Frikallo commited on
Commit
61e039b
·
1 Parent(s): 775425c

End of training

Browse files
Files changed (3) hide show
  1. all_results.json +8 -0
  2. train_results.json +8 -0
  3. trainer_state.json +3661 -0
all_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "train_loss": 2.9040869920020134,
4
+ "train_runtime": 3297.2859,
5
+ "train_samples": 3031,
6
+ "train_samples_per_second": 0.919,
7
+ "train_steps_per_second": 0.919
8
+ }
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "train_loss": 2.9040869920020134,
4
+ "train_runtime": 3297.2859,
5
+ "train_samples": 3031,
6
+ "train_samples_per_second": 0.919,
7
+ "train_steps_per_second": 0.919
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,3661 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.0,
5
+ "global_step": 3031,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 0.00013697367205542726,
13
+ "loss": 4.515,
14
+ "step": 5
15
+ },
16
+ {
17
+ "epoch": 0.0,
18
+ "learning_rate": 0.0001367473441108545,
19
+ "loss": 4.4104,
20
+ "step": 10
21
+ },
22
+ {
23
+ "epoch": 0.0,
24
+ "learning_rate": 0.00013652101616628176,
25
+ "loss": 4.1934,
26
+ "step": 15
27
+ },
28
+ {
29
+ "epoch": 0.01,
30
+ "learning_rate": 0.000136294688221709,
31
+ "loss": 4.2499,
32
+ "step": 20
33
+ },
34
+ {
35
+ "epoch": 0.01,
36
+ "learning_rate": 0.00013606836027713626,
37
+ "loss": 4.1236,
38
+ "step": 25
39
+ },
40
+ {
41
+ "epoch": 0.01,
42
+ "learning_rate": 0.00013584203233256352,
43
+ "loss": 4.1098,
44
+ "step": 30
45
+ },
46
+ {
47
+ "epoch": 0.01,
48
+ "learning_rate": 0.00013561570438799076,
49
+ "loss": 4.0061,
50
+ "step": 35
51
+ },
52
+ {
53
+ "epoch": 0.01,
54
+ "learning_rate": 0.000135389376443418,
55
+ "loss": 3.8647,
56
+ "step": 40
57
+ },
58
+ {
59
+ "epoch": 0.01,
60
+ "learning_rate": 0.00013516304849884526,
61
+ "loss": 3.9263,
62
+ "step": 45
63
+ },
64
+ {
65
+ "epoch": 0.02,
66
+ "learning_rate": 0.00013493672055427252,
67
+ "loss": 3.8562,
68
+ "step": 50
69
+ },
70
+ {
71
+ "epoch": 0.02,
72
+ "learning_rate": 0.00013471039260969976,
73
+ "loss": 3.9549,
74
+ "step": 55
75
+ },
76
+ {
77
+ "epoch": 0.02,
78
+ "learning_rate": 0.00013448406466512702,
79
+ "loss": 3.947,
80
+ "step": 60
81
+ },
82
+ {
83
+ "epoch": 0.02,
84
+ "learning_rate": 0.00013425773672055428,
85
+ "loss": 3.9132,
86
+ "step": 65
87
+ },
88
+ {
89
+ "epoch": 0.02,
90
+ "learning_rate": 0.00013403140877598152,
91
+ "loss": 3.7056,
92
+ "step": 70
93
+ },
94
+ {
95
+ "epoch": 0.02,
96
+ "learning_rate": 0.00013380508083140878,
97
+ "loss": 3.9002,
98
+ "step": 75
99
+ },
100
+ {
101
+ "epoch": 0.03,
102
+ "learning_rate": 0.00013357875288683602,
103
+ "loss": 3.9315,
104
+ "step": 80
105
+ },
106
+ {
107
+ "epoch": 0.03,
108
+ "learning_rate": 0.00013335242494226328,
109
+ "loss": 3.8103,
110
+ "step": 85
111
+ },
112
+ {
113
+ "epoch": 0.03,
114
+ "learning_rate": 0.00013312609699769055,
115
+ "loss": 3.7449,
116
+ "step": 90
117
+ },
118
+ {
119
+ "epoch": 0.03,
120
+ "learning_rate": 0.00013289976905311778,
121
+ "loss": 3.7089,
122
+ "step": 95
123
+ },
124
+ {
125
+ "epoch": 0.03,
126
+ "learning_rate": 0.00013267344110854504,
127
+ "loss": 3.8787,
128
+ "step": 100
129
+ },
130
+ {
131
+ "epoch": 0.03,
132
+ "learning_rate": 0.00013244711316397228,
133
+ "loss": 3.797,
134
+ "step": 105
135
+ },
136
+ {
137
+ "epoch": 0.04,
138
+ "learning_rate": 0.00013222078521939954,
139
+ "loss": 3.5114,
140
+ "step": 110
141
+ },
142
+ {
143
+ "epoch": 0.04,
144
+ "learning_rate": 0.00013199445727482678,
145
+ "loss": 3.8432,
146
+ "step": 115
147
+ },
148
+ {
149
+ "epoch": 0.04,
150
+ "learning_rate": 0.00013176812933025404,
151
+ "loss": 3.4919,
152
+ "step": 120
153
+ },
154
+ {
155
+ "epoch": 0.04,
156
+ "learning_rate": 0.00013154180138568128,
157
+ "loss": 3.683,
158
+ "step": 125
159
+ },
160
+ {
161
+ "epoch": 0.04,
162
+ "learning_rate": 0.00013131547344110854,
163
+ "loss": 3.9541,
164
+ "step": 130
165
+ },
166
+ {
167
+ "epoch": 0.04,
168
+ "learning_rate": 0.0001310891454965358,
169
+ "loss": 3.6708,
170
+ "step": 135
171
+ },
172
+ {
173
+ "epoch": 0.05,
174
+ "learning_rate": 0.00013086281755196304,
175
+ "loss": 3.7126,
176
+ "step": 140
177
+ },
178
+ {
179
+ "epoch": 0.05,
180
+ "learning_rate": 0.0001306364896073903,
181
+ "loss": 3.6123,
182
+ "step": 145
183
+ },
184
+ {
185
+ "epoch": 0.05,
186
+ "learning_rate": 0.00013041016166281757,
187
+ "loss": 3.7501,
188
+ "step": 150
189
+ },
190
+ {
191
+ "epoch": 0.05,
192
+ "learning_rate": 0.0001301838337182448,
193
+ "loss": 3.6422,
194
+ "step": 155
195
+ },
196
+ {
197
+ "epoch": 0.05,
198
+ "learning_rate": 0.00012995750577367206,
199
+ "loss": 3.5964,
200
+ "step": 160
201
+ },
202
+ {
203
+ "epoch": 0.05,
204
+ "learning_rate": 0.0001297311778290993,
205
+ "loss": 3.7605,
206
+ "step": 165
207
+ },
208
+ {
209
+ "epoch": 0.06,
210
+ "learning_rate": 0.00012950484988452656,
211
+ "loss": 3.5969,
212
+ "step": 170
213
+ },
214
+ {
215
+ "epoch": 0.06,
216
+ "learning_rate": 0.00012927852193995383,
217
+ "loss": 3.6422,
218
+ "step": 175
219
+ },
220
+ {
221
+ "epoch": 0.06,
222
+ "learning_rate": 0.00012905219399538106,
223
+ "loss": 3.3317,
224
+ "step": 180
225
+ },
226
+ {
227
+ "epoch": 0.06,
228
+ "learning_rate": 0.00012882586605080833,
229
+ "loss": 3.3237,
230
+ "step": 185
231
+ },
232
+ {
233
+ "epoch": 0.06,
234
+ "learning_rate": 0.00012859953810623556,
235
+ "loss": 3.511,
236
+ "step": 190
237
+ },
238
+ {
239
+ "epoch": 0.06,
240
+ "learning_rate": 0.00012837321016166282,
241
+ "loss": 3.634,
242
+ "step": 195
243
+ },
244
+ {
245
+ "epoch": 0.07,
246
+ "learning_rate": 0.00012814688221709006,
247
+ "loss": 3.2707,
248
+ "step": 200
249
+ },
250
+ {
251
+ "epoch": 0.07,
252
+ "learning_rate": 0.00012792055427251732,
253
+ "loss": 3.7581,
254
+ "step": 205
255
+ },
256
+ {
257
+ "epoch": 0.07,
258
+ "learning_rate": 0.00012769422632794456,
259
+ "loss": 3.7056,
260
+ "step": 210
261
+ },
262
+ {
263
+ "epoch": 0.07,
264
+ "learning_rate": 0.00012746789838337182,
265
+ "loss": 3.784,
266
+ "step": 215
267
+ },
268
+ {
269
+ "epoch": 0.07,
270
+ "learning_rate": 0.00012724157043879909,
271
+ "loss": 2.9932,
272
+ "step": 220
273
+ },
274
+ {
275
+ "epoch": 0.07,
276
+ "learning_rate": 0.00012701524249422632,
277
+ "loss": 3.6009,
278
+ "step": 225
279
+ },
280
+ {
281
+ "epoch": 0.08,
282
+ "learning_rate": 0.00012678891454965358,
283
+ "loss": 3.7976,
284
+ "step": 230
285
+ },
286
+ {
287
+ "epoch": 0.08,
288
+ "learning_rate": 0.00012656258660508085,
289
+ "loss": 3.7625,
290
+ "step": 235
291
+ },
292
+ {
293
+ "epoch": 0.08,
294
+ "learning_rate": 0.00012633625866050808,
295
+ "loss": 3.264,
296
+ "step": 240
297
+ },
298
+ {
299
+ "epoch": 0.08,
300
+ "learning_rate": 0.00012610993071593535,
301
+ "loss": 3.7345,
302
+ "step": 245
303
+ },
304
+ {
305
+ "epoch": 0.08,
306
+ "learning_rate": 0.00012588360277136258,
307
+ "loss": 3.5308,
308
+ "step": 250
309
+ },
310
+ {
311
+ "epoch": 0.08,
312
+ "learning_rate": 0.00012565727482678985,
313
+ "loss": 3.7175,
314
+ "step": 255
315
+ },
316
+ {
317
+ "epoch": 0.09,
318
+ "learning_rate": 0.0001254309468822171,
319
+ "loss": 3.5114,
320
+ "step": 260
321
+ },
322
+ {
323
+ "epoch": 0.09,
324
+ "learning_rate": 0.00012520461893764434,
325
+ "loss": 3.7169,
326
+ "step": 265
327
+ },
328
+ {
329
+ "epoch": 0.09,
330
+ "learning_rate": 0.00012497829099307158,
331
+ "loss": 3.4318,
332
+ "step": 270
333
+ },
334
+ {
335
+ "epoch": 0.09,
336
+ "learning_rate": 0.00012475196304849884,
337
+ "loss": 3.3658,
338
+ "step": 275
339
+ },
340
+ {
341
+ "epoch": 0.09,
342
+ "learning_rate": 0.0001245256351039261,
343
+ "loss": 3.7082,
344
+ "step": 280
345
+ },
346
+ {
347
+ "epoch": 0.09,
348
+ "learning_rate": 0.00012429930715935334,
349
+ "loss": 3.5838,
350
+ "step": 285
351
+ },
352
+ {
353
+ "epoch": 0.1,
354
+ "learning_rate": 0.0001240729792147806,
355
+ "loss": 2.8508,
356
+ "step": 290
357
+ },
358
+ {
359
+ "epoch": 0.1,
360
+ "learning_rate": 0.00012384665127020784,
361
+ "loss": 3.2041,
362
+ "step": 295
363
+ },
364
+ {
365
+ "epoch": 0.1,
366
+ "learning_rate": 0.0001236203233256351,
367
+ "loss": 3.1151,
368
+ "step": 300
369
+ },
370
+ {
371
+ "epoch": 0.1,
372
+ "learning_rate": 0.00012339399538106237,
373
+ "loss": 3.5128,
374
+ "step": 305
375
+ },
376
+ {
377
+ "epoch": 0.1,
378
+ "learning_rate": 0.0001231676674364896,
379
+ "loss": 3.6807,
380
+ "step": 310
381
+ },
382
+ {
383
+ "epoch": 0.1,
384
+ "learning_rate": 0.00012294133949191687,
385
+ "loss": 3.6973,
386
+ "step": 315
387
+ },
388
+ {
389
+ "epoch": 0.11,
390
+ "learning_rate": 0.00012271501154734413,
391
+ "loss": 3.7338,
392
+ "step": 320
393
+ },
394
+ {
395
+ "epoch": 0.11,
396
+ "learning_rate": 0.00012248868360277136,
397
+ "loss": 3.3445,
398
+ "step": 325
399
+ },
400
+ {
401
+ "epoch": 0.11,
402
+ "learning_rate": 0.00012226235565819863,
403
+ "loss": 3.6292,
404
+ "step": 330
405
+ },
406
+ {
407
+ "epoch": 0.11,
408
+ "learning_rate": 0.00012203602771362588,
409
+ "loss": 3.6886,
410
+ "step": 335
411
+ },
412
+ {
413
+ "epoch": 0.11,
414
+ "learning_rate": 0.00012180969976905311,
415
+ "loss": 2.6932,
416
+ "step": 340
417
+ },
418
+ {
419
+ "epoch": 0.11,
420
+ "learning_rate": 0.00012158337182448036,
421
+ "loss": 3.4789,
422
+ "step": 345
423
+ },
424
+ {
425
+ "epoch": 0.12,
426
+ "learning_rate": 0.00012135704387990761,
427
+ "loss": 3.4813,
428
+ "step": 350
429
+ },
430
+ {
431
+ "epoch": 0.12,
432
+ "learning_rate": 0.00012113071593533487,
433
+ "loss": 3.6185,
434
+ "step": 355
435
+ },
436
+ {
437
+ "epoch": 0.12,
438
+ "learning_rate": 0.00012090438799076212,
439
+ "loss": 3.5236,
440
+ "step": 360
441
+ },
442
+ {
443
+ "epoch": 0.12,
444
+ "learning_rate": 0.00012067806004618937,
445
+ "loss": 3.6228,
446
+ "step": 365
447
+ },
448
+ {
449
+ "epoch": 0.12,
450
+ "learning_rate": 0.00012045173210161662,
451
+ "loss": 3.414,
452
+ "step": 370
453
+ },
454
+ {
455
+ "epoch": 0.12,
456
+ "learning_rate": 0.00012022540415704389,
457
+ "loss": 3.6309,
458
+ "step": 375
459
+ },
460
+ {
461
+ "epoch": 0.13,
462
+ "learning_rate": 0.00011999907621247114,
463
+ "loss": 3.5011,
464
+ "step": 380
465
+ },
466
+ {
467
+ "epoch": 0.13,
468
+ "learning_rate": 0.00011977274826789839,
469
+ "loss": 3.5267,
470
+ "step": 385
471
+ },
472
+ {
473
+ "epoch": 0.13,
474
+ "learning_rate": 0.00011954642032332563,
475
+ "loss": 3.5156,
476
+ "step": 390
477
+ },
478
+ {
479
+ "epoch": 0.13,
480
+ "learning_rate": 0.0001193200923787529,
481
+ "loss": 3.3563,
482
+ "step": 395
483
+ },
484
+ {
485
+ "epoch": 0.13,
486
+ "learning_rate": 0.00011909376443418015,
487
+ "loss": 3.6401,
488
+ "step": 400
489
+ },
490
+ {
491
+ "epoch": 0.13,
492
+ "learning_rate": 0.0001188674364896074,
493
+ "loss": 3.4525,
494
+ "step": 405
495
+ },
496
+ {
497
+ "epoch": 0.14,
498
+ "learning_rate": 0.00011864110854503465,
499
+ "loss": 3.5465,
500
+ "step": 410
501
+ },
502
+ {
503
+ "epoch": 0.14,
504
+ "learning_rate": 0.00011841478060046188,
505
+ "loss": 2.87,
506
+ "step": 415
507
+ },
508
+ {
509
+ "epoch": 0.14,
510
+ "learning_rate": 0.00011818845265588914,
511
+ "loss": 2.6848,
512
+ "step": 420
513
+ },
514
+ {
515
+ "epoch": 0.14,
516
+ "learning_rate": 0.0001179621247113164,
517
+ "loss": 3.5327,
518
+ "step": 425
519
+ },
520
+ {
521
+ "epoch": 0.14,
522
+ "learning_rate": 0.00011773579676674364,
523
+ "loss": 3.1602,
524
+ "step": 430
525
+ },
526
+ {
527
+ "epoch": 0.14,
528
+ "learning_rate": 0.0001175094688221709,
529
+ "loss": 3.1387,
530
+ "step": 435
531
+ },
532
+ {
533
+ "epoch": 0.15,
534
+ "learning_rate": 0.00011728314087759816,
535
+ "loss": 3.1799,
536
+ "step": 440
537
+ },
538
+ {
539
+ "epoch": 0.15,
540
+ "learning_rate": 0.0001170568129330254,
541
+ "loss": 3.1679,
542
+ "step": 445
543
+ },
544
+ {
545
+ "epoch": 0.15,
546
+ "learning_rate": 0.00011683048498845266,
547
+ "loss": 3.0457,
548
+ "step": 450
549
+ },
550
+ {
551
+ "epoch": 0.15,
552
+ "learning_rate": 0.0001166041570438799,
553
+ "loss": 3.5954,
554
+ "step": 455
555
+ },
556
+ {
557
+ "epoch": 0.15,
558
+ "learning_rate": 0.00011637782909930717,
559
+ "loss": 3.008,
560
+ "step": 460
561
+ },
562
+ {
563
+ "epoch": 0.15,
564
+ "learning_rate": 0.00011615150115473442,
565
+ "loss": 3.6305,
566
+ "step": 465
567
+ },
568
+ {
569
+ "epoch": 0.16,
570
+ "learning_rate": 0.00011592517321016167,
571
+ "loss": 3.056,
572
+ "step": 470
573
+ },
574
+ {
575
+ "epoch": 0.16,
576
+ "learning_rate": 0.00011569884526558892,
577
+ "loss": 3.0358,
578
+ "step": 475
579
+ },
580
+ {
581
+ "epoch": 0.16,
582
+ "learning_rate": 0.00011547251732101618,
583
+ "loss": 3.3535,
584
+ "step": 480
585
+ },
586
+ {
587
+ "epoch": 0.16,
588
+ "learning_rate": 0.00011524618937644343,
589
+ "loss": 3.5352,
590
+ "step": 485
591
+ },
592
+ {
593
+ "epoch": 0.16,
594
+ "learning_rate": 0.00011501986143187066,
595
+ "loss": 3.5372,
596
+ "step": 490
597
+ },
598
+ {
599
+ "epoch": 0.16,
600
+ "learning_rate": 0.00011479353348729791,
601
+ "loss": 2.9131,
602
+ "step": 495
603
+ },
604
+ {
605
+ "epoch": 0.16,
606
+ "learning_rate": 0.00011456720554272516,
607
+ "loss": 3.4774,
608
+ "step": 500
609
+ },
610
+ {
611
+ "epoch": 0.17,
612
+ "learning_rate": 0.00011434087759815243,
613
+ "loss": 3.5785,
614
+ "step": 505
615
+ },
616
+ {
617
+ "epoch": 0.17,
618
+ "learning_rate": 0.00011411454965357968,
619
+ "loss": 3.0545,
620
+ "step": 510
621
+ },
622
+ {
623
+ "epoch": 0.17,
624
+ "learning_rate": 0.00011388822170900693,
625
+ "loss": 3.6184,
626
+ "step": 515
627
+ },
628
+ {
629
+ "epoch": 0.17,
630
+ "learning_rate": 0.00011366189376443417,
631
+ "loss": 3.536,
632
+ "step": 520
633
+ },
634
+ {
635
+ "epoch": 0.17,
636
+ "learning_rate": 0.00011343556581986144,
637
+ "loss": 3.4078,
638
+ "step": 525
639
+ },
640
+ {
641
+ "epoch": 0.17,
642
+ "learning_rate": 0.00011320923787528869,
643
+ "loss": 3.0253,
644
+ "step": 530
645
+ },
646
+ {
647
+ "epoch": 0.18,
648
+ "learning_rate": 0.00011298290993071594,
649
+ "loss": 3.5612,
650
+ "step": 535
651
+ },
652
+ {
653
+ "epoch": 0.18,
654
+ "learning_rate": 0.00011275658198614319,
655
+ "loss": 3.609,
656
+ "step": 540
657
+ },
658
+ {
659
+ "epoch": 0.18,
660
+ "learning_rate": 0.00011253025404157045,
661
+ "loss": 3.48,
662
+ "step": 545
663
+ },
664
+ {
665
+ "epoch": 0.18,
666
+ "learning_rate": 0.0001123039260969977,
667
+ "loss": 3.0283,
668
+ "step": 550
669
+ },
670
+ {
671
+ "epoch": 0.18,
672
+ "learning_rate": 0.00011207759815242495,
673
+ "loss": 2.9956,
674
+ "step": 555
675
+ },
676
+ {
677
+ "epoch": 0.18,
678
+ "learning_rate": 0.0001118512702078522,
679
+ "loss": 3.4985,
680
+ "step": 560
681
+ },
682
+ {
683
+ "epoch": 0.19,
684
+ "learning_rate": 0.00011162494226327943,
685
+ "loss": 3.6792,
686
+ "step": 565
687
+ },
688
+ {
689
+ "epoch": 0.19,
690
+ "learning_rate": 0.0001113986143187067,
691
+ "loss": 3.0146,
692
+ "step": 570
693
+ },
694
+ {
695
+ "epoch": 0.19,
696
+ "learning_rate": 0.00011117228637413395,
697
+ "loss": 3.3542,
698
+ "step": 575
699
+ },
700
+ {
701
+ "epoch": 0.19,
702
+ "learning_rate": 0.0001109459584295612,
703
+ "loss": 3.4283,
704
+ "step": 580
705
+ },
706
+ {
707
+ "epoch": 0.19,
708
+ "learning_rate": 0.00011071963048498844,
709
+ "loss": 3.4591,
710
+ "step": 585
711
+ },
712
+ {
713
+ "epoch": 0.19,
714
+ "learning_rate": 0.00011049330254041571,
715
+ "loss": 2.725,
716
+ "step": 590
717
+ },
718
+ {
719
+ "epoch": 0.2,
720
+ "learning_rate": 0.00011026697459584296,
721
+ "loss": 2.8868,
722
+ "step": 595
723
+ },
724
+ {
725
+ "epoch": 0.2,
726
+ "learning_rate": 0.0001100406466512702,
727
+ "loss": 2.981,
728
+ "step": 600
729
+ },
730
+ {
731
+ "epoch": 0.2,
732
+ "learning_rate": 0.00010981431870669746,
733
+ "loss": 3.4859,
734
+ "step": 605
735
+ },
736
+ {
737
+ "epoch": 0.2,
738
+ "learning_rate": 0.00010958799076212472,
739
+ "loss": 3.2683,
740
+ "step": 610
741
+ },
742
+ {
743
+ "epoch": 0.2,
744
+ "learning_rate": 0.00010936166281755197,
745
+ "loss": 3.5555,
746
+ "step": 615
747
+ },
748
+ {
749
+ "epoch": 0.2,
750
+ "learning_rate": 0.00010913533487297922,
751
+ "loss": 2.9897,
752
+ "step": 620
753
+ },
754
+ {
755
+ "epoch": 0.21,
756
+ "learning_rate": 0.00010890900692840647,
757
+ "loss": 3.227,
758
+ "step": 625
759
+ },
760
+ {
761
+ "epoch": 0.21,
762
+ "learning_rate": 0.00010868267898383373,
763
+ "loss": 3.2891,
764
+ "step": 630
765
+ },
766
+ {
767
+ "epoch": 0.21,
768
+ "learning_rate": 0.00010845635103926098,
769
+ "loss": 3.5478,
770
+ "step": 635
771
+ },
772
+ {
773
+ "epoch": 0.21,
774
+ "learning_rate": 0.00010823002309468822,
775
+ "loss": 2.9625,
776
+ "step": 640
777
+ },
778
+ {
779
+ "epoch": 0.21,
780
+ "learning_rate": 0.00010800369515011547,
781
+ "loss": 3.2351,
782
+ "step": 645
783
+ },
784
+ {
785
+ "epoch": 0.21,
786
+ "learning_rate": 0.00010777736720554271,
787
+ "loss": 3.4041,
788
+ "step": 650
789
+ },
790
+ {
791
+ "epoch": 0.22,
792
+ "learning_rate": 0.00010755103926096998,
793
+ "loss": 3.5005,
794
+ "step": 655
795
+ },
796
+ {
797
+ "epoch": 0.22,
798
+ "learning_rate": 0.00010732471131639723,
799
+ "loss": 3.4141,
800
+ "step": 660
801
+ },
802
+ {
803
+ "epoch": 0.22,
804
+ "learning_rate": 0.00010709838337182448,
805
+ "loss": 3.0204,
806
+ "step": 665
807
+ },
808
+ {
809
+ "epoch": 0.22,
810
+ "learning_rate": 0.00010687205542725173,
811
+ "loss": 2.8737,
812
+ "step": 670
813
+ },
814
+ {
815
+ "epoch": 0.22,
816
+ "learning_rate": 0.00010664572748267899,
817
+ "loss": 3.44,
818
+ "step": 675
819
+ },
820
+ {
821
+ "epoch": 0.22,
822
+ "learning_rate": 0.00010641939953810624,
823
+ "loss": 3.4723,
824
+ "step": 680
825
+ },
826
+ {
827
+ "epoch": 0.23,
828
+ "learning_rate": 0.00010619307159353349,
829
+ "loss": 2.8187,
830
+ "step": 685
831
+ },
832
+ {
833
+ "epoch": 0.23,
834
+ "learning_rate": 0.00010596674364896074,
835
+ "loss": 3.5171,
836
+ "step": 690
837
+ },
838
+ {
839
+ "epoch": 0.23,
840
+ "learning_rate": 0.000105740415704388,
841
+ "loss": 3.346,
842
+ "step": 695
843
+ },
844
+ {
845
+ "epoch": 0.23,
846
+ "learning_rate": 0.00010551408775981525,
847
+ "loss": 3.3495,
848
+ "step": 700
849
+ },
850
+ {
851
+ "epoch": 0.23,
852
+ "learning_rate": 0.0001052877598152425,
853
+ "loss": 3.4582,
854
+ "step": 705
855
+ },
856
+ {
857
+ "epoch": 0.23,
858
+ "learning_rate": 0.00010506143187066975,
859
+ "loss": 3.4312,
860
+ "step": 710
861
+ },
862
+ {
863
+ "epoch": 0.24,
864
+ "learning_rate": 0.00010483510392609698,
865
+ "loss": 2.9329,
866
+ "step": 715
867
+ },
868
+ {
869
+ "epoch": 0.24,
870
+ "learning_rate": 0.00010460877598152425,
871
+ "loss": 3.4131,
872
+ "step": 720
873
+ },
874
+ {
875
+ "epoch": 0.24,
876
+ "learning_rate": 0.0001043824480369515,
877
+ "loss": 3.432,
878
+ "step": 725
879
+ },
880
+ {
881
+ "epoch": 0.24,
882
+ "learning_rate": 0.00010415612009237875,
883
+ "loss": 3.4073,
884
+ "step": 730
885
+ },
886
+ {
887
+ "epoch": 0.24,
888
+ "learning_rate": 0.000103929792147806,
889
+ "loss": 3.0032,
890
+ "step": 735
891
+ },
892
+ {
893
+ "epoch": 0.24,
894
+ "learning_rate": 0.00010370346420323326,
895
+ "loss": 3.3989,
896
+ "step": 740
897
+ },
898
+ {
899
+ "epoch": 0.25,
900
+ "learning_rate": 0.00010347713625866051,
901
+ "loss": 2.5839,
902
+ "step": 745
903
+ },
904
+ {
905
+ "epoch": 0.25,
906
+ "learning_rate": 0.00010325080831408776,
907
+ "loss": 3.5058,
908
+ "step": 750
909
+ },
910
+ {
911
+ "epoch": 0.25,
912
+ "learning_rate": 0.00010302448036951501,
913
+ "loss": 2.8574,
914
+ "step": 755
915
+ },
916
+ {
917
+ "epoch": 0.25,
918
+ "learning_rate": 0.00010279815242494227,
919
+ "loss": 3.2324,
920
+ "step": 760
921
+ },
922
+ {
923
+ "epoch": 0.25,
924
+ "learning_rate": 0.00010257182448036952,
925
+ "loss": 3.4433,
926
+ "step": 765
927
+ },
928
+ {
929
+ "epoch": 0.25,
930
+ "learning_rate": 0.00010234549653579677,
931
+ "loss": 3.2637,
932
+ "step": 770
933
+ },
934
+ {
935
+ "epoch": 0.26,
936
+ "learning_rate": 0.00010211916859122402,
937
+ "loss": 3.4732,
938
+ "step": 775
939
+ },
940
+ {
941
+ "epoch": 0.26,
942
+ "learning_rate": 0.00010189284064665128,
943
+ "loss": 3.215,
944
+ "step": 780
945
+ },
946
+ {
947
+ "epoch": 0.26,
948
+ "learning_rate": 0.00010166651270207853,
949
+ "loss": 3.0964,
950
+ "step": 785
951
+ },
952
+ {
953
+ "epoch": 0.26,
954
+ "learning_rate": 0.00010144018475750577,
955
+ "loss": 3.2448,
956
+ "step": 790
957
+ },
958
+ {
959
+ "epoch": 0.26,
960
+ "learning_rate": 0.00010121385681293302,
961
+ "loss": 3.3717,
962
+ "step": 795
963
+ },
964
+ {
965
+ "epoch": 0.26,
966
+ "learning_rate": 0.00010098752886836027,
967
+ "loss": 3.4094,
968
+ "step": 800
969
+ },
970
+ {
971
+ "epoch": 0.27,
972
+ "learning_rate": 0.00010076120092378753,
973
+ "loss": 2.8995,
974
+ "step": 805
975
+ },
976
+ {
977
+ "epoch": 0.27,
978
+ "learning_rate": 0.00010053487297921478,
979
+ "loss": 3.4194,
980
+ "step": 810
981
+ },
982
+ {
983
+ "epoch": 0.27,
984
+ "learning_rate": 0.00010030854503464203,
985
+ "loss": 2.8647,
986
+ "step": 815
987
+ },
988
+ {
989
+ "epoch": 0.27,
990
+ "learning_rate": 0.00010008221709006928,
991
+ "loss": 2.649,
992
+ "step": 820
993
+ },
994
+ {
995
+ "epoch": 0.27,
996
+ "learning_rate": 9.985588914549654e-05,
997
+ "loss": 2.593,
998
+ "step": 825
999
+ },
1000
+ {
1001
+ "epoch": 0.27,
1002
+ "learning_rate": 9.962956120092379e-05,
1003
+ "loss": 2.7692,
1004
+ "step": 830
1005
+ },
1006
+ {
1007
+ "epoch": 0.28,
1008
+ "learning_rate": 9.940323325635104e-05,
1009
+ "loss": 2.923,
1010
+ "step": 835
1011
+ },
1012
+ {
1013
+ "epoch": 0.28,
1014
+ "learning_rate": 9.917690531177829e-05,
1015
+ "loss": 2.7793,
1016
+ "step": 840
1017
+ },
1018
+ {
1019
+ "epoch": 0.28,
1020
+ "learning_rate": 9.895057736720555e-05,
1021
+ "loss": 2.3173,
1022
+ "step": 845
1023
+ },
1024
+ {
1025
+ "epoch": 0.28,
1026
+ "learning_rate": 9.87242494226328e-05,
1027
+ "loss": 3.3614,
1028
+ "step": 850
1029
+ },
1030
+ {
1031
+ "epoch": 0.28,
1032
+ "learning_rate": 9.849792147806005e-05,
1033
+ "loss": 2.8285,
1034
+ "step": 855
1035
+ },
1036
+ {
1037
+ "epoch": 0.28,
1038
+ "learning_rate": 9.82715935334873e-05,
1039
+ "loss": 3.1578,
1040
+ "step": 860
1041
+ },
1042
+ {
1043
+ "epoch": 0.29,
1044
+ "learning_rate": 9.804526558891455e-05,
1045
+ "loss": 3.2701,
1046
+ "step": 865
1047
+ },
1048
+ {
1049
+ "epoch": 0.29,
1050
+ "learning_rate": 9.78189376443418e-05,
1051
+ "loss": 3.1701,
1052
+ "step": 870
1053
+ },
1054
+ {
1055
+ "epoch": 0.29,
1056
+ "learning_rate": 9.759260969976905e-05,
1057
+ "loss": 3.3629,
1058
+ "step": 875
1059
+ },
1060
+ {
1061
+ "epoch": 0.29,
1062
+ "learning_rate": 9.73662817551963e-05,
1063
+ "loss": 3.2969,
1064
+ "step": 880
1065
+ },
1066
+ {
1067
+ "epoch": 0.29,
1068
+ "learning_rate": 9.713995381062355e-05,
1069
+ "loss": 2.4782,
1070
+ "step": 885
1071
+ },
1072
+ {
1073
+ "epoch": 0.29,
1074
+ "learning_rate": 9.691362586605081e-05,
1075
+ "loss": 3.3053,
1076
+ "step": 890
1077
+ },
1078
+ {
1079
+ "epoch": 0.3,
1080
+ "learning_rate": 9.668729792147806e-05,
1081
+ "loss": 3.3026,
1082
+ "step": 895
1083
+ },
1084
+ {
1085
+ "epoch": 0.3,
1086
+ "learning_rate": 9.646096997690531e-05,
1087
+ "loss": 3.353,
1088
+ "step": 900
1089
+ },
1090
+ {
1091
+ "epoch": 0.3,
1092
+ "learning_rate": 9.623464203233256e-05,
1093
+ "loss": 2.7781,
1094
+ "step": 905
1095
+ },
1096
+ {
1097
+ "epoch": 0.3,
1098
+ "learning_rate": 9.600831408775982e-05,
1099
+ "loss": 2.8703,
1100
+ "step": 910
1101
+ },
1102
+ {
1103
+ "epoch": 0.3,
1104
+ "learning_rate": 9.578198614318707e-05,
1105
+ "loss": 3.1661,
1106
+ "step": 915
1107
+ },
1108
+ {
1109
+ "epoch": 0.3,
1110
+ "learning_rate": 9.555565819861432e-05,
1111
+ "loss": 2.7715,
1112
+ "step": 920
1113
+ },
1114
+ {
1115
+ "epoch": 0.31,
1116
+ "learning_rate": 9.532933025404157e-05,
1117
+ "loss": 2.8061,
1118
+ "step": 925
1119
+ },
1120
+ {
1121
+ "epoch": 0.31,
1122
+ "learning_rate": 9.510300230946883e-05,
1123
+ "loss": 3.2015,
1124
+ "step": 930
1125
+ },
1126
+ {
1127
+ "epoch": 0.31,
1128
+ "learning_rate": 9.487667436489608e-05,
1129
+ "loss": 2.4464,
1130
+ "step": 935
1131
+ },
1132
+ {
1133
+ "epoch": 0.31,
1134
+ "learning_rate": 9.465034642032332e-05,
1135
+ "loss": 2.5126,
1136
+ "step": 940
1137
+ },
1138
+ {
1139
+ "epoch": 0.31,
1140
+ "learning_rate": 9.442401847575057e-05,
1141
+ "loss": 3.3578,
1142
+ "step": 945
1143
+ },
1144
+ {
1145
+ "epoch": 0.31,
1146
+ "learning_rate": 9.419769053117783e-05,
1147
+ "loss": 3.2997,
1148
+ "step": 950
1149
+ },
1150
+ {
1151
+ "epoch": 0.32,
1152
+ "learning_rate": 9.397136258660508e-05,
1153
+ "loss": 3.0528,
1154
+ "step": 955
1155
+ },
1156
+ {
1157
+ "epoch": 0.32,
1158
+ "learning_rate": 9.374503464203233e-05,
1159
+ "loss": 3.2615,
1160
+ "step": 960
1161
+ },
1162
+ {
1163
+ "epoch": 0.32,
1164
+ "learning_rate": 9.351870669745958e-05,
1165
+ "loss": 2.2974,
1166
+ "step": 965
1167
+ },
1168
+ {
1169
+ "epoch": 0.32,
1170
+ "learning_rate": 9.329237875288683e-05,
1171
+ "loss": 2.4089,
1172
+ "step": 970
1173
+ },
1174
+ {
1175
+ "epoch": 0.32,
1176
+ "learning_rate": 9.306605080831409e-05,
1177
+ "loss": 3.3305,
1178
+ "step": 975
1179
+ },
1180
+ {
1181
+ "epoch": 0.32,
1182
+ "learning_rate": 9.283972286374134e-05,
1183
+ "loss": 2.9349,
1184
+ "step": 980
1185
+ },
1186
+ {
1187
+ "epoch": 0.32,
1188
+ "learning_rate": 9.261339491916859e-05,
1189
+ "loss": 3.026,
1190
+ "step": 985
1191
+ },
1192
+ {
1193
+ "epoch": 0.33,
1194
+ "learning_rate": 9.238706697459584e-05,
1195
+ "loss": 3.2153,
1196
+ "step": 990
1197
+ },
1198
+ {
1199
+ "epoch": 0.33,
1200
+ "learning_rate": 9.21607390300231e-05,
1201
+ "loss": 2.754,
1202
+ "step": 995
1203
+ },
1204
+ {
1205
+ "epoch": 0.33,
1206
+ "learning_rate": 9.193441108545035e-05,
1207
+ "loss": 3.3203,
1208
+ "step": 1000
1209
+ },
1210
+ {
1211
+ "epoch": 0.33,
1212
+ "learning_rate": 9.17080831408776e-05,
1213
+ "loss": 3.3236,
1214
+ "step": 1005
1215
+ },
1216
+ {
1217
+ "epoch": 0.33,
1218
+ "learning_rate": 9.148175519630485e-05,
1219
+ "loss": 3.2072,
1220
+ "step": 1010
1221
+ },
1222
+ {
1223
+ "epoch": 0.33,
1224
+ "learning_rate": 9.12554272517321e-05,
1225
+ "loss": 3.271,
1226
+ "step": 1015
1227
+ },
1228
+ {
1229
+ "epoch": 0.34,
1230
+ "learning_rate": 9.102909930715935e-05,
1231
+ "loss": 3.3537,
1232
+ "step": 1020
1233
+ },
1234
+ {
1235
+ "epoch": 0.34,
1236
+ "learning_rate": 9.08027713625866e-05,
1237
+ "loss": 3.0024,
1238
+ "step": 1025
1239
+ },
1240
+ {
1241
+ "epoch": 0.34,
1242
+ "learning_rate": 9.057644341801385e-05,
1243
+ "loss": 3.4014,
1244
+ "step": 1030
1245
+ },
1246
+ {
1247
+ "epoch": 0.34,
1248
+ "learning_rate": 9.035011547344111e-05,
1249
+ "loss": 2.6294,
1250
+ "step": 1035
1251
+ },
1252
+ {
1253
+ "epoch": 0.34,
1254
+ "learning_rate": 9.012378752886836e-05,
1255
+ "loss": 2.736,
1256
+ "step": 1040
1257
+ },
1258
+ {
1259
+ "epoch": 0.34,
1260
+ "learning_rate": 8.989745958429561e-05,
1261
+ "loss": 2.1099,
1262
+ "step": 1045
1263
+ },
1264
+ {
1265
+ "epoch": 0.35,
1266
+ "learning_rate": 8.967113163972286e-05,
1267
+ "loss": 3.1103,
1268
+ "step": 1050
1269
+ },
1270
+ {
1271
+ "epoch": 0.35,
1272
+ "learning_rate": 8.944480369515012e-05,
1273
+ "loss": 2.7629,
1274
+ "step": 1055
1275
+ },
1276
+ {
1277
+ "epoch": 0.35,
1278
+ "learning_rate": 8.921847575057737e-05,
1279
+ "loss": 2.9411,
1280
+ "step": 1060
1281
+ },
1282
+ {
1283
+ "epoch": 0.35,
1284
+ "learning_rate": 8.899214780600462e-05,
1285
+ "loss": 2.6607,
1286
+ "step": 1065
1287
+ },
1288
+ {
1289
+ "epoch": 0.35,
1290
+ "learning_rate": 8.876581986143187e-05,
1291
+ "loss": 1.6034,
1292
+ "step": 1070
1293
+ },
1294
+ {
1295
+ "epoch": 0.35,
1296
+ "learning_rate": 8.853949191685912e-05,
1297
+ "loss": 2.7623,
1298
+ "step": 1075
1299
+ },
1300
+ {
1301
+ "epoch": 0.36,
1302
+ "learning_rate": 8.831316397228638e-05,
1303
+ "loss": 3.3491,
1304
+ "step": 1080
1305
+ },
1306
+ {
1307
+ "epoch": 0.36,
1308
+ "learning_rate": 8.808683602771363e-05,
1309
+ "loss": 3.3623,
1310
+ "step": 1085
1311
+ },
1312
+ {
1313
+ "epoch": 0.36,
1314
+ "learning_rate": 8.786050808314087e-05,
1315
+ "loss": 3.3125,
1316
+ "step": 1090
1317
+ },
1318
+ {
1319
+ "epoch": 0.36,
1320
+ "learning_rate": 8.763418013856812e-05,
1321
+ "loss": 3.2459,
1322
+ "step": 1095
1323
+ },
1324
+ {
1325
+ "epoch": 0.36,
1326
+ "learning_rate": 8.740785219399538e-05,
1327
+ "loss": 2.9697,
1328
+ "step": 1100
1329
+ },
1330
+ {
1331
+ "epoch": 0.36,
1332
+ "learning_rate": 8.718152424942263e-05,
1333
+ "loss": 2.063,
1334
+ "step": 1105
1335
+ },
1336
+ {
1337
+ "epoch": 0.37,
1338
+ "learning_rate": 8.695519630484988e-05,
1339
+ "loss": 2.6997,
1340
+ "step": 1110
1341
+ },
1342
+ {
1343
+ "epoch": 0.37,
1344
+ "learning_rate": 8.672886836027713e-05,
1345
+ "loss": 3.2616,
1346
+ "step": 1115
1347
+ },
1348
+ {
1349
+ "epoch": 0.37,
1350
+ "learning_rate": 8.65025404157044e-05,
1351
+ "loss": 3.327,
1352
+ "step": 1120
1353
+ },
1354
+ {
1355
+ "epoch": 0.37,
1356
+ "learning_rate": 8.627621247113164e-05,
1357
+ "loss": 2.6713,
1358
+ "step": 1125
1359
+ },
1360
+ {
1361
+ "epoch": 0.37,
1362
+ "learning_rate": 8.604988452655889e-05,
1363
+ "loss": 2.8985,
1364
+ "step": 1130
1365
+ },
1366
+ {
1367
+ "epoch": 0.37,
1368
+ "learning_rate": 8.582355658198614e-05,
1369
+ "loss": 3.3272,
1370
+ "step": 1135
1371
+ },
1372
+ {
1373
+ "epoch": 0.38,
1374
+ "learning_rate": 8.55972286374134e-05,
1375
+ "loss": 3.2654,
1376
+ "step": 1140
1377
+ },
1378
+ {
1379
+ "epoch": 0.38,
1380
+ "learning_rate": 8.537090069284065e-05,
1381
+ "loss": 3.2365,
1382
+ "step": 1145
1383
+ },
1384
+ {
1385
+ "epoch": 0.38,
1386
+ "learning_rate": 8.51445727482679e-05,
1387
+ "loss": 2.3625,
1388
+ "step": 1150
1389
+ },
1390
+ {
1391
+ "epoch": 0.38,
1392
+ "learning_rate": 8.491824480369515e-05,
1393
+ "loss": 3.3359,
1394
+ "step": 1155
1395
+ },
1396
+ {
1397
+ "epoch": 0.38,
1398
+ "learning_rate": 8.46919168591224e-05,
1399
+ "loss": 3.2978,
1400
+ "step": 1160
1401
+ },
1402
+ {
1403
+ "epoch": 0.38,
1404
+ "learning_rate": 8.446558891454965e-05,
1405
+ "loss": 3.2987,
1406
+ "step": 1165
1407
+ },
1408
+ {
1409
+ "epoch": 0.39,
1410
+ "learning_rate": 8.42392609699769e-05,
1411
+ "loss": 3.1829,
1412
+ "step": 1170
1413
+ },
1414
+ {
1415
+ "epoch": 0.39,
1416
+ "learning_rate": 8.401293302540415e-05,
1417
+ "loss": 2.9166,
1418
+ "step": 1175
1419
+ },
1420
+ {
1421
+ "epoch": 0.39,
1422
+ "learning_rate": 8.37866050808314e-05,
1423
+ "loss": 2.8781,
1424
+ "step": 1180
1425
+ },
1426
+ {
1427
+ "epoch": 0.39,
1428
+ "learning_rate": 8.356027713625866e-05,
1429
+ "loss": 3.311,
1430
+ "step": 1185
1431
+ },
1432
+ {
1433
+ "epoch": 0.39,
1434
+ "learning_rate": 8.333394919168591e-05,
1435
+ "loss": 3.2549,
1436
+ "step": 1190
1437
+ },
1438
+ {
1439
+ "epoch": 0.39,
1440
+ "learning_rate": 8.310762124711316e-05,
1441
+ "loss": 3.3449,
1442
+ "step": 1195
1443
+ },
1444
+ {
1445
+ "epoch": 0.4,
1446
+ "learning_rate": 8.288129330254041e-05,
1447
+ "loss": 3.1834,
1448
+ "step": 1200
1449
+ },
1450
+ {
1451
+ "epoch": 0.4,
1452
+ "learning_rate": 8.265496535796768e-05,
1453
+ "loss": 2.6987,
1454
+ "step": 1205
1455
+ },
1456
+ {
1457
+ "epoch": 0.4,
1458
+ "learning_rate": 8.242863741339492e-05,
1459
+ "loss": 3.2679,
1460
+ "step": 1210
1461
+ },
1462
+ {
1463
+ "epoch": 0.4,
1464
+ "learning_rate": 8.220230946882217e-05,
1465
+ "loss": 2.2549,
1466
+ "step": 1215
1467
+ },
1468
+ {
1469
+ "epoch": 0.4,
1470
+ "learning_rate": 8.197598152424942e-05,
1471
+ "loss": 2.9217,
1472
+ "step": 1220
1473
+ },
1474
+ {
1475
+ "epoch": 0.4,
1476
+ "learning_rate": 8.174965357967669e-05,
1477
+ "loss": 3.2378,
1478
+ "step": 1225
1479
+ },
1480
+ {
1481
+ "epoch": 0.41,
1482
+ "learning_rate": 8.152332563510394e-05,
1483
+ "loss": 3.2458,
1484
+ "step": 1230
1485
+ },
1486
+ {
1487
+ "epoch": 0.41,
1488
+ "learning_rate": 8.129699769053119e-05,
1489
+ "loss": 2.5979,
1490
+ "step": 1235
1491
+ },
1492
+ {
1493
+ "epoch": 0.41,
1494
+ "learning_rate": 8.107066974595842e-05,
1495
+ "loss": 3.141,
1496
+ "step": 1240
1497
+ },
1498
+ {
1499
+ "epoch": 0.41,
1500
+ "learning_rate": 8.084434180138567e-05,
1501
+ "loss": 3.2136,
1502
+ "step": 1245
1503
+ },
1504
+ {
1505
+ "epoch": 0.41,
1506
+ "learning_rate": 8.061801385681293e-05,
1507
+ "loss": 2.5998,
1508
+ "step": 1250
1509
+ },
1510
+ {
1511
+ "epoch": 0.41,
1512
+ "learning_rate": 8.039168591224018e-05,
1513
+ "loss": 1.9745,
1514
+ "step": 1255
1515
+ },
1516
+ {
1517
+ "epoch": 0.42,
1518
+ "learning_rate": 8.016535796766743e-05,
1519
+ "loss": 3.2393,
1520
+ "step": 1260
1521
+ },
1522
+ {
1523
+ "epoch": 0.42,
1524
+ "learning_rate": 7.993903002309468e-05,
1525
+ "loss": 1.985,
1526
+ "step": 1265
1527
+ },
1528
+ {
1529
+ "epoch": 0.42,
1530
+ "learning_rate": 7.971270207852195e-05,
1531
+ "loss": 3.1701,
1532
+ "step": 1270
1533
+ },
1534
+ {
1535
+ "epoch": 0.42,
1536
+ "learning_rate": 7.94863741339492e-05,
1537
+ "loss": 3.2786,
1538
+ "step": 1275
1539
+ },
1540
+ {
1541
+ "epoch": 0.42,
1542
+ "learning_rate": 7.926004618937644e-05,
1543
+ "loss": 3.3514,
1544
+ "step": 1280
1545
+ },
1546
+ {
1547
+ "epoch": 0.42,
1548
+ "learning_rate": 7.90337182448037e-05,
1549
+ "loss": 2.6101,
1550
+ "step": 1285
1551
+ },
1552
+ {
1553
+ "epoch": 0.43,
1554
+ "learning_rate": 7.880739030023096e-05,
1555
+ "loss": 3.2465,
1556
+ "step": 1290
1557
+ },
1558
+ {
1559
+ "epoch": 0.43,
1560
+ "learning_rate": 7.85810623556582e-05,
1561
+ "loss": 3.1456,
1562
+ "step": 1295
1563
+ },
1564
+ {
1565
+ "epoch": 0.43,
1566
+ "learning_rate": 7.835473441108546e-05,
1567
+ "loss": 3.1528,
1568
+ "step": 1300
1569
+ },
1570
+ {
1571
+ "epoch": 0.43,
1572
+ "learning_rate": 7.81284064665127e-05,
1573
+ "loss": 2.3822,
1574
+ "step": 1305
1575
+ },
1576
+ {
1577
+ "epoch": 0.43,
1578
+ "learning_rate": 7.790207852193997e-05,
1579
+ "loss": 3.1659,
1580
+ "step": 1310
1581
+ },
1582
+ {
1583
+ "epoch": 0.43,
1584
+ "learning_rate": 7.76757505773672e-05,
1585
+ "loss": 2.6111,
1586
+ "step": 1315
1587
+ },
1588
+ {
1589
+ "epoch": 0.44,
1590
+ "learning_rate": 7.744942263279445e-05,
1591
+ "loss": 2.6443,
1592
+ "step": 1320
1593
+ },
1594
+ {
1595
+ "epoch": 0.44,
1596
+ "learning_rate": 7.72230946882217e-05,
1597
+ "loss": 2.184,
1598
+ "step": 1325
1599
+ },
1600
+ {
1601
+ "epoch": 0.44,
1602
+ "learning_rate": 7.699676674364895e-05,
1603
+ "loss": 3.2053,
1604
+ "step": 1330
1605
+ },
1606
+ {
1607
+ "epoch": 0.44,
1608
+ "learning_rate": 7.677043879907622e-05,
1609
+ "loss": 3.323,
1610
+ "step": 1335
1611
+ },
1612
+ {
1613
+ "epoch": 0.44,
1614
+ "learning_rate": 7.654411085450346e-05,
1615
+ "loss": 2.2732,
1616
+ "step": 1340
1617
+ },
1618
+ {
1619
+ "epoch": 0.44,
1620
+ "learning_rate": 7.631778290993071e-05,
1621
+ "loss": 3.1413,
1622
+ "step": 1345
1623
+ },
1624
+ {
1625
+ "epoch": 0.45,
1626
+ "learning_rate": 7.609145496535796e-05,
1627
+ "loss": 3.2421,
1628
+ "step": 1350
1629
+ },
1630
+ {
1631
+ "epoch": 0.45,
1632
+ "learning_rate": 7.586512702078523e-05,
1633
+ "loss": 3.1688,
1634
+ "step": 1355
1635
+ },
1636
+ {
1637
+ "epoch": 0.45,
1638
+ "learning_rate": 7.563879907621248e-05,
1639
+ "loss": 3.1884,
1640
+ "step": 1360
1641
+ },
1642
+ {
1643
+ "epoch": 0.45,
1644
+ "learning_rate": 7.541247113163973e-05,
1645
+ "loss": 3.1239,
1646
+ "step": 1365
1647
+ },
1648
+ {
1649
+ "epoch": 0.45,
1650
+ "learning_rate": 7.518614318706697e-05,
1651
+ "loss": 2.6056,
1652
+ "step": 1370
1653
+ },
1654
+ {
1655
+ "epoch": 0.45,
1656
+ "learning_rate": 7.495981524249424e-05,
1657
+ "loss": 3.1442,
1658
+ "step": 1375
1659
+ },
1660
+ {
1661
+ "epoch": 0.46,
1662
+ "learning_rate": 7.473348729792149e-05,
1663
+ "loss": 3.2654,
1664
+ "step": 1380
1665
+ },
1666
+ {
1667
+ "epoch": 0.46,
1668
+ "learning_rate": 7.450715935334874e-05,
1669
+ "loss": 3.1673,
1670
+ "step": 1385
1671
+ },
1672
+ {
1673
+ "epoch": 0.46,
1674
+ "learning_rate": 7.428083140877597e-05,
1675
+ "loss": 2.8148,
1676
+ "step": 1390
1677
+ },
1678
+ {
1679
+ "epoch": 0.46,
1680
+ "learning_rate": 7.405450346420322e-05,
1681
+ "loss": 2.5837,
1682
+ "step": 1395
1683
+ },
1684
+ {
1685
+ "epoch": 0.46,
1686
+ "learning_rate": 7.382817551963049e-05,
1687
+ "loss": 2.2148,
1688
+ "step": 1400
1689
+ },
1690
+ {
1691
+ "epoch": 0.46,
1692
+ "learning_rate": 7.360184757505773e-05,
1693
+ "loss": 3.1805,
1694
+ "step": 1405
1695
+ },
1696
+ {
1697
+ "epoch": 0.47,
1698
+ "learning_rate": 7.337551963048498e-05,
1699
+ "loss": 2.798,
1700
+ "step": 1410
1701
+ },
1702
+ {
1703
+ "epoch": 0.47,
1704
+ "learning_rate": 7.314919168591223e-05,
1705
+ "loss": 3.2318,
1706
+ "step": 1415
1707
+ },
1708
+ {
1709
+ "epoch": 0.47,
1710
+ "learning_rate": 7.29228637413395e-05,
1711
+ "loss": 2.8976,
1712
+ "step": 1420
1713
+ },
1714
+ {
1715
+ "epoch": 0.47,
1716
+ "learning_rate": 7.269653579676675e-05,
1717
+ "loss": 3.0862,
1718
+ "step": 1425
1719
+ },
1720
+ {
1721
+ "epoch": 0.47,
1722
+ "learning_rate": 7.2470207852194e-05,
1723
+ "loss": 2.5277,
1724
+ "step": 1430
1725
+ },
1726
+ {
1727
+ "epoch": 0.47,
1728
+ "learning_rate": 7.224387990762124e-05,
1729
+ "loss": 2.0882,
1730
+ "step": 1435
1731
+ },
1732
+ {
1733
+ "epoch": 0.48,
1734
+ "learning_rate": 7.201755196304851e-05,
1735
+ "loss": 3.0711,
1736
+ "step": 1440
1737
+ },
1738
+ {
1739
+ "epoch": 0.48,
1740
+ "learning_rate": 7.179122401847576e-05,
1741
+ "loss": 2.6206,
1742
+ "step": 1445
1743
+ },
1744
+ {
1745
+ "epoch": 0.48,
1746
+ "learning_rate": 7.156489607390301e-05,
1747
+ "loss": 2.0745,
1748
+ "step": 1450
1749
+ },
1750
+ {
1751
+ "epoch": 0.48,
1752
+ "learning_rate": 7.133856812933026e-05,
1753
+ "loss": 2.5786,
1754
+ "step": 1455
1755
+ },
1756
+ {
1757
+ "epoch": 0.48,
1758
+ "learning_rate": 7.111224018475752e-05,
1759
+ "loss": 2.6242,
1760
+ "step": 1460
1761
+ },
1762
+ {
1763
+ "epoch": 0.48,
1764
+ "learning_rate": 7.088591224018476e-05,
1765
+ "loss": 2.506,
1766
+ "step": 1465
1767
+ },
1768
+ {
1769
+ "epoch": 0.48,
1770
+ "learning_rate": 7.0659584295612e-05,
1771
+ "loss": 2.5967,
1772
+ "step": 1470
1773
+ },
1774
+ {
1775
+ "epoch": 0.49,
1776
+ "learning_rate": 7.043325635103925e-05,
1777
+ "loss": 3.1206,
1778
+ "step": 1475
1779
+ },
1780
+ {
1781
+ "epoch": 0.49,
1782
+ "learning_rate": 7.02069284064665e-05,
1783
+ "loss": 3.1769,
1784
+ "step": 1480
1785
+ },
1786
+ {
1787
+ "epoch": 0.49,
1788
+ "learning_rate": 6.998060046189377e-05,
1789
+ "loss": 2.7589,
1790
+ "step": 1485
1791
+ },
1792
+ {
1793
+ "epoch": 0.49,
1794
+ "learning_rate": 6.975427251732102e-05,
1795
+ "loss": 2.892,
1796
+ "step": 1490
1797
+ },
1798
+ {
1799
+ "epoch": 0.49,
1800
+ "learning_rate": 6.952794457274827e-05,
1801
+ "loss": 2.7783,
1802
+ "step": 1495
1803
+ },
1804
+ {
1805
+ "epoch": 0.49,
1806
+ "learning_rate": 6.930161662817551e-05,
1807
+ "loss": 3.0331,
1808
+ "step": 1500
1809
+ },
1810
+ {
1811
+ "epoch": 0.5,
1812
+ "learning_rate": 6.907528868360278e-05,
1813
+ "loss": 3.1017,
1814
+ "step": 1505
1815
+ },
1816
+ {
1817
+ "epoch": 0.5,
1818
+ "learning_rate": 6.884896073903003e-05,
1819
+ "loss": 2.909,
1820
+ "step": 1510
1821
+ },
1822
+ {
1823
+ "epoch": 0.5,
1824
+ "learning_rate": 6.862263279445728e-05,
1825
+ "loss": 2.5939,
1826
+ "step": 1515
1827
+ },
1828
+ {
1829
+ "epoch": 0.5,
1830
+ "learning_rate": 6.839630484988453e-05,
1831
+ "loss": 3.1691,
1832
+ "step": 1520
1833
+ },
1834
+ {
1835
+ "epoch": 0.5,
1836
+ "learning_rate": 6.816997690531178e-05,
1837
+ "loss": 2.9704,
1838
+ "step": 1525
1839
+ },
1840
+ {
1841
+ "epoch": 0.5,
1842
+ "learning_rate": 6.794364896073903e-05,
1843
+ "loss": 2.3905,
1844
+ "step": 1530
1845
+ },
1846
+ {
1847
+ "epoch": 0.51,
1848
+ "learning_rate": 6.771732101616629e-05,
1849
+ "loss": 2.1867,
1850
+ "step": 1535
1851
+ },
1852
+ {
1853
+ "epoch": 0.51,
1854
+ "learning_rate": 6.749099307159354e-05,
1855
+ "loss": 2.147,
1856
+ "step": 1540
1857
+ },
1858
+ {
1859
+ "epoch": 0.51,
1860
+ "learning_rate": 6.726466512702079e-05,
1861
+ "loss": 2.7461,
1862
+ "step": 1545
1863
+ },
1864
+ {
1865
+ "epoch": 0.51,
1866
+ "learning_rate": 6.703833718244804e-05,
1867
+ "loss": 3.1617,
1868
+ "step": 1550
1869
+ },
1870
+ {
1871
+ "epoch": 0.51,
1872
+ "learning_rate": 6.681200923787529e-05,
1873
+ "loss": 3.1386,
1874
+ "step": 1555
1875
+ },
1876
+ {
1877
+ "epoch": 0.51,
1878
+ "learning_rate": 6.658568129330254e-05,
1879
+ "loss": 2.7372,
1880
+ "step": 1560
1881
+ },
1882
+ {
1883
+ "epoch": 0.52,
1884
+ "learning_rate": 6.635935334872979e-05,
1885
+ "loss": 3.1669,
1886
+ "step": 1565
1887
+ },
1888
+ {
1889
+ "epoch": 0.52,
1890
+ "learning_rate": 6.613302540415705e-05,
1891
+ "loss": 2.5412,
1892
+ "step": 1570
1893
+ },
1894
+ {
1895
+ "epoch": 0.52,
1896
+ "learning_rate": 6.59066974595843e-05,
1897
+ "loss": 2.7079,
1898
+ "step": 1575
1899
+ },
1900
+ {
1901
+ "epoch": 0.52,
1902
+ "learning_rate": 6.568036951501155e-05,
1903
+ "loss": 2.5812,
1904
+ "step": 1580
1905
+ },
1906
+ {
1907
+ "epoch": 0.52,
1908
+ "learning_rate": 6.54540415704388e-05,
1909
+ "loss": 2.575,
1910
+ "step": 1585
1911
+ },
1912
+ {
1913
+ "epoch": 0.52,
1914
+ "learning_rate": 6.522771362586606e-05,
1915
+ "loss": 3.1151,
1916
+ "step": 1590
1917
+ },
1918
+ {
1919
+ "epoch": 0.53,
1920
+ "learning_rate": 6.50013856812933e-05,
1921
+ "loss": 3.0427,
1922
+ "step": 1595
1923
+ },
1924
+ {
1925
+ "epoch": 0.53,
1926
+ "learning_rate": 6.477505773672056e-05,
1927
+ "loss": 3.0496,
1928
+ "step": 1600
1929
+ },
1930
+ {
1931
+ "epoch": 0.53,
1932
+ "learning_rate": 6.454872979214781e-05,
1933
+ "loss": 3.1504,
1934
+ "step": 1605
1935
+ },
1936
+ {
1937
+ "epoch": 0.53,
1938
+ "learning_rate": 6.432240184757506e-05,
1939
+ "loss": 2.7546,
1940
+ "step": 1610
1941
+ },
1942
+ {
1943
+ "epoch": 0.53,
1944
+ "learning_rate": 6.40960739030023e-05,
1945
+ "loss": 2.5357,
1946
+ "step": 1615
1947
+ },
1948
+ {
1949
+ "epoch": 0.53,
1950
+ "learning_rate": 6.386974595842957e-05,
1951
+ "loss": 2.3116,
1952
+ "step": 1620
1953
+ },
1954
+ {
1955
+ "epoch": 0.54,
1956
+ "learning_rate": 6.364341801385682e-05,
1957
+ "loss": 2.7059,
1958
+ "step": 1625
1959
+ },
1960
+ {
1961
+ "epoch": 0.54,
1962
+ "learning_rate": 6.341709006928407e-05,
1963
+ "loss": 2.5225,
1964
+ "step": 1630
1965
+ },
1966
+ {
1967
+ "epoch": 0.54,
1968
+ "learning_rate": 6.319076212471132e-05,
1969
+ "loss": 2.5516,
1970
+ "step": 1635
1971
+ },
1972
+ {
1973
+ "epoch": 0.54,
1974
+ "learning_rate": 6.296443418013857e-05,
1975
+ "loss": 2.677,
1976
+ "step": 1640
1977
+ },
1978
+ {
1979
+ "epoch": 0.54,
1980
+ "learning_rate": 6.273810623556582e-05,
1981
+ "loss": 1.9059,
1982
+ "step": 1645
1983
+ },
1984
+ {
1985
+ "epoch": 0.54,
1986
+ "learning_rate": 6.251177829099307e-05,
1987
+ "loss": 1.3935,
1988
+ "step": 1650
1989
+ },
1990
+ {
1991
+ "epoch": 0.55,
1992
+ "learning_rate": 6.228545034642033e-05,
1993
+ "loss": 2.6359,
1994
+ "step": 1655
1995
+ },
1996
+ {
1997
+ "epoch": 0.55,
1998
+ "learning_rate": 6.205912240184758e-05,
1999
+ "loss": 2.7623,
2000
+ "step": 1660
2001
+ },
2002
+ {
2003
+ "epoch": 0.55,
2004
+ "learning_rate": 6.183279445727483e-05,
2005
+ "loss": 3.0703,
2006
+ "step": 1665
2007
+ },
2008
+ {
2009
+ "epoch": 0.55,
2010
+ "learning_rate": 6.160646651270208e-05,
2011
+ "loss": 2.7344,
2012
+ "step": 1670
2013
+ },
2014
+ {
2015
+ "epoch": 0.55,
2016
+ "learning_rate": 6.138013856812933e-05,
2017
+ "loss": 3.0174,
2018
+ "step": 1675
2019
+ },
2020
+ {
2021
+ "epoch": 0.55,
2022
+ "learning_rate": 6.115381062355658e-05,
2023
+ "loss": 2.5429,
2024
+ "step": 1680
2025
+ },
2026
+ {
2027
+ "epoch": 0.56,
2028
+ "learning_rate": 6.092748267898383e-05,
2029
+ "loss": 3.092,
2030
+ "step": 1685
2031
+ },
2032
+ {
2033
+ "epoch": 0.56,
2034
+ "learning_rate": 6.070115473441109e-05,
2035
+ "loss": 2.0041,
2036
+ "step": 1690
2037
+ },
2038
+ {
2039
+ "epoch": 0.56,
2040
+ "learning_rate": 6.047482678983834e-05,
2041
+ "loss": 1.9942,
2042
+ "step": 1695
2043
+ },
2044
+ {
2045
+ "epoch": 0.56,
2046
+ "learning_rate": 6.0248498845265595e-05,
2047
+ "loss": 3.0518,
2048
+ "step": 1700
2049
+ },
2050
+ {
2051
+ "epoch": 0.56,
2052
+ "learning_rate": 6.0022170900692844e-05,
2053
+ "loss": 3.1302,
2054
+ "step": 1705
2055
+ },
2056
+ {
2057
+ "epoch": 0.56,
2058
+ "learning_rate": 5.979584295612009e-05,
2059
+ "loss": 2.6788,
2060
+ "step": 1710
2061
+ },
2062
+ {
2063
+ "epoch": 0.57,
2064
+ "learning_rate": 5.956951501154734e-05,
2065
+ "loss": 2.6504,
2066
+ "step": 1715
2067
+ },
2068
+ {
2069
+ "epoch": 0.57,
2070
+ "learning_rate": 5.934318706697459e-05,
2071
+ "loss": 3.1506,
2072
+ "step": 1720
2073
+ },
2074
+ {
2075
+ "epoch": 0.57,
2076
+ "learning_rate": 5.911685912240185e-05,
2077
+ "loss": 2.5277,
2078
+ "step": 1725
2079
+ },
2080
+ {
2081
+ "epoch": 0.57,
2082
+ "learning_rate": 5.88905311778291e-05,
2083
+ "loss": 2.7044,
2084
+ "step": 1730
2085
+ },
2086
+ {
2087
+ "epoch": 0.57,
2088
+ "learning_rate": 5.8664203233256355e-05,
2089
+ "loss": 3.0895,
2090
+ "step": 1735
2091
+ },
2092
+ {
2093
+ "epoch": 0.57,
2094
+ "learning_rate": 5.8437875288683604e-05,
2095
+ "loss": 2.7463,
2096
+ "step": 1740
2097
+ },
2098
+ {
2099
+ "epoch": 0.58,
2100
+ "learning_rate": 5.8211547344110853e-05,
2101
+ "loss": 2.5867,
2102
+ "step": 1745
2103
+ },
2104
+ {
2105
+ "epoch": 0.58,
2106
+ "learning_rate": 5.79852193995381e-05,
2107
+ "loss": 2.5577,
2108
+ "step": 1750
2109
+ },
2110
+ {
2111
+ "epoch": 0.58,
2112
+ "learning_rate": 5.775889145496536e-05,
2113
+ "loss": 2.0084,
2114
+ "step": 1755
2115
+ },
2116
+ {
2117
+ "epoch": 0.58,
2118
+ "learning_rate": 5.753256351039261e-05,
2119
+ "loss": 2.656,
2120
+ "step": 1760
2121
+ },
2122
+ {
2123
+ "epoch": 0.58,
2124
+ "learning_rate": 5.7306235565819865e-05,
2125
+ "loss": 3.0814,
2126
+ "step": 1765
2127
+ },
2128
+ {
2129
+ "epoch": 0.58,
2130
+ "learning_rate": 5.7079907621247114e-05,
2131
+ "loss": 3.0512,
2132
+ "step": 1770
2133
+ },
2134
+ {
2135
+ "epoch": 0.59,
2136
+ "learning_rate": 5.685357967667437e-05,
2137
+ "loss": 2.4837,
2138
+ "step": 1775
2139
+ },
2140
+ {
2141
+ "epoch": 0.59,
2142
+ "learning_rate": 5.662725173210162e-05,
2143
+ "loss": 3.0879,
2144
+ "step": 1780
2145
+ },
2146
+ {
2147
+ "epoch": 0.59,
2148
+ "learning_rate": 5.640092378752886e-05,
2149
+ "loss": 3.0925,
2150
+ "step": 1785
2151
+ },
2152
+ {
2153
+ "epoch": 0.59,
2154
+ "learning_rate": 5.617459584295612e-05,
2155
+ "loss": 2.9465,
2156
+ "step": 1790
2157
+ },
2158
+ {
2159
+ "epoch": 0.59,
2160
+ "learning_rate": 5.594826789838337e-05,
2161
+ "loss": 2.1648,
2162
+ "step": 1795
2163
+ },
2164
+ {
2165
+ "epoch": 0.59,
2166
+ "learning_rate": 5.5721939953810625e-05,
2167
+ "loss": 2.4771,
2168
+ "step": 1800
2169
+ },
2170
+ {
2171
+ "epoch": 0.6,
2172
+ "learning_rate": 5.5495612009237874e-05,
2173
+ "loss": 3.0954,
2174
+ "step": 1805
2175
+ },
2176
+ {
2177
+ "epoch": 0.6,
2178
+ "learning_rate": 5.526928406466513e-05,
2179
+ "loss": 2.481,
2180
+ "step": 1810
2181
+ },
2182
+ {
2183
+ "epoch": 0.6,
2184
+ "learning_rate": 5.504295612009238e-05,
2185
+ "loss": 3.0379,
2186
+ "step": 1815
2187
+ },
2188
+ {
2189
+ "epoch": 0.6,
2190
+ "learning_rate": 5.481662817551963e-05,
2191
+ "loss": 2.5866,
2192
+ "step": 1820
2193
+ },
2194
+ {
2195
+ "epoch": 0.6,
2196
+ "learning_rate": 5.459030023094688e-05,
2197
+ "loss": 2.4673,
2198
+ "step": 1825
2199
+ },
2200
+ {
2201
+ "epoch": 0.6,
2202
+ "learning_rate": 5.4363972286374135e-05,
2203
+ "loss": 2.0428,
2204
+ "step": 1830
2205
+ },
2206
+ {
2207
+ "epoch": 0.61,
2208
+ "learning_rate": 5.4137644341801384e-05,
2209
+ "loss": 2.9469,
2210
+ "step": 1835
2211
+ },
2212
+ {
2213
+ "epoch": 0.61,
2214
+ "learning_rate": 5.391131639722864e-05,
2215
+ "loss": 3.065,
2216
+ "step": 1840
2217
+ },
2218
+ {
2219
+ "epoch": 0.61,
2220
+ "learning_rate": 5.368498845265589e-05,
2221
+ "loss": 2.3777,
2222
+ "step": 1845
2223
+ },
2224
+ {
2225
+ "epoch": 0.61,
2226
+ "learning_rate": 5.3458660508083146e-05,
2227
+ "loss": 2.4645,
2228
+ "step": 1850
2229
+ },
2230
+ {
2231
+ "epoch": 0.61,
2232
+ "learning_rate": 5.3232332563510396e-05,
2233
+ "loss": 2.9935,
2234
+ "step": 1855
2235
+ },
2236
+ {
2237
+ "epoch": 0.61,
2238
+ "learning_rate": 5.300600461893764e-05,
2239
+ "loss": 3.0377,
2240
+ "step": 1860
2241
+ },
2242
+ {
2243
+ "epoch": 0.62,
2244
+ "learning_rate": 5.2779676674364895e-05,
2245
+ "loss": 2.2265,
2246
+ "step": 1865
2247
+ },
2248
+ {
2249
+ "epoch": 0.62,
2250
+ "learning_rate": 5.2553348729792144e-05,
2251
+ "loss": 2.0621,
2252
+ "step": 1870
2253
+ },
2254
+ {
2255
+ "epoch": 0.62,
2256
+ "learning_rate": 5.23270207852194e-05,
2257
+ "loss": 2.285,
2258
+ "step": 1875
2259
+ },
2260
+ {
2261
+ "epoch": 0.62,
2262
+ "learning_rate": 5.210069284064665e-05,
2263
+ "loss": 2.605,
2264
+ "step": 1880
2265
+ },
2266
+ {
2267
+ "epoch": 0.62,
2268
+ "learning_rate": 5.1874364896073906e-05,
2269
+ "loss": 2.4744,
2270
+ "step": 1885
2271
+ },
2272
+ {
2273
+ "epoch": 0.62,
2274
+ "learning_rate": 5.1648036951501155e-05,
2275
+ "loss": 2.3932,
2276
+ "step": 1890
2277
+ },
2278
+ {
2279
+ "epoch": 0.63,
2280
+ "learning_rate": 5.1421709006928405e-05,
2281
+ "loss": 3.0259,
2282
+ "step": 1895
2283
+ },
2284
+ {
2285
+ "epoch": 0.63,
2286
+ "learning_rate": 5.1195381062355654e-05,
2287
+ "loss": 2.9768,
2288
+ "step": 1900
2289
+ },
2290
+ {
2291
+ "epoch": 0.63,
2292
+ "learning_rate": 5.096905311778291e-05,
2293
+ "loss": 2.4951,
2294
+ "step": 1905
2295
+ },
2296
+ {
2297
+ "epoch": 0.63,
2298
+ "learning_rate": 5.074272517321016e-05,
2299
+ "loss": 2.9348,
2300
+ "step": 1910
2301
+ },
2302
+ {
2303
+ "epoch": 0.63,
2304
+ "learning_rate": 5.0516397228637416e-05,
2305
+ "loss": 3.0326,
2306
+ "step": 1915
2307
+ },
2308
+ {
2309
+ "epoch": 0.63,
2310
+ "learning_rate": 5.0290069284064666e-05,
2311
+ "loss": 2.544,
2312
+ "step": 1920
2313
+ },
2314
+ {
2315
+ "epoch": 0.64,
2316
+ "learning_rate": 5.006374133949192e-05,
2317
+ "loss": 2.9794,
2318
+ "step": 1925
2319
+ },
2320
+ {
2321
+ "epoch": 0.64,
2322
+ "learning_rate": 4.983741339491917e-05,
2323
+ "loss": 2.2389,
2324
+ "step": 1930
2325
+ },
2326
+ {
2327
+ "epoch": 0.64,
2328
+ "learning_rate": 4.961108545034642e-05,
2329
+ "loss": 2.5411,
2330
+ "step": 1935
2331
+ },
2332
+ {
2333
+ "epoch": 0.64,
2334
+ "learning_rate": 4.938475750577367e-05,
2335
+ "loss": 2.1246,
2336
+ "step": 1940
2337
+ },
2338
+ {
2339
+ "epoch": 0.64,
2340
+ "learning_rate": 4.915842956120092e-05,
2341
+ "loss": 2.549,
2342
+ "step": 1945
2343
+ },
2344
+ {
2345
+ "epoch": 0.64,
2346
+ "learning_rate": 4.8932101616628176e-05,
2347
+ "loss": 2.9752,
2348
+ "step": 1950
2349
+ },
2350
+ {
2351
+ "epoch": 0.65,
2352
+ "learning_rate": 4.8705773672055425e-05,
2353
+ "loss": 2.2107,
2354
+ "step": 1955
2355
+ },
2356
+ {
2357
+ "epoch": 0.65,
2358
+ "learning_rate": 4.847944572748268e-05,
2359
+ "loss": 2.9383,
2360
+ "step": 1960
2361
+ },
2362
+ {
2363
+ "epoch": 0.65,
2364
+ "learning_rate": 4.825311778290993e-05,
2365
+ "loss": 2.4591,
2366
+ "step": 1965
2367
+ },
2368
+ {
2369
+ "epoch": 0.65,
2370
+ "learning_rate": 4.802678983833718e-05,
2371
+ "loss": 2.5564,
2372
+ "step": 1970
2373
+ },
2374
+ {
2375
+ "epoch": 0.65,
2376
+ "learning_rate": 4.780046189376443e-05,
2377
+ "loss": 2.6209,
2378
+ "step": 1975
2379
+ },
2380
+ {
2381
+ "epoch": 0.65,
2382
+ "learning_rate": 4.7574133949191686e-05,
2383
+ "loss": 2.4038,
2384
+ "step": 1980
2385
+ },
2386
+ {
2387
+ "epoch": 0.65,
2388
+ "learning_rate": 4.7347806004618936e-05,
2389
+ "loss": 2.5304,
2390
+ "step": 1985
2391
+ },
2392
+ {
2393
+ "epoch": 0.66,
2394
+ "learning_rate": 4.712147806004619e-05,
2395
+ "loss": 2.6632,
2396
+ "step": 1990
2397
+ },
2398
+ {
2399
+ "epoch": 0.66,
2400
+ "learning_rate": 4.689515011547344e-05,
2401
+ "loss": 1.937,
2402
+ "step": 1995
2403
+ },
2404
+ {
2405
+ "epoch": 0.66,
2406
+ "learning_rate": 4.66688221709007e-05,
2407
+ "loss": 2.4658,
2408
+ "step": 2000
2409
+ },
2410
+ {
2411
+ "epoch": 0.66,
2412
+ "learning_rate": 4.644249422632795e-05,
2413
+ "loss": 2.5939,
2414
+ "step": 2005
2415
+ },
2416
+ {
2417
+ "epoch": 0.66,
2418
+ "learning_rate": 4.6216166281755197e-05,
2419
+ "loss": 2.618,
2420
+ "step": 2010
2421
+ },
2422
+ {
2423
+ "epoch": 0.66,
2424
+ "learning_rate": 4.5989838337182446e-05,
2425
+ "loss": 3.0384,
2426
+ "step": 2015
2427
+ },
2428
+ {
2429
+ "epoch": 0.67,
2430
+ "learning_rate": 4.57635103926097e-05,
2431
+ "loss": 2.9713,
2432
+ "step": 2020
2433
+ },
2434
+ {
2435
+ "epoch": 0.67,
2436
+ "learning_rate": 4.553718244803695e-05,
2437
+ "loss": 3.0171,
2438
+ "step": 2025
2439
+ },
2440
+ {
2441
+ "epoch": 0.67,
2442
+ "learning_rate": 4.53108545034642e-05,
2443
+ "loss": 3.0126,
2444
+ "step": 2030
2445
+ },
2446
+ {
2447
+ "epoch": 0.67,
2448
+ "learning_rate": 4.508452655889146e-05,
2449
+ "loss": 2.9481,
2450
+ "step": 2035
2451
+ },
2452
+ {
2453
+ "epoch": 0.67,
2454
+ "learning_rate": 4.485819861431871e-05,
2455
+ "loss": 2.4466,
2456
+ "step": 2040
2457
+ },
2458
+ {
2459
+ "epoch": 0.67,
2460
+ "learning_rate": 4.4631870669745956e-05,
2461
+ "loss": 2.0695,
2462
+ "step": 2045
2463
+ },
2464
+ {
2465
+ "epoch": 0.68,
2466
+ "learning_rate": 4.4405542725173206e-05,
2467
+ "loss": 3.0076,
2468
+ "step": 2050
2469
+ },
2470
+ {
2471
+ "epoch": 0.68,
2472
+ "learning_rate": 4.417921478060046e-05,
2473
+ "loss": 3.0097,
2474
+ "step": 2055
2475
+ },
2476
+ {
2477
+ "epoch": 0.68,
2478
+ "learning_rate": 4.395288683602771e-05,
2479
+ "loss": 2.4857,
2480
+ "step": 2060
2481
+ },
2482
+ {
2483
+ "epoch": 0.68,
2484
+ "learning_rate": 4.372655889145497e-05,
2485
+ "loss": 3.0361,
2486
+ "step": 2065
2487
+ },
2488
+ {
2489
+ "epoch": 0.68,
2490
+ "learning_rate": 4.350023094688222e-05,
2491
+ "loss": 1.9166,
2492
+ "step": 2070
2493
+ },
2494
+ {
2495
+ "epoch": 0.68,
2496
+ "learning_rate": 4.327390300230947e-05,
2497
+ "loss": 2.9817,
2498
+ "step": 2075
2499
+ },
2500
+ {
2501
+ "epoch": 0.69,
2502
+ "learning_rate": 4.304757505773672e-05,
2503
+ "loss": 2.9798,
2504
+ "step": 2080
2505
+ },
2506
+ {
2507
+ "epoch": 0.69,
2508
+ "learning_rate": 4.282124711316397e-05,
2509
+ "loss": 2.4675,
2510
+ "step": 2085
2511
+ },
2512
+ {
2513
+ "epoch": 0.69,
2514
+ "learning_rate": 4.259491916859122e-05,
2515
+ "loss": 3.0161,
2516
+ "step": 2090
2517
+ },
2518
+ {
2519
+ "epoch": 0.69,
2520
+ "learning_rate": 4.236859122401848e-05,
2521
+ "loss": 1.9453,
2522
+ "step": 2095
2523
+ },
2524
+ {
2525
+ "epoch": 0.69,
2526
+ "learning_rate": 4.214226327944573e-05,
2527
+ "loss": 3.0623,
2528
+ "step": 2100
2529
+ },
2530
+ {
2531
+ "epoch": 0.69,
2532
+ "learning_rate": 4.1915935334872984e-05,
2533
+ "loss": 3.0624,
2534
+ "step": 2105
2535
+ },
2536
+ {
2537
+ "epoch": 0.7,
2538
+ "learning_rate": 4.168960739030023e-05,
2539
+ "loss": 2.4797,
2540
+ "step": 2110
2541
+ },
2542
+ {
2543
+ "epoch": 0.7,
2544
+ "learning_rate": 4.146327944572748e-05,
2545
+ "loss": 2.0303,
2546
+ "step": 2115
2547
+ },
2548
+ {
2549
+ "epoch": 0.7,
2550
+ "learning_rate": 4.123695150115473e-05,
2551
+ "loss": 2.9842,
2552
+ "step": 2120
2553
+ },
2554
+ {
2555
+ "epoch": 0.7,
2556
+ "learning_rate": 4.101062355658198e-05,
2557
+ "loss": 2.5711,
2558
+ "step": 2125
2559
+ },
2560
+ {
2561
+ "epoch": 0.7,
2562
+ "learning_rate": 4.078429561200924e-05,
2563
+ "loss": 2.409,
2564
+ "step": 2130
2565
+ },
2566
+ {
2567
+ "epoch": 0.7,
2568
+ "learning_rate": 4.055796766743649e-05,
2569
+ "loss": 2.473,
2570
+ "step": 2135
2571
+ },
2572
+ {
2573
+ "epoch": 0.71,
2574
+ "learning_rate": 4.033163972286374e-05,
2575
+ "loss": 1.9119,
2576
+ "step": 2140
2577
+ },
2578
+ {
2579
+ "epoch": 0.71,
2580
+ "learning_rate": 4.010531177829099e-05,
2581
+ "loss": 2.9008,
2582
+ "step": 2145
2583
+ },
2584
+ {
2585
+ "epoch": 0.71,
2586
+ "learning_rate": 3.987898383371825e-05,
2587
+ "loss": 3.032,
2588
+ "step": 2150
2589
+ },
2590
+ {
2591
+ "epoch": 0.71,
2592
+ "learning_rate": 3.96526558891455e-05,
2593
+ "loss": 2.3587,
2594
+ "step": 2155
2595
+ },
2596
+ {
2597
+ "epoch": 0.71,
2598
+ "learning_rate": 3.942632794457275e-05,
2599
+ "loss": 2.9925,
2600
+ "step": 2160
2601
+ },
2602
+ {
2603
+ "epoch": 0.71,
2604
+ "learning_rate": 3.92e-05,
2605
+ "loss": 2.9025,
2606
+ "step": 2165
2607
+ },
2608
+ {
2609
+ "epoch": 0.72,
2610
+ "learning_rate": 3.8973672055427254e-05,
2611
+ "loss": 3.0899,
2612
+ "step": 2170
2613
+ },
2614
+ {
2615
+ "epoch": 0.72,
2616
+ "learning_rate": 3.87473441108545e-05,
2617
+ "loss": 2.9319,
2618
+ "step": 2175
2619
+ },
2620
+ {
2621
+ "epoch": 0.72,
2622
+ "learning_rate": 3.852101616628176e-05,
2623
+ "loss": 2.453,
2624
+ "step": 2180
2625
+ },
2626
+ {
2627
+ "epoch": 0.72,
2628
+ "learning_rate": 3.829468822170901e-05,
2629
+ "loss": 2.0663,
2630
+ "step": 2185
2631
+ },
2632
+ {
2633
+ "epoch": 0.72,
2634
+ "learning_rate": 3.8068360277136265e-05,
2635
+ "loss": 2.4464,
2636
+ "step": 2190
2637
+ },
2638
+ {
2639
+ "epoch": 0.72,
2640
+ "learning_rate": 3.784203233256351e-05,
2641
+ "loss": 2.5029,
2642
+ "step": 2195
2643
+ },
2644
+ {
2645
+ "epoch": 0.73,
2646
+ "learning_rate": 3.761570438799076e-05,
2647
+ "loss": 2.4315,
2648
+ "step": 2200
2649
+ },
2650
+ {
2651
+ "epoch": 0.73,
2652
+ "learning_rate": 3.738937644341801e-05,
2653
+ "loss": 2.9723,
2654
+ "step": 2205
2655
+ },
2656
+ {
2657
+ "epoch": 0.73,
2658
+ "learning_rate": 3.716304849884526e-05,
2659
+ "loss": 2.9545,
2660
+ "step": 2210
2661
+ },
2662
+ {
2663
+ "epoch": 0.73,
2664
+ "learning_rate": 3.693672055427252e-05,
2665
+ "loss": 2.4394,
2666
+ "step": 2215
2667
+ },
2668
+ {
2669
+ "epoch": 0.73,
2670
+ "learning_rate": 3.671039260969977e-05,
2671
+ "loss": 2.8852,
2672
+ "step": 2220
2673
+ },
2674
+ {
2675
+ "epoch": 0.73,
2676
+ "learning_rate": 3.6484064665127025e-05,
2677
+ "loss": 2.5524,
2678
+ "step": 2225
2679
+ },
2680
+ {
2681
+ "epoch": 0.74,
2682
+ "learning_rate": 3.6257736720554274e-05,
2683
+ "loss": 2.5424,
2684
+ "step": 2230
2685
+ },
2686
+ {
2687
+ "epoch": 0.74,
2688
+ "learning_rate": 3.6031408775981524e-05,
2689
+ "loss": 2.4416,
2690
+ "step": 2235
2691
+ },
2692
+ {
2693
+ "epoch": 0.74,
2694
+ "learning_rate": 3.580508083140877e-05,
2695
+ "loss": 2.3582,
2696
+ "step": 2240
2697
+ },
2698
+ {
2699
+ "epoch": 0.74,
2700
+ "learning_rate": 3.557875288683603e-05,
2701
+ "loss": 2.9595,
2702
+ "step": 2245
2703
+ },
2704
+ {
2705
+ "epoch": 0.74,
2706
+ "learning_rate": 3.535242494226328e-05,
2707
+ "loss": 3.0225,
2708
+ "step": 2250
2709
+ },
2710
+ {
2711
+ "epoch": 0.74,
2712
+ "learning_rate": 3.5126096997690535e-05,
2713
+ "loss": 2.0518,
2714
+ "step": 2255
2715
+ },
2716
+ {
2717
+ "epoch": 0.75,
2718
+ "learning_rate": 3.4899769053117784e-05,
2719
+ "loss": 2.9843,
2720
+ "step": 2260
2721
+ },
2722
+ {
2723
+ "epoch": 0.75,
2724
+ "learning_rate": 3.467344110854504e-05,
2725
+ "loss": 2.3971,
2726
+ "step": 2265
2727
+ },
2728
+ {
2729
+ "epoch": 0.75,
2730
+ "learning_rate": 3.444711316397228e-05,
2731
+ "loss": 2.9604,
2732
+ "step": 2270
2733
+ },
2734
+ {
2735
+ "epoch": 0.75,
2736
+ "learning_rate": 3.422078521939954e-05,
2737
+ "loss": 1.526,
2738
+ "step": 2275
2739
+ },
2740
+ {
2741
+ "epoch": 0.75,
2742
+ "learning_rate": 3.399445727482679e-05,
2743
+ "loss": 2.5448,
2744
+ "step": 2280
2745
+ },
2746
+ {
2747
+ "epoch": 0.75,
2748
+ "learning_rate": 3.376812933025404e-05,
2749
+ "loss": 2.5325,
2750
+ "step": 2285
2751
+ },
2752
+ {
2753
+ "epoch": 0.76,
2754
+ "learning_rate": 3.3541801385681295e-05,
2755
+ "loss": 2.0032,
2756
+ "step": 2290
2757
+ },
2758
+ {
2759
+ "epoch": 0.76,
2760
+ "learning_rate": 3.3315473441108544e-05,
2761
+ "loss": 2.447,
2762
+ "step": 2295
2763
+ },
2764
+ {
2765
+ "epoch": 0.76,
2766
+ "learning_rate": 3.3089145496535794e-05,
2767
+ "loss": 2.9568,
2768
+ "step": 2300
2769
+ },
2770
+ {
2771
+ "epoch": 0.76,
2772
+ "learning_rate": 3.286281755196305e-05,
2773
+ "loss": 2.5495,
2774
+ "step": 2305
2775
+ },
2776
+ {
2777
+ "epoch": 0.76,
2778
+ "learning_rate": 3.26364896073903e-05,
2779
+ "loss": 2.4766,
2780
+ "step": 2310
2781
+ },
2782
+ {
2783
+ "epoch": 0.76,
2784
+ "learning_rate": 3.2410161662817556e-05,
2785
+ "loss": 2.9513,
2786
+ "step": 2315
2787
+ },
2788
+ {
2789
+ "epoch": 0.77,
2790
+ "learning_rate": 3.2183833718244805e-05,
2791
+ "loss": 1.9875,
2792
+ "step": 2320
2793
+ },
2794
+ {
2795
+ "epoch": 0.77,
2796
+ "learning_rate": 3.1957505773672054e-05,
2797
+ "loss": 2.0134,
2798
+ "step": 2325
2799
+ },
2800
+ {
2801
+ "epoch": 0.77,
2802
+ "learning_rate": 3.173117782909931e-05,
2803
+ "loss": 2.5515,
2804
+ "step": 2330
2805
+ },
2806
+ {
2807
+ "epoch": 0.77,
2808
+ "learning_rate": 3.150484988452656e-05,
2809
+ "loss": 2.9005,
2810
+ "step": 2335
2811
+ },
2812
+ {
2813
+ "epoch": 0.77,
2814
+ "learning_rate": 3.127852193995381e-05,
2815
+ "loss": 2.8825,
2816
+ "step": 2340
2817
+ },
2818
+ {
2819
+ "epoch": 0.77,
2820
+ "learning_rate": 3.1052193995381066e-05,
2821
+ "loss": 1.8196,
2822
+ "step": 2345
2823
+ },
2824
+ {
2825
+ "epoch": 0.78,
2826
+ "learning_rate": 3.0825866050808315e-05,
2827
+ "loss": 2.8672,
2828
+ "step": 2350
2829
+ },
2830
+ {
2831
+ "epoch": 0.78,
2832
+ "learning_rate": 3.0599538106235565e-05,
2833
+ "loss": 2.5313,
2834
+ "step": 2355
2835
+ },
2836
+ {
2837
+ "epoch": 0.78,
2838
+ "learning_rate": 3.0373210161662818e-05,
2839
+ "loss": 2.4912,
2840
+ "step": 2360
2841
+ },
2842
+ {
2843
+ "epoch": 0.78,
2844
+ "learning_rate": 3.014688221709007e-05,
2845
+ "loss": 2.851,
2846
+ "step": 2365
2847
+ },
2848
+ {
2849
+ "epoch": 0.78,
2850
+ "learning_rate": 2.9920554272517323e-05,
2851
+ "loss": 2.9382,
2852
+ "step": 2370
2853
+ },
2854
+ {
2855
+ "epoch": 0.78,
2856
+ "learning_rate": 2.9694226327944573e-05,
2857
+ "loss": 2.8546,
2858
+ "step": 2375
2859
+ },
2860
+ {
2861
+ "epoch": 0.79,
2862
+ "learning_rate": 2.9467898383371826e-05,
2863
+ "loss": 2.9912,
2864
+ "step": 2380
2865
+ },
2866
+ {
2867
+ "epoch": 0.79,
2868
+ "learning_rate": 2.924157043879908e-05,
2869
+ "loss": 2.4831,
2870
+ "step": 2385
2871
+ },
2872
+ {
2873
+ "epoch": 0.79,
2874
+ "learning_rate": 2.901524249422633e-05,
2875
+ "loss": 2.96,
2876
+ "step": 2390
2877
+ },
2878
+ {
2879
+ "epoch": 0.79,
2880
+ "learning_rate": 2.8788914549653577e-05,
2881
+ "loss": 2.916,
2882
+ "step": 2395
2883
+ },
2884
+ {
2885
+ "epoch": 0.79,
2886
+ "learning_rate": 2.856258660508083e-05,
2887
+ "loss": 2.9423,
2888
+ "step": 2400
2889
+ },
2890
+ {
2891
+ "epoch": 0.79,
2892
+ "learning_rate": 2.8336258660508083e-05,
2893
+ "loss": 2.3475,
2894
+ "step": 2405
2895
+ },
2896
+ {
2897
+ "epoch": 0.8,
2898
+ "learning_rate": 2.8109930715935336e-05,
2899
+ "loss": 2.4757,
2900
+ "step": 2410
2901
+ },
2902
+ {
2903
+ "epoch": 0.8,
2904
+ "learning_rate": 2.7883602771362585e-05,
2905
+ "loss": 2.497,
2906
+ "step": 2415
2907
+ },
2908
+ {
2909
+ "epoch": 0.8,
2910
+ "learning_rate": 2.7657274826789838e-05,
2911
+ "loss": 2.8617,
2912
+ "step": 2420
2913
+ },
2914
+ {
2915
+ "epoch": 0.8,
2916
+ "learning_rate": 2.743094688221709e-05,
2917
+ "loss": 2.5424,
2918
+ "step": 2425
2919
+ },
2920
+ {
2921
+ "epoch": 0.8,
2922
+ "learning_rate": 2.720461893764434e-05,
2923
+ "loss": 1.9343,
2924
+ "step": 2430
2925
+ },
2926
+ {
2927
+ "epoch": 0.8,
2928
+ "learning_rate": 2.6978290993071593e-05,
2929
+ "loss": 2.3538,
2930
+ "step": 2435
2931
+ },
2932
+ {
2933
+ "epoch": 0.81,
2934
+ "learning_rate": 2.6751963048498846e-05,
2935
+ "loss": 2.457,
2936
+ "step": 2440
2937
+ },
2938
+ {
2939
+ "epoch": 0.81,
2940
+ "learning_rate": 2.65256351039261e-05,
2941
+ "loss": 1.8267,
2942
+ "step": 2445
2943
+ },
2944
+ {
2945
+ "epoch": 0.81,
2946
+ "learning_rate": 2.629930715935335e-05,
2947
+ "loss": 2.8956,
2948
+ "step": 2450
2949
+ },
2950
+ {
2951
+ "epoch": 0.81,
2952
+ "learning_rate": 2.60729792147806e-05,
2953
+ "loss": 2.4804,
2954
+ "step": 2455
2955
+ },
2956
+ {
2957
+ "epoch": 0.81,
2958
+ "learning_rate": 2.5846651270207854e-05,
2959
+ "loss": 0.9658,
2960
+ "step": 2460
2961
+ },
2962
+ {
2963
+ "epoch": 0.81,
2964
+ "learning_rate": 2.5620323325635107e-05,
2965
+ "loss": 2.5179,
2966
+ "step": 2465
2967
+ },
2968
+ {
2969
+ "epoch": 0.81,
2970
+ "learning_rate": 2.5393995381062353e-05,
2971
+ "loss": 2.0064,
2972
+ "step": 2470
2973
+ },
2974
+ {
2975
+ "epoch": 0.82,
2976
+ "learning_rate": 2.5167667436489606e-05,
2977
+ "loss": 2.9194,
2978
+ "step": 2475
2979
+ },
2980
+ {
2981
+ "epoch": 0.82,
2982
+ "learning_rate": 2.494133949191686e-05,
2983
+ "loss": 2.9696,
2984
+ "step": 2480
2985
+ },
2986
+ {
2987
+ "epoch": 0.82,
2988
+ "learning_rate": 2.471501154734411e-05,
2989
+ "loss": 2.3919,
2990
+ "step": 2485
2991
+ },
2992
+ {
2993
+ "epoch": 0.82,
2994
+ "learning_rate": 2.448868360277136e-05,
2995
+ "loss": 2.9989,
2996
+ "step": 2490
2997
+ },
2998
+ {
2999
+ "epoch": 0.82,
3000
+ "learning_rate": 2.4262355658198614e-05,
3001
+ "loss": 1.9785,
3002
+ "step": 2495
3003
+ },
3004
+ {
3005
+ "epoch": 0.82,
3006
+ "learning_rate": 2.4036027713625867e-05,
3007
+ "loss": 2.4736,
3008
+ "step": 2500
3009
+ },
3010
+ {
3011
+ "epoch": 0.83,
3012
+ "learning_rate": 2.3809699769053116e-05,
3013
+ "loss": 2.9508,
3014
+ "step": 2505
3015
+ },
3016
+ {
3017
+ "epoch": 0.83,
3018
+ "learning_rate": 2.358337182448037e-05,
3019
+ "loss": 3.0043,
3020
+ "step": 2510
3021
+ },
3022
+ {
3023
+ "epoch": 0.83,
3024
+ "learning_rate": 2.3357043879907622e-05,
3025
+ "loss": 2.3829,
3026
+ "step": 2515
3027
+ },
3028
+ {
3029
+ "epoch": 0.83,
3030
+ "learning_rate": 2.3130715935334875e-05,
3031
+ "loss": 2.4771,
3032
+ "step": 2520
3033
+ },
3034
+ {
3035
+ "epoch": 0.83,
3036
+ "learning_rate": 2.2904387990762124e-05,
3037
+ "loss": 2.4709,
3038
+ "step": 2525
3039
+ },
3040
+ {
3041
+ "epoch": 0.83,
3042
+ "learning_rate": 2.2678060046189377e-05,
3043
+ "loss": 2.8935,
3044
+ "step": 2530
3045
+ },
3046
+ {
3047
+ "epoch": 0.84,
3048
+ "learning_rate": 2.245173210161663e-05,
3049
+ "loss": 2.9177,
3050
+ "step": 2535
3051
+ },
3052
+ {
3053
+ "epoch": 0.84,
3054
+ "learning_rate": 2.2225404157043883e-05,
3055
+ "loss": 2.9019,
3056
+ "step": 2540
3057
+ },
3058
+ {
3059
+ "epoch": 0.84,
3060
+ "learning_rate": 2.1999076212471132e-05,
3061
+ "loss": 2.944,
3062
+ "step": 2545
3063
+ },
3064
+ {
3065
+ "epoch": 0.84,
3066
+ "learning_rate": 2.1772748267898385e-05,
3067
+ "loss": 2.4892,
3068
+ "step": 2550
3069
+ },
3070
+ {
3071
+ "epoch": 0.84,
3072
+ "learning_rate": 2.1546420323325638e-05,
3073
+ "loss": 2.8733,
3074
+ "step": 2555
3075
+ },
3076
+ {
3077
+ "epoch": 0.84,
3078
+ "learning_rate": 2.1320092378752887e-05,
3079
+ "loss": 1.6182,
3080
+ "step": 2560
3081
+ },
3082
+ {
3083
+ "epoch": 0.85,
3084
+ "learning_rate": 2.1093764434180137e-05,
3085
+ "loss": 2.8951,
3086
+ "step": 2565
3087
+ },
3088
+ {
3089
+ "epoch": 0.85,
3090
+ "learning_rate": 2.086743648960739e-05,
3091
+ "loss": 2.903,
3092
+ "step": 2570
3093
+ },
3094
+ {
3095
+ "epoch": 0.85,
3096
+ "learning_rate": 2.0641108545034642e-05,
3097
+ "loss": 2.3775,
3098
+ "step": 2575
3099
+ },
3100
+ {
3101
+ "epoch": 0.85,
3102
+ "learning_rate": 2.0414780600461892e-05,
3103
+ "loss": 2.5025,
3104
+ "step": 2580
3105
+ },
3106
+ {
3107
+ "epoch": 0.85,
3108
+ "learning_rate": 2.0188452655889145e-05,
3109
+ "loss": 2.3309,
3110
+ "step": 2585
3111
+ },
3112
+ {
3113
+ "epoch": 0.85,
3114
+ "learning_rate": 1.9962124711316398e-05,
3115
+ "loss": 2.5212,
3116
+ "step": 2590
3117
+ },
3118
+ {
3119
+ "epoch": 0.86,
3120
+ "learning_rate": 1.973579676674365e-05,
3121
+ "loss": 2.8385,
3122
+ "step": 2595
3123
+ },
3124
+ {
3125
+ "epoch": 0.86,
3126
+ "learning_rate": 1.95094688221709e-05,
3127
+ "loss": 2.3192,
3128
+ "step": 2600
3129
+ },
3130
+ {
3131
+ "epoch": 0.86,
3132
+ "learning_rate": 1.9283140877598153e-05,
3133
+ "loss": 2.8028,
3134
+ "step": 2605
3135
+ },
3136
+ {
3137
+ "epoch": 0.86,
3138
+ "learning_rate": 1.9056812933025406e-05,
3139
+ "loss": 2.9434,
3140
+ "step": 2610
3141
+ },
3142
+ {
3143
+ "epoch": 0.86,
3144
+ "learning_rate": 1.883048498845266e-05,
3145
+ "loss": 2.9547,
3146
+ "step": 2615
3147
+ },
3148
+ {
3149
+ "epoch": 0.86,
3150
+ "learning_rate": 1.8604157043879908e-05,
3151
+ "loss": 1.8949,
3152
+ "step": 2620
3153
+ },
3154
+ {
3155
+ "epoch": 0.87,
3156
+ "learning_rate": 1.837782909930716e-05,
3157
+ "loss": 1.8706,
3158
+ "step": 2625
3159
+ },
3160
+ {
3161
+ "epoch": 0.87,
3162
+ "learning_rate": 1.8151501154734413e-05,
3163
+ "loss": 2.424,
3164
+ "step": 2630
3165
+ },
3166
+ {
3167
+ "epoch": 0.87,
3168
+ "learning_rate": 1.7925173210161666e-05,
3169
+ "loss": 2.3591,
3170
+ "step": 2635
3171
+ },
3172
+ {
3173
+ "epoch": 0.87,
3174
+ "learning_rate": 1.7698845265588912e-05,
3175
+ "loss": 2.8878,
3176
+ "step": 2640
3177
+ },
3178
+ {
3179
+ "epoch": 0.87,
3180
+ "learning_rate": 1.7472517321016165e-05,
3181
+ "loss": 2.8651,
3182
+ "step": 2645
3183
+ },
3184
+ {
3185
+ "epoch": 0.87,
3186
+ "learning_rate": 1.7246189376443418e-05,
3187
+ "loss": 2.8761,
3188
+ "step": 2650
3189
+ },
3190
+ {
3191
+ "epoch": 0.88,
3192
+ "learning_rate": 1.701986143187067e-05,
3193
+ "loss": 2.833,
3194
+ "step": 2655
3195
+ },
3196
+ {
3197
+ "epoch": 0.88,
3198
+ "learning_rate": 1.679353348729792e-05,
3199
+ "loss": 2.395,
3200
+ "step": 2660
3201
+ },
3202
+ {
3203
+ "epoch": 0.88,
3204
+ "learning_rate": 1.6567205542725173e-05,
3205
+ "loss": 2.9417,
3206
+ "step": 2665
3207
+ },
3208
+ {
3209
+ "epoch": 0.88,
3210
+ "learning_rate": 1.6340877598152426e-05,
3211
+ "loss": 2.9002,
3212
+ "step": 2670
3213
+ },
3214
+ {
3215
+ "epoch": 0.88,
3216
+ "learning_rate": 1.611454965357968e-05,
3217
+ "loss": 2.3858,
3218
+ "step": 2675
3219
+ },
3220
+ {
3221
+ "epoch": 0.88,
3222
+ "learning_rate": 1.588822170900693e-05,
3223
+ "loss": 2.4055,
3224
+ "step": 2680
3225
+ },
3226
+ {
3227
+ "epoch": 0.89,
3228
+ "learning_rate": 1.566189376443418e-05,
3229
+ "loss": 2.8661,
3230
+ "step": 2685
3231
+ },
3232
+ {
3233
+ "epoch": 0.89,
3234
+ "learning_rate": 1.543556581986143e-05,
3235
+ "loss": 2.4005,
3236
+ "step": 2690
3237
+ },
3238
+ {
3239
+ "epoch": 0.89,
3240
+ "learning_rate": 1.5209237875288685e-05,
3241
+ "loss": 2.8337,
3242
+ "step": 2695
3243
+ },
3244
+ {
3245
+ "epoch": 0.89,
3246
+ "learning_rate": 1.4982909930715935e-05,
3247
+ "loss": 1.8415,
3248
+ "step": 2700
3249
+ },
3250
+ {
3251
+ "epoch": 0.89,
3252
+ "learning_rate": 1.4756581986143188e-05,
3253
+ "loss": 2.8588,
3254
+ "step": 2705
3255
+ },
3256
+ {
3257
+ "epoch": 0.89,
3258
+ "learning_rate": 1.4530254041570439e-05,
3259
+ "loss": 2.8498,
3260
+ "step": 2710
3261
+ },
3262
+ {
3263
+ "epoch": 0.9,
3264
+ "learning_rate": 1.430392609699769e-05,
3265
+ "loss": 2.3665,
3266
+ "step": 2715
3267
+ },
3268
+ {
3269
+ "epoch": 0.9,
3270
+ "learning_rate": 1.4077598152424943e-05,
3271
+ "loss": 2.9143,
3272
+ "step": 2720
3273
+ },
3274
+ {
3275
+ "epoch": 0.9,
3276
+ "learning_rate": 1.3851270207852194e-05,
3277
+ "loss": 2.8927,
3278
+ "step": 2725
3279
+ },
3280
+ {
3281
+ "epoch": 0.9,
3282
+ "learning_rate": 1.3624942263279447e-05,
3283
+ "loss": 2.2419,
3284
+ "step": 2730
3285
+ },
3286
+ {
3287
+ "epoch": 0.9,
3288
+ "learning_rate": 1.3398614318706696e-05,
3289
+ "loss": 2.8982,
3290
+ "step": 2735
3291
+ },
3292
+ {
3293
+ "epoch": 0.9,
3294
+ "learning_rate": 1.3172286374133949e-05,
3295
+ "loss": 2.3598,
3296
+ "step": 2740
3297
+ },
3298
+ {
3299
+ "epoch": 0.91,
3300
+ "learning_rate": 1.29459584295612e-05,
3301
+ "loss": 1.9034,
3302
+ "step": 2745
3303
+ },
3304
+ {
3305
+ "epoch": 0.91,
3306
+ "learning_rate": 1.2719630484988453e-05,
3307
+ "loss": 2.8746,
3308
+ "step": 2750
3309
+ },
3310
+ {
3311
+ "epoch": 0.91,
3312
+ "learning_rate": 1.2493302540415704e-05,
3313
+ "loss": 2.5453,
3314
+ "step": 2755
3315
+ },
3316
+ {
3317
+ "epoch": 0.91,
3318
+ "learning_rate": 1.2266974595842957e-05,
3319
+ "loss": 2.3748,
3320
+ "step": 2760
3321
+ },
3322
+ {
3323
+ "epoch": 0.91,
3324
+ "learning_rate": 1.2040646651270208e-05,
3325
+ "loss": 2.3395,
3326
+ "step": 2765
3327
+ },
3328
+ {
3329
+ "epoch": 0.91,
3330
+ "learning_rate": 1.1814318706697461e-05,
3331
+ "loss": 1.8988,
3332
+ "step": 2770
3333
+ },
3334
+ {
3335
+ "epoch": 0.92,
3336
+ "learning_rate": 1.1587990762124712e-05,
3337
+ "loss": 2.832,
3338
+ "step": 2775
3339
+ },
3340
+ {
3341
+ "epoch": 0.92,
3342
+ "learning_rate": 1.1361662817551963e-05,
3343
+ "loss": 2.9164,
3344
+ "step": 2780
3345
+ },
3346
+ {
3347
+ "epoch": 0.92,
3348
+ "learning_rate": 1.1135334872979214e-05,
3349
+ "loss": 1.3357,
3350
+ "step": 2785
3351
+ },
3352
+ {
3353
+ "epoch": 0.92,
3354
+ "learning_rate": 1.0909006928406465e-05,
3355
+ "loss": 2.3844,
3356
+ "step": 2790
3357
+ },
3358
+ {
3359
+ "epoch": 0.92,
3360
+ "learning_rate": 1.0682678983833718e-05,
3361
+ "loss": 2.3568,
3362
+ "step": 2795
3363
+ },
3364
+ {
3365
+ "epoch": 0.92,
3366
+ "learning_rate": 1.045635103926097e-05,
3367
+ "loss": 2.9659,
3368
+ "step": 2800
3369
+ },
3370
+ {
3371
+ "epoch": 0.93,
3372
+ "learning_rate": 1.0230023094688222e-05,
3373
+ "loss": 2.3948,
3374
+ "step": 2805
3375
+ },
3376
+ {
3377
+ "epoch": 0.93,
3378
+ "learning_rate": 1.0003695150115473e-05,
3379
+ "loss": 2.3828,
3380
+ "step": 2810
3381
+ },
3382
+ {
3383
+ "epoch": 0.93,
3384
+ "learning_rate": 9.777367205542726e-06,
3385
+ "loss": 2.3914,
3386
+ "step": 2815
3387
+ },
3388
+ {
3389
+ "epoch": 0.93,
3390
+ "learning_rate": 9.551039260969976e-06,
3391
+ "loss": 2.4423,
3392
+ "step": 2820
3393
+ },
3394
+ {
3395
+ "epoch": 0.93,
3396
+ "learning_rate": 9.324711316397229e-06,
3397
+ "loss": 2.4067,
3398
+ "step": 2825
3399
+ },
3400
+ {
3401
+ "epoch": 0.93,
3402
+ "learning_rate": 9.09838337182448e-06,
3403
+ "loss": 2.8074,
3404
+ "step": 2830
3405
+ },
3406
+ {
3407
+ "epoch": 0.94,
3408
+ "learning_rate": 8.872055427251733e-06,
3409
+ "loss": 2.9147,
3410
+ "step": 2835
3411
+ },
3412
+ {
3413
+ "epoch": 0.94,
3414
+ "learning_rate": 8.645727482678984e-06,
3415
+ "loss": 2.4484,
3416
+ "step": 2840
3417
+ },
3418
+ {
3419
+ "epoch": 0.94,
3420
+ "learning_rate": 8.419399538106235e-06,
3421
+ "loss": 2.8331,
3422
+ "step": 2845
3423
+ },
3424
+ {
3425
+ "epoch": 0.94,
3426
+ "learning_rate": 8.193071593533488e-06,
3427
+ "loss": 2.3605,
3428
+ "step": 2850
3429
+ },
3430
+ {
3431
+ "epoch": 0.94,
3432
+ "learning_rate": 7.966743648960739e-06,
3433
+ "loss": 2.0343,
3434
+ "step": 2855
3435
+ },
3436
+ {
3437
+ "epoch": 0.94,
3438
+ "learning_rate": 7.74041570438799e-06,
3439
+ "loss": 2.8557,
3440
+ "step": 2860
3441
+ },
3442
+ {
3443
+ "epoch": 0.95,
3444
+ "learning_rate": 7.514087759815243e-06,
3445
+ "loss": 2.4906,
3446
+ "step": 2865
3447
+ },
3448
+ {
3449
+ "epoch": 0.95,
3450
+ "learning_rate": 7.287759815242494e-06,
3451
+ "loss": 2.2705,
3452
+ "step": 2870
3453
+ },
3454
+ {
3455
+ "epoch": 0.95,
3456
+ "learning_rate": 7.061431870669746e-06,
3457
+ "loss": 2.1079,
3458
+ "step": 2875
3459
+ },
3460
+ {
3461
+ "epoch": 0.95,
3462
+ "learning_rate": 6.835103926096998e-06,
3463
+ "loss": 2.9184,
3464
+ "step": 2880
3465
+ },
3466
+ {
3467
+ "epoch": 0.95,
3468
+ "learning_rate": 6.60877598152425e-06,
3469
+ "loss": 2.7964,
3470
+ "step": 2885
3471
+ },
3472
+ {
3473
+ "epoch": 0.95,
3474
+ "learning_rate": 6.382448036951501e-06,
3475
+ "loss": 2.2701,
3476
+ "step": 2890
3477
+ },
3478
+ {
3479
+ "epoch": 0.96,
3480
+ "learning_rate": 6.156120092378753e-06,
3481
+ "loss": 2.4188,
3482
+ "step": 2895
3483
+ },
3484
+ {
3485
+ "epoch": 0.96,
3486
+ "learning_rate": 5.929792147806005e-06,
3487
+ "loss": 2.4344,
3488
+ "step": 2900
3489
+ },
3490
+ {
3491
+ "epoch": 0.96,
3492
+ "learning_rate": 5.703464203233256e-06,
3493
+ "loss": 2.0036,
3494
+ "step": 2905
3495
+ },
3496
+ {
3497
+ "epoch": 0.96,
3498
+ "learning_rate": 5.4771362586605075e-06,
3499
+ "loss": 2.3556,
3500
+ "step": 2910
3501
+ },
3502
+ {
3503
+ "epoch": 0.96,
3504
+ "learning_rate": 5.2508083140877595e-06,
3505
+ "loss": 2.8956,
3506
+ "step": 2915
3507
+ },
3508
+ {
3509
+ "epoch": 0.96,
3510
+ "learning_rate": 5.0244803695150115e-06,
3511
+ "loss": 2.9212,
3512
+ "step": 2920
3513
+ },
3514
+ {
3515
+ "epoch": 0.97,
3516
+ "learning_rate": 4.7981524249422635e-06,
3517
+ "loss": 2.325,
3518
+ "step": 2925
3519
+ },
3520
+ {
3521
+ "epoch": 0.97,
3522
+ "learning_rate": 4.571824480369515e-06,
3523
+ "loss": 2.9001,
3524
+ "step": 2930
3525
+ },
3526
+ {
3527
+ "epoch": 0.97,
3528
+ "learning_rate": 4.345496535796767e-06,
3529
+ "loss": 2.0078,
3530
+ "step": 2935
3531
+ },
3532
+ {
3533
+ "epoch": 0.97,
3534
+ "learning_rate": 4.119168591224019e-06,
3535
+ "loss": 1.8776,
3536
+ "step": 2940
3537
+ },
3538
+ {
3539
+ "epoch": 0.97,
3540
+ "learning_rate": 3.892840646651271e-06,
3541
+ "loss": 2.315,
3542
+ "step": 2945
3543
+ },
3544
+ {
3545
+ "epoch": 0.97,
3546
+ "learning_rate": 3.666512702078522e-06,
3547
+ "loss": 1.9461,
3548
+ "step": 2950
3549
+ },
3550
+ {
3551
+ "epoch": 0.97,
3552
+ "learning_rate": 3.4401847575057737e-06,
3553
+ "loss": 2.8444,
3554
+ "step": 2955
3555
+ },
3556
+ {
3557
+ "epoch": 0.98,
3558
+ "learning_rate": 3.2138568129330253e-06,
3559
+ "loss": 2.7964,
3560
+ "step": 2960
3561
+ },
3562
+ {
3563
+ "epoch": 0.98,
3564
+ "learning_rate": 2.987528868360277e-06,
3565
+ "loss": 2.2962,
3566
+ "step": 2965
3567
+ },
3568
+ {
3569
+ "epoch": 0.98,
3570
+ "learning_rate": 2.761200923787529e-06,
3571
+ "loss": 2.9162,
3572
+ "step": 2970
3573
+ },
3574
+ {
3575
+ "epoch": 0.98,
3576
+ "learning_rate": 2.5348729792147804e-06,
3577
+ "loss": 2.3761,
3578
+ "step": 2975
3579
+ },
3580
+ {
3581
+ "epoch": 0.98,
3582
+ "learning_rate": 2.3085450346420324e-06,
3583
+ "loss": 2.7675,
3584
+ "step": 2980
3585
+ },
3586
+ {
3587
+ "epoch": 0.98,
3588
+ "learning_rate": 2.082217090069284e-06,
3589
+ "loss": 2.3303,
3590
+ "step": 2985
3591
+ },
3592
+ {
3593
+ "epoch": 0.99,
3594
+ "learning_rate": 1.8558891454965358e-06,
3595
+ "loss": 2.3253,
3596
+ "step": 2990
3597
+ },
3598
+ {
3599
+ "epoch": 0.99,
3600
+ "learning_rate": 1.6295612009237876e-06,
3601
+ "loss": 2.8782,
3602
+ "step": 2995
3603
+ },
3604
+ {
3605
+ "epoch": 0.99,
3606
+ "learning_rate": 1.4032332563510394e-06,
3607
+ "loss": 2.7999,
3608
+ "step": 3000
3609
+ },
3610
+ {
3611
+ "epoch": 0.99,
3612
+ "learning_rate": 1.1769053117782911e-06,
3613
+ "loss": 2.7688,
3614
+ "step": 3005
3615
+ },
3616
+ {
3617
+ "epoch": 0.99,
3618
+ "learning_rate": 9.505773672055428e-07,
3619
+ "loss": 2.8891,
3620
+ "step": 3010
3621
+ },
3622
+ {
3623
+ "epoch": 0.99,
3624
+ "learning_rate": 7.242494226327945e-07,
3625
+ "loss": 2.4505,
3626
+ "step": 3015
3627
+ },
3628
+ {
3629
+ "epoch": 1.0,
3630
+ "learning_rate": 4.979214780600462e-07,
3631
+ "loss": 2.8581,
3632
+ "step": 3020
3633
+ },
3634
+ {
3635
+ "epoch": 1.0,
3636
+ "learning_rate": 2.715935334872979e-07,
3637
+ "loss": 2.3888,
3638
+ "step": 3025
3639
+ },
3640
+ {
3641
+ "epoch": 1.0,
3642
+ "learning_rate": 4.5265588914549656e-08,
3643
+ "loss": 2.2638,
3644
+ "step": 3030
3645
+ },
3646
+ {
3647
+ "epoch": 1.0,
3648
+ "step": 3031,
3649
+ "total_flos": 5629783610228736.0,
3650
+ "train_loss": 2.9040869920020134,
3651
+ "train_runtime": 3297.2859,
3652
+ "train_samples_per_second": 0.919,
3653
+ "train_steps_per_second": 0.919
3654
+ }
3655
+ ],
3656
+ "max_steps": 3031,
3657
+ "num_train_epochs": 1,
3658
+ "total_flos": 5629783610228736.0,
3659
+ "trial_name": null,
3660
+ "trial_params": null
3661
+ }