": 0.1134,
285
+ "step": 460
286
+ },
287
+ {
288
+ "epoch": 0.44,
289
+ "learning_rate": 3.175675675675676e-05,
290
+ "loss": 0.1223,
291
+ "step": 470
292
+ },
293
+ {
294
+ "epoch": 0.45,
295
+ "learning_rate": 3.2432432432432436e-05,
296
+ "loss": 0.0932,
297
+ "step": 480
298
+ },
299
+ {
300
+ "epoch": 0.46,
301
+ "learning_rate": 3.310810810810811e-05,
302
+ "loss": 0.0883,
303
+ "step": 490
304
+ },
305
+ {
306
+ "epoch": 0.47,
307
+ "learning_rate": 3.3783783783783784e-05,
308
+ "loss": 0.1624,
309
+ "step": 500
310
+ },
311
+ {
312
+ "epoch": 0.48,
313
+ "learning_rate": 3.445945945945946e-05,
314
+ "loss": 0.1123,
315
+ "step": 510
316
+ },
317
+ {
318
+ "epoch": 0.49,
319
+ "learning_rate": 3.513513513513514e-05,
320
+ "loss": 0.1121,
321
+ "step": 520
322
+ },
323
+ {
324
+ "epoch": 0.5,
325
+ "learning_rate": 3.581081081081081e-05,
326
+ "loss": 0.0872,
327
+ "step": 530
328
+ },
329
+ {
330
+ "epoch": 0.51,
331
+ "learning_rate": 3.648648648648649e-05,
332
+ "loss": 0.0877,
333
+ "step": 540
334
+ },
335
+ {
336
+ "epoch": 0.52,
337
+ "learning_rate": 3.7162162162162165e-05,
338
+ "loss": 0.0739,
339
+ "step": 550
340
+ },
341
+ {
342
+ "epoch": 0.53,
343
+ "learning_rate": 3.783783783783784e-05,
344
+ "loss": 0.1002,
345
+ "step": 560
346
+ },
347
+ {
348
+ "epoch": 0.54,
349
+ "learning_rate": 3.851351351351351e-05,
350
+ "loss": 0.0866,
351
+ "step": 570
352
+ },
353
+ {
354
+ "epoch": 0.55,
355
+ "learning_rate": 3.918918918918919e-05,
356
+ "loss": 0.1211,
357
+ "step": 580
358
+ },
359
+ {
360
+ "epoch": 0.56,
361
+ "learning_rate": 3.986486486486487e-05,
362
+ "loss": 0.1612,
363
+ "step": 590
364
+ },
365
+ {
366
+ "epoch": 0.57,
367
+ "learning_rate": 4.0540540540540545e-05,
368
+ "loss": 0.1543,
369
+ "step": 600
370
+ },
371
+ {
372
+ "epoch": 0.58,
373
+ "learning_rate": 4.1216216216216216e-05,
374
+ "loss": 0.144,
375
+ "step": 610
376
+ },
377
+ {
378
+ "epoch": 0.59,
379
+ "learning_rate": 4.189189189189189e-05,
380
+ "loss": 0.0848,
381
+ "step": 620
382
+ },
383
+ {
384
+ "epoch": 0.6,
385
+ "learning_rate": 4.256756756756757e-05,
386
+ "loss": 0.0958,
387
+ "step": 630
388
+ },
389
+ {
390
+ "epoch": 0.61,
391
+ "learning_rate": 4.324324324324325e-05,
392
+ "loss": 0.0994,
393
+ "step": 640
394
+ },
395
+ {
396
+ "epoch": 0.61,
397
+ "learning_rate": 4.391891891891892e-05,
398
+ "loss": 0.0862,
399
+ "step": 650
400
+ },
401
+ {
402
+ "epoch": 0.62,
403
+ "learning_rate": 4.4594594594594596e-05,
404
+ "loss": 0.0659,
405
+ "step": 660
406
+ },
407
+ {
408
+ "epoch": 0.63,
409
+ "learning_rate": 4.5270270270270274e-05,
410
+ "loss": 0.1152,
411
+ "step": 670
412
+ },
413
+ {
414
+ "epoch": 0.64,
415
+ "learning_rate": 4.594594594594595e-05,
416
+ "loss": 0.0698,
417
+ "step": 680
418
+ },
419
+ {
420
+ "epoch": 0.65,
421
+ "learning_rate": 4.662162162162162e-05,
422
+ "loss": 0.0585,
423
+ "step": 690
424
+ },
425
+ {
426
+ "epoch": 0.66,
427
+ "learning_rate": 4.72972972972973e-05,
428
+ "loss": 0.1281,
429
+ "step": 700
430
+ },
431
+ {
432
+ "epoch": 0.67,
433
+ "learning_rate": 4.797297297297298e-05,
434
+ "loss": 0.1298,
435
+ "step": 710
436
+ },
437
+ {
438
+ "epoch": 0.68,
439
+ "learning_rate": 4.8648648648648654e-05,
440
+ "loss": 0.0814,
441
+ "step": 720
442
+ },
443
+ {
444
+ "epoch": 0.69,
445
+ "learning_rate": 4.9324324324324325e-05,
446
+ "loss": 0.0842,
447
+ "step": 730
448
+ },
449
+ {
450
+ "epoch": 0.7,
451
+ "learning_rate": 5e-05,
452
+ "loss": 0.0946,
453
+ "step": 740
454
+ },
455
+ {
456
+ "epoch": 0.71,
457
+ "learning_rate": 4.992491365069831e-05,
458
+ "loss": 0.0675,
459
+ "step": 750
460
+ },
461
+ {
462
+ "epoch": 0.72,
463
+ "learning_rate": 4.984982730139661e-05,
464
+ "loss": 0.073,
465
+ "step": 760
466
+ },
467
+ {
468
+ "epoch": 0.73,
469
+ "learning_rate": 4.977474095209491e-05,
470
+ "loss": 0.075,
471
+ "step": 770
472
+ },
473
+ {
474
+ "epoch": 0.74,
475
+ "learning_rate": 4.9699654602793215e-05,
476
+ "loss": 0.1122,
477
+ "step": 780
478
+ },
479
+ {
480
+ "epoch": 0.75,
481
+ "learning_rate": 4.9624568253491515e-05,
482
+ "loss": 0.0666,
483
+ "step": 790
484
+ },
485
+ {
486
+ "epoch": 0.76,
487
+ "learning_rate": 4.954948190418982e-05,
488
+ "loss": 0.0837,
489
+ "step": 800
490
+ },
491
+ {
492
+ "epoch": 0.77,
493
+ "learning_rate": 4.947439555488813e-05,
494
+ "loss": 0.0829,
495
+ "step": 810
496
+ },
497
+ {
498
+ "epoch": 0.78,
499
+ "learning_rate": 4.939930920558643e-05,
500
+ "loss": 0.0832,
501
+ "step": 820
502
+ },
503
+ {
504
+ "epoch": 0.78,
505
+ "learning_rate": 4.932422285628473e-05,
506
+ "loss": 0.0992,
507
+ "step": 830
508
+ },
509
+ {
510
+ "epoch": 0.79,
511
+ "learning_rate": 4.9249136506983033e-05,
512
+ "loss": 0.0731,
513
+ "step": 840
514
+ },
515
+ {
516
+ "epoch": 0.8,
517
+ "learning_rate": 4.917405015768133e-05,
518
+ "loss": 0.0756,
519
+ "step": 850
520
+ },
521
+ {
522
+ "epoch": 0.81,
523
+ "learning_rate": 4.909896380837964e-05,
524
+ "loss": 0.0959,
525
+ "step": 860
526
+ },
527
+ {
528
+ "epoch": 0.82,
529
+ "learning_rate": 4.9023877459077946e-05,
530
+ "loss": 0.0818,
531
+ "step": 870
532
+ },
533
+ {
534
+ "epoch": 0.83,
535
+ "learning_rate": 4.8948791109776246e-05,
536
+ "loss": 0.0808,
537
+ "step": 880
538
+ },
539
+ {
540
+ "epoch": 0.84,
541
+ "learning_rate": 4.8873704760474546e-05,
542
+ "loss": 0.0608,
543
+ "step": 890
544
+ },
545
+ {
546
+ "epoch": 0.85,
547
+ "learning_rate": 4.879861841117285e-05,
548
+ "loss": 0.0643,
549
+ "step": 900
550
+ },
551
+ {
552
+ "epoch": 0.86,
553
+ "learning_rate": 4.872353206187115e-05,
554
+ "loss": 0.0814,
555
+ "step": 910
556
+ },
557
+ {
558
+ "epoch": 0.87,
559
+ "learning_rate": 4.864844571256946e-05,
560
+ "loss": 0.0921,
561
+ "step": 920
562
+ },
563
+ {
564
+ "epoch": 0.88,
565
+ "learning_rate": 4.8573359363267765e-05,
566
+ "loss": 0.1271,
567
+ "step": 930
568
+ },
569
+ {
570
+ "epoch": 0.89,
571
+ "learning_rate": 4.8498273013966065e-05,
572
+ "loss": 0.0768,
573
+ "step": 940
574
+ },
575
+ {
576
+ "epoch": 0.9,
577
+ "learning_rate": 4.8423186664664364e-05,
578
+ "loss": 0.0867,
579
+ "step": 950
580
+ },
581
+ {
582
+ "epoch": 0.91,
583
+ "learning_rate": 4.834810031536267e-05,
584
+ "loss": 0.0827,
585
+ "step": 960
586
+ },
587
+ {
588
+ "epoch": 0.92,
589
+ "learning_rate": 4.827301396606097e-05,
590
+ "loss": 0.0758,
591
+ "step": 970
592
+ },
593
+ {
594
+ "epoch": 0.93,
595
+ "learning_rate": 4.819792761675928e-05,
596
+ "loss": 0.0841,
597
+ "step": 980
598
+ },
599
+ {
600
+ "epoch": 0.94,
601
+ "learning_rate": 4.8122841267457583e-05,
602
+ "loss": 0.0629,
603
+ "step": 990
604
+ },
605
+ {
606
+ "epoch": 0.95,
607
+ "learning_rate": 4.8047754918155876e-05,
608
+ "loss": 0.0795,
609
+ "step": 1000
610
+ },
611
+ {
612
+ "epoch": 0.95,
613
+ "learning_rate": 4.797266856885418e-05,
614
+ "loss": 0.0634,
615
+ "step": 1010
616
+ },
617
+ {
618
+ "epoch": 0.96,
619
+ "learning_rate": 4.789758221955249e-05,
620
+ "loss": 0.0348,
621
+ "step": 1020
622
+ },
623
+ {
624
+ "epoch": 0.97,
625
+ "learning_rate": 4.782249587025079e-05,
626
+ "loss": 0.055,
627
+ "step": 1030
628
+ },
629
+ {
630
+ "epoch": 0.98,
631
+ "learning_rate": 4.7747409520949096e-05,
632
+ "loss": 0.0801,
633
+ "step": 1040
634
+ },
635
+ {
636
+ "epoch": 0.99,
637
+ "learning_rate": 4.7672323171647395e-05,
638
+ "loss": 0.0631,
639
+ "step": 1050
640
+ },
641
+ {
642
+ "epoch": 1.0,
643
+ "eval_accuracy": 0.9956364598025388,
644
+ "eval_loss": 0.014884675852954388,
645
+ "eval_runtime": 140.2456,
646
+ "eval_samples_per_second": 161.773,
647
+ "eval_steps_per_second": 10.111,
648
+ "step": 1057
649
+ },
650
+ {
651
+ "epoch": 1.0,
652
+ "learning_rate": 4.7597236822345695e-05,
653
+ "loss": 0.0548,
654
+ "step": 1060
655
+ },
656
+ {
657
+ "epoch": 1.01,
658
+ "learning_rate": 4.7522150473044e-05,
659
+ "loss": 0.0842,
660
+ "step": 1070
661
+ },
662
+ {
663
+ "epoch": 1.02,
664
+ "learning_rate": 4.744706412374231e-05,
665
+ "loss": 0.0395,
666
+ "step": 1080
667
+ },
668
+ {
669
+ "epoch": 1.03,
670
+ "learning_rate": 4.737197777444061e-05,
671
+ "loss": 0.107,
672
+ "step": 1090
673
+ },
674
+ {
675
+ "epoch": 1.04,
676
+ "learning_rate": 4.7296891425138914e-05,
677
+ "loss": 0.0405,
678
+ "step": 1100
679
+ },
680
+ {
681
+ "epoch": 1.05,
682
+ "learning_rate": 4.7221805075837214e-05,
683
+ "loss": 0.063,
684
+ "step": 1110
685
+ },
686
+ {
687
+ "epoch": 1.06,
688
+ "learning_rate": 4.7146718726535514e-05,
689
+ "loss": 0.0759,
690
+ "step": 1120
691
+ },
692
+ {
693
+ "epoch": 1.07,
694
+ "learning_rate": 4.707163237723382e-05,
695
+ "loss": 0.0515,
696
+ "step": 1130
697
+ },
698
+ {
699
+ "epoch": 1.08,
700
+ "learning_rate": 4.699654602793213e-05,
701
+ "loss": 0.0705,
702
+ "step": 1140
703
+ },
704
+ {
705
+ "epoch": 1.09,
706
+ "learning_rate": 4.6921459678630426e-05,
707
+ "loss": 0.0376,
708
+ "step": 1150
709
+ },
710
+ {
711
+ "epoch": 1.1,
712
+ "learning_rate": 4.684637332932873e-05,
713
+ "loss": 0.0673,
714
+ "step": 1160
715
+ },
716
+ {
717
+ "epoch": 1.11,
718
+ "learning_rate": 4.677128698002703e-05,
719
+ "loss": 0.0396,
720
+ "step": 1170
721
+ },
722
+ {
723
+ "epoch": 1.12,
724
+ "learning_rate": 4.669620063072533e-05,
725
+ "loss": 0.0377,
726
+ "step": 1180
727
+ },
728
+ {
729
+ "epoch": 1.13,
730
+ "learning_rate": 4.662111428142364e-05,
731
+ "loss": 0.0449,
732
+ "step": 1190
733
+ },
734
+ {
735
+ "epoch": 1.13,
736
+ "learning_rate": 4.6546027932121945e-05,
737
+ "loss": 0.0576,
738
+ "step": 1200
739
+ },
740
+ {
741
+ "epoch": 1.14,
742
+ "learning_rate": 4.6470941582820245e-05,
743
+ "loss": 0.0491,
744
+ "step": 1210
745
+ },
746
+ {
747
+ "epoch": 1.15,
748
+ "learning_rate": 4.639585523351855e-05,
749
+ "loss": 0.0871,
750
+ "step": 1220
751
+ },
752
+ {
753
+ "epoch": 1.16,
754
+ "learning_rate": 4.632076888421685e-05,
755
+ "loss": 0.0526,
756
+ "step": 1230
757
+ },
758
+ {
759
+ "epoch": 1.17,
760
+ "learning_rate": 4.624568253491515e-05,
761
+ "loss": 0.0298,
762
+ "step": 1240
763
+ },
764
+ {
765
+ "epoch": 1.18,
766
+ "learning_rate": 4.617059618561346e-05,
767
+ "loss": 0.0664,
768
+ "step": 1250
769
+ },
770
+ {
771
+ "epoch": 1.19,
772
+ "learning_rate": 4.6095509836311764e-05,
773
+ "loss": 0.047,
774
+ "step": 1260
775
+ },
776
+ {
777
+ "epoch": 1.2,
778
+ "learning_rate": 4.6020423487010064e-05,
779
+ "loss": 0.0402,
780
+ "step": 1270
781
+ },
782
+ {
783
+ "epoch": 1.21,
784
+ "learning_rate": 4.594533713770836e-05,
785
+ "loss": 0.0639,
786
+ "step": 1280
787
+ },
788
+ {
789
+ "epoch": 1.22,
790
+ "learning_rate": 4.587025078840667e-05,
791
+ "loss": 0.0495,
792
+ "step": 1290
793
+ },
794
+ {
795
+ "epoch": 1.23,
796
+ "learning_rate": 4.579516443910497e-05,
797
+ "loss": 0.0488,
798
+ "step": 1300
799
+ },
800
+ {
801
+ "epoch": 1.24,
802
+ "learning_rate": 4.5720078089803276e-05,
803
+ "loss": 0.0386,
804
+ "step": 1310
805
+ },
806
+ {
807
+ "epoch": 1.25,
808
+ "learning_rate": 4.564499174050158e-05,
809
+ "loss": 0.0486,
810
+ "step": 1320
811
+ },
812
+ {
813
+ "epoch": 1.26,
814
+ "learning_rate": 4.556990539119988e-05,
815
+ "loss": 0.0412,
816
+ "step": 1330
817
+ },
818
+ {
819
+ "epoch": 1.27,
820
+ "learning_rate": 4.549481904189818e-05,
821
+ "loss": 0.0489,
822
+ "step": 1340
823
+ },
824
+ {
825
+ "epoch": 1.28,
826
+ "learning_rate": 4.541973269259649e-05,
827
+ "loss": 0.0606,
828
+ "step": 1350
829
+ },
830
+ {
831
+ "epoch": 1.29,
832
+ "learning_rate": 4.534464634329479e-05,
833
+ "loss": 0.0416,
834
+ "step": 1360
835
+ },
836
+ {
837
+ "epoch": 1.3,
838
+ "learning_rate": 4.5269559993993095e-05,
839
+ "loss": 0.0629,
840
+ "step": 1370
841
+ },
842
+ {
843
+ "epoch": 1.3,
844
+ "learning_rate": 4.51944736446914e-05,
845
+ "loss": 0.0532,
846
+ "step": 1380
847
+ },
848
+ {
849
+ "epoch": 1.31,
850
+ "learning_rate": 4.51193872953897e-05,
851
+ "loss": 0.0354,
852
+ "step": 1390
853
+ },
854
+ {
855
+ "epoch": 1.32,
856
+ "learning_rate": 4.5044300946088e-05,
857
+ "loss": 0.0454,
858
+ "step": 1400
859
+ },
860
+ {
861
+ "epoch": 1.33,
862
+ "learning_rate": 4.496921459678631e-05,
863
+ "loss": 0.0555,
864
+ "step": 1410
865
+ },
866
+ {
867
+ "epoch": 1.34,
868
+ "learning_rate": 4.489412824748461e-05,
869
+ "loss": 0.0524,
870
+ "step": 1420
871
+ },
872
+ {
873
+ "epoch": 1.35,
874
+ "learning_rate": 4.481904189818291e-05,
875
+ "loss": 0.0599,
876
+ "step": 1430
877
+ },
878
+ {
879
+ "epoch": 1.36,
880
+ "learning_rate": 4.474395554888122e-05,
881
+ "loss": 0.0421,
882
+ "step": 1440
883
+ },
884
+ {
885
+ "epoch": 1.37,
886
+ "learning_rate": 4.466886919957951e-05,
887
+ "loss": 0.0329,
888
+ "step": 1450
889
+ },
890
+ {
891
+ "epoch": 1.38,
892
+ "learning_rate": 4.459378285027782e-05,
893
+ "loss": 0.0337,
894
+ "step": 1460
895
+ },
896
+ {
897
+ "epoch": 1.39,
898
+ "learning_rate": 4.4518696500976126e-05,
899
+ "loss": 0.0431,
900
+ "step": 1470
901
+ },
902
+ {
903
+ "epoch": 1.4,
904
+ "learning_rate": 4.4443610151674425e-05,
905
+ "loss": 0.0255,
906
+ "step": 1480
907
+ },
908
+ {
909
+ "epoch": 1.41,
910
+ "learning_rate": 4.436852380237273e-05,
911
+ "loss": 0.0564,
912
+ "step": 1490
913
+ },
914
+ {
915
+ "epoch": 1.42,
916
+ "learning_rate": 4.429343745307103e-05,
917
+ "loss": 0.0652,
918
+ "step": 1500
919
+ },
920
+ {
921
+ "epoch": 1.43,
922
+ "learning_rate": 4.421835110376933e-05,
923
+ "loss": 0.0407,
924
+ "step": 1510
925
+ },
926
+ {
927
+ "epoch": 1.44,
928
+ "learning_rate": 4.414326475446764e-05,
929
+ "loss": 0.043,
930
+ "step": 1520
931
+ },
932
+ {
933
+ "epoch": 1.45,
934
+ "learning_rate": 4.4068178405165944e-05,
935
+ "loss": 0.045,
936
+ "step": 1530
937
+ },
938
+ {
939
+ "epoch": 1.46,
940
+ "learning_rate": 4.3993092055864244e-05,
941
+ "loss": 0.0472,
942
+ "step": 1540
943
+ },
944
+ {
945
+ "epoch": 1.47,
946
+ "learning_rate": 4.391800570656255e-05,
947
+ "loss": 0.0467,
948
+ "step": 1550
949
+ },
950
+ {
951
+ "epoch": 1.47,
952
+ "learning_rate": 4.384291935726085e-05,
953
+ "loss": 0.0713,
954
+ "step": 1560
955
+ },
956
+ {
957
+ "epoch": 1.48,
958
+ "learning_rate": 4.376783300795915e-05,
959
+ "loss": 0.0413,
960
+ "step": 1570
961
+ },
962
+ {
963
+ "epoch": 1.49,
964
+ "learning_rate": 4.3692746658657457e-05,
965
+ "loss": 0.0303,
966
+ "step": 1580
967
+ },
968
+ {
969
+ "epoch": 1.5,
970
+ "learning_rate": 4.361766030935576e-05,
971
+ "loss": 0.0404,
972
+ "step": 1590
973
+ },
974
+ {
975
+ "epoch": 1.51,
976
+ "learning_rate": 4.354257396005406e-05,
977
+ "loss": 0.0643,
978
+ "step": 1600
979
+ },
980
+ {
981
+ "epoch": 1.52,
982
+ "learning_rate": 4.346748761075237e-05,
983
+ "loss": 0.0234,
984
+ "step": 1610
985
+ },
986
+ {
987
+ "epoch": 1.53,
988
+ "learning_rate": 4.339240126145067e-05,
989
+ "loss": 0.0297,
990
+ "step": 1620
991
+ },
992
+ {
993
+ "epoch": 1.54,
994
+ "learning_rate": 4.331731491214897e-05,
995
+ "loss": 0.0436,
996
+ "step": 1630
997
+ },
998
+ {
999
+ "epoch": 1.55,
1000
+ "learning_rate": 4.3242228562847275e-05,
1001
+ "loss": 0.0133,
1002
+ "step": 1640
1003
+ },
1004
+ {
1005
+ "epoch": 1.56,
1006
+ "learning_rate": 4.316714221354558e-05,
1007
+ "loss": 0.0261,
1008
+ "step": 1650
1009
+ },
1010
+ {
1011
+ "epoch": 1.57,
1012
+ "learning_rate": 4.309205586424388e-05,
1013
+ "loss": 0.0124,
1014
+ "step": 1660
1015
+ },
1016
+ {
1017
+ "epoch": 1.58,
1018
+ "learning_rate": 4.301696951494219e-05,
1019
+ "loss": 0.0378,
1020
+ "step": 1670
1021
+ },
1022
+ {
1023
+ "epoch": 1.59,
1024
+ "learning_rate": 4.294188316564049e-05,
1025
+ "loss": 0.008,
1026
+ "step": 1680
1027
+ },
1028
+ {
1029
+ "epoch": 1.6,
1030
+ "learning_rate": 4.286679681633879e-05,
1031
+ "loss": 0.0167,
1032
+ "step": 1690
1033
+ },
1034
+ {
1035
+ "epoch": 1.61,
1036
+ "learning_rate": 4.2791710467037094e-05,
1037
+ "loss": 0.0244,
1038
+ "step": 1700
1039
+ },
1040
+ {
1041
+ "epoch": 1.62,
1042
+ "learning_rate": 4.27166241177354e-05,
1043
+ "loss": 0.0674,
1044
+ "step": 1710
1045
+ },
1046
+ {
1047
+ "epoch": 1.63,
1048
+ "learning_rate": 4.26415377684337e-05,
1049
+ "loss": 0.0151,
1050
+ "step": 1720
1051
+ },
1052
+ {
1053
+ "epoch": 1.64,
1054
+ "learning_rate": 4.2566451419132e-05,
1055
+ "loss": 0.0234,
1056
+ "step": 1730
1057
+ },
1058
+ {
1059
+ "epoch": 1.65,
1060
+ "learning_rate": 4.2491365069830306e-05,
1061
+ "loss": 0.0581,
1062
+ "step": 1740
1063
+ },
1064
+ {
1065
+ "epoch": 1.65,
1066
+ "learning_rate": 4.2416278720528606e-05,
1067
+ "loss": 0.0108,
1068
+ "step": 1750
1069
+ },
1070
+ {
1071
+ "epoch": 1.66,
1072
+ "learning_rate": 4.234119237122691e-05,
1073
+ "loss": 0.0245,
1074
+ "step": 1760
1075
+ },
1076
+ {
1077
+ "epoch": 1.67,
1078
+ "learning_rate": 4.226610602192522e-05,
1079
+ "loss": 0.0422,
1080
+ "step": 1770
1081
+ },
1082
+ {
1083
+ "epoch": 1.68,
1084
+ "learning_rate": 4.219101967262352e-05,
1085
+ "loss": 0.0572,
1086
+ "step": 1780
1087
+ },
1088
+ {
1089
+ "epoch": 1.69,
1090
+ "learning_rate": 4.211593332332182e-05,
1091
+ "loss": 0.0315,
1092
+ "step": 1790
1093
+ },
1094
+ {
1095
+ "epoch": 1.7,
1096
+ "learning_rate": 4.2040846974020125e-05,
1097
+ "loss": 0.049,
1098
+ "step": 1800
1099
+ },
1100
+ {
1101
+ "epoch": 1.71,
1102
+ "learning_rate": 4.196576062471843e-05,
1103
+ "loss": 0.0336,
1104
+ "step": 1810
1105
+ },
1106
+ {
1107
+ "epoch": 1.72,
1108
+ "learning_rate": 4.189067427541673e-05,
1109
+ "loss": 0.0319,
1110
+ "step": 1820
1111
+ },
1112
+ {
1113
+ "epoch": 1.73,
1114
+ "learning_rate": 4.181558792611504e-05,
1115
+ "loss": 0.0303,
1116
+ "step": 1830
1117
+ },
1118
+ {
1119
+ "epoch": 1.74,
1120
+ "learning_rate": 4.174050157681334e-05,
1121
+ "loss": 0.0187,
1122
+ "step": 1840
1123
+ },
1124
+ {
1125
+ "epoch": 1.75,
1126
+ "learning_rate": 4.166541522751164e-05,
1127
+ "loss": 0.0784,
1128
+ "step": 1850
1129
+ },
1130
+ {
1131
+ "epoch": 1.76,
1132
+ "learning_rate": 4.1590328878209944e-05,
1133
+ "loss": 0.0926,
1134
+ "step": 1860
1135
+ },
1136
+ {
1137
+ "epoch": 1.77,
1138
+ "learning_rate": 4.151524252890825e-05,
1139
+ "loss": 0.0233,
1140
+ "step": 1870
1141
+ },
1142
+ {
1143
+ "epoch": 1.78,
1144
+ "learning_rate": 4.144015617960655e-05,
1145
+ "loss": 0.0394,
1146
+ "step": 1880
1147
+ },
1148
+ {
1149
+ "epoch": 1.79,
1150
+ "learning_rate": 4.1365069830304856e-05,
1151
+ "loss": 0.0489,
1152
+ "step": 1890
1153
+ },
1154
+ {
1155
+ "epoch": 1.8,
1156
+ "learning_rate": 4.1289983481003156e-05,
1157
+ "loss": 0.0413,
1158
+ "step": 1900
1159
+ },
1160
+ {
1161
+ "epoch": 1.81,
1162
+ "learning_rate": 4.1214897131701456e-05,
1163
+ "loss": 0.0216,
1164
+ "step": 1910
1165
+ },
1166
+ {
1167
+ "epoch": 1.82,
1168
+ "learning_rate": 4.113981078239976e-05,
1169
+ "loss": 0.0327,
1170
+ "step": 1920
1171
+ },
1172
+ {
1173
+ "epoch": 1.82,
1174
+ "learning_rate": 4.106472443309807e-05,
1175
+ "loss": 0.0488,
1176
+ "step": 1930
1177
+ },
1178
+ {
1179
+ "epoch": 1.83,
1180
+ "learning_rate": 4.098963808379637e-05,
1181
+ "loss": 0.0747,
1182
+ "step": 1940
1183
+ },
1184
+ {
1185
+ "epoch": 1.84,
1186
+ "learning_rate": 4.0914551734494675e-05,
1187
+ "loss": 0.0521,
1188
+ "step": 1950
1189
+ },
1190
+ {
1191
+ "epoch": 1.85,
1192
+ "learning_rate": 4.0839465385192975e-05,
1193
+ "loss": 0.0289,
1194
+ "step": 1960
1195
+ },
1196
+ {
1197
+ "epoch": 1.86,
1198
+ "learning_rate": 4.0764379035891274e-05,
1199
+ "loss": 0.0182,
1200
+ "step": 1970
1201
+ },
1202
+ {
1203
+ "epoch": 1.87,
1204
+ "learning_rate": 4.068929268658958e-05,
1205
+ "loss": 0.0306,
1206
+ "step": 1980
1207
+ },
1208
+ {
1209
+ "epoch": 1.88,
1210
+ "learning_rate": 4.061420633728789e-05,
1211
+ "loss": 0.0121,
1212
+ "step": 1990
1213
+ },
1214
+ {
1215
+ "epoch": 1.89,
1216
+ "learning_rate": 4.053911998798619e-05,
1217
+ "loss": 0.0302,
1218
+ "step": 2000
1219
+ },
1220
+ {
1221
+ "epoch": 1.9,
1222
+ "learning_rate": 4.046403363868449e-05,
1223
+ "loss": 0.0219,
1224
+ "step": 2010
1225
+ },
1226
+ {
1227
+ "epoch": 1.91,
1228
+ "learning_rate": 4.038894728938279e-05,
1229
+ "loss": 0.0211,
1230
+ "step": 2020
1231
+ },
1232
+ {
1233
+ "epoch": 1.92,
1234
+ "learning_rate": 4.031386094008109e-05,
1235
+ "loss": 0.0186,
1236
+ "step": 2030
1237
+ },
1238
+ {
1239
+ "epoch": 1.93,
1240
+ "learning_rate": 4.02387745907794e-05,
1241
+ "loss": 0.0346,
1242
+ "step": 2040
1243
+ },
1244
+ {
1245
+ "epoch": 1.94,
1246
+ "learning_rate": 4.0163688241477706e-05,
1247
+ "loss": 0.0366,
1248
+ "step": 2050
1249
+ },
1250
+ {
1251
+ "epoch": 1.95,
1252
+ "learning_rate": 4.0088601892176006e-05,
1253
+ "loss": 0.0418,
1254
+ "step": 2060
1255
+ },
1256
+ {
1257
+ "epoch": 1.96,
1258
+ "learning_rate": 4.0013515542874305e-05,
1259
+ "loss": 0.0429,
1260
+ "step": 2070
1261
+ },
1262
+ {
1263
+ "epoch": 1.97,
1264
+ "learning_rate": 3.993842919357261e-05,
1265
+ "loss": 0.0283,
1266
+ "step": 2080
1267
+ },
1268
+ {
1269
+ "epoch": 1.98,
1270
+ "learning_rate": 3.986334284427091e-05,
1271
+ "loss": 0.013,
1272
+ "step": 2090
1273
+ },
1274
+ {
1275
+ "epoch": 1.99,
1276
+ "learning_rate": 3.978825649496922e-05,
1277
+ "loss": 0.0157,
1278
+ "step": 2100
1279
+ },
1280
+ {
1281
+ "epoch": 1.99,
1282
+ "learning_rate": 3.9713170145667525e-05,
1283
+ "loss": 0.0254,
1284
+ "step": 2110
1285
+ },
1286
+ {
1287
+ "epoch": 2.0,
1288
+ "eval_accuracy": 0.998589562764457,
1289
+ "eval_loss": 0.0042145391926169395,
1290
+ "eval_runtime": 138.9707,
1291
+ "eval_samples_per_second": 163.257,
1292
+ "eval_steps_per_second": 10.204,
1293
+ "step": 2115
1294
+ },
1295
+ {
1296
+ "epoch": 2.0,
1297
+ "learning_rate": 3.9638083796365824e-05,
1298
+ "loss": 0.0268,
1299
+ "step": 2120
1300
+ },
1301
+ {
1302
+ "epoch": 2.01,
1303
+ "learning_rate": 3.9562997447064124e-05,
1304
+ "loss": 0.0169,
1305
+ "step": 2130
1306
+ },
1307
+ {
1308
+ "epoch": 2.02,
1309
+ "learning_rate": 3.948791109776243e-05,
1310
+ "loss": 0.0421,
1311
+ "step": 2140
1312
+ },
1313
+ {
1314
+ "epoch": 2.03,
1315
+ "learning_rate": 3.941282474846073e-05,
1316
+ "loss": 0.0083,
1317
+ "step": 2150
1318
+ },
1319
+ {
1320
+ "epoch": 2.04,
1321
+ "learning_rate": 3.933773839915904e-05,
1322
+ "loss": 0.0348,
1323
+ "step": 2160
1324
+ },
1325
+ {
1326
+ "epoch": 2.05,
1327
+ "learning_rate": 3.926265204985734e-05,
1328
+ "loss": 0.0357,
1329
+ "step": 2170
1330
+ },
1331
+ {
1332
+ "epoch": 2.06,
1333
+ "learning_rate": 3.918756570055564e-05,
1334
+ "loss": 0.0524,
1335
+ "step": 2180
1336
+ },
1337
+ {
1338
+ "epoch": 2.07,
1339
+ "learning_rate": 3.911247935125394e-05,
1340
+ "loss": 0.0376,
1341
+ "step": 2190
1342
+ },
1343
+ {
1344
+ "epoch": 2.08,
1345
+ "learning_rate": 3.903739300195225e-05,
1346
+ "loss": 0.0447,
1347
+ "step": 2200
1348
+ },
1349
+ {
1350
+ "epoch": 2.09,
1351
+ "learning_rate": 3.896230665265055e-05,
1352
+ "loss": 0.045,
1353
+ "step": 2210
1354
+ },
1355
+ {
1356
+ "epoch": 2.1,
1357
+ "learning_rate": 3.8887220303348855e-05,
1358
+ "loss": 0.0199,
1359
+ "step": 2220
1360
+ },
1361
+ {
1362
+ "epoch": 2.11,
1363
+ "learning_rate": 3.8812133954047155e-05,
1364
+ "loss": 0.0499,
1365
+ "step": 2230
1366
+ },
1367
+ {
1368
+ "epoch": 2.12,
1369
+ "learning_rate": 3.8737047604745455e-05,
1370
+ "loss": 0.0406,
1371
+ "step": 2240
1372
+ },
1373
+ {
1374
+ "epoch": 2.13,
1375
+ "learning_rate": 3.866196125544376e-05,
1376
+ "loss": 0.0305,
1377
+ "step": 2250
1378
+ },
1379
+ {
1380
+ "epoch": 2.14,
1381
+ "learning_rate": 3.858687490614207e-05,
1382
+ "loss": 0.007,
1383
+ "step": 2260
1384
+ },
1385
+ {
1386
+ "epoch": 2.15,
1387
+ "learning_rate": 3.851178855684037e-05,
1388
+ "loss": 0.0193,
1389
+ "step": 2270
1390
+ },
1391
+ {
1392
+ "epoch": 2.16,
1393
+ "learning_rate": 3.8436702207538674e-05,
1394
+ "loss": 0.0384,
1395
+ "step": 2280
1396
+ },
1397
+ {
1398
+ "epoch": 2.16,
1399
+ "learning_rate": 3.8361615858236974e-05,
1400
+ "loss": 0.0372,
1401
+ "step": 2290
1402
+ },
1403
+ {
1404
+ "epoch": 2.17,
1405
+ "learning_rate": 3.8286529508935273e-05,
1406
+ "loss": 0.0535,
1407
+ "step": 2300
1408
+ },
1409
+ {
1410
+ "epoch": 2.18,
1411
+ "learning_rate": 3.821144315963358e-05,
1412
+ "loss": 0.0611,
1413
+ "step": 2310
1414
+ },
1415
+ {
1416
+ "epoch": 2.19,
1417
+ "learning_rate": 3.8136356810331886e-05,
1418
+ "loss": 0.027,
1419
+ "step": 2320
1420
+ },
1421
+ {
1422
+ "epoch": 2.2,
1423
+ "learning_rate": 3.8061270461030186e-05,
1424
+ "loss": 0.0239,
1425
+ "step": 2330
1426
+ },
1427
+ {
1428
+ "epoch": 2.21,
1429
+ "learning_rate": 3.798618411172849e-05,
1430
+ "loss": 0.0183,
1431
+ "step": 2340
1432
+ },
1433
+ {
1434
+ "epoch": 2.22,
1435
+ "learning_rate": 3.791109776242679e-05,
1436
+ "loss": 0.0107,
1437
+ "step": 2350
1438
+ },
1439
+ {
1440
+ "epoch": 2.23,
1441
+ "learning_rate": 3.783601141312509e-05,
1442
+ "loss": 0.0374,
1443
+ "step": 2360
1444
+ },
1445
+ {
1446
+ "epoch": 2.24,
1447
+ "learning_rate": 3.77609250638234e-05,
1448
+ "loss": 0.0199,
1449
+ "step": 2370
1450
+ },
1451
+ {
1452
+ "epoch": 2.25,
1453
+ "learning_rate": 3.7685838714521705e-05,
1454
+ "loss": 0.0367,
1455
+ "step": 2380
1456
+ },
1457
+ {
1458
+ "epoch": 2.26,
1459
+ "learning_rate": 3.7610752365220005e-05,
1460
+ "loss": 0.0269,
1461
+ "step": 2390
1462
+ },
1463
+ {
1464
+ "epoch": 2.27,
1465
+ "learning_rate": 3.753566601591831e-05,
1466
+ "loss": 0.0529,
1467
+ "step": 2400
1468
+ },
1469
+ {
1470
+ "epoch": 2.28,
1471
+ "learning_rate": 3.746057966661661e-05,
1472
+ "loss": 0.0197,
1473
+ "step": 2410
1474
+ },
1475
+ {
1476
+ "epoch": 2.29,
1477
+ "learning_rate": 3.738549331731491e-05,
1478
+ "loss": 0.0146,
1479
+ "step": 2420
1480
+ },
1481
+ {
1482
+ "epoch": 2.3,
1483
+ "learning_rate": 3.731040696801322e-05,
1484
+ "loss": 0.0214,
1485
+ "step": 2430
1486
+ },
1487
+ {
1488
+ "epoch": 2.31,
1489
+ "learning_rate": 3.7235320618711524e-05,
1490
+ "loss": 0.0171,
1491
+ "step": 2440
1492
+ },
1493
+ {
1494
+ "epoch": 2.32,
1495
+ "learning_rate": 3.7160234269409823e-05,
1496
+ "loss": 0.0322,
1497
+ "step": 2450
1498
+ },
1499
+ {
1500
+ "epoch": 2.33,
1501
+ "learning_rate": 3.708514792010812e-05,
1502
+ "loss": 0.0193,
1503
+ "step": 2460
1504
+ },
1505
+ {
1506
+ "epoch": 2.34,
1507
+ "learning_rate": 3.701006157080643e-05,
1508
+ "loss": 0.0121,
1509
+ "step": 2470
1510
+ },
1511
+ {
1512
+ "epoch": 2.34,
1513
+ "learning_rate": 3.693497522150473e-05,
1514
+ "loss": 0.0144,
1515
+ "step": 2480
1516
+ },
1517
+ {
1518
+ "epoch": 2.35,
1519
+ "learning_rate": 3.6859888872203036e-05,
1520
+ "loss": 0.0289,
1521
+ "step": 2490
1522
+ },
1523
+ {
1524
+ "epoch": 2.36,
1525
+ "learning_rate": 3.678480252290134e-05,
1526
+ "loss": 0.0248,
1527
+ "step": 2500
1528
+ },
1529
+ {
1530
+ "epoch": 2.37,
1531
+ "learning_rate": 3.670971617359964e-05,
1532
+ "loss": 0.0157,
1533
+ "step": 2510
1534
+ },
1535
+ {
1536
+ "epoch": 2.38,
1537
+ "learning_rate": 3.663462982429794e-05,
1538
+ "loss": 0.0261,
1539
+ "step": 2520
1540
+ },
1541
+ {
1542
+ "epoch": 2.39,
1543
+ "learning_rate": 3.655954347499625e-05,
1544
+ "loss": 0.0223,
1545
+ "step": 2530
1546
+ },
1547
+ {
1548
+ "epoch": 2.4,
1549
+ "learning_rate": 3.648445712569455e-05,
1550
+ "loss": 0.017,
1551
+ "step": 2540
1552
+ },
1553
+ {
1554
+ "epoch": 2.41,
1555
+ "learning_rate": 3.6409370776392855e-05,
1556
+ "loss": 0.0128,
1557
+ "step": 2550
1558
+ },
1559
+ {
1560
+ "epoch": 2.42,
1561
+ "learning_rate": 3.633428442709116e-05,
1562
+ "loss": 0.0206,
1563
+ "step": 2560
1564
+ },
1565
+ {
1566
+ "epoch": 2.43,
1567
+ "learning_rate": 3.625919807778946e-05,
1568
+ "loss": 0.0297,
1569
+ "step": 2570
1570
+ },
1571
+ {
1572
+ "epoch": 2.44,
1573
+ "learning_rate": 3.618411172848776e-05,
1574
+ "loss": 0.0299,
1575
+ "step": 2580
1576
+ },
1577
+ {
1578
+ "epoch": 2.45,
1579
+ "learning_rate": 3.610902537918607e-05,
1580
+ "loss": 0.019,
1581
+ "step": 2590
1582
+ },
1583
+ {
1584
+ "epoch": 2.46,
1585
+ "learning_rate": 3.603393902988437e-05,
1586
+ "loss": 0.0265,
1587
+ "step": 2600
1588
+ },
1589
+ {
1590
+ "epoch": 2.47,
1591
+ "learning_rate": 3.595885268058267e-05,
1592
+ "loss": 0.0039,
1593
+ "step": 2610
1594
+ },
1595
+ {
1596
+ "epoch": 2.48,
1597
+ "learning_rate": 3.588376633128098e-05,
1598
+ "loss": 0.0169,
1599
+ "step": 2620
1600
+ },
1601
+ {
1602
+ "epoch": 2.49,
1603
+ "learning_rate": 3.580867998197928e-05,
1604
+ "loss": 0.0377,
1605
+ "step": 2630
1606
+ },
1607
+ {
1608
+ "epoch": 2.5,
1609
+ "learning_rate": 3.573359363267758e-05,
1610
+ "loss": 0.0057,
1611
+ "step": 2640
1612
+ },
1613
+ {
1614
+ "epoch": 2.51,
1615
+ "learning_rate": 3.5658507283375886e-05,
1616
+ "loss": 0.0567,
1617
+ "step": 2650
1618
+ },
1619
+ {
1620
+ "epoch": 2.51,
1621
+ "learning_rate": 3.5583420934074185e-05,
1622
+ "loss": 0.0279,
1623
+ "step": 2660
1624
+ },
1625
+ {
1626
+ "epoch": 2.52,
1627
+ "learning_rate": 3.550833458477249e-05,
1628
+ "loss": 0.0146,
1629
+ "step": 2670
1630
+ },
1631
+ {
1632
+ "epoch": 2.53,
1633
+ "learning_rate": 3.54332482354708e-05,
1634
+ "loss": 0.0091,
1635
+ "step": 2680
1636
+ },
1637
+ {
1638
+ "epoch": 2.54,
1639
+ "learning_rate": 3.535816188616909e-05,
1640
+ "loss": 0.0222,
1641
+ "step": 2690
1642
+ },
1643
+ {
1644
+ "epoch": 2.55,
1645
+ "learning_rate": 3.52830755368674e-05,
1646
+ "loss": 0.0205,
1647
+ "step": 2700
1648
+ },
1649
+ {
1650
+ "epoch": 2.56,
1651
+ "learning_rate": 3.5207989187565704e-05,
1652
+ "loss": 0.0134,
1653
+ "step": 2710
1654
+ },
1655
+ {
1656
+ "epoch": 2.57,
1657
+ "learning_rate": 3.5132902838264004e-05,
1658
+ "loss": 0.0132,
1659
+ "step": 2720
1660
+ },
1661
+ {
1662
+ "epoch": 2.58,
1663
+ "learning_rate": 3.505781648896231e-05,
1664
+ "loss": 0.0219,
1665
+ "step": 2730
1666
+ },
1667
+ {
1668
+ "epoch": 2.59,
1669
+ "learning_rate": 3.498273013966061e-05,
1670
+ "loss": 0.0024,
1671
+ "step": 2740
1672
+ },
1673
+ {
1674
+ "epoch": 2.6,
1675
+ "learning_rate": 3.490764379035891e-05,
1676
+ "loss": 0.0214,
1677
+ "step": 2750
1678
+ },
1679
+ {
1680
+ "epoch": 2.61,
1681
+ "learning_rate": 3.4832557441057216e-05,
1682
+ "loss": 0.0126,
1683
+ "step": 2760
1684
+ },
1685
+ {
1686
+ "epoch": 2.62,
1687
+ "learning_rate": 3.475747109175552e-05,
1688
+ "loss": 0.0065,
1689
+ "step": 2770
1690
+ },
1691
+ {
1692
+ "epoch": 2.63,
1693
+ "learning_rate": 3.468238474245382e-05,
1694
+ "loss": 0.0141,
1695
+ "step": 2780
1696
+ },
1697
+ {
1698
+ "epoch": 2.64,
1699
+ "learning_rate": 3.460729839315213e-05,
1700
+ "loss": 0.0346,
1701
+ "step": 2790
1702
+ },
1703
+ {
1704
+ "epoch": 2.65,
1705
+ "learning_rate": 3.453221204385043e-05,
1706
+ "loss": 0.0123,
1707
+ "step": 2800
1708
+ },
1709
+ {
1710
+ "epoch": 2.66,
1711
+ "learning_rate": 3.445712569454873e-05,
1712
+ "loss": 0.0111,
1713
+ "step": 2810
1714
+ },
1715
+ {
1716
+ "epoch": 2.67,
1717
+ "learning_rate": 3.4382039345247035e-05,
1718
+ "loss": 0.0212,
1719
+ "step": 2820
1720
+ },
1721
+ {
1722
+ "epoch": 2.68,
1723
+ "learning_rate": 3.430695299594534e-05,
1724
+ "loss": 0.0129,
1725
+ "step": 2830
1726
+ },
1727
+ {
1728
+ "epoch": 2.68,
1729
+ "learning_rate": 3.423186664664364e-05,
1730
+ "loss": 0.0221,
1731
+ "step": 2840
1732
+ },
1733
+ {
1734
+ "epoch": 2.69,
1735
+ "learning_rate": 3.415678029734195e-05,
1736
+ "loss": 0.0353,
1737
+ "step": 2850
1738
+ },
1739
+ {
1740
+ "epoch": 2.7,
1741
+ "learning_rate": 3.408169394804025e-05,
1742
+ "loss": 0.0134,
1743
+ "step": 2860
1744
+ },
1745
+ {
1746
+ "epoch": 2.71,
1747
+ "learning_rate": 3.400660759873855e-05,
1748
+ "loss": 0.0152,
1749
+ "step": 2870
1750
+ },
1751
+ {
1752
+ "epoch": 2.72,
1753
+ "learning_rate": 3.3931521249436854e-05,
1754
+ "loss": 0.0215,
1755
+ "step": 2880
1756
+ },
1757
+ {
1758
+ "epoch": 2.73,
1759
+ "learning_rate": 3.385643490013516e-05,
1760
+ "loss": 0.0188,
1761
+ "step": 2890
1762
+ },
1763
+ {
1764
+ "epoch": 2.74,
1765
+ "learning_rate": 3.378134855083346e-05,
1766
+ "loss": 0.0246,
1767
+ "step": 2900
1768
+ },
1769
+ {
1770
+ "epoch": 2.75,
1771
+ "learning_rate": 3.3706262201531766e-05,
1772
+ "loss": 0.0219,
1773
+ "step": 2910
1774
+ },
1775
+ {
1776
+ "epoch": 2.76,
1777
+ "learning_rate": 3.3631175852230066e-05,
1778
+ "loss": 0.0367,
1779
+ "step": 2920
1780
+ },
1781
+ {
1782
+ "epoch": 2.77,
1783
+ "learning_rate": 3.3556089502928366e-05,
1784
+ "loss": 0.0231,
1785
+ "step": 2930
1786
+ },
1787
+ {
1788
+ "epoch": 2.78,
1789
+ "learning_rate": 3.348100315362667e-05,
1790
+ "loss": 0.019,
1791
+ "step": 2940
1792
+ },
1793
+ {
1794
+ "epoch": 2.79,
1795
+ "learning_rate": 3.340591680432498e-05,
1796
+ "loss": 0.0128,
1797
+ "step": 2950
1798
+ },
1799
+ {
1800
+ "epoch": 2.8,
1801
+ "learning_rate": 3.333083045502328e-05,
1802
+ "loss": 0.0333,
1803
+ "step": 2960
1804
+ },
1805
+ {
1806
+ "epoch": 2.81,
1807
+ "learning_rate": 3.325574410572158e-05,
1808
+ "loss": 0.0098,
1809
+ "step": 2970
1810
+ },
1811
+ {
1812
+ "epoch": 2.82,
1813
+ "learning_rate": 3.3180657756419885e-05,
1814
+ "loss": 0.0255,
1815
+ "step": 2980
1816
+ },
1817
+ {
1818
+ "epoch": 2.83,
1819
+ "learning_rate": 3.3105571407118184e-05,
1820
+ "loss": 0.0163,
1821
+ "step": 2990
1822
+ },
1823
+ {
1824
+ "epoch": 2.84,
1825
+ "learning_rate": 3.303048505781649e-05,
1826
+ "loss": 0.0123,
1827
+ "step": 3000
1828
+ },
1829
+ {
1830
+ "epoch": 2.85,
1831
+ "learning_rate": 3.29553987085148e-05,
1832
+ "loss": 0.0083,
1833
+ "step": 3010
1834
+ },
1835
+ {
1836
+ "epoch": 2.86,
1837
+ "learning_rate": 3.28803123592131e-05,
1838
+ "loss": 0.005,
1839
+ "step": 3020
1840
+ },
1841
+ {
1842
+ "epoch": 2.86,
1843
+ "learning_rate": 3.28052260099114e-05,
1844
+ "loss": 0.0026,
1845
+ "step": 3030
1846
+ },
1847
+ {
1848
+ "epoch": 2.87,
1849
+ "learning_rate": 3.27301396606097e-05,
1850
+ "loss": 0.0241,
1851
+ "step": 3040
1852
+ },
1853
+ {
1854
+ "epoch": 2.88,
1855
+ "learning_rate": 3.2655053311308e-05,
1856
+ "loss": 0.0106,
1857
+ "step": 3050
1858
+ },
1859
+ {
1860
+ "epoch": 2.89,
1861
+ "learning_rate": 3.257996696200631e-05,
1862
+ "loss": 0.0071,
1863
+ "step": 3060
1864
+ },
1865
+ {
1866
+ "epoch": 2.9,
1867
+ "learning_rate": 3.2504880612704616e-05,
1868
+ "loss": 0.0364,
1869
+ "step": 3070
1870
+ },
1871
+ {
1872
+ "epoch": 2.91,
1873
+ "learning_rate": 3.2429794263402916e-05,
1874
+ "loss": 0.0113,
1875
+ "step": 3080
1876
+ },
1877
+ {
1878
+ "epoch": 2.92,
1879
+ "learning_rate": 3.2354707914101215e-05,
1880
+ "loss": 0.0066,
1881
+ "step": 3090
1882
+ },
1883
+ {
1884
+ "epoch": 2.93,
1885
+ "learning_rate": 3.227962156479952e-05,
1886
+ "loss": 0.0267,
1887
+ "step": 3100
1888
+ },
1889
+ {
1890
+ "epoch": 2.94,
1891
+ "learning_rate": 3.220453521549782e-05,
1892
+ "loss": 0.0248,
1893
+ "step": 3110
1894
+ },
1895
+ {
1896
+ "epoch": 2.95,
1897
+ "learning_rate": 3.212944886619613e-05,
1898
+ "loss": 0.0371,
1899
+ "step": 3120
1900
+ },
1901
+ {
1902
+ "epoch": 2.96,
1903
+ "learning_rate": 3.2054362516894435e-05,
1904
+ "loss": 0.0124,
1905
+ "step": 3130
1906
+ },
1907
+ {
1908
+ "epoch": 2.97,
1909
+ "learning_rate": 3.1979276167592734e-05,
1910
+ "loss": 0.0237,
1911
+ "step": 3140
1912
+ },
1913
+ {
1914
+ "epoch": 2.98,
1915
+ "learning_rate": 3.1904189818291034e-05,
1916
+ "loss": 0.0228,
1917
+ "step": 3150
1918
+ },
1919
+ {
1920
+ "epoch": 2.99,
1921
+ "learning_rate": 3.182910346898934e-05,
1922
+ "loss": 0.0079,
1923
+ "step": 3160
1924
+ },
1925
+ {
1926
+ "epoch": 3.0,
1927
+ "learning_rate": 3.175401711968764e-05,
1928
+ "loss": 0.026,
1929
+ "step": 3170
1930
+ },
1931
+ {
1932
+ "epoch": 3.0,
1933
+ "eval_accuracy": 0.9989862482369535,
1934
+ "eval_loss": 0.0020762328058481216,
1935
+ "eval_runtime": 141.2388,
1936
+ "eval_samples_per_second": 160.636,
1937
+ "eval_steps_per_second": 10.04,
1938
+ "step": 3173
1939
+ },
1940
+ {
1941
+ "epoch": 3.01,
1942
+ "learning_rate": 3.167893077038595e-05,
1943
+ "loss": 0.0156,
1944
+ "step": 3180
1945
+ },
1946
+ {
1947
+ "epoch": 3.02,
1948
+ "learning_rate": 3.1603844421084247e-05,
1949
+ "loss": 0.0103,
1950
+ "step": 3190
1951
+ },
1952
+ {
1953
+ "epoch": 3.03,
1954
+ "learning_rate": 3.1528758071782546e-05,
1955
+ "loss": 0.0328,
1956
+ "step": 3200
1957
+ },
1958
+ {
1959
+ "epoch": 3.03,
1960
+ "learning_rate": 3.145367172248085e-05,
1961
+ "loss": 0.0233,
1962
+ "step": 3210
1963
+ },
1964
+ {
1965
+ "epoch": 3.04,
1966
+ "learning_rate": 3.137858537317916e-05,
1967
+ "loss": 0.0165,
1968
+ "step": 3220
1969
+ },
1970
+ {
1971
+ "epoch": 3.05,
1972
+ "learning_rate": 3.130349902387746e-05,
1973
+ "loss": 0.0089,
1974
+ "step": 3230
1975
+ },
1976
+ {
1977
+ "epoch": 3.06,
1978
+ "learning_rate": 3.1228412674575765e-05,
1979
+ "loss": 0.0021,
1980
+ "step": 3240
1981
+ },
1982
+ {
1983
+ "epoch": 3.07,
1984
+ "learning_rate": 3.1153326325274065e-05,
1985
+ "loss": 0.0015,
1986
+ "step": 3250
1987
+ },
1988
+ {
1989
+ "epoch": 3.08,
1990
+ "learning_rate": 3.1078239975972365e-05,
1991
+ "loss": 0.0224,
1992
+ "step": 3260
1993
+ },
1994
+ {
1995
+ "epoch": 3.09,
1996
+ "learning_rate": 3.100315362667067e-05,
1997
+ "loss": 0.01,
1998
+ "step": 3270
1999
+ },
2000
+ {
2001
+ "epoch": 3.1,
2002
+ "learning_rate": 3.092806727736898e-05,
2003
+ "loss": 0.0097,
2004
+ "step": 3280
2005
+ },
2006
+ {
2007
+ "epoch": 3.11,
2008
+ "learning_rate": 3.085298092806728e-05,
2009
+ "loss": 0.0223,
2010
+ "step": 3290
2011
+ },
2012
+ {
2013
+ "epoch": 3.12,
2014
+ "learning_rate": 3.0777894578765584e-05,
2015
+ "loss": 0.015,
2016
+ "step": 3300
2017
+ },
2018
+ {
2019
+ "epoch": 3.13,
2020
+ "learning_rate": 3.0702808229463884e-05,
2021
+ "loss": 0.0105,
2022
+ "step": 3310
2023
+ },
2024
+ {
2025
+ "epoch": 3.14,
2026
+ "learning_rate": 3.0627721880162184e-05,
2027
+ "loss": 0.0237,
2028
+ "step": 3320
2029
+ },
2030
+ {
2031
+ "epoch": 3.15,
2032
+ "learning_rate": 3.055263553086049e-05,
2033
+ "loss": 0.0421,
2034
+ "step": 3330
2035
+ },
2036
+ {
2037
+ "epoch": 3.16,
2038
+ "learning_rate": 3.0477549181558797e-05,
2039
+ "loss": 0.0199,
2040
+ "step": 3340
2041
+ },
2042
+ {
2043
+ "epoch": 3.17,
2044
+ "learning_rate": 3.0402462832257096e-05,
2045
+ "loss": 0.0229,
2046
+ "step": 3350
2047
+ },
2048
+ {
2049
+ "epoch": 3.18,
2050
+ "learning_rate": 3.03273764829554e-05,
2051
+ "loss": 0.0093,
2052
+ "step": 3360
2053
+ },
2054
+ {
2055
+ "epoch": 3.19,
2056
+ "learning_rate": 3.0252290133653706e-05,
2057
+ "loss": 0.0202,
2058
+ "step": 3370
2059
+ },
2060
+ {
2061
+ "epoch": 3.2,
2062
+ "learning_rate": 3.0177203784352006e-05,
2063
+ "loss": 0.0228,
2064
+ "step": 3380
2065
+ },
2066
+ {
2067
+ "epoch": 3.2,
2068
+ "learning_rate": 3.010211743505031e-05,
2069
+ "loss": 0.014,
2070
+ "step": 3390
2071
+ },
2072
+ {
2073
+ "epoch": 3.21,
2074
+ "learning_rate": 3.0027031085748615e-05,
2075
+ "loss": 0.0093,
2076
+ "step": 3400
2077
+ },
2078
+ {
2079
+ "epoch": 3.22,
2080
+ "learning_rate": 2.995194473644691e-05,
2081
+ "loss": 0.0041,
2082
+ "step": 3410
2083
+ },
2084
+ {
2085
+ "epoch": 3.23,
2086
+ "learning_rate": 2.9876858387145218e-05,
2087
+ "loss": 0.0102,
2088
+ "step": 3420
2089
+ },
2090
+ {
2091
+ "epoch": 3.24,
2092
+ "learning_rate": 2.980177203784352e-05,
2093
+ "loss": 0.0138,
2094
+ "step": 3430
2095
+ },
2096
+ {
2097
+ "epoch": 3.25,
2098
+ "learning_rate": 2.972668568854182e-05,
2099
+ "loss": 0.0183,
2100
+ "step": 3440
2101
+ },
2102
+ {
2103
+ "epoch": 3.26,
2104
+ "learning_rate": 2.9651599339240127e-05,
2105
+ "loss": 0.01,
2106
+ "step": 3450
2107
+ },
2108
+ {
2109
+ "epoch": 3.27,
2110
+ "learning_rate": 2.957651298993843e-05,
2111
+ "loss": 0.0218,
2112
+ "step": 3460
2113
+ },
2114
+ {
2115
+ "epoch": 3.28,
2116
+ "learning_rate": 2.950142664063673e-05,
2117
+ "loss": 0.0285,
2118
+ "step": 3470
2119
+ },
2120
+ {
2121
+ "epoch": 3.29,
2122
+ "learning_rate": 2.9426340291335037e-05,
2123
+ "loss": 0.015,
2124
+ "step": 3480
2125
+ },
2126
+ {
2127
+ "epoch": 3.3,
2128
+ "learning_rate": 2.935125394203334e-05,
2129
+ "loss": 0.0077,
2130
+ "step": 3490
2131
+ },
2132
+ {
2133
+ "epoch": 3.31,
2134
+ "learning_rate": 2.927616759273164e-05,
2135
+ "loss": 0.0106,
2136
+ "step": 3500
2137
+ },
2138
+ {
2139
+ "epoch": 3.32,
2140
+ "learning_rate": 2.9201081243429946e-05,
2141
+ "loss": 0.0087,
2142
+ "step": 3510
2143
+ },
2144
+ {
2145
+ "epoch": 3.33,
2146
+ "learning_rate": 2.912599489412825e-05,
2147
+ "loss": 0.023,
2148
+ "step": 3520
2149
+ },
2150
+ {
2151
+ "epoch": 3.34,
2152
+ "learning_rate": 2.905090854482655e-05,
2153
+ "loss": 0.0262,
2154
+ "step": 3530
2155
+ },
2156
+ {
2157
+ "epoch": 3.35,
2158
+ "learning_rate": 2.8975822195524855e-05,
2159
+ "loss": 0.0358,
2160
+ "step": 3540
2161
+ },
2162
+ {
2163
+ "epoch": 3.36,
2164
+ "learning_rate": 2.890073584622316e-05,
2165
+ "loss": 0.0093,
2166
+ "step": 3550
2167
+ },
2168
+ {
2169
+ "epoch": 3.37,
2170
+ "learning_rate": 2.8825649496921458e-05,
2171
+ "loss": 0.0179,
2172
+ "step": 3560
2173
+ },
2174
+ {
2175
+ "epoch": 3.38,
2176
+ "learning_rate": 2.8750563147619765e-05,
2177
+ "loss": 0.0172,
2178
+ "step": 3570
2179
+ },
2180
+ {
2181
+ "epoch": 3.38,
2182
+ "learning_rate": 2.8675476798318068e-05,
2183
+ "loss": 0.0043,
2184
+ "step": 3580
2185
+ },
2186
+ {
2187
+ "epoch": 3.39,
2188
+ "learning_rate": 2.8600390449016367e-05,
2189
+ "loss": 0.0134,
2190
+ "step": 3590
2191
+ },
2192
+ {
2193
+ "epoch": 3.4,
2194
+ "learning_rate": 2.8525304099714674e-05,
2195
+ "loss": 0.0317,
2196
+ "step": 3600
2197
+ },
2198
+ {
2199
+ "epoch": 3.41,
2200
+ "learning_rate": 2.8450217750412977e-05,
2201
+ "loss": 0.0153,
2202
+ "step": 3610
2203
+ },
2204
+ {
2205
+ "epoch": 3.42,
2206
+ "learning_rate": 2.8375131401111277e-05,
2207
+ "loss": 0.0106,
2208
+ "step": 3620
2209
+ },
2210
+ {
2211
+ "epoch": 3.43,
2212
+ "learning_rate": 2.8300045051809583e-05,
2213
+ "loss": 0.0121,
2214
+ "step": 3630
2215
+ },
2216
+ {
2217
+ "epoch": 3.44,
2218
+ "learning_rate": 2.8224958702507886e-05,
2219
+ "loss": 0.0071,
2220
+ "step": 3640
2221
+ },
2222
+ {
2223
+ "epoch": 3.45,
2224
+ "learning_rate": 2.8149872353206186e-05,
2225
+ "loss": 0.0006,
2226
+ "step": 3650
2227
+ },
2228
+ {
2229
+ "epoch": 3.46,
2230
+ "learning_rate": 2.807478600390449e-05,
2231
+ "loss": 0.0144,
2232
+ "step": 3660
2233
+ },
2234
+ {
2235
+ "epoch": 3.47,
2236
+ "learning_rate": 2.7999699654602796e-05,
2237
+ "loss": 0.0075,
2238
+ "step": 3670
2239
+ },
2240
+ {
2241
+ "epoch": 3.48,
2242
+ "learning_rate": 2.7924613305301095e-05,
2243
+ "loss": 0.0089,
2244
+ "step": 3680
2245
+ },
2246
+ {
2247
+ "epoch": 3.49,
2248
+ "learning_rate": 2.78495269559994e-05,
2249
+ "loss": 0.0118,
2250
+ "step": 3690
2251
+ },
2252
+ {
2253
+ "epoch": 3.5,
2254
+ "learning_rate": 2.7774440606697705e-05,
2255
+ "loss": 0.0035,
2256
+ "step": 3700
2257
+ },
2258
+ {
2259
+ "epoch": 3.51,
2260
+ "learning_rate": 2.7699354257396005e-05,
2261
+ "loss": 0.0066,
2262
+ "step": 3710
2263
+ },
2264
+ {
2265
+ "epoch": 3.52,
2266
+ "learning_rate": 2.7624267908094308e-05,
2267
+ "loss": 0.022,
2268
+ "step": 3720
2269
+ },
2270
+ {
2271
+ "epoch": 3.53,
2272
+ "learning_rate": 2.7549181558792614e-05,
2273
+ "loss": 0.0018,
2274
+ "step": 3730
2275
+ },
2276
+ {
2277
+ "epoch": 3.54,
2278
+ "learning_rate": 2.7474095209490914e-05,
2279
+ "loss": 0.0117,
2280
+ "step": 3740
2281
+ },
2282
+ {
2283
+ "epoch": 3.55,
2284
+ "learning_rate": 2.7399008860189217e-05,
2285
+ "loss": 0.0137,
2286
+ "step": 3750
2287
+ },
2288
+ {
2289
+ "epoch": 3.55,
2290
+ "learning_rate": 2.7323922510887524e-05,
2291
+ "loss": 0.0143,
2292
+ "step": 3760
2293
+ },
2294
+ {
2295
+ "epoch": 3.56,
2296
+ "learning_rate": 2.7248836161585823e-05,
2297
+ "loss": 0.0214,
2298
+ "step": 3770
2299
+ },
2300
+ {
2301
+ "epoch": 3.57,
2302
+ "learning_rate": 2.7173749812284126e-05,
2303
+ "loss": 0.0051,
2304
+ "step": 3780
2305
+ },
2306
+ {
2307
+ "epoch": 3.58,
2308
+ "learning_rate": 2.7098663462982433e-05,
2309
+ "loss": 0.0197,
2310
+ "step": 3790
2311
+ },
2312
+ {
2313
+ "epoch": 3.59,
2314
+ "learning_rate": 2.7023577113680733e-05,
2315
+ "loss": 0.0081,
2316
+ "step": 3800
2317
+ },
2318
+ {
2319
+ "epoch": 3.6,
2320
+ "learning_rate": 2.6948490764379036e-05,
2321
+ "loss": 0.0225,
2322
+ "step": 3810
2323
+ },
2324
+ {
2325
+ "epoch": 3.61,
2326
+ "learning_rate": 2.6873404415077342e-05,
2327
+ "loss": 0.0101,
2328
+ "step": 3820
2329
+ },
2330
+ {
2331
+ "epoch": 3.62,
2332
+ "learning_rate": 2.6798318065775642e-05,
2333
+ "loss": 0.0162,
2334
+ "step": 3830
2335
+ },
2336
+ {
2337
+ "epoch": 3.63,
2338
+ "learning_rate": 2.6723231716473945e-05,
2339
+ "loss": 0.0171,
2340
+ "step": 3840
2341
+ },
2342
+ {
2343
+ "epoch": 3.64,
2344
+ "learning_rate": 2.664814536717225e-05,
2345
+ "loss": 0.0091,
2346
+ "step": 3850
2347
+ },
2348
+ {
2349
+ "epoch": 3.65,
2350
+ "learning_rate": 2.657305901787055e-05,
2351
+ "loss": 0.0066,
2352
+ "step": 3860
2353
+ },
2354
+ {
2355
+ "epoch": 3.66,
2356
+ "learning_rate": 2.6497972668568854e-05,
2357
+ "loss": 0.0077,
2358
+ "step": 3870
2359
+ },
2360
+ {
2361
+ "epoch": 3.67,
2362
+ "learning_rate": 2.642288631926716e-05,
2363
+ "loss": 0.0278,
2364
+ "step": 3880
2365
+ },
2366
+ {
2367
+ "epoch": 3.68,
2368
+ "learning_rate": 2.6347799969965464e-05,
2369
+ "loss": 0.0168,
2370
+ "step": 3890
2371
+ },
2372
+ {
2373
+ "epoch": 3.69,
2374
+ "learning_rate": 2.6272713620663764e-05,
2375
+ "loss": 0.0441,
2376
+ "step": 3900
2377
+ },
2378
+ {
2379
+ "epoch": 3.7,
2380
+ "learning_rate": 2.619762727136207e-05,
2381
+ "loss": 0.0035,
2382
+ "step": 3910
2383
+ },
2384
+ {
2385
+ "epoch": 3.71,
2386
+ "learning_rate": 2.6122540922060373e-05,
2387
+ "loss": 0.0035,
2388
+ "step": 3920
2389
+ },
2390
+ {
2391
+ "epoch": 3.72,
2392
+ "learning_rate": 2.6047454572758673e-05,
2393
+ "loss": 0.0053,
2394
+ "step": 3930
2395
+ },
2396
+ {
2397
+ "epoch": 3.72,
2398
+ "learning_rate": 2.5972368223456976e-05,
2399
+ "loss": 0.0435,
2400
+ "step": 3940
2401
+ },
2402
+ {
2403
+ "epoch": 3.73,
2404
+ "learning_rate": 2.5897281874155283e-05,
2405
+ "loss": 0.0188,
2406
+ "step": 3950
2407
+ },
2408
+ {
2409
+ "epoch": 3.74,
2410
+ "learning_rate": 2.5822195524853582e-05,
2411
+ "loss": 0.0321,
2412
+ "step": 3960
2413
+ },
2414
+ {
2415
+ "epoch": 3.75,
2416
+ "learning_rate": 2.5747109175551885e-05,
2417
+ "loss": 0.0138,
2418
+ "step": 3970
2419
+ },
2420
+ {
2421
+ "epoch": 3.76,
2422
+ "learning_rate": 2.5672022826250192e-05,
2423
+ "loss": 0.0047,
2424
+ "step": 3980
2425
+ },
2426
+ {
2427
+ "epoch": 3.77,
2428
+ "learning_rate": 2.559693647694849e-05,
2429
+ "loss": 0.0135,
2430
+ "step": 3990
2431
+ },
2432
+ {
2433
+ "epoch": 3.78,
2434
+ "learning_rate": 2.5521850127646795e-05,
2435
+ "loss": 0.0233,
2436
+ "step": 4000
2437
+ },
2438
+ {
2439
+ "epoch": 3.79,
2440
+ "learning_rate": 2.54467637783451e-05,
2441
+ "loss": 0.0101,
2442
+ "step": 4010
2443
+ },
2444
+ {
2445
+ "epoch": 3.8,
2446
+ "learning_rate": 2.53716774290434e-05,
2447
+ "loss": 0.0113,
2448
+ "step": 4020
2449
+ },
2450
+ {
2451
+ "epoch": 3.81,
2452
+ "learning_rate": 2.5296591079741704e-05,
2453
+ "loss": 0.0016,
2454
+ "step": 4030
2455
+ },
2456
+ {
2457
+ "epoch": 3.82,
2458
+ "learning_rate": 2.522150473044001e-05,
2459
+ "loss": 0.0193,
2460
+ "step": 4040
2461
+ },
2462
+ {
2463
+ "epoch": 3.83,
2464
+ "learning_rate": 2.514641838113831e-05,
2465
+ "loss": 0.0206,
2466
+ "step": 4050
2467
+ },
2468
+ {
2469
+ "epoch": 3.84,
2470
+ "learning_rate": 2.5071332031836613e-05,
2471
+ "loss": 0.0178,
2472
+ "step": 4060
2473
+ },
2474
+ {
2475
+ "epoch": 3.85,
2476
+ "learning_rate": 2.4996245682534917e-05,
2477
+ "loss": 0.0219,
2478
+ "step": 4070
2479
+ },
2480
+ {
2481
+ "epoch": 3.86,
2482
+ "learning_rate": 2.492115933323322e-05,
2483
+ "loss": 0.0288,
2484
+ "step": 4080
2485
+ },
2486
+ {
2487
+ "epoch": 3.87,
2488
+ "learning_rate": 2.4846072983931523e-05,
2489
+ "loss": 0.0052,
2490
+ "step": 4090
2491
+ },
2492
+ {
2493
+ "epoch": 3.88,
2494
+ "learning_rate": 2.4770986634629826e-05,
2495
+ "loss": 0.0076,
2496
+ "step": 4100
2497
+ },
2498
+ {
2499
+ "epoch": 3.89,
2500
+ "learning_rate": 2.469590028532813e-05,
2501
+ "loss": 0.0154,
2502
+ "step": 4110
2503
+ },
2504
+ {
2505
+ "epoch": 3.9,
2506
+ "learning_rate": 2.4620813936026432e-05,
2507
+ "loss": 0.0056,
2508
+ "step": 4120
2509
+ },
2510
+ {
2511
+ "epoch": 3.9,
2512
+ "learning_rate": 2.4545727586724735e-05,
2513
+ "loss": 0.0021,
2514
+ "step": 4130
2515
+ },
2516
+ {
2517
+ "epoch": 3.91,
2518
+ "learning_rate": 2.4470641237423035e-05,
2519
+ "loss": 0.0372,
2520
+ "step": 4140
2521
+ },
2522
+ {
2523
+ "epoch": 3.92,
2524
+ "learning_rate": 2.439555488812134e-05,
2525
+ "loss": 0.0164,
2526
+ "step": 4150
2527
+ },
2528
+ {
2529
+ "epoch": 3.93,
2530
+ "learning_rate": 2.4320468538819644e-05,
2531
+ "loss": 0.0104,
2532
+ "step": 4160
2533
+ },
2534
+ {
2535
+ "epoch": 3.94,
2536
+ "learning_rate": 2.4245382189517944e-05,
2537
+ "loss": 0.0054,
2538
+ "step": 4170
2539
+ },
2540
+ {
2541
+ "epoch": 3.95,
2542
+ "learning_rate": 2.417029584021625e-05,
2543
+ "loss": 0.0135,
2544
+ "step": 4180
2545
+ },
2546
+ {
2547
+ "epoch": 3.96,
2548
+ "learning_rate": 2.4095209490914554e-05,
2549
+ "loss": 0.0029,
2550
+ "step": 4190
2551
+ },
2552
+ {
2553
+ "epoch": 3.97,
2554
+ "learning_rate": 2.4020123141612854e-05,
2555
+ "loss": 0.0338,
2556
+ "step": 4200
2557
+ },
2558
+ {
2559
+ "epoch": 3.98,
2560
+ "learning_rate": 2.394503679231116e-05,
2561
+ "loss": 0.0105,
2562
+ "step": 4210
2563
+ },
2564
+ {
2565
+ "epoch": 3.99,
2566
+ "learning_rate": 2.3869950443009463e-05,
2567
+ "loss": 0.0274,
2568
+ "step": 4220
2569
+ },
2570
+ {
2571
+ "epoch": 4.0,
2572
+ "learning_rate": 2.3794864093707763e-05,
2573
+ "loss": 0.0092,
2574
+ "step": 4230
2575
+ },
2576
+ {
2577
+ "epoch": 4.0,
2578
+ "eval_accuracy": 0.998589562764457,
2579
+ "eval_loss": 0.0034797259140759706,
2580
+ "eval_runtime": 143.9645,
2581
+ "eval_samples_per_second": 157.594,
2582
+ "eval_steps_per_second": 9.85,
2583
+ "step": 4231
2584
+ },
2585
+ {
2586
+ "epoch": 4.01,
2587
+ "learning_rate": 2.371977774440607e-05,
2588
+ "loss": 0.039,
2589
+ "step": 4240
2590
+ },
2591
+ {
2592
+ "epoch": 4.02,
2593
+ "learning_rate": 2.3644691395104372e-05,
2594
+ "loss": 0.0054,
2595
+ "step": 4250
2596
+ },
2597
+ {
2598
+ "epoch": 4.03,
2599
+ "learning_rate": 2.3569605045802672e-05,
2600
+ "loss": 0.0071,
2601
+ "step": 4260
2602
+ },
2603
+ {
2604
+ "epoch": 4.04,
2605
+ "learning_rate": 2.349451869650098e-05,
2606
+ "loss": 0.0221,
2607
+ "step": 4270
2608
+ },
2609
+ {
2610
+ "epoch": 4.05,
2611
+ "learning_rate": 2.341943234719928e-05,
2612
+ "loss": 0.0041,
2613
+ "step": 4280
2614
+ },
2615
+ {
2616
+ "epoch": 4.06,
2617
+ "learning_rate": 2.334434599789758e-05,
2618
+ "loss": 0.0036,
2619
+ "step": 4290
2620
+ },
2621
+ {
2622
+ "epoch": 4.07,
2623
+ "learning_rate": 2.3269259648595888e-05,
2624
+ "loss": 0.0103,
2625
+ "step": 4300
2626
+ },
2627
+ {
2628
+ "epoch": 4.07,
2629
+ "learning_rate": 2.3194173299294188e-05,
2630
+ "loss": 0.0055,
2631
+ "step": 4310
2632
+ },
2633
+ {
2634
+ "epoch": 4.08,
2635
+ "learning_rate": 2.311908694999249e-05,
2636
+ "loss": 0.0093,
2637
+ "step": 4320
2638
+ },
2639
+ {
2640
+ "epoch": 4.09,
2641
+ "learning_rate": 2.3044000600690797e-05,
2642
+ "loss": 0.0075,
2643
+ "step": 4330
2644
+ },
2645
+ {
2646
+ "epoch": 4.1,
2647
+ "learning_rate": 2.2968914251389097e-05,
2648
+ "loss": 0.0111,
2649
+ "step": 4340
2650
+ },
2651
+ {
2652
+ "epoch": 4.11,
2653
+ "learning_rate": 2.28938279020874e-05,
2654
+ "loss": 0.0026,
2655
+ "step": 4350
2656
+ },
2657
+ {
2658
+ "epoch": 4.12,
2659
+ "learning_rate": 2.2818741552785707e-05,
2660
+ "loss": 0.0029,
2661
+ "step": 4360
2662
+ },
2663
+ {
2664
+ "epoch": 4.13,
2665
+ "learning_rate": 2.2743655203484006e-05,
2666
+ "loss": 0.0074,
2667
+ "step": 4370
2668
+ },
2669
+ {
2670
+ "epoch": 4.14,
2671
+ "learning_rate": 2.266856885418231e-05,
2672
+ "loss": 0.005,
2673
+ "step": 4380
2674
+ },
2675
+ {
2676
+ "epoch": 4.15,
2677
+ "learning_rate": 2.2593482504880616e-05,
2678
+ "loss": 0.0013,
2679
+ "step": 4390
2680
+ },
2681
+ {
2682
+ "epoch": 4.16,
2683
+ "learning_rate": 2.2518396155578916e-05,
2684
+ "loss": 0.013,
2685
+ "step": 4400
2686
+ },
2687
+ {
2688
+ "epoch": 4.17,
2689
+ "learning_rate": 2.2443309806277222e-05,
2690
+ "loss": 0.0326,
2691
+ "step": 4410
2692
+ },
2693
+ {
2694
+ "epoch": 4.18,
2695
+ "learning_rate": 2.2368223456975522e-05,
2696
+ "loss": 0.0193,
2697
+ "step": 4420
2698
+ },
2699
+ {
2700
+ "epoch": 4.19,
2701
+ "learning_rate": 2.2293137107673825e-05,
2702
+ "loss": 0.0016,
2703
+ "step": 4430
2704
+ },
2705
+ {
2706
+ "epoch": 4.2,
2707
+ "learning_rate": 2.221805075837213e-05,
2708
+ "loss": 0.0208,
2709
+ "step": 4440
2710
+ },
2711
+ {
2712
+ "epoch": 4.21,
2713
+ "learning_rate": 2.214296440907043e-05,
2714
+ "loss": 0.0152,
2715
+ "step": 4450
2716
+ },
2717
+ {
2718
+ "epoch": 4.22,
2719
+ "learning_rate": 2.2067878059768734e-05,
2720
+ "loss": 0.0088,
2721
+ "step": 4460
2722
+ },
2723
+ {
2724
+ "epoch": 4.23,
2725
+ "learning_rate": 2.199279171046704e-05,
2726
+ "loss": 0.0002,
2727
+ "step": 4470
2728
+ },
2729
+ {
2730
+ "epoch": 4.24,
2731
+ "learning_rate": 2.191770536116534e-05,
2732
+ "loss": 0.0147,
2733
+ "step": 4480
2734
+ },
2735
+ {
2736
+ "epoch": 4.24,
2737
+ "learning_rate": 2.1842619011863644e-05,
2738
+ "loss": 0.0065,
2739
+ "step": 4490
2740
+ },
2741
+ {
2742
+ "epoch": 4.25,
2743
+ "learning_rate": 2.176753266256195e-05,
2744
+ "loss": 0.0216,
2745
+ "step": 4500
2746
+ },
2747
+ {
2748
+ "epoch": 4.26,
2749
+ "learning_rate": 2.169244631326025e-05,
2750
+ "loss": 0.0007,
2751
+ "step": 4510
2752
+ },
2753
+ {
2754
+ "epoch": 4.27,
2755
+ "learning_rate": 2.1617359963958553e-05,
2756
+ "loss": 0.0139,
2757
+ "step": 4520
2758
+ },
2759
+ {
2760
+ "epoch": 4.28,
2761
+ "learning_rate": 2.1542273614656856e-05,
2762
+ "loss": 0.0051,
2763
+ "step": 4530
2764
+ },
2765
+ {
2766
+ "epoch": 4.29,
2767
+ "learning_rate": 2.146718726535516e-05,
2768
+ "loss": 0.0014,
2769
+ "step": 4540
2770
+ },
2771
+ {
2772
+ "epoch": 4.3,
2773
+ "learning_rate": 2.1392100916053462e-05,
2774
+ "loss": 0.018,
2775
+ "step": 4550
2776
+ },
2777
+ {
2778
+ "epoch": 4.31,
2779
+ "learning_rate": 2.1317014566751765e-05,
2780
+ "loss": 0.0083,
2781
+ "step": 4560
2782
+ },
2783
+ {
2784
+ "epoch": 4.32,
2785
+ "learning_rate": 2.124192821745007e-05,
2786
+ "loss": 0.0148,
2787
+ "step": 4570
2788
+ },
2789
+ {
2790
+ "epoch": 4.33,
2791
+ "learning_rate": 2.116684186814837e-05,
2792
+ "loss": 0.0037,
2793
+ "step": 4580
2794
+ },
2795
+ {
2796
+ "epoch": 4.34,
2797
+ "learning_rate": 2.1091755518846675e-05,
2798
+ "loss": 0.0099,
2799
+ "step": 4590
2800
+ },
2801
+ {
2802
+ "epoch": 4.35,
2803
+ "learning_rate": 2.1016669169544978e-05,
2804
+ "loss": 0.0006,
2805
+ "step": 4600
2806
+ },
2807
+ {
2808
+ "epoch": 4.36,
2809
+ "learning_rate": 2.094158282024328e-05,
2810
+ "loss": 0.0028,
2811
+ "step": 4610
2812
+ },
2813
+ {
2814
+ "epoch": 4.37,
2815
+ "learning_rate": 2.0866496470941584e-05,
2816
+ "loss": 0.0173,
2817
+ "step": 4620
2818
+ },
2819
+ {
2820
+ "epoch": 4.38,
2821
+ "learning_rate": 2.0791410121639887e-05,
2822
+ "loss": 0.0137,
2823
+ "step": 4630
2824
+ },
2825
+ {
2826
+ "epoch": 4.39,
2827
+ "learning_rate": 2.071632377233819e-05,
2828
+ "loss": 0.0071,
2829
+ "step": 4640
2830
+ },
2831
+ {
2832
+ "epoch": 4.4,
2833
+ "learning_rate": 2.0641237423036493e-05,
2834
+ "loss": 0.0182,
2835
+ "step": 4650
2836
+ },
2837
+ {
2838
+ "epoch": 4.41,
2839
+ "learning_rate": 2.0566151073734796e-05,
2840
+ "loss": 0.0082,
2841
+ "step": 4660
2842
+ },
2843
+ {
2844
+ "epoch": 4.42,
2845
+ "learning_rate": 2.04910647244331e-05,
2846
+ "loss": 0.0144,
2847
+ "step": 4670
2848
+ },
2849
+ {
2850
+ "epoch": 4.42,
2851
+ "learning_rate": 2.0415978375131403e-05,
2852
+ "loss": 0.01,
2853
+ "step": 4680
2854
+ },
2855
+ {
2856
+ "epoch": 4.43,
2857
+ "learning_rate": 2.0340892025829706e-05,
2858
+ "loss": 0.0041,
2859
+ "step": 4690
2860
+ },
2861
+ {
2862
+ "epoch": 4.44,
2863
+ "learning_rate": 2.026580567652801e-05,
2864
+ "loss": 0.007,
2865
+ "step": 4700
2866
+ },
2867
+ {
2868
+ "epoch": 4.45,
2869
+ "learning_rate": 2.0190719327226312e-05,
2870
+ "loss": 0.0112,
2871
+ "step": 4710
2872
+ },
2873
+ {
2874
+ "epoch": 4.46,
2875
+ "learning_rate": 2.0115632977924615e-05,
2876
+ "loss": 0.0176,
2877
+ "step": 4720
2878
+ },
2879
+ {
2880
+ "epoch": 4.47,
2881
+ "learning_rate": 2.0040546628622918e-05,
2882
+ "loss": 0.0215,
2883
+ "step": 4730
2884
+ },
2885
+ {
2886
+ "epoch": 4.48,
2887
+ "learning_rate": 1.996546027932122e-05,
2888
+ "loss": 0.0269,
2889
+ "step": 4740
2890
+ },
2891
+ {
2892
+ "epoch": 4.49,
2893
+ "learning_rate": 1.9890373930019524e-05,
2894
+ "loss": 0.0113,
2895
+ "step": 4750
2896
+ },
2897
+ {
2898
+ "epoch": 4.5,
2899
+ "learning_rate": 1.9815287580717824e-05,
2900
+ "loss": 0.0051,
2901
+ "step": 4760
2902
+ },
2903
+ {
2904
+ "epoch": 4.51,
2905
+ "learning_rate": 1.974020123141613e-05,
2906
+ "loss": 0.0006,
2907
+ "step": 4770
2908
+ },
2909
+ {
2910
+ "epoch": 4.52,
2911
+ "learning_rate": 1.9665114882114434e-05,
2912
+ "loss": 0.0155,
2913
+ "step": 4780
2914
+ },
2915
+ {
2916
+ "epoch": 4.53,
2917
+ "learning_rate": 1.9590028532812733e-05,
2918
+ "loss": 0.0042,
2919
+ "step": 4790
2920
+ },
2921
+ {
2922
+ "epoch": 4.54,
2923
+ "learning_rate": 1.951494218351104e-05,
2924
+ "loss": 0.0053,
2925
+ "step": 4800
2926
+ },
2927
+ {
2928
+ "epoch": 4.55,
2929
+ "learning_rate": 1.9439855834209343e-05,
2930
+ "loss": 0.0004,
2931
+ "step": 4810
2932
+ },
2933
+ {
2934
+ "epoch": 4.56,
2935
+ "learning_rate": 1.9364769484907643e-05,
2936
+ "loss": 0.0044,
2937
+ "step": 4820
2938
+ },
2939
+ {
2940
+ "epoch": 4.57,
2941
+ "learning_rate": 1.928968313560595e-05,
2942
+ "loss": 0.0321,
2943
+ "step": 4830
2944
+ },
2945
+ {
2946
+ "epoch": 4.58,
2947
+ "learning_rate": 1.9214596786304252e-05,
2948
+ "loss": 0.0099,
2949
+ "step": 4840
2950
+ },
2951
+ {
2952
+ "epoch": 4.59,
2953
+ "learning_rate": 1.9139510437002552e-05,
2954
+ "loss": 0.0006,
2955
+ "step": 4850
2956
+ },
2957
+ {
2958
+ "epoch": 4.59,
2959
+ "learning_rate": 1.906442408770086e-05,
2960
+ "loss": 0.0023,
2961
+ "step": 4860
2962
+ },
2963
+ {
2964
+ "epoch": 4.6,
2965
+ "learning_rate": 1.898933773839916e-05,
2966
+ "loss": 0.0096,
2967
+ "step": 4870
2968
+ },
2969
+ {
2970
+ "epoch": 4.61,
2971
+ "learning_rate": 1.891425138909746e-05,
2972
+ "loss": 0.0157,
2973
+ "step": 4880
2974
+ },
2975
+ {
2976
+ "epoch": 4.62,
2977
+ "learning_rate": 1.8839165039795768e-05,
2978
+ "loss": 0.0054,
2979
+ "step": 4890
2980
+ },
2981
+ {
2982
+ "epoch": 4.63,
2983
+ "learning_rate": 1.8764078690494068e-05,
2984
+ "loss": 0.0087,
2985
+ "step": 4900
2986
+ },
2987
+ {
2988
+ "epoch": 4.64,
2989
+ "learning_rate": 1.868899234119237e-05,
2990
+ "loss": 0.0009,
2991
+ "step": 4910
2992
+ },
2993
+ {
2994
+ "epoch": 4.65,
2995
+ "learning_rate": 1.8613905991890677e-05,
2996
+ "loss": 0.0026,
2997
+ "step": 4920
2998
+ },
2999
+ {
3000
+ "epoch": 4.66,
3001
+ "learning_rate": 1.8538819642588977e-05,
3002
+ "loss": 0.0179,
3003
+ "step": 4930
3004
+ },
3005
+ {
3006
+ "epoch": 4.67,
3007
+ "learning_rate": 1.846373329328728e-05,
3008
+ "loss": 0.0242,
3009
+ "step": 4940
3010
+ },
3011
+ {
3012
+ "epoch": 4.68,
3013
+ "learning_rate": 1.8388646943985587e-05,
3014
+ "loss": 0.0053,
3015
+ "step": 4950
3016
+ },
3017
+ {
3018
+ "epoch": 4.69,
3019
+ "learning_rate": 1.8313560594683886e-05,
3020
+ "loss": 0.0039,
3021
+ "step": 4960
3022
+ },
3023
+ {
3024
+ "epoch": 4.7,
3025
+ "learning_rate": 1.823847424538219e-05,
3026
+ "loss": 0.0025,
3027
+ "step": 4970
3028
+ },
3029
+ {
3030
+ "epoch": 4.71,
3031
+ "learning_rate": 1.8163387896080496e-05,
3032
+ "loss": 0.0048,
3033
+ "step": 4980
3034
+ },
3035
+ {
3036
+ "epoch": 4.72,
3037
+ "learning_rate": 1.8088301546778796e-05,
3038
+ "loss": 0.0135,
3039
+ "step": 4990
3040
+ },
3041
+ {
3042
+ "epoch": 4.73,
3043
+ "learning_rate": 1.80132151974771e-05,
3044
+ "loss": 0.0196,
3045
+ "step": 5000
3046
+ },
3047
+ {
3048
+ "epoch": 4.74,
3049
+ "learning_rate": 1.7938128848175402e-05,
3050
+ "loss": 0.0046,
3051
+ "step": 5010
3052
+ },
3053
+ {
3054
+ "epoch": 4.75,
3055
+ "learning_rate": 1.7863042498873705e-05,
3056
+ "loss": 0.0023,
3057
+ "step": 5020
3058
+ },
3059
+ {
3060
+ "epoch": 4.76,
3061
+ "learning_rate": 1.7787956149572008e-05,
3062
+ "loss": 0.0026,
3063
+ "step": 5030
3064
+ },
3065
+ {
3066
+ "epoch": 4.76,
3067
+ "learning_rate": 1.771286980027031e-05,
3068
+ "loss": 0.0131,
3069
+ "step": 5040
3070
+ },
3071
+ {
3072
+ "epoch": 4.77,
3073
+ "learning_rate": 1.7637783450968614e-05,
3074
+ "loss": 0.0076,
3075
+ "step": 5050
3076
+ },
3077
+ {
3078
+ "epoch": 4.78,
3079
+ "learning_rate": 1.7562697101666917e-05,
3080
+ "loss": 0.0258,
3081
+ "step": 5060
3082
+ },
3083
+ {
3084
+ "epoch": 4.79,
3085
+ "learning_rate": 1.748761075236522e-05,
3086
+ "loss": 0.0074,
3087
+ "step": 5070
3088
+ },
3089
+ {
3090
+ "epoch": 4.8,
3091
+ "learning_rate": 1.7412524403063524e-05,
3092
+ "loss": 0.017,
3093
+ "step": 5080
3094
+ },
3095
+ {
3096
+ "epoch": 4.81,
3097
+ "learning_rate": 1.7337438053761827e-05,
3098
+ "loss": 0.0075,
3099
+ "step": 5090
3100
+ },
3101
+ {
3102
+ "epoch": 4.82,
3103
+ "learning_rate": 1.726235170446013e-05,
3104
+ "loss": 0.0028,
3105
+ "step": 5100
3106
+ },
3107
+ {
3108
+ "epoch": 4.83,
3109
+ "learning_rate": 1.7187265355158433e-05,
3110
+ "loss": 0.0031,
3111
+ "step": 5110
3112
+ },
3113
+ {
3114
+ "epoch": 4.84,
3115
+ "learning_rate": 1.7112179005856736e-05,
3116
+ "loss": 0.0054,
3117
+ "step": 5120
3118
+ },
3119
+ {
3120
+ "epoch": 4.85,
3121
+ "learning_rate": 1.703709265655504e-05,
3122
+ "loss": 0.007,
3123
+ "step": 5130
3124
+ },
3125
+ {
3126
+ "epoch": 4.86,
3127
+ "learning_rate": 1.6962006307253342e-05,
3128
+ "loss": 0.0041,
3129
+ "step": 5140
3130
+ },
3131
+ {
3132
+ "epoch": 4.87,
3133
+ "learning_rate": 1.6886919957951645e-05,
3134
+ "loss": 0.0012,
3135
+ "step": 5150
3136
+ },
3137
+ {
3138
+ "epoch": 4.88,
3139
+ "learning_rate": 1.681183360864995e-05,
3140
+ "loss": 0.0091,
3141
+ "step": 5160
3142
+ },
3143
+ {
3144
+ "epoch": 4.89,
3145
+ "learning_rate": 1.673674725934825e-05,
3146
+ "loss": 0.0124,
3147
+ "step": 5170
3148
+ },
3149
+ {
3150
+ "epoch": 4.9,
3151
+ "learning_rate": 1.6661660910046555e-05,
3152
+ "loss": 0.0044,
3153
+ "step": 5180
3154
+ },
3155
+ {
3156
+ "epoch": 4.91,
3157
+ "learning_rate": 1.6586574560744858e-05,
3158
+ "loss": 0.0044,
3159
+ "step": 5190
3160
+ },
3161
+ {
3162
+ "epoch": 4.92,
3163
+ "learning_rate": 1.651148821144316e-05,
3164
+ "loss": 0.0012,
3165
+ "step": 5200
3166
+ },
3167
+ {
3168
+ "epoch": 4.93,
3169
+ "learning_rate": 1.6436401862141464e-05,
3170
+ "loss": 0.0055,
3171
+ "step": 5210
3172
+ },
3173
+ {
3174
+ "epoch": 4.94,
3175
+ "learning_rate": 1.6361315512839767e-05,
3176
+ "loss": 0.0035,
3177
+ "step": 5220
3178
+ },
3179
+ {
3180
+ "epoch": 4.94,
3181
+ "learning_rate": 1.628622916353807e-05,
3182
+ "loss": 0.001,
3183
+ "step": 5230
3184
+ },
3185
+ {
3186
+ "epoch": 4.95,
3187
+ "learning_rate": 1.621114281423637e-05,
3188
+ "loss": 0.0032,
3189
+ "step": 5240
3190
+ },
3191
+ {
3192
+ "epoch": 4.96,
3193
+ "learning_rate": 1.6136056464934676e-05,
3194
+ "loss": 0.0053,
3195
+ "step": 5250
3196
+ },
3197
+ {
3198
+ "epoch": 4.97,
3199
+ "learning_rate": 1.606097011563298e-05,
3200
+ "loss": 0.008,
3201
+ "step": 5260
3202
+ },
3203
+ {
3204
+ "epoch": 4.98,
3205
+ "learning_rate": 1.598588376633128e-05,
3206
+ "loss": 0.0014,
3207
+ "step": 5270
3208
+ },
3209
+ {
3210
+ "epoch": 4.99,
3211
+ "learning_rate": 1.5910797417029586e-05,
3212
+ "loss": 0.0151,
3213
+ "step": 5280
3214
+ },
3215
+ {
3216
+ "epoch": 5.0,
3217
+ "eval_accuracy": 0.999691466854725,
3218
+ "eval_loss": 0.0013564619002863765,
3219
+ "eval_runtime": 140.0749,
3220
+ "eval_samples_per_second": 161.971,
3221
+ "eval_steps_per_second": 10.123,
3222
+ "step": 5288
3223
+ },
3224
+ {
3225
+ "epoch": 5.0,
3226
+ "learning_rate": 1.583571106772789e-05,
3227
+ "loss": 0.0003,
3228
+ "step": 5290
3229
+ },
3230
+ {
3231
+ "epoch": 5.01,
3232
+ "learning_rate": 1.576062471842619e-05,
3233
+ "loss": 0.0006,
3234
+ "step": 5300
3235
+ },
3236
+ {
3237
+ "epoch": 5.02,
3238
+ "learning_rate": 1.5685538369124495e-05,
3239
+ "loss": 0.011,
3240
+ "step": 5310
3241
+ },
3242
+ {
3243
+ "epoch": 5.03,
3244
+ "learning_rate": 1.5610452019822798e-05,
3245
+ "loss": 0.0008,
3246
+ "step": 5320
3247
+ },
3248
+ {
3249
+ "epoch": 5.04,
3250
+ "learning_rate": 1.5535365670521098e-05,
3251
+ "loss": 0.0018,
3252
+ "step": 5330
3253
+ },
3254
+ {
3255
+ "epoch": 5.05,
3256
+ "learning_rate": 1.5460279321219404e-05,
3257
+ "loss": 0.0065,
3258
+ "step": 5340
3259
+ },
3260
+ {
3261
+ "epoch": 5.06,
3262
+ "learning_rate": 1.5385192971917707e-05,
3263
+ "loss": 0.0143,
3264
+ "step": 5350
3265
+ },
3266
+ {
3267
+ "epoch": 5.07,
3268
+ "learning_rate": 1.5310106622616007e-05,
3269
+ "loss": 0.0116,
3270
+ "step": 5360
3271
+ },
3272
+ {
3273
+ "epoch": 5.08,
3274
+ "learning_rate": 1.5235020273314312e-05,
3275
+ "loss": 0.0088,
3276
+ "step": 5370
3277
+ },
3278
+ {
3279
+ "epoch": 5.09,
3280
+ "learning_rate": 1.5159933924012615e-05,
3281
+ "loss": 0.0109,
3282
+ "step": 5380
3283
+ },
3284
+ {
3285
+ "epoch": 5.1,
3286
+ "learning_rate": 1.5084847574710916e-05,
3287
+ "loss": 0.0005,
3288
+ "step": 5390
3289
+ },
3290
+ {
3291
+ "epoch": 5.11,
3292
+ "learning_rate": 1.5009761225409221e-05,
3293
+ "loss": 0.0016,
3294
+ "step": 5400
3295
+ },
3296
+ {
3297
+ "epoch": 5.11,
3298
+ "learning_rate": 1.4934674876107524e-05,
3299
+ "loss": 0.0008,
3300
+ "step": 5410
3301
+ },
3302
+ {
3303
+ "epoch": 5.12,
3304
+ "learning_rate": 1.4859588526805826e-05,
3305
+ "loss": 0.0101,
3306
+ "step": 5420
3307
+ },
3308
+ {
3309
+ "epoch": 5.13,
3310
+ "learning_rate": 1.478450217750413e-05,
3311
+ "loss": 0.003,
3312
+ "step": 5430
3313
+ },
3314
+ {
3315
+ "epoch": 5.14,
3316
+ "learning_rate": 1.4709415828202434e-05,
3317
+ "loss": 0.0046,
3318
+ "step": 5440
3319
+ },
3320
+ {
3321
+ "epoch": 5.15,
3322
+ "learning_rate": 1.4634329478900738e-05,
3323
+ "loss": 0.0041,
3324
+ "step": 5450
3325
+ },
3326
+ {
3327
+ "epoch": 5.16,
3328
+ "learning_rate": 1.455924312959904e-05,
3329
+ "loss": 0.0011,
3330
+ "step": 5460
3331
+ },
3332
+ {
3333
+ "epoch": 5.17,
3334
+ "learning_rate": 1.4484156780297343e-05,
3335
+ "loss": 0.0071,
3336
+ "step": 5470
3337
+ },
3338
+ {
3339
+ "epoch": 5.18,
3340
+ "learning_rate": 1.4409070430995648e-05,
3341
+ "loss": 0.0191,
3342
+ "step": 5480
3343
+ },
3344
+ {
3345
+ "epoch": 5.19,
3346
+ "learning_rate": 1.433398408169395e-05,
3347
+ "loss": 0.0264,
3348
+ "step": 5490
3349
+ },
3350
+ {
3351
+ "epoch": 5.2,
3352
+ "learning_rate": 1.425889773239225e-05,
3353
+ "loss": 0.0038,
3354
+ "step": 5500
3355
+ },
3356
+ {
3357
+ "epoch": 5.21,
3358
+ "learning_rate": 1.4183811383090555e-05,
3359
+ "loss": 0.0149,
3360
+ "step": 5510
3361
+ },
3362
+ {
3363
+ "epoch": 5.22,
3364
+ "learning_rate": 1.4108725033788858e-05,
3365
+ "loss": 0.0009,
3366
+ "step": 5520
3367
+ },
3368
+ {
3369
+ "epoch": 5.23,
3370
+ "learning_rate": 1.403363868448716e-05,
3371
+ "loss": 0.0013,
3372
+ "step": 5530
3373
+ },
3374
+ {
3375
+ "epoch": 5.24,
3376
+ "learning_rate": 1.3958552335185465e-05,
3377
+ "loss": 0.0194,
3378
+ "step": 5540
3379
+ },
3380
+ {
3381
+ "epoch": 5.25,
3382
+ "learning_rate": 1.3883465985883768e-05,
3383
+ "loss": 0.0003,
3384
+ "step": 5550
3385
+ },
3386
+ {
3387
+ "epoch": 5.26,
3388
+ "learning_rate": 1.380837963658207e-05,
3389
+ "loss": 0.0105,
3390
+ "step": 5560
3391
+ },
3392
+ {
3393
+ "epoch": 5.27,
3394
+ "learning_rate": 1.3733293287280374e-05,
3395
+ "loss": 0.0018,
3396
+ "step": 5570
3397
+ },
3398
+ {
3399
+ "epoch": 5.28,
3400
+ "learning_rate": 1.3658206937978677e-05,
3401
+ "loss": 0.01,
3402
+ "step": 5580
3403
+ },
3404
+ {
3405
+ "epoch": 5.28,
3406
+ "learning_rate": 1.3583120588676979e-05,
3407
+ "loss": 0.0092,
3408
+ "step": 5590
3409
+ },
3410
+ {
3411
+ "epoch": 5.29,
3412
+ "learning_rate": 1.3508034239375283e-05,
3413
+ "loss": 0.0136,
3414
+ "step": 5600
3415
+ },
3416
+ {
3417
+ "epoch": 5.3,
3418
+ "learning_rate": 1.3432947890073585e-05,
3419
+ "loss": 0.0057,
3420
+ "step": 5610
3421
+ },
3422
+ {
3423
+ "epoch": 5.31,
3424
+ "learning_rate": 1.3357861540771888e-05,
3425
+ "loss": 0.0099,
3426
+ "step": 5620
3427
+ },
3428
+ {
3429
+ "epoch": 5.32,
3430
+ "learning_rate": 1.3282775191470193e-05,
3431
+ "loss": 0.0096,
3432
+ "step": 5630
3433
+ },
3434
+ {
3435
+ "epoch": 5.33,
3436
+ "learning_rate": 1.3207688842168494e-05,
3437
+ "loss": 0.0002,
3438
+ "step": 5640
3439
+ },
3440
+ {
3441
+ "epoch": 5.34,
3442
+ "learning_rate": 1.3132602492866797e-05,
3443
+ "loss": 0.0012,
3444
+ "step": 5650
3445
+ },
3446
+ {
3447
+ "epoch": 5.35,
3448
+ "learning_rate": 1.3057516143565102e-05,
3449
+ "loss": 0.0177,
3450
+ "step": 5660
3451
+ },
3452
+ {
3453
+ "epoch": 5.36,
3454
+ "learning_rate": 1.2982429794263403e-05,
3455
+ "loss": 0.0013,
3456
+ "step": 5670
3457
+ },
3458
+ {
3459
+ "epoch": 5.37,
3460
+ "learning_rate": 1.2907343444961707e-05,
3461
+ "loss": 0.0001,
3462
+ "step": 5680
3463
+ },
3464
+ {
3465
+ "epoch": 5.38,
3466
+ "learning_rate": 1.2832257095660011e-05,
3467
+ "loss": 0.0017,
3468
+ "step": 5690
3469
+ },
3470
+ {
3471
+ "epoch": 5.39,
3472
+ "learning_rate": 1.2757170746358313e-05,
3473
+ "loss": 0.0097,
3474
+ "step": 5700
3475
+ },
3476
+ {
3477
+ "epoch": 5.4,
3478
+ "learning_rate": 1.2682084397056616e-05,
3479
+ "loss": 0.0005,
3480
+ "step": 5710
3481
+ },
3482
+ {
3483
+ "epoch": 5.41,
3484
+ "learning_rate": 1.260699804775492e-05,
3485
+ "loss": 0.0081,
3486
+ "step": 5720
3487
+ },
3488
+ {
3489
+ "epoch": 5.42,
3490
+ "learning_rate": 1.2531911698453222e-05,
3491
+ "loss": 0.0017,
3492
+ "step": 5730
3493
+ },
3494
+ {
3495
+ "epoch": 5.43,
3496
+ "learning_rate": 1.2456825349151525e-05,
3497
+ "loss": 0.0257,
3498
+ "step": 5740
3499
+ },
3500
+ {
3501
+ "epoch": 5.44,
3502
+ "learning_rate": 1.2381738999849828e-05,
3503
+ "loss": 0.0081,
3504
+ "step": 5750
3505
+ },
3506
+ {
3507
+ "epoch": 5.45,
3508
+ "learning_rate": 1.2306652650548131e-05,
3509
+ "loss": 0.0062,
3510
+ "step": 5760
3511
+ },
3512
+ {
3513
+ "epoch": 5.45,
3514
+ "learning_rate": 1.2231566301246434e-05,
3515
+ "loss": 0.0007,
3516
+ "step": 5770
3517
+ },
3518
+ {
3519
+ "epoch": 5.46,
3520
+ "learning_rate": 1.2156479951944736e-05,
3521
+ "loss": 0.0011,
3522
+ "step": 5780
3523
+ },
3524
+ {
3525
+ "epoch": 5.47,
3526
+ "learning_rate": 1.208139360264304e-05,
3527
+ "loss": 0.0105,
3528
+ "step": 5790
3529
+ },
3530
+ {
3531
+ "epoch": 5.48,
3532
+ "learning_rate": 1.2006307253341344e-05,
3533
+ "loss": 0.0042,
3534
+ "step": 5800
3535
+ },
3536
+ {
3537
+ "epoch": 5.49,
3538
+ "learning_rate": 1.1931220904039645e-05,
3539
+ "loss": 0.0002,
3540
+ "step": 5810
3541
+ },
3542
+ {
3543
+ "epoch": 5.5,
3544
+ "learning_rate": 1.185613455473795e-05,
3545
+ "loss": 0.0002,
3546
+ "step": 5820
3547
+ },
3548
+ {
3549
+ "epoch": 5.51,
3550
+ "learning_rate": 1.1781048205436253e-05,
3551
+ "loss": 0.0016,
3552
+ "step": 5830
3553
+ },
3554
+ {
3555
+ "epoch": 5.52,
3556
+ "learning_rate": 1.1705961856134555e-05,
3557
+ "loss": 0.0159,
3558
+ "step": 5840
3559
+ },
3560
+ {
3561
+ "epoch": 5.53,
3562
+ "learning_rate": 1.1630875506832858e-05,
3563
+ "loss": 0.0162,
3564
+ "step": 5850
3565
+ },
3566
+ {
3567
+ "epoch": 5.54,
3568
+ "learning_rate": 1.1555789157531162e-05,
3569
+ "loss": 0.001,
3570
+ "step": 5860
3571
+ },
3572
+ {
3573
+ "epoch": 5.55,
3574
+ "learning_rate": 1.1480702808229464e-05,
3575
+ "loss": 0.0006,
3576
+ "step": 5870
3577
+ },
3578
+ {
3579
+ "epoch": 5.56,
3580
+ "learning_rate": 1.1405616458927767e-05,
3581
+ "loss": 0.0029,
3582
+ "step": 5880
3583
+ },
3584
+ {
3585
+ "epoch": 5.57,
3586
+ "learning_rate": 1.1330530109626072e-05,
3587
+ "loss": 0.0055,
3588
+ "step": 5890
3589
+ },
3590
+ {
3591
+ "epoch": 5.58,
3592
+ "learning_rate": 1.1255443760324373e-05,
3593
+ "loss": 0.0006,
3594
+ "step": 5900
3595
+ },
3596
+ {
3597
+ "epoch": 5.59,
3598
+ "learning_rate": 1.1180357411022676e-05,
3599
+ "loss": 0.0012,
3600
+ "step": 5910
3601
+ },
3602
+ {
3603
+ "epoch": 5.6,
3604
+ "learning_rate": 1.110527106172098e-05,
3605
+ "loss": 0.0101,
3606
+ "step": 5920
3607
+ },
3608
+ {
3609
+ "epoch": 5.61,
3610
+ "learning_rate": 1.1030184712419282e-05,
3611
+ "loss": 0.0011,
3612
+ "step": 5930
3613
+ },
3614
+ {
3615
+ "epoch": 5.62,
3616
+ "learning_rate": 1.0955098363117586e-05,
3617
+ "loss": 0.0122,
3618
+ "step": 5940
3619
+ },
3620
+ {
3621
+ "epoch": 5.63,
3622
+ "learning_rate": 1.0880012013815889e-05,
3623
+ "loss": 0.0078,
3624
+ "step": 5950
3625
+ },
3626
+ {
3627
+ "epoch": 5.63,
3628
+ "learning_rate": 1.0804925664514192e-05,
3629
+ "loss": 0.0123,
3630
+ "step": 5960
3631
+ },
3632
+ {
3633
+ "epoch": 5.64,
3634
+ "learning_rate": 1.0729839315212495e-05,
3635
+ "loss": 0.0103,
3636
+ "step": 5970
3637
+ },
3638
+ {
3639
+ "epoch": 5.65,
3640
+ "learning_rate": 1.0654752965910798e-05,
3641
+ "loss": 0.0003,
3642
+ "step": 5980
3643
+ },
3644
+ {
3645
+ "epoch": 5.66,
3646
+ "learning_rate": 1.0579666616609101e-05,
3647
+ "loss": 0.0025,
3648
+ "step": 5990
3649
+ },
3650
+ {
3651
+ "epoch": 5.67,
3652
+ "learning_rate": 1.0504580267307404e-05,
3653
+ "loss": 0.0207,
3654
+ "step": 6000
3655
+ },
3656
+ {
3657
+ "epoch": 5.68,
3658
+ "learning_rate": 1.0429493918005707e-05,
3659
+ "loss": 0.0033,
3660
+ "step": 6010
3661
+ },
3662
+ {
3663
+ "epoch": 5.69,
3664
+ "learning_rate": 1.0354407568704009e-05,
3665
+ "loss": 0.0127,
3666
+ "step": 6020
3667
+ },
3668
+ {
3669
+ "epoch": 5.7,
3670
+ "learning_rate": 1.0279321219402314e-05,
3671
+ "loss": 0.0007,
3672
+ "step": 6030
3673
+ },
3674
+ {
3675
+ "epoch": 5.71,
3676
+ "learning_rate": 1.0204234870100617e-05,
3677
+ "loss": 0.0018,
3678
+ "step": 6040
3679
+ },
3680
+ {
3681
+ "epoch": 5.72,
3682
+ "learning_rate": 1.0129148520798918e-05,
3683
+ "loss": 0.0006,
3684
+ "step": 6050
3685
+ },
3686
+ {
3687
+ "epoch": 5.73,
3688
+ "learning_rate": 1.0054062171497223e-05,
3689
+ "loss": 0.0148,
3690
+ "step": 6060
3691
+ },
3692
+ {
3693
+ "epoch": 5.74,
3694
+ "learning_rate": 9.978975822195526e-06,
3695
+ "loss": 0.0027,
3696
+ "step": 6070
3697
+ },
3698
+ {
3699
+ "epoch": 5.75,
3700
+ "learning_rate": 9.903889472893827e-06,
3701
+ "loss": 0.0057,
3702
+ "step": 6080
3703
+ },
3704
+ {
3705
+ "epoch": 5.76,
3706
+ "learning_rate": 9.82880312359213e-06,
3707
+ "loss": 0.0073,
3708
+ "step": 6090
3709
+ },
3710
+ {
3711
+ "epoch": 5.77,
3712
+ "learning_rate": 9.753716774290435e-06,
3713
+ "loss": 0.0093,
3714
+ "step": 6100
3715
+ },
3716
+ {
3717
+ "epoch": 5.78,
3718
+ "learning_rate": 9.678630424988737e-06,
3719
+ "loss": 0.0014,
3720
+ "step": 6110
3721
+ },
3722
+ {
3723
+ "epoch": 5.79,
3724
+ "learning_rate": 9.60354407568704e-06,
3725
+ "loss": 0.0019,
3726
+ "step": 6120
3727
+ },
3728
+ {
3729
+ "epoch": 5.8,
3730
+ "learning_rate": 9.528457726385345e-06,
3731
+ "loss": 0.0001,
3732
+ "step": 6130
3733
+ },
3734
+ {
3735
+ "epoch": 5.8,
3736
+ "learning_rate": 9.453371377083646e-06,
3737
+ "loss": 0.0002,
3738
+ "step": 6140
3739
+ },
3740
+ {
3741
+ "epoch": 5.81,
3742
+ "learning_rate": 9.378285027781949e-06,
3743
+ "loss": 0.0002,
3744
+ "step": 6150
3745
+ },
3746
+ {
3747
+ "epoch": 5.82,
3748
+ "learning_rate": 9.303198678480252e-06,
3749
+ "loss": 0.0005,
3750
+ "step": 6160
3751
+ },
3752
+ {
3753
+ "epoch": 5.83,
3754
+ "learning_rate": 9.228112329178555e-06,
3755
+ "loss": 0.0,
3756
+ "step": 6170
3757
+ },
3758
+ {
3759
+ "epoch": 5.84,
3760
+ "learning_rate": 9.153025979876858e-06,
3761
+ "loss": 0.0139,
3762
+ "step": 6180
3763
+ },
3764
+ {
3765
+ "epoch": 5.85,
3766
+ "learning_rate": 9.077939630575162e-06,
3767
+ "loss": 0.0017,
3768
+ "step": 6190
3769
+ },
3770
+ {
3771
+ "epoch": 5.86,
3772
+ "learning_rate": 9.002853281273465e-06,
3773
+ "loss": 0.0007,
3774
+ "step": 6200
3775
+ },
3776
+ {
3777
+ "epoch": 5.87,
3778
+ "learning_rate": 8.927766931971768e-06,
3779
+ "loss": 0.001,
3780
+ "step": 6210
3781
+ },
3782
+ {
3783
+ "epoch": 5.88,
3784
+ "learning_rate": 8.852680582670071e-06,
3785
+ "loss": 0.001,
3786
+ "step": 6220
3787
+ },
3788
+ {
3789
+ "epoch": 5.89,
3790
+ "learning_rate": 8.777594233368374e-06,
3791
+ "loss": 0.0001,
3792
+ "step": 6230
3793
+ },
3794
+ {
3795
+ "epoch": 5.9,
3796
+ "learning_rate": 8.702507884066677e-06,
3797
+ "loss": 0.0031,
3798
+ "step": 6240
3799
+ },
3800
+ {
3801
+ "epoch": 5.91,
3802
+ "learning_rate": 8.62742153476498e-06,
3803
+ "loss": 0.0006,
3804
+ "step": 6250
3805
+ },
3806
+ {
3807
+ "epoch": 5.92,
3808
+ "learning_rate": 8.552335185463283e-06,
3809
+ "loss": 0.0002,
3810
+ "step": 6260
3811
+ },
3812
+ {
3813
+ "epoch": 5.93,
3814
+ "learning_rate": 8.477248836161586e-06,
3815
+ "loss": 0.0006,
3816
+ "step": 6270
3817
+ },
3818
+ {
3819
+ "epoch": 5.94,
3820
+ "learning_rate": 8.40216248685989e-06,
3821
+ "loss": 0.0132,
3822
+ "step": 6280
3823
+ },
3824
+ {
3825
+ "epoch": 5.95,
3826
+ "learning_rate": 8.327076137558193e-06,
3827
+ "loss": 0.0005,
3828
+ "step": 6290
3829
+ },
3830
+ {
3831
+ "epoch": 5.96,
3832
+ "learning_rate": 8.251989788256496e-06,
3833
+ "loss": 0.0008,
3834
+ "step": 6300
3835
+ },
3836
+ {
3837
+ "epoch": 5.97,
3838
+ "learning_rate": 8.176903438954799e-06,
3839
+ "loss": 0.0057,
3840
+ "step": 6310
3841
+ },
3842
+ {
3843
+ "epoch": 5.97,
3844
+ "learning_rate": 8.101817089653102e-06,
3845
+ "loss": 0.0019,
3846
+ "step": 6320
3847
+ },
3848
+ {
3849
+ "epoch": 5.98,
3850
+ "learning_rate": 8.026730740351403e-06,
3851
+ "loss": 0.0057,
3852
+ "step": 6330
3853
+ },
3854
+ {
3855
+ "epoch": 5.99,
3856
+ "learning_rate": 7.951644391049708e-06,
3857
+ "loss": 0.0075,
3858
+ "step": 6340
3859
+ },
3860
+ {
3861
+ "epoch": 6.0,
3862
+ "eval_accuracy": 0.9989421720733427,
3863
+ "eval_loss": 0.00466978270560503,
3864
+ "eval_runtime": 139.7307,
3865
+ "eval_samples_per_second": 162.369,
3866
+ "eval_steps_per_second": 10.148,
3867
+ "step": 6346
3868
+ },
3869
+ {
3870
+ "epoch": 6.0,
3871
+ "learning_rate": 7.876558041748011e-06,
3872
+ "loss": 0.0058,
3873
+ "step": 6350
3874
+ },
3875
+ {
3876
+ "epoch": 6.01,
3877
+ "learning_rate": 7.801471692446313e-06,
3878
+ "loss": 0.0091,
3879
+ "step": 6360
3880
+ },
3881
+ {
3882
+ "epoch": 6.02,
3883
+ "learning_rate": 7.726385343144617e-06,
3884
+ "loss": 0.0014,
3885
+ "step": 6370
3886
+ },
3887
+ {
3888
+ "epoch": 6.03,
3889
+ "learning_rate": 7.65129899384292e-06,
3890
+ "loss": 0.0027,
3891
+ "step": 6380
3892
+ },
3893
+ {
3894
+ "epoch": 6.04,
3895
+ "learning_rate": 7.576212644541222e-06,
3896
+ "loss": 0.0005,
3897
+ "step": 6390
3898
+ },
3899
+ {
3900
+ "epoch": 6.05,
3901
+ "learning_rate": 7.501126295239526e-06,
3902
+ "loss": 0.0166,
3903
+ "step": 6400
3904
+ },
3905
+ {
3906
+ "epoch": 6.06,
3907
+ "learning_rate": 7.426039945937829e-06,
3908
+ "loss": 0.0002,
3909
+ "step": 6410
3910
+ },
3911
+ {
3912
+ "epoch": 6.07,
3913
+ "learning_rate": 7.350953596636131e-06,
3914
+ "loss": 0.0002,
3915
+ "step": 6420
3916
+ },
3917
+ {
3918
+ "epoch": 6.08,
3919
+ "learning_rate": 7.275867247334435e-06,
3920
+ "loss": 0.0011,
3921
+ "step": 6430
3922
+ },
3923
+ {
3924
+ "epoch": 6.09,
3925
+ "learning_rate": 7.200780898032738e-06,
3926
+ "loss": 0.0153,
3927
+ "step": 6440
3928
+ },
3929
+ {
3930
+ "epoch": 6.1,
3931
+ "learning_rate": 7.125694548731041e-06,
3932
+ "loss": 0.0097,
3933
+ "step": 6450
3934
+ },
3935
+ {
3936
+ "epoch": 6.11,
3937
+ "learning_rate": 7.050608199429344e-06,
3938
+ "loss": 0.0045,
3939
+ "step": 6460
3940
+ },
3941
+ {
3942
+ "epoch": 6.12,
3943
+ "learning_rate": 6.975521850127648e-06,
3944
+ "loss": 0.0083,
3945
+ "step": 6470
3946
+ },
3947
+ {
3948
+ "epoch": 6.13,
3949
+ "learning_rate": 6.90043550082595e-06,
3950
+ "loss": 0.002,
3951
+ "step": 6480
3952
+ },
3953
+ {
3954
+ "epoch": 6.14,
3955
+ "learning_rate": 6.825349151524253e-06,
3956
+ "loss": 0.0017,
3957
+ "step": 6490
3958
+ },
3959
+ {
3960
+ "epoch": 6.15,
3961
+ "learning_rate": 6.750262802222557e-06,
3962
+ "loss": 0.0045,
3963
+ "step": 6500
3964
+ },
3965
+ {
3966
+ "epoch": 6.15,
3967
+ "learning_rate": 6.675176452920858e-06,
3968
+ "loss": 0.0071,
3969
+ "step": 6510
3970
+ },
3971
+ {
3972
+ "epoch": 6.16,
3973
+ "learning_rate": 6.600090103619162e-06,
3974
+ "loss": 0.0072,
3975
+ "step": 6520
3976
+ },
3977
+ {
3978
+ "epoch": 6.17,
3979
+ "learning_rate": 6.5250037543174655e-06,
3980
+ "loss": 0.0015,
3981
+ "step": 6530
3982
+ },
3983
+ {
3984
+ "epoch": 6.18,
3985
+ "learning_rate": 6.449917405015768e-06,
3986
+ "loss": 0.0003,
3987
+ "step": 6540
3988
+ },
3989
+ {
3990
+ "epoch": 6.19,
3991
+ "learning_rate": 6.374831055714072e-06,
3992
+ "loss": 0.0063,
3993
+ "step": 6550
3994
+ },
3995
+ {
3996
+ "epoch": 6.2,
3997
+ "learning_rate": 6.299744706412375e-06,
3998
+ "loss": 0.0001,
3999
+ "step": 6560
4000
+ },
4001
+ {
4002
+ "epoch": 6.21,
4003
+ "learning_rate": 6.224658357110678e-06,
4004
+ "loss": 0.0003,
4005
+ "step": 6570
4006
+ },
4007
+ {
4008
+ "epoch": 6.22,
4009
+ "learning_rate": 6.14957200780898e-06,
4010
+ "loss": 0.0001,
4011
+ "step": 6580
4012
+ },
4013
+ {
4014
+ "epoch": 6.23,
4015
+ "learning_rate": 6.074485658507283e-06,
4016
+ "loss": 0.0051,
4017
+ "step": 6590
4018
+ },
4019
+ {
4020
+ "epoch": 6.24,
4021
+ "learning_rate": 5.999399309205587e-06,
4022
+ "loss": 0.0071,
4023
+ "step": 6600
4024
+ },
4025
+ {
4026
+ "epoch": 6.25,
4027
+ "learning_rate": 5.9243129599038895e-06,
4028
+ "loss": 0.0002,
4029
+ "step": 6610
4030
+ },
4031
+ {
4032
+ "epoch": 6.26,
4033
+ "learning_rate": 5.8492266106021934e-06,
4034
+ "loss": 0.0003,
4035
+ "step": 6620
4036
+ },
4037
+ {
4038
+ "epoch": 6.27,
4039
+ "learning_rate": 5.774140261300496e-06,
4040
+ "loss": 0.0025,
4041
+ "step": 6630
4042
+ },
4043
+ {
4044
+ "epoch": 6.28,
4045
+ "learning_rate": 5.699053911998799e-06,
4046
+ "loss": 0.0025,
4047
+ "step": 6640
4048
+ },
4049
+ {
4050
+ "epoch": 6.29,
4051
+ "learning_rate": 5.623967562697102e-06,
4052
+ "loss": 0.0006,
4053
+ "step": 6650
4054
+ },
4055
+ {
4056
+ "epoch": 6.3,
4057
+ "learning_rate": 5.548881213395405e-06,
4058
+ "loss": 0.001,
4059
+ "step": 6660
4060
+ },
4061
+ {
4062
+ "epoch": 6.31,
4063
+ "learning_rate": 5.473794864093708e-06,
4064
+ "loss": 0.016,
4065
+ "step": 6670
4066
+ },
4067
+ {
4068
+ "epoch": 6.32,
4069
+ "learning_rate": 5.398708514792011e-06,
4070
+ "loss": 0.0093,
4071
+ "step": 6680
4072
+ },
4073
+ {
4074
+ "epoch": 6.32,
4075
+ "learning_rate": 5.323622165490314e-06,
4076
+ "loss": 0.0156,
4077
+ "step": 6690
4078
+ },
4079
+ {
4080
+ "epoch": 6.33,
4081
+ "learning_rate": 5.248535816188617e-06,
4082
+ "loss": 0.0004,
4083
+ "step": 6700
4084
+ },
4085
+ {
4086
+ "epoch": 6.34,
4087
+ "learning_rate": 5.1734494668869205e-06,
4088
+ "loss": 0.0026,
4089
+ "step": 6710
4090
+ },
4091
+ {
4092
+ "epoch": 6.35,
4093
+ "learning_rate": 5.098363117585224e-06,
4094
+ "loss": 0.0001,
4095
+ "step": 6720
4096
+ },
4097
+ {
4098
+ "epoch": 6.36,
4099
+ "learning_rate": 5.023276768283526e-06,
4100
+ "loss": 0.0025,
4101
+ "step": 6730
4102
+ },
4103
+ {
4104
+ "epoch": 6.37,
4105
+ "learning_rate": 4.94819041898183e-06,
4106
+ "loss": 0.0004,
4107
+ "step": 6740
4108
+ },
4109
+ {
4110
+ "epoch": 6.38,
4111
+ "learning_rate": 4.873104069680132e-06,
4112
+ "loss": 0.0014,
4113
+ "step": 6750
4114
+ },
4115
+ {
4116
+ "epoch": 6.39,
4117
+ "learning_rate": 4.798017720378435e-06,
4118
+ "loss": 0.0109,
4119
+ "step": 6760
4120
+ },
4121
+ {
4122
+ "epoch": 6.4,
4123
+ "learning_rate": 4.722931371076738e-06,
4124
+ "loss": 0.0008,
4125
+ "step": 6770
4126
+ },
4127
+ {
4128
+ "epoch": 6.41,
4129
+ "learning_rate": 4.6478450217750414e-06,
4130
+ "loss": 0.0032,
4131
+ "step": 6780
4132
+ },
4133
+ {
4134
+ "epoch": 6.42,
4135
+ "learning_rate": 4.5727586724733445e-06,
4136
+ "loss": 0.005,
4137
+ "step": 6790
4138
+ },
4139
+ {
4140
+ "epoch": 6.43,
4141
+ "learning_rate": 4.497672323171648e-06,
4142
+ "loss": 0.0094,
4143
+ "step": 6800
4144
+ },
4145
+ {
4146
+ "epoch": 6.44,
4147
+ "learning_rate": 4.422585973869951e-06,
4148
+ "loss": 0.0002,
4149
+ "step": 6810
4150
+ },
4151
+ {
4152
+ "epoch": 6.45,
4153
+ "learning_rate": 4.347499624568253e-06,
4154
+ "loss": 0.004,
4155
+ "step": 6820
4156
+ },
4157
+ {
4158
+ "epoch": 6.46,
4159
+ "learning_rate": 4.272413275266557e-06,
4160
+ "loss": 0.0004,
4161
+ "step": 6830
4162
+ },
4163
+ {
4164
+ "epoch": 6.47,
4165
+ "learning_rate": 4.19732692596486e-06,
4166
+ "loss": 0.0002,
4167
+ "step": 6840
4168
+ },
4169
+ {
4170
+ "epoch": 6.48,
4171
+ "learning_rate": 4.122240576663162e-06,
4172
+ "loss": 0.0012,
4173
+ "step": 6850
4174
+ },
4175
+ {
4176
+ "epoch": 6.49,
4177
+ "learning_rate": 4.047154227361466e-06,
4178
+ "loss": 0.0002,
4179
+ "step": 6860
4180
+ },
4181
+ {
4182
+ "epoch": 6.49,
4183
+ "learning_rate": 3.9720678780597685e-06,
4184
+ "loss": 0.006,
4185
+ "step": 6870
4186
+ },
4187
+ {
4188
+ "epoch": 6.5,
4189
+ "learning_rate": 3.8969815287580725e-06,
4190
+ "loss": 0.0004,
4191
+ "step": 6880
4192
+ },
4193
+ {
4194
+ "epoch": 6.51,
4195
+ "learning_rate": 3.821895179456375e-06,
4196
+ "loss": 0.0097,
4197
+ "step": 6890
4198
+ },
4199
+ {
4200
+ "epoch": 6.52,
4201
+ "learning_rate": 3.746808830154678e-06,
4202
+ "loss": 0.0,
4203
+ "step": 6900
4204
+ },
4205
+ {
4206
+ "epoch": 6.53,
4207
+ "learning_rate": 3.6717224808529814e-06,
4208
+ "loss": 0.0003,
4209
+ "step": 6910
4210
+ },
4211
+ {
4212
+ "epoch": 6.54,
4213
+ "learning_rate": 3.596636131551284e-06,
4214
+ "loss": 0.0007,
4215
+ "step": 6920
4216
+ },
4217
+ {
4218
+ "epoch": 6.55,
4219
+ "learning_rate": 3.521549782249587e-06,
4220
+ "loss": 0.0031,
4221
+ "step": 6930
4222
+ },
4223
+ {
4224
+ "epoch": 6.56,
4225
+ "learning_rate": 3.4464634329478907e-06,
4226
+ "loss": 0.0044,
4227
+ "step": 6940
4228
+ },
4229
+ {
4230
+ "epoch": 6.57,
4231
+ "learning_rate": 3.3713770836461934e-06,
4232
+ "loss": 0.0003,
4233
+ "step": 6950
4234
+ },
4235
+ {
4236
+ "epoch": 6.58,
4237
+ "learning_rate": 3.296290734344496e-06,
4238
+ "loss": 0.0017,
4239
+ "step": 6960
4240
+ },
4241
+ {
4242
+ "epoch": 6.59,
4243
+ "learning_rate": 3.2212043850427996e-06,
4244
+ "loss": 0.0102,
4245
+ "step": 6970
4246
+ },
4247
+ {
4248
+ "epoch": 6.6,
4249
+ "learning_rate": 3.1461180357411023e-06,
4250
+ "loss": 0.0006,
4251
+ "step": 6980
4252
+ },
4253
+ {
4254
+ "epoch": 6.61,
4255
+ "learning_rate": 3.0710316864394054e-06,
4256
+ "loss": 0.0013,
4257
+ "step": 6990
4258
+ },
4259
+ {
4260
+ "epoch": 6.62,
4261
+ "learning_rate": 2.9959453371377085e-06,
4262
+ "loss": 0.0023,
4263
+ "step": 7000
4264
+ },
4265
+ {
4266
+ "epoch": 6.63,
4267
+ "learning_rate": 2.9208589878360116e-06,
4268
+ "loss": 0.0001,
4269
+ "step": 7010
4270
+ },
4271
+ {
4272
+ "epoch": 6.64,
4273
+ "learning_rate": 2.8457726385343147e-06,
4274
+ "loss": 0.002,
4275
+ "step": 7020
4276
+ },
4277
+ {
4278
+ "epoch": 6.65,
4279
+ "learning_rate": 2.7706862892326174e-06,
4280
+ "loss": 0.0011,
4281
+ "step": 7030
4282
+ },
4283
+ {
4284
+ "epoch": 6.66,
4285
+ "learning_rate": 2.6955999399309205e-06,
4286
+ "loss": 0.0118,
4287
+ "step": 7040
4288
+ },
4289
+ {
4290
+ "epoch": 6.67,
4291
+ "learning_rate": 2.6205135906292236e-06,
4292
+ "loss": 0.0001,
4293
+ "step": 7050
4294
+ },
4295
+ {
4296
+ "epoch": 6.67,
4297
+ "learning_rate": 2.545427241327527e-06,
4298
+ "loss": 0.0055,
4299
+ "step": 7060
4300
+ },
4301
+ {
4302
+ "epoch": 6.68,
4303
+ "learning_rate": 2.47034089202583e-06,
4304
+ "loss": 0.002,
4305
+ "step": 7070
4306
+ },
4307
+ {
4308
+ "epoch": 6.69,
4309
+ "learning_rate": 2.395254542724133e-06,
4310
+ "loss": 0.0186,
4311
+ "step": 7080
4312
+ },
4313
+ {
4314
+ "epoch": 6.7,
4315
+ "learning_rate": 2.320168193422436e-06,
4316
+ "loss": 0.0146,
4317
+ "step": 7090
4318
+ },
4319
+ {
4320
+ "epoch": 6.71,
4321
+ "learning_rate": 2.2450818441207387e-06,
4322
+ "loss": 0.0062,
4323
+ "step": 7100
4324
+ },
4325
+ {
4326
+ "epoch": 6.72,
4327
+ "learning_rate": 2.169995494819042e-06,
4328
+ "loss": 0.008,
4329
+ "step": 7110
4330
+ },
4331
+ {
4332
+ "epoch": 6.73,
4333
+ "learning_rate": 2.0949091455173454e-06,
4334
+ "loss": 0.0009,
4335
+ "step": 7120
4336
+ },
4337
+ {
4338
+ "epoch": 6.74,
4339
+ "learning_rate": 2.019822796215648e-06,
4340
+ "loss": 0.0026,
4341
+ "step": 7130
4342
+ },
4343
+ {
4344
+ "epoch": 6.75,
4345
+ "learning_rate": 1.944736446913951e-06,
4346
+ "loss": 0.0001,
4347
+ "step": 7140
4348
+ },
4349
+ {
4350
+ "epoch": 6.76,
4351
+ "learning_rate": 1.8696500976122543e-06,
4352
+ "loss": 0.0001,
4353
+ "step": 7150
4354
+ },
4355
+ {
4356
+ "epoch": 6.77,
4357
+ "learning_rate": 1.7945637483105572e-06,
4358
+ "loss": 0.002,
4359
+ "step": 7160
4360
+ },
4361
+ {
4362
+ "epoch": 6.78,
4363
+ "learning_rate": 1.7194773990088603e-06,
4364
+ "loss": 0.0009,
4365
+ "step": 7170
4366
+ },
4367
+ {
4368
+ "epoch": 6.79,
4369
+ "learning_rate": 1.6443910497071634e-06,
4370
+ "loss": 0.024,
4371
+ "step": 7180
4372
+ },
4373
+ {
4374
+ "epoch": 6.8,
4375
+ "learning_rate": 1.5693047004054665e-06,
4376
+ "loss": 0.0039,
4377
+ "step": 7190
4378
+ },
4379
+ {
4380
+ "epoch": 6.81,
4381
+ "learning_rate": 1.4942183511037694e-06,
4382
+ "loss": 0.0015,
4383
+ "step": 7200
4384
+ },
4385
+ {
4386
+ "epoch": 6.82,
4387
+ "learning_rate": 1.4191320018020725e-06,
4388
+ "loss": 0.0037,
4389
+ "step": 7210
4390
+ },
4391
+ {
4392
+ "epoch": 6.83,
4393
+ "learning_rate": 1.3440456525003754e-06,
4394
+ "loss": 0.0001,
4395
+ "step": 7220
4396
+ },
4397
+ {
4398
+ "epoch": 6.84,
4399
+ "learning_rate": 1.2689593031986787e-06,
4400
+ "loss": 0.0157,
4401
+ "step": 7230
4402
+ },
4403
+ {
4404
+ "epoch": 6.84,
4405
+ "learning_rate": 1.1938729538969816e-06,
4406
+ "loss": 0.0002,
4407
+ "step": 7240
4408
+ },
4409
+ {
4410
+ "epoch": 6.85,
4411
+ "learning_rate": 1.1187866045952845e-06,
4412
+ "loss": 0.0176,
4413
+ "step": 7250
4414
+ },
4415
+ {
4416
+ "epoch": 6.86,
4417
+ "learning_rate": 1.0437002552935878e-06,
4418
+ "loss": 0.0001,
4419
+ "step": 7260
4420
+ },
4421
+ {
4422
+ "epoch": 6.87,
4423
+ "learning_rate": 9.686139059918907e-07,
4424
+ "loss": 0.0,
4425
+ "step": 7270
4426
+ },
4427
+ {
4428
+ "epoch": 6.88,
4429
+ "learning_rate": 8.935275566901939e-07,
4430
+ "loss": 0.0025,
4431
+ "step": 7280
4432
+ },
4433
+ {
4434
+ "epoch": 6.89,
4435
+ "learning_rate": 8.184412073884968e-07,
4436
+ "loss": 0.0006,
4437
+ "step": 7290
4438
+ },
4439
+ {
4440
+ "epoch": 6.9,
4441
+ "learning_rate": 7.433548580867999e-07,
4442
+ "loss": 0.0052,
4443
+ "step": 7300
4444
+ },
4445
+ {
4446
+ "epoch": 6.91,
4447
+ "learning_rate": 6.682685087851029e-07,
4448
+ "loss": 0.0001,
4449
+ "step": 7310
4450
+ },
4451
+ {
4452
+ "epoch": 6.92,
4453
+ "learning_rate": 5.931821594834059e-07,
4454
+ "loss": 0.0019,
4455
+ "step": 7320
4456
+ },
4457
+ {
4458
+ "epoch": 6.93,
4459
+ "learning_rate": 5.18095810181709e-07,
4460
+ "loss": 0.0004,
4461
+ "step": 7330
4462
+ },
4463
+ {
4464
+ "epoch": 6.94,
4465
+ "learning_rate": 4.4300946088001206e-07,
4466
+ "loss": 0.0128,
4467
+ "step": 7340
4468
+ },
4469
+ {
4470
+ "epoch": 6.95,
4471
+ "learning_rate": 3.6792311157831506e-07,
4472
+ "loss": 0.0044,
4473
+ "step": 7350
4474
+ },
4475
+ {
4476
+ "epoch": 6.96,
4477
+ "learning_rate": 2.928367622766181e-07,
4478
+ "loss": 0.0002,
4479
+ "step": 7360
4480
+ },
4481
+ {
4482
+ "epoch": 6.97,
4483
+ "learning_rate": 2.1775041297492117e-07,
4484
+ "loss": 0.0002,
4485
+ "step": 7370
4486
+ },
4487
+ {
4488
+ "epoch": 6.98,
4489
+ "learning_rate": 1.4266406367322422e-07,
4490
+ "loss": 0.0105,
4491
+ "step": 7380
4492
+ },
4493
+ {
4494
+ "epoch": 6.99,
4495
+ "learning_rate": 6.757771437152727e-08,
4496
+ "loss": 0.0001,
4497
+ "step": 7390
4498
+ },
4499
+ {
4500
+ "epoch": 7.0,
4501
+ "eval_accuracy": 0.9997796191819464,
4502
+ "eval_loss": 0.0009974736021831632,
4503
+ "eval_runtime": 143.004,
4504
+ "eval_samples_per_second": 158.653,
4505
+ "eval_steps_per_second": 9.916,
4506
+ "step": 7399
4507
+ },
4508
+ {
4509
+ "epoch": 7.0,
4510
+ "step": 7399,
4511
+ "total_flos": 3.714527647408435e+19,
4512
+ "train_loss": 0.14141770749387125,
4513
+ "train_runtime": 7823.1917,
4514
+ "train_samples_per_second": 60.567,
4515
+ "train_steps_per_second": 0.946
4516
+ }
4517
+ ],
4518
+ "logging_steps": 10,
4519
+ "max_steps": 7399,
4520
+ "num_train_epochs": 7,
4521
+ "save_steps": 500,
4522
+ "total_flos": 3.714527647408435e+19,
4523
+ "trial_name": null,
4524
+ "trial_params": null
4525
+ }