devkushal75 commited on
Commit
c9ddb93
·
1 Parent(s): 575ea35

Upload trainer_state.json

Browse files
Files changed (1) hide show
  1. trainer_state.json +3616 -0
trainer_state.json ADDED
@@ -0,0 +1,3616 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 48.0,
5
+ "global_step": 6000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.08,
12
+ "learning_rate": 1.0000000000000002e-06,
13
+ "loss": 3.6493,
14
+ "step": 10
15
+ },
16
+ {
17
+ "epoch": 0.16,
18
+ "learning_rate": 2.0000000000000003e-06,
19
+ "loss": 3.6538,
20
+ "step": 20
21
+ },
22
+ {
23
+ "epoch": 0.24,
24
+ "learning_rate": 3e-06,
25
+ "loss": 3.6342,
26
+ "step": 30
27
+ },
28
+ {
29
+ "epoch": 0.32,
30
+ "learning_rate": 4.000000000000001e-06,
31
+ "loss": 3.5774,
32
+ "step": 40
33
+ },
34
+ {
35
+ "epoch": 0.4,
36
+ "learning_rate": 5e-06,
37
+ "loss": 3.5288,
38
+ "step": 50
39
+ },
40
+ {
41
+ "epoch": 0.48,
42
+ "learning_rate": 6e-06,
43
+ "loss": 3.452,
44
+ "step": 60
45
+ },
46
+ {
47
+ "epoch": 0.56,
48
+ "learning_rate": 7.000000000000001e-06,
49
+ "loss": 3.3685,
50
+ "step": 70
51
+ },
52
+ {
53
+ "epoch": 0.64,
54
+ "learning_rate": 8.000000000000001e-06,
55
+ "loss": 3.2667,
56
+ "step": 80
57
+ },
58
+ {
59
+ "epoch": 0.72,
60
+ "learning_rate": 9e-06,
61
+ "loss": 3.2796,
62
+ "step": 90
63
+ },
64
+ {
65
+ "epoch": 0.8,
66
+ "learning_rate": 1e-05,
67
+ "loss": 3.1435,
68
+ "step": 100
69
+ },
70
+ {
71
+ "epoch": 0.88,
72
+ "learning_rate": 1.1000000000000001e-05,
73
+ "loss": 3.1206,
74
+ "step": 110
75
+ },
76
+ {
77
+ "epoch": 0.96,
78
+ "learning_rate": 1.2e-05,
79
+ "loss": 3.0387,
80
+ "step": 120
81
+ },
82
+ {
83
+ "epoch": 1.04,
84
+ "learning_rate": 1.3000000000000001e-05,
85
+ "loss": 2.9759,
86
+ "step": 130
87
+ },
88
+ {
89
+ "epoch": 1.12,
90
+ "learning_rate": 1.4000000000000001e-05,
91
+ "loss": 3.0132,
92
+ "step": 140
93
+ },
94
+ {
95
+ "epoch": 1.2,
96
+ "learning_rate": 1.5e-05,
97
+ "loss": 2.9027,
98
+ "step": 150
99
+ },
100
+ {
101
+ "epoch": 1.28,
102
+ "learning_rate": 1.6000000000000003e-05,
103
+ "loss": 2.815,
104
+ "step": 160
105
+ },
106
+ {
107
+ "epoch": 1.36,
108
+ "learning_rate": 1.7000000000000003e-05,
109
+ "loss": 2.8081,
110
+ "step": 170
111
+ },
112
+ {
113
+ "epoch": 1.44,
114
+ "learning_rate": 1.8e-05,
115
+ "loss": 2.8926,
116
+ "step": 180
117
+ },
118
+ {
119
+ "epoch": 1.52,
120
+ "learning_rate": 1.9e-05,
121
+ "loss": 2.8096,
122
+ "step": 190
123
+ },
124
+ {
125
+ "epoch": 1.6,
126
+ "learning_rate": 2e-05,
127
+ "loss": 2.7153,
128
+ "step": 200
129
+ },
130
+ {
131
+ "epoch": 1.68,
132
+ "learning_rate": 2.1e-05,
133
+ "loss": 2.7193,
134
+ "step": 210
135
+ },
136
+ {
137
+ "epoch": 1.76,
138
+ "learning_rate": 2.2000000000000003e-05,
139
+ "loss": 2.5646,
140
+ "step": 220
141
+ },
142
+ {
143
+ "epoch": 1.84,
144
+ "learning_rate": 2.3000000000000003e-05,
145
+ "loss": 2.7471,
146
+ "step": 230
147
+ },
148
+ {
149
+ "epoch": 1.92,
150
+ "learning_rate": 2.4e-05,
151
+ "loss": 2.7368,
152
+ "step": 240
153
+ },
154
+ {
155
+ "epoch": 2.0,
156
+ "learning_rate": 2.5e-05,
157
+ "loss": 2.6167,
158
+ "step": 250
159
+ },
160
+ {
161
+ "epoch": 2.08,
162
+ "learning_rate": 2.6000000000000002e-05,
163
+ "loss": 2.492,
164
+ "step": 260
165
+ },
166
+ {
167
+ "epoch": 2.16,
168
+ "learning_rate": 2.7000000000000002e-05,
169
+ "loss": 2.5191,
170
+ "step": 270
171
+ },
172
+ {
173
+ "epoch": 2.24,
174
+ "learning_rate": 2.8000000000000003e-05,
175
+ "loss": 2.5223,
176
+ "step": 280
177
+ },
178
+ {
179
+ "epoch": 2.32,
180
+ "learning_rate": 2.9e-05,
181
+ "loss": 2.4294,
182
+ "step": 290
183
+ },
184
+ {
185
+ "epoch": 2.4,
186
+ "learning_rate": 3e-05,
187
+ "loss": 2.3719,
188
+ "step": 300
189
+ },
190
+ {
191
+ "epoch": 2.48,
192
+ "learning_rate": 3.1e-05,
193
+ "loss": 2.4148,
194
+ "step": 310
195
+ },
196
+ {
197
+ "epoch": 2.56,
198
+ "learning_rate": 3.2000000000000005e-05,
199
+ "loss": 2.4787,
200
+ "step": 320
201
+ },
202
+ {
203
+ "epoch": 2.64,
204
+ "learning_rate": 3.3e-05,
205
+ "loss": 2.4021,
206
+ "step": 330
207
+ },
208
+ {
209
+ "epoch": 2.72,
210
+ "learning_rate": 3.4000000000000007e-05,
211
+ "loss": 2.3982,
212
+ "step": 340
213
+ },
214
+ {
215
+ "epoch": 2.8,
216
+ "learning_rate": 3.5e-05,
217
+ "loss": 2.28,
218
+ "step": 350
219
+ },
220
+ {
221
+ "epoch": 2.88,
222
+ "learning_rate": 3.6e-05,
223
+ "loss": 2.4302,
224
+ "step": 360
225
+ },
226
+ {
227
+ "epoch": 2.96,
228
+ "learning_rate": 3.7e-05,
229
+ "loss": 2.2597,
230
+ "step": 370
231
+ },
232
+ {
233
+ "epoch": 3.04,
234
+ "learning_rate": 3.8e-05,
235
+ "loss": 2.0781,
236
+ "step": 380
237
+ },
238
+ {
239
+ "epoch": 3.12,
240
+ "learning_rate": 3.9000000000000006e-05,
241
+ "loss": 2.0256,
242
+ "step": 390
243
+ },
244
+ {
245
+ "epoch": 3.2,
246
+ "learning_rate": 4e-05,
247
+ "loss": 2.1566,
248
+ "step": 400
249
+ },
250
+ {
251
+ "epoch": 3.28,
252
+ "learning_rate": 4.1e-05,
253
+ "loss": 2.0487,
254
+ "step": 410
255
+ },
256
+ {
257
+ "epoch": 3.36,
258
+ "learning_rate": 4.2e-05,
259
+ "loss": 2.0661,
260
+ "step": 420
261
+ },
262
+ {
263
+ "epoch": 3.44,
264
+ "learning_rate": 4.3e-05,
265
+ "loss": 2.0816,
266
+ "step": 430
267
+ },
268
+ {
269
+ "epoch": 3.52,
270
+ "learning_rate": 4.4000000000000006e-05,
271
+ "loss": 2.1713,
272
+ "step": 440
273
+ },
274
+ {
275
+ "epoch": 3.6,
276
+ "learning_rate": 4.5e-05,
277
+ "loss": 2.151,
278
+ "step": 450
279
+ },
280
+ {
281
+ "epoch": 3.68,
282
+ "learning_rate": 4.600000000000001e-05,
283
+ "loss": 1.9657,
284
+ "step": 460
285
+ },
286
+ {
287
+ "epoch": 3.76,
288
+ "learning_rate": 4.7e-05,
289
+ "loss": 2.0014,
290
+ "step": 470
291
+ },
292
+ {
293
+ "epoch": 3.84,
294
+ "learning_rate": 4.8e-05,
295
+ "loss": 1.9713,
296
+ "step": 480
297
+ },
298
+ {
299
+ "epoch": 3.92,
300
+ "learning_rate": 4.9e-05,
301
+ "loss": 1.936,
302
+ "step": 490
303
+ },
304
+ {
305
+ "epoch": 4.0,
306
+ "learning_rate": 5e-05,
307
+ "loss": 1.9813,
308
+ "step": 500
309
+ },
310
+ {
311
+ "epoch": 4.08,
312
+ "learning_rate": 4.9913043478260876e-05,
313
+ "loss": 1.7488,
314
+ "step": 510
315
+ },
316
+ {
317
+ "epoch": 4.16,
318
+ "learning_rate": 4.9826086956521736e-05,
319
+ "loss": 1.7985,
320
+ "step": 520
321
+ },
322
+ {
323
+ "epoch": 4.24,
324
+ "learning_rate": 4.973913043478261e-05,
325
+ "loss": 1.7398,
326
+ "step": 530
327
+ },
328
+ {
329
+ "epoch": 4.32,
330
+ "learning_rate": 4.9652173913043483e-05,
331
+ "loss": 1.5946,
332
+ "step": 540
333
+ },
334
+ {
335
+ "epoch": 4.4,
336
+ "learning_rate": 4.956521739130435e-05,
337
+ "loss": 1.879,
338
+ "step": 550
339
+ },
340
+ {
341
+ "epoch": 4.48,
342
+ "learning_rate": 4.947826086956522e-05,
343
+ "loss": 1.8759,
344
+ "step": 560
345
+ },
346
+ {
347
+ "epoch": 4.56,
348
+ "learning_rate": 4.939130434782609e-05,
349
+ "loss": 1.7932,
350
+ "step": 570
351
+ },
352
+ {
353
+ "epoch": 4.64,
354
+ "learning_rate": 4.930434782608696e-05,
355
+ "loss": 1.7335,
356
+ "step": 580
357
+ },
358
+ {
359
+ "epoch": 4.72,
360
+ "learning_rate": 4.9217391304347824e-05,
361
+ "loss": 1.7787,
362
+ "step": 590
363
+ },
364
+ {
365
+ "epoch": 4.8,
366
+ "learning_rate": 4.91304347826087e-05,
367
+ "loss": 1.8258,
368
+ "step": 600
369
+ },
370
+ {
371
+ "epoch": 4.88,
372
+ "learning_rate": 4.904347826086957e-05,
373
+ "loss": 1.7937,
374
+ "step": 610
375
+ },
376
+ {
377
+ "epoch": 4.96,
378
+ "learning_rate": 4.895652173913044e-05,
379
+ "loss": 1.734,
380
+ "step": 620
381
+ },
382
+ {
383
+ "epoch": 5.04,
384
+ "learning_rate": 4.8869565217391305e-05,
385
+ "loss": 1.5941,
386
+ "step": 630
387
+ },
388
+ {
389
+ "epoch": 5.12,
390
+ "learning_rate": 4.878260869565218e-05,
391
+ "loss": 1.6789,
392
+ "step": 640
393
+ },
394
+ {
395
+ "epoch": 5.2,
396
+ "learning_rate": 4.8695652173913046e-05,
397
+ "loss": 1.4186,
398
+ "step": 650
399
+ },
400
+ {
401
+ "epoch": 5.28,
402
+ "learning_rate": 4.860869565217391e-05,
403
+ "loss": 1.5457,
404
+ "step": 660
405
+ },
406
+ {
407
+ "epoch": 5.36,
408
+ "learning_rate": 4.8521739130434786e-05,
409
+ "loss": 1.5282,
410
+ "step": 670
411
+ },
412
+ {
413
+ "epoch": 5.44,
414
+ "learning_rate": 4.843478260869565e-05,
415
+ "loss": 1.5594,
416
+ "step": 680
417
+ },
418
+ {
419
+ "epoch": 5.52,
420
+ "learning_rate": 4.834782608695652e-05,
421
+ "loss": 1.7068,
422
+ "step": 690
423
+ },
424
+ {
425
+ "epoch": 5.6,
426
+ "learning_rate": 4.8260869565217394e-05,
427
+ "loss": 1.5521,
428
+ "step": 700
429
+ },
430
+ {
431
+ "epoch": 5.68,
432
+ "learning_rate": 4.817391304347826e-05,
433
+ "loss": 1.5973,
434
+ "step": 710
435
+ },
436
+ {
437
+ "epoch": 5.76,
438
+ "learning_rate": 4.8086956521739134e-05,
439
+ "loss": 1.5842,
440
+ "step": 720
441
+ },
442
+ {
443
+ "epoch": 5.84,
444
+ "learning_rate": 4.8e-05,
445
+ "loss": 1.5819,
446
+ "step": 730
447
+ },
448
+ {
449
+ "epoch": 5.92,
450
+ "learning_rate": 4.7913043478260875e-05,
451
+ "loss": 1.6313,
452
+ "step": 740
453
+ },
454
+ {
455
+ "epoch": 6.0,
456
+ "learning_rate": 4.782608695652174e-05,
457
+ "loss": 1.5262,
458
+ "step": 750
459
+ },
460
+ {
461
+ "epoch": 6.08,
462
+ "learning_rate": 4.773913043478261e-05,
463
+ "loss": 1.3722,
464
+ "step": 760
465
+ },
466
+ {
467
+ "epoch": 6.16,
468
+ "learning_rate": 4.765217391304348e-05,
469
+ "loss": 1.2905,
470
+ "step": 770
471
+ },
472
+ {
473
+ "epoch": 6.24,
474
+ "learning_rate": 4.756521739130435e-05,
475
+ "loss": 1.4479,
476
+ "step": 780
477
+ },
478
+ {
479
+ "epoch": 6.32,
480
+ "learning_rate": 4.747826086956522e-05,
481
+ "loss": 1.424,
482
+ "step": 790
483
+ },
484
+ {
485
+ "epoch": 6.4,
486
+ "learning_rate": 4.739130434782609e-05,
487
+ "loss": 1.3006,
488
+ "step": 800
489
+ },
490
+ {
491
+ "epoch": 6.48,
492
+ "learning_rate": 4.7304347826086956e-05,
493
+ "loss": 1.4776,
494
+ "step": 810
495
+ },
496
+ {
497
+ "epoch": 6.56,
498
+ "learning_rate": 4.721739130434783e-05,
499
+ "loss": 1.4928,
500
+ "step": 820
501
+ },
502
+ {
503
+ "epoch": 6.64,
504
+ "learning_rate": 4.71304347826087e-05,
505
+ "loss": 1.4294,
506
+ "step": 830
507
+ },
508
+ {
509
+ "epoch": 6.72,
510
+ "learning_rate": 4.7043478260869564e-05,
511
+ "loss": 1.386,
512
+ "step": 840
513
+ },
514
+ {
515
+ "epoch": 6.8,
516
+ "learning_rate": 4.695652173913044e-05,
517
+ "loss": 1.4618,
518
+ "step": 850
519
+ },
520
+ {
521
+ "epoch": 6.88,
522
+ "learning_rate": 4.686956521739131e-05,
523
+ "loss": 1.4186,
524
+ "step": 860
525
+ },
526
+ {
527
+ "epoch": 6.96,
528
+ "learning_rate": 4.678260869565218e-05,
529
+ "loss": 1.6092,
530
+ "step": 870
531
+ },
532
+ {
533
+ "epoch": 7.04,
534
+ "learning_rate": 4.6695652173913045e-05,
535
+ "loss": 1.3375,
536
+ "step": 880
537
+ },
538
+ {
539
+ "epoch": 7.12,
540
+ "learning_rate": 4.660869565217392e-05,
541
+ "loss": 1.2338,
542
+ "step": 890
543
+ },
544
+ {
545
+ "epoch": 7.2,
546
+ "learning_rate": 4.6521739130434785e-05,
547
+ "loss": 1.3258,
548
+ "step": 900
549
+ },
550
+ {
551
+ "epoch": 7.28,
552
+ "learning_rate": 4.643478260869565e-05,
553
+ "loss": 1.2257,
554
+ "step": 910
555
+ },
556
+ {
557
+ "epoch": 7.36,
558
+ "learning_rate": 4.6347826086956526e-05,
559
+ "loss": 1.2538,
560
+ "step": 920
561
+ },
562
+ {
563
+ "epoch": 7.44,
564
+ "learning_rate": 4.62608695652174e-05,
565
+ "loss": 1.3203,
566
+ "step": 930
567
+ },
568
+ {
569
+ "epoch": 7.52,
570
+ "learning_rate": 4.617391304347826e-05,
571
+ "loss": 1.2229,
572
+ "step": 940
573
+ },
574
+ {
575
+ "epoch": 7.6,
576
+ "learning_rate": 4.608695652173913e-05,
577
+ "loss": 1.31,
578
+ "step": 950
579
+ },
580
+ {
581
+ "epoch": 7.68,
582
+ "learning_rate": 4.600000000000001e-05,
583
+ "loss": 1.3021,
584
+ "step": 960
585
+ },
586
+ {
587
+ "epoch": 7.76,
588
+ "learning_rate": 4.591304347826087e-05,
589
+ "loss": 1.3159,
590
+ "step": 970
591
+ },
592
+ {
593
+ "epoch": 7.84,
594
+ "learning_rate": 4.582608695652174e-05,
595
+ "loss": 1.3946,
596
+ "step": 980
597
+ },
598
+ {
599
+ "epoch": 7.92,
600
+ "learning_rate": 4.5739130434782614e-05,
601
+ "loss": 1.4731,
602
+ "step": 990
603
+ },
604
+ {
605
+ "epoch": 8.0,
606
+ "learning_rate": 4.565217391304348e-05,
607
+ "loss": 1.4642,
608
+ "step": 1000
609
+ },
610
+ {
611
+ "epoch": 8.08,
612
+ "learning_rate": 4.556521739130435e-05,
613
+ "loss": 1.173,
614
+ "step": 1010
615
+ },
616
+ {
617
+ "epoch": 8.16,
618
+ "learning_rate": 4.547826086956522e-05,
619
+ "loss": 1.1411,
620
+ "step": 1020
621
+ },
622
+ {
623
+ "epoch": 8.24,
624
+ "learning_rate": 4.539130434782609e-05,
625
+ "loss": 1.1576,
626
+ "step": 1030
627
+ },
628
+ {
629
+ "epoch": 8.32,
630
+ "learning_rate": 4.5304347826086955e-05,
631
+ "loss": 1.1805,
632
+ "step": 1040
633
+ },
634
+ {
635
+ "epoch": 8.4,
636
+ "learning_rate": 4.521739130434783e-05,
637
+ "loss": 1.1661,
638
+ "step": 1050
639
+ },
640
+ {
641
+ "epoch": 8.48,
642
+ "learning_rate": 4.51304347826087e-05,
643
+ "loss": 1.1878,
644
+ "step": 1060
645
+ },
646
+ {
647
+ "epoch": 8.56,
648
+ "learning_rate": 4.504347826086956e-05,
649
+ "loss": 1.2674,
650
+ "step": 1070
651
+ },
652
+ {
653
+ "epoch": 8.64,
654
+ "learning_rate": 4.4956521739130436e-05,
655
+ "loss": 1.1968,
656
+ "step": 1080
657
+ },
658
+ {
659
+ "epoch": 8.72,
660
+ "learning_rate": 4.486956521739131e-05,
661
+ "loss": 1.3326,
662
+ "step": 1090
663
+ },
664
+ {
665
+ "epoch": 8.8,
666
+ "learning_rate": 4.478260869565218e-05,
667
+ "loss": 1.1905,
668
+ "step": 1100
669
+ },
670
+ {
671
+ "epoch": 8.88,
672
+ "learning_rate": 4.4695652173913044e-05,
673
+ "loss": 1.3127,
674
+ "step": 1110
675
+ },
676
+ {
677
+ "epoch": 8.96,
678
+ "learning_rate": 4.460869565217392e-05,
679
+ "loss": 1.3091,
680
+ "step": 1120
681
+ },
682
+ {
683
+ "epoch": 9.04,
684
+ "learning_rate": 4.4521739130434784e-05,
685
+ "loss": 1.1955,
686
+ "step": 1130
687
+ },
688
+ {
689
+ "epoch": 9.12,
690
+ "learning_rate": 4.443478260869565e-05,
691
+ "loss": 1.0431,
692
+ "step": 1140
693
+ },
694
+ {
695
+ "epoch": 9.2,
696
+ "learning_rate": 4.4347826086956525e-05,
697
+ "loss": 1.0506,
698
+ "step": 1150
699
+ },
700
+ {
701
+ "epoch": 9.28,
702
+ "learning_rate": 4.426086956521739e-05,
703
+ "loss": 1.0924,
704
+ "step": 1160
705
+ },
706
+ {
707
+ "epoch": 9.36,
708
+ "learning_rate": 4.4173913043478265e-05,
709
+ "loss": 1.1698,
710
+ "step": 1170
711
+ },
712
+ {
713
+ "epoch": 9.44,
714
+ "learning_rate": 4.408695652173913e-05,
715
+ "loss": 1.2168,
716
+ "step": 1180
717
+ },
718
+ {
719
+ "epoch": 9.52,
720
+ "learning_rate": 4.4000000000000006e-05,
721
+ "loss": 1.1931,
722
+ "step": 1190
723
+ },
724
+ {
725
+ "epoch": 9.6,
726
+ "learning_rate": 4.391304347826087e-05,
727
+ "loss": 1.2466,
728
+ "step": 1200
729
+ },
730
+ {
731
+ "epoch": 9.68,
732
+ "learning_rate": 4.382608695652174e-05,
733
+ "loss": 1.1826,
734
+ "step": 1210
735
+ },
736
+ {
737
+ "epoch": 9.76,
738
+ "learning_rate": 4.373913043478261e-05,
739
+ "loss": 1.1994,
740
+ "step": 1220
741
+ },
742
+ {
743
+ "epoch": 9.84,
744
+ "learning_rate": 4.365217391304348e-05,
745
+ "loss": 1.116,
746
+ "step": 1230
747
+ },
748
+ {
749
+ "epoch": 9.92,
750
+ "learning_rate": 4.3565217391304353e-05,
751
+ "loss": 1.1418,
752
+ "step": 1240
753
+ },
754
+ {
755
+ "epoch": 10.0,
756
+ "learning_rate": 4.347826086956522e-05,
757
+ "loss": 1.2596,
758
+ "step": 1250
759
+ },
760
+ {
761
+ "epoch": 10.08,
762
+ "learning_rate": 4.339130434782609e-05,
763
+ "loss": 1.0032,
764
+ "step": 1260
765
+ },
766
+ {
767
+ "epoch": 10.16,
768
+ "learning_rate": 4.330434782608696e-05,
769
+ "loss": 1.0681,
770
+ "step": 1270
771
+ },
772
+ {
773
+ "epoch": 10.24,
774
+ "learning_rate": 4.321739130434783e-05,
775
+ "loss": 1.0991,
776
+ "step": 1280
777
+ },
778
+ {
779
+ "epoch": 10.32,
780
+ "learning_rate": 4.3130434782608695e-05,
781
+ "loss": 0.9904,
782
+ "step": 1290
783
+ },
784
+ {
785
+ "epoch": 10.4,
786
+ "learning_rate": 4.304347826086957e-05,
787
+ "loss": 1.037,
788
+ "step": 1300
789
+ },
790
+ {
791
+ "epoch": 10.48,
792
+ "learning_rate": 4.2956521739130435e-05,
793
+ "loss": 1.1417,
794
+ "step": 1310
795
+ },
796
+ {
797
+ "epoch": 10.56,
798
+ "learning_rate": 4.286956521739131e-05,
799
+ "loss": 1.0919,
800
+ "step": 1320
801
+ },
802
+ {
803
+ "epoch": 10.64,
804
+ "learning_rate": 4.2782608695652176e-05,
805
+ "loss": 1.0887,
806
+ "step": 1330
807
+ },
808
+ {
809
+ "epoch": 10.72,
810
+ "learning_rate": 4.269565217391305e-05,
811
+ "loss": 1.1425,
812
+ "step": 1340
813
+ },
814
+ {
815
+ "epoch": 10.8,
816
+ "learning_rate": 4.2608695652173916e-05,
817
+ "loss": 1.2263,
818
+ "step": 1350
819
+ },
820
+ {
821
+ "epoch": 10.88,
822
+ "learning_rate": 4.252173913043478e-05,
823
+ "loss": 1.2244,
824
+ "step": 1360
825
+ },
826
+ {
827
+ "epoch": 10.96,
828
+ "learning_rate": 4.2434782608695657e-05,
829
+ "loss": 1.0981,
830
+ "step": 1370
831
+ },
832
+ {
833
+ "epoch": 11.04,
834
+ "learning_rate": 4.2347826086956523e-05,
835
+ "loss": 1.1665,
836
+ "step": 1380
837
+ },
838
+ {
839
+ "epoch": 11.12,
840
+ "learning_rate": 4.226086956521739e-05,
841
+ "loss": 0.9729,
842
+ "step": 1390
843
+ },
844
+ {
845
+ "epoch": 11.2,
846
+ "learning_rate": 4.2173913043478264e-05,
847
+ "loss": 1.0235,
848
+ "step": 1400
849
+ },
850
+ {
851
+ "epoch": 11.28,
852
+ "learning_rate": 4.208695652173914e-05,
853
+ "loss": 1.0718,
854
+ "step": 1410
855
+ },
856
+ {
857
+ "epoch": 11.36,
858
+ "learning_rate": 4.2e-05,
859
+ "loss": 1.1057,
860
+ "step": 1420
861
+ },
862
+ {
863
+ "epoch": 11.44,
864
+ "learning_rate": 4.191304347826087e-05,
865
+ "loss": 1.0105,
866
+ "step": 1430
867
+ },
868
+ {
869
+ "epoch": 11.52,
870
+ "learning_rate": 4.1826086956521745e-05,
871
+ "loss": 1.0284,
872
+ "step": 1440
873
+ },
874
+ {
875
+ "epoch": 11.6,
876
+ "learning_rate": 4.1739130434782605e-05,
877
+ "loss": 0.9386,
878
+ "step": 1450
879
+ },
880
+ {
881
+ "epoch": 11.68,
882
+ "learning_rate": 4.165217391304348e-05,
883
+ "loss": 1.0335,
884
+ "step": 1460
885
+ },
886
+ {
887
+ "epoch": 11.76,
888
+ "learning_rate": 4.156521739130435e-05,
889
+ "loss": 1.125,
890
+ "step": 1470
891
+ },
892
+ {
893
+ "epoch": 11.84,
894
+ "learning_rate": 4.147826086956522e-05,
895
+ "loss": 1.1251,
896
+ "step": 1480
897
+ },
898
+ {
899
+ "epoch": 11.92,
900
+ "learning_rate": 4.1391304347826086e-05,
901
+ "loss": 1.1962,
902
+ "step": 1490
903
+ },
904
+ {
905
+ "epoch": 12.0,
906
+ "learning_rate": 4.130434782608696e-05,
907
+ "loss": 1.1286,
908
+ "step": 1500
909
+ },
910
+ {
911
+ "epoch": 12.08,
912
+ "learning_rate": 4.1217391304347827e-05,
913
+ "loss": 0.9152,
914
+ "step": 1510
915
+ },
916
+ {
917
+ "epoch": 12.16,
918
+ "learning_rate": 4.1130434782608693e-05,
919
+ "loss": 0.9163,
920
+ "step": 1520
921
+ },
922
+ {
923
+ "epoch": 12.24,
924
+ "learning_rate": 4.104347826086957e-05,
925
+ "loss": 0.959,
926
+ "step": 1530
927
+ },
928
+ {
929
+ "epoch": 12.32,
930
+ "learning_rate": 4.095652173913044e-05,
931
+ "loss": 1.006,
932
+ "step": 1540
933
+ },
934
+ {
935
+ "epoch": 12.4,
936
+ "learning_rate": 4.086956521739131e-05,
937
+ "loss": 1.0188,
938
+ "step": 1550
939
+ },
940
+ {
941
+ "epoch": 12.48,
942
+ "learning_rate": 4.0782608695652174e-05,
943
+ "loss": 0.9931,
944
+ "step": 1560
945
+ },
946
+ {
947
+ "epoch": 12.56,
948
+ "learning_rate": 4.069565217391305e-05,
949
+ "loss": 1.0445,
950
+ "step": 1570
951
+ },
952
+ {
953
+ "epoch": 12.64,
954
+ "learning_rate": 4.0608695652173915e-05,
955
+ "loss": 1.0901,
956
+ "step": 1580
957
+ },
958
+ {
959
+ "epoch": 12.72,
960
+ "learning_rate": 4.052173913043478e-05,
961
+ "loss": 0.9693,
962
+ "step": 1590
963
+ },
964
+ {
965
+ "epoch": 12.8,
966
+ "learning_rate": 4.0434782608695655e-05,
967
+ "loss": 1.1191,
968
+ "step": 1600
969
+ },
970
+ {
971
+ "epoch": 12.88,
972
+ "learning_rate": 4.034782608695652e-05,
973
+ "loss": 1.1188,
974
+ "step": 1610
975
+ },
976
+ {
977
+ "epoch": 12.96,
978
+ "learning_rate": 4.026086956521739e-05,
979
+ "loss": 1.1422,
980
+ "step": 1620
981
+ },
982
+ {
983
+ "epoch": 13.04,
984
+ "learning_rate": 4.017391304347826e-05,
985
+ "loss": 1.0245,
986
+ "step": 1630
987
+ },
988
+ {
989
+ "epoch": 13.12,
990
+ "learning_rate": 4.008695652173913e-05,
991
+ "loss": 0.9048,
992
+ "step": 1640
993
+ },
994
+ {
995
+ "epoch": 13.2,
996
+ "learning_rate": 4e-05,
997
+ "loss": 0.925,
998
+ "step": 1650
999
+ },
1000
+ {
1001
+ "epoch": 13.28,
1002
+ "learning_rate": 3.991304347826087e-05,
1003
+ "loss": 0.8803,
1004
+ "step": 1660
1005
+ },
1006
+ {
1007
+ "epoch": 13.36,
1008
+ "learning_rate": 3.9826086956521744e-05,
1009
+ "loss": 0.9284,
1010
+ "step": 1670
1011
+ },
1012
+ {
1013
+ "epoch": 13.44,
1014
+ "learning_rate": 3.973913043478261e-05,
1015
+ "loss": 0.9985,
1016
+ "step": 1680
1017
+ },
1018
+ {
1019
+ "epoch": 13.52,
1020
+ "learning_rate": 3.965217391304348e-05,
1021
+ "loss": 1.0461,
1022
+ "step": 1690
1023
+ },
1024
+ {
1025
+ "epoch": 13.6,
1026
+ "learning_rate": 3.956521739130435e-05,
1027
+ "loss": 0.9542,
1028
+ "step": 1700
1029
+ },
1030
+ {
1031
+ "epoch": 13.68,
1032
+ "learning_rate": 3.947826086956522e-05,
1033
+ "loss": 0.9822,
1034
+ "step": 1710
1035
+ },
1036
+ {
1037
+ "epoch": 13.76,
1038
+ "learning_rate": 3.939130434782609e-05,
1039
+ "loss": 1.0781,
1040
+ "step": 1720
1041
+ },
1042
+ {
1043
+ "epoch": 13.84,
1044
+ "learning_rate": 3.930434782608696e-05,
1045
+ "loss": 1.0549,
1046
+ "step": 1730
1047
+ },
1048
+ {
1049
+ "epoch": 13.92,
1050
+ "learning_rate": 3.9217391304347825e-05,
1051
+ "loss": 1.0676,
1052
+ "step": 1740
1053
+ },
1054
+ {
1055
+ "epoch": 14.0,
1056
+ "learning_rate": 3.91304347826087e-05,
1057
+ "loss": 1.1258,
1058
+ "step": 1750
1059
+ },
1060
+ {
1061
+ "epoch": 14.08,
1062
+ "learning_rate": 3.9043478260869566e-05,
1063
+ "loss": 0.8544,
1064
+ "step": 1760
1065
+ },
1066
+ {
1067
+ "epoch": 14.16,
1068
+ "learning_rate": 3.895652173913043e-05,
1069
+ "loss": 0.9229,
1070
+ "step": 1770
1071
+ },
1072
+ {
1073
+ "epoch": 14.24,
1074
+ "learning_rate": 3.8869565217391306e-05,
1075
+ "loss": 0.9222,
1076
+ "step": 1780
1077
+ },
1078
+ {
1079
+ "epoch": 14.32,
1080
+ "learning_rate": 3.878260869565218e-05,
1081
+ "loss": 0.9293,
1082
+ "step": 1790
1083
+ },
1084
+ {
1085
+ "epoch": 14.4,
1086
+ "learning_rate": 3.869565217391305e-05,
1087
+ "loss": 0.9189,
1088
+ "step": 1800
1089
+ },
1090
+ {
1091
+ "epoch": 14.48,
1092
+ "learning_rate": 3.8608695652173914e-05,
1093
+ "loss": 0.9336,
1094
+ "step": 1810
1095
+ },
1096
+ {
1097
+ "epoch": 14.56,
1098
+ "learning_rate": 3.852173913043479e-05,
1099
+ "loss": 0.9787,
1100
+ "step": 1820
1101
+ },
1102
+ {
1103
+ "epoch": 14.64,
1104
+ "learning_rate": 3.8434782608695654e-05,
1105
+ "loss": 1.0282,
1106
+ "step": 1830
1107
+ },
1108
+ {
1109
+ "epoch": 14.72,
1110
+ "learning_rate": 3.834782608695652e-05,
1111
+ "loss": 0.9987,
1112
+ "step": 1840
1113
+ },
1114
+ {
1115
+ "epoch": 14.8,
1116
+ "learning_rate": 3.8260869565217395e-05,
1117
+ "loss": 1.0231,
1118
+ "step": 1850
1119
+ },
1120
+ {
1121
+ "epoch": 14.88,
1122
+ "learning_rate": 3.817391304347827e-05,
1123
+ "loss": 1.0009,
1124
+ "step": 1860
1125
+ },
1126
+ {
1127
+ "epoch": 14.96,
1128
+ "learning_rate": 3.808695652173913e-05,
1129
+ "loss": 1.0475,
1130
+ "step": 1870
1131
+ },
1132
+ {
1133
+ "epoch": 15.04,
1134
+ "learning_rate": 3.8e-05,
1135
+ "loss": 0.9605,
1136
+ "step": 1880
1137
+ },
1138
+ {
1139
+ "epoch": 15.12,
1140
+ "learning_rate": 3.7913043478260876e-05,
1141
+ "loss": 0.8575,
1142
+ "step": 1890
1143
+ },
1144
+ {
1145
+ "epoch": 15.2,
1146
+ "learning_rate": 3.7826086956521736e-05,
1147
+ "loss": 0.8875,
1148
+ "step": 1900
1149
+ },
1150
+ {
1151
+ "epoch": 15.28,
1152
+ "learning_rate": 3.773913043478261e-05,
1153
+ "loss": 0.9066,
1154
+ "step": 1910
1155
+ },
1156
+ {
1157
+ "epoch": 15.36,
1158
+ "learning_rate": 3.765217391304348e-05,
1159
+ "loss": 0.9074,
1160
+ "step": 1920
1161
+ },
1162
+ {
1163
+ "epoch": 15.44,
1164
+ "learning_rate": 3.756521739130435e-05,
1165
+ "loss": 0.9326,
1166
+ "step": 1930
1167
+ },
1168
+ {
1169
+ "epoch": 15.52,
1170
+ "learning_rate": 3.747826086956522e-05,
1171
+ "loss": 0.8979,
1172
+ "step": 1940
1173
+ },
1174
+ {
1175
+ "epoch": 15.6,
1176
+ "learning_rate": 3.739130434782609e-05,
1177
+ "loss": 0.9302,
1178
+ "step": 1950
1179
+ },
1180
+ {
1181
+ "epoch": 15.68,
1182
+ "learning_rate": 3.730434782608696e-05,
1183
+ "loss": 0.9764,
1184
+ "step": 1960
1185
+ },
1186
+ {
1187
+ "epoch": 15.76,
1188
+ "learning_rate": 3.7217391304347824e-05,
1189
+ "loss": 0.9775,
1190
+ "step": 1970
1191
+ },
1192
+ {
1193
+ "epoch": 15.84,
1194
+ "learning_rate": 3.71304347826087e-05,
1195
+ "loss": 0.9807,
1196
+ "step": 1980
1197
+ },
1198
+ {
1199
+ "epoch": 15.92,
1200
+ "learning_rate": 3.704347826086957e-05,
1201
+ "loss": 1.0364,
1202
+ "step": 1990
1203
+ },
1204
+ {
1205
+ "epoch": 16.0,
1206
+ "learning_rate": 3.695652173913043e-05,
1207
+ "loss": 1.0756,
1208
+ "step": 2000
1209
+ },
1210
+ {
1211
+ "epoch": 16.08,
1212
+ "learning_rate": 3.6869565217391305e-05,
1213
+ "loss": 0.8401,
1214
+ "step": 2010
1215
+ },
1216
+ {
1217
+ "epoch": 16.16,
1218
+ "learning_rate": 3.678260869565218e-05,
1219
+ "loss": 0.8194,
1220
+ "step": 2020
1221
+ },
1222
+ {
1223
+ "epoch": 16.24,
1224
+ "learning_rate": 3.6695652173913046e-05,
1225
+ "loss": 0.9139,
1226
+ "step": 2030
1227
+ },
1228
+ {
1229
+ "epoch": 16.32,
1230
+ "learning_rate": 3.660869565217391e-05,
1231
+ "loss": 0.8998,
1232
+ "step": 2040
1233
+ },
1234
+ {
1235
+ "epoch": 16.4,
1236
+ "learning_rate": 3.6521739130434786e-05,
1237
+ "loss": 0.9026,
1238
+ "step": 2050
1239
+ },
1240
+ {
1241
+ "epoch": 16.48,
1242
+ "learning_rate": 3.643478260869565e-05,
1243
+ "loss": 0.9809,
1244
+ "step": 2060
1245
+ },
1246
+ {
1247
+ "epoch": 16.56,
1248
+ "learning_rate": 3.634782608695652e-05,
1249
+ "loss": 0.9175,
1250
+ "step": 2070
1251
+ },
1252
+ {
1253
+ "epoch": 16.64,
1254
+ "learning_rate": 3.6260869565217394e-05,
1255
+ "loss": 0.953,
1256
+ "step": 2080
1257
+ },
1258
+ {
1259
+ "epoch": 16.72,
1260
+ "learning_rate": 3.617391304347826e-05,
1261
+ "loss": 0.9542,
1262
+ "step": 2090
1263
+ },
1264
+ {
1265
+ "epoch": 16.8,
1266
+ "learning_rate": 3.6086956521739134e-05,
1267
+ "loss": 1.0157,
1268
+ "step": 2100
1269
+ },
1270
+ {
1271
+ "epoch": 16.88,
1272
+ "learning_rate": 3.6e-05,
1273
+ "loss": 0.9836,
1274
+ "step": 2110
1275
+ },
1276
+ {
1277
+ "epoch": 16.96,
1278
+ "learning_rate": 3.5913043478260875e-05,
1279
+ "loss": 1.043,
1280
+ "step": 2120
1281
+ },
1282
+ {
1283
+ "epoch": 17.04,
1284
+ "learning_rate": 3.582608695652174e-05,
1285
+ "loss": 0.8193,
1286
+ "step": 2130
1287
+ },
1288
+ {
1289
+ "epoch": 17.12,
1290
+ "learning_rate": 3.573913043478261e-05,
1291
+ "loss": 0.8391,
1292
+ "step": 2140
1293
+ },
1294
+ {
1295
+ "epoch": 17.2,
1296
+ "learning_rate": 3.565217391304348e-05,
1297
+ "loss": 0.8211,
1298
+ "step": 2150
1299
+ },
1300
+ {
1301
+ "epoch": 17.28,
1302
+ "learning_rate": 3.556521739130435e-05,
1303
+ "loss": 0.8447,
1304
+ "step": 2160
1305
+ },
1306
+ {
1307
+ "epoch": 17.36,
1308
+ "learning_rate": 3.5478260869565216e-05,
1309
+ "loss": 0.8472,
1310
+ "step": 2170
1311
+ },
1312
+ {
1313
+ "epoch": 17.44,
1314
+ "learning_rate": 3.539130434782609e-05,
1315
+ "loss": 0.8897,
1316
+ "step": 2180
1317
+ },
1318
+ {
1319
+ "epoch": 17.52,
1320
+ "learning_rate": 3.5304347826086956e-05,
1321
+ "loss": 0.8916,
1322
+ "step": 2190
1323
+ },
1324
+ {
1325
+ "epoch": 17.6,
1326
+ "learning_rate": 3.521739130434783e-05,
1327
+ "loss": 0.9339,
1328
+ "step": 2200
1329
+ },
1330
+ {
1331
+ "epoch": 17.68,
1332
+ "learning_rate": 3.51304347826087e-05,
1333
+ "loss": 0.9641,
1334
+ "step": 2210
1335
+ },
1336
+ {
1337
+ "epoch": 17.76,
1338
+ "learning_rate": 3.5043478260869564e-05,
1339
+ "loss": 0.9489,
1340
+ "step": 2220
1341
+ },
1342
+ {
1343
+ "epoch": 17.84,
1344
+ "learning_rate": 3.495652173913044e-05,
1345
+ "loss": 0.9923,
1346
+ "step": 2230
1347
+ },
1348
+ {
1349
+ "epoch": 17.92,
1350
+ "learning_rate": 3.4869565217391304e-05,
1351
+ "loss": 0.954,
1352
+ "step": 2240
1353
+ },
1354
+ {
1355
+ "epoch": 18.0,
1356
+ "learning_rate": 3.478260869565218e-05,
1357
+ "loss": 0.9964,
1358
+ "step": 2250
1359
+ },
1360
+ {
1361
+ "epoch": 18.08,
1362
+ "learning_rate": 3.4695652173913045e-05,
1363
+ "loss": 0.8709,
1364
+ "step": 2260
1365
+ },
1366
+ {
1367
+ "epoch": 18.16,
1368
+ "learning_rate": 3.460869565217392e-05,
1369
+ "loss": 0.8586,
1370
+ "step": 2270
1371
+ },
1372
+ {
1373
+ "epoch": 18.24,
1374
+ "learning_rate": 3.4521739130434785e-05,
1375
+ "loss": 0.799,
1376
+ "step": 2280
1377
+ },
1378
+ {
1379
+ "epoch": 18.32,
1380
+ "learning_rate": 3.443478260869565e-05,
1381
+ "loss": 0.7715,
1382
+ "step": 2290
1383
+ },
1384
+ {
1385
+ "epoch": 18.4,
1386
+ "learning_rate": 3.4347826086956526e-05,
1387
+ "loss": 0.8854,
1388
+ "step": 2300
1389
+ },
1390
+ {
1391
+ "epoch": 18.48,
1392
+ "learning_rate": 3.426086956521739e-05,
1393
+ "loss": 0.9446,
1394
+ "step": 2310
1395
+ },
1396
+ {
1397
+ "epoch": 18.56,
1398
+ "learning_rate": 3.417391304347826e-05,
1399
+ "loss": 0.892,
1400
+ "step": 2320
1401
+ },
1402
+ {
1403
+ "epoch": 18.64,
1404
+ "learning_rate": 3.408695652173913e-05,
1405
+ "loss": 0.9439,
1406
+ "step": 2330
1407
+ },
1408
+ {
1409
+ "epoch": 18.72,
1410
+ "learning_rate": 3.4000000000000007e-05,
1411
+ "loss": 0.8857,
1412
+ "step": 2340
1413
+ },
1414
+ {
1415
+ "epoch": 18.8,
1416
+ "learning_rate": 3.3913043478260867e-05,
1417
+ "loss": 0.9497,
1418
+ "step": 2350
1419
+ },
1420
+ {
1421
+ "epoch": 18.88,
1422
+ "learning_rate": 3.382608695652174e-05,
1423
+ "loss": 0.9948,
1424
+ "step": 2360
1425
+ },
1426
+ {
1427
+ "epoch": 18.96,
1428
+ "learning_rate": 3.3739130434782614e-05,
1429
+ "loss": 0.9594,
1430
+ "step": 2370
1431
+ },
1432
+ {
1433
+ "epoch": 19.04,
1434
+ "learning_rate": 3.365217391304348e-05,
1435
+ "loss": 0.9142,
1436
+ "step": 2380
1437
+ },
1438
+ {
1439
+ "epoch": 19.12,
1440
+ "learning_rate": 3.356521739130435e-05,
1441
+ "loss": 0.8201,
1442
+ "step": 2390
1443
+ },
1444
+ {
1445
+ "epoch": 19.2,
1446
+ "learning_rate": 3.347826086956522e-05,
1447
+ "loss": 0.8425,
1448
+ "step": 2400
1449
+ },
1450
+ {
1451
+ "epoch": 19.28,
1452
+ "learning_rate": 3.339130434782609e-05,
1453
+ "loss": 0.7843,
1454
+ "step": 2410
1455
+ },
1456
+ {
1457
+ "epoch": 19.36,
1458
+ "learning_rate": 3.3304347826086955e-05,
1459
+ "loss": 0.9149,
1460
+ "step": 2420
1461
+ },
1462
+ {
1463
+ "epoch": 19.44,
1464
+ "learning_rate": 3.321739130434783e-05,
1465
+ "loss": 0.7852,
1466
+ "step": 2430
1467
+ },
1468
+ {
1469
+ "epoch": 19.52,
1470
+ "learning_rate": 3.31304347826087e-05,
1471
+ "loss": 0.8849,
1472
+ "step": 2440
1473
+ },
1474
+ {
1475
+ "epoch": 19.6,
1476
+ "learning_rate": 3.304347826086956e-05,
1477
+ "loss": 0.9418,
1478
+ "step": 2450
1479
+ },
1480
+ {
1481
+ "epoch": 19.68,
1482
+ "learning_rate": 3.2956521739130436e-05,
1483
+ "loss": 0.938,
1484
+ "step": 2460
1485
+ },
1486
+ {
1487
+ "epoch": 19.76,
1488
+ "learning_rate": 3.286956521739131e-05,
1489
+ "loss": 0.9208,
1490
+ "step": 2470
1491
+ },
1492
+ {
1493
+ "epoch": 19.84,
1494
+ "learning_rate": 3.278260869565217e-05,
1495
+ "loss": 0.9258,
1496
+ "step": 2480
1497
+ },
1498
+ {
1499
+ "epoch": 19.92,
1500
+ "learning_rate": 3.269565217391304e-05,
1501
+ "loss": 0.8696,
1502
+ "step": 2490
1503
+ },
1504
+ {
1505
+ "epoch": 20.0,
1506
+ "learning_rate": 3.260869565217392e-05,
1507
+ "loss": 0.9085,
1508
+ "step": 2500
1509
+ },
1510
+ {
1511
+ "epoch": 20.08,
1512
+ "learning_rate": 3.2521739130434784e-05,
1513
+ "loss": 0.8414,
1514
+ "step": 2510
1515
+ },
1516
+ {
1517
+ "epoch": 20.16,
1518
+ "learning_rate": 3.243478260869565e-05,
1519
+ "loss": 0.8509,
1520
+ "step": 2520
1521
+ },
1522
+ {
1523
+ "epoch": 20.24,
1524
+ "learning_rate": 3.2347826086956524e-05,
1525
+ "loss": 0.7594,
1526
+ "step": 2530
1527
+ },
1528
+ {
1529
+ "epoch": 20.32,
1530
+ "learning_rate": 3.226086956521739e-05,
1531
+ "loss": 0.8413,
1532
+ "step": 2540
1533
+ },
1534
+ {
1535
+ "epoch": 20.4,
1536
+ "learning_rate": 3.217391304347826e-05,
1537
+ "loss": 0.8918,
1538
+ "step": 2550
1539
+ },
1540
+ {
1541
+ "epoch": 20.48,
1542
+ "learning_rate": 3.208695652173913e-05,
1543
+ "loss": 0.8074,
1544
+ "step": 2560
1545
+ },
1546
+ {
1547
+ "epoch": 20.56,
1548
+ "learning_rate": 3.2000000000000005e-05,
1549
+ "loss": 0.8728,
1550
+ "step": 2570
1551
+ },
1552
+ {
1553
+ "epoch": 20.64,
1554
+ "learning_rate": 3.191304347826087e-05,
1555
+ "loss": 0.8559,
1556
+ "step": 2580
1557
+ },
1558
+ {
1559
+ "epoch": 20.72,
1560
+ "learning_rate": 3.182608695652174e-05,
1561
+ "loss": 0.8778,
1562
+ "step": 2590
1563
+ },
1564
+ {
1565
+ "epoch": 20.8,
1566
+ "learning_rate": 3.173913043478261e-05,
1567
+ "loss": 0.9155,
1568
+ "step": 2600
1569
+ },
1570
+ {
1571
+ "epoch": 20.88,
1572
+ "learning_rate": 3.165217391304348e-05,
1573
+ "loss": 0.9692,
1574
+ "step": 2610
1575
+ },
1576
+ {
1577
+ "epoch": 20.96,
1578
+ "learning_rate": 3.1565217391304346e-05,
1579
+ "loss": 0.9685,
1580
+ "step": 2620
1581
+ },
1582
+ {
1583
+ "epoch": 21.04,
1584
+ "learning_rate": 3.147826086956522e-05,
1585
+ "loss": 0.9113,
1586
+ "step": 2630
1587
+ },
1588
+ {
1589
+ "epoch": 21.12,
1590
+ "learning_rate": 3.139130434782609e-05,
1591
+ "loss": 0.7723,
1592
+ "step": 2640
1593
+ },
1594
+ {
1595
+ "epoch": 21.2,
1596
+ "learning_rate": 3.130434782608696e-05,
1597
+ "loss": 0.8294,
1598
+ "step": 2650
1599
+ },
1600
+ {
1601
+ "epoch": 21.28,
1602
+ "learning_rate": 3.121739130434783e-05,
1603
+ "loss": 0.7949,
1604
+ "step": 2660
1605
+ },
1606
+ {
1607
+ "epoch": 21.36,
1608
+ "learning_rate": 3.1130434782608694e-05,
1609
+ "loss": 0.8509,
1610
+ "step": 2670
1611
+ },
1612
+ {
1613
+ "epoch": 21.44,
1614
+ "learning_rate": 3.104347826086957e-05,
1615
+ "loss": 0.8727,
1616
+ "step": 2680
1617
+ },
1618
+ {
1619
+ "epoch": 21.52,
1620
+ "learning_rate": 3.0956521739130435e-05,
1621
+ "loss": 0.8094,
1622
+ "step": 2690
1623
+ },
1624
+ {
1625
+ "epoch": 21.6,
1626
+ "learning_rate": 3.086956521739131e-05,
1627
+ "loss": 0.8458,
1628
+ "step": 2700
1629
+ },
1630
+ {
1631
+ "epoch": 21.68,
1632
+ "learning_rate": 3.0782608695652175e-05,
1633
+ "loss": 0.8416,
1634
+ "step": 2710
1635
+ },
1636
+ {
1637
+ "epoch": 21.76,
1638
+ "learning_rate": 3.069565217391305e-05,
1639
+ "loss": 0.8336,
1640
+ "step": 2720
1641
+ },
1642
+ {
1643
+ "epoch": 21.84,
1644
+ "learning_rate": 3.0608695652173916e-05,
1645
+ "loss": 0.9844,
1646
+ "step": 2730
1647
+ },
1648
+ {
1649
+ "epoch": 21.92,
1650
+ "learning_rate": 3.052173913043478e-05,
1651
+ "loss": 0.8814,
1652
+ "step": 2740
1653
+ },
1654
+ {
1655
+ "epoch": 22.0,
1656
+ "learning_rate": 3.0434782608695656e-05,
1657
+ "loss": 0.917,
1658
+ "step": 2750
1659
+ },
1660
+ {
1661
+ "epoch": 22.08,
1662
+ "learning_rate": 3.034782608695652e-05,
1663
+ "loss": 0.8319,
1664
+ "step": 2760
1665
+ },
1666
+ {
1667
+ "epoch": 22.16,
1668
+ "learning_rate": 3.0260869565217393e-05,
1669
+ "loss": 0.7807,
1670
+ "step": 2770
1671
+ },
1672
+ {
1673
+ "epoch": 22.24,
1674
+ "learning_rate": 3.0173913043478264e-05,
1675
+ "loss": 0.8231,
1676
+ "step": 2780
1677
+ },
1678
+ {
1679
+ "epoch": 22.32,
1680
+ "learning_rate": 3.008695652173913e-05,
1681
+ "loss": 0.7917,
1682
+ "step": 2790
1683
+ },
1684
+ {
1685
+ "epoch": 22.4,
1686
+ "learning_rate": 3e-05,
1687
+ "loss": 0.8658,
1688
+ "step": 2800
1689
+ },
1690
+ {
1691
+ "epoch": 22.48,
1692
+ "learning_rate": 2.991304347826087e-05,
1693
+ "loss": 0.8434,
1694
+ "step": 2810
1695
+ },
1696
+ {
1697
+ "epoch": 22.56,
1698
+ "learning_rate": 2.982608695652174e-05,
1699
+ "loss": 0.7944,
1700
+ "step": 2820
1701
+ },
1702
+ {
1703
+ "epoch": 22.64,
1704
+ "learning_rate": 2.9739130434782608e-05,
1705
+ "loss": 0.8425,
1706
+ "step": 2830
1707
+ },
1708
+ {
1709
+ "epoch": 22.72,
1710
+ "learning_rate": 2.965217391304348e-05,
1711
+ "loss": 0.9204,
1712
+ "step": 2840
1713
+ },
1714
+ {
1715
+ "epoch": 22.8,
1716
+ "learning_rate": 2.9565217391304352e-05,
1717
+ "loss": 0.8919,
1718
+ "step": 2850
1719
+ },
1720
+ {
1721
+ "epoch": 22.88,
1722
+ "learning_rate": 2.9478260869565215e-05,
1723
+ "loss": 0.8238,
1724
+ "step": 2860
1725
+ },
1726
+ {
1727
+ "epoch": 22.96,
1728
+ "learning_rate": 2.939130434782609e-05,
1729
+ "loss": 0.8917,
1730
+ "step": 2870
1731
+ },
1732
+ {
1733
+ "epoch": 23.04,
1734
+ "learning_rate": 2.930434782608696e-05,
1735
+ "loss": 0.886,
1736
+ "step": 2880
1737
+ },
1738
+ {
1739
+ "epoch": 23.12,
1740
+ "learning_rate": 2.921739130434783e-05,
1741
+ "loss": 0.8199,
1742
+ "step": 2890
1743
+ },
1744
+ {
1745
+ "epoch": 23.2,
1746
+ "learning_rate": 2.9130434782608696e-05,
1747
+ "loss": 0.7968,
1748
+ "step": 2900
1749
+ },
1750
+ {
1751
+ "epoch": 23.28,
1752
+ "learning_rate": 2.9043478260869567e-05,
1753
+ "loss": 0.8332,
1754
+ "step": 2910
1755
+ },
1756
+ {
1757
+ "epoch": 23.36,
1758
+ "learning_rate": 2.8956521739130437e-05,
1759
+ "loss": 0.8201,
1760
+ "step": 2920
1761
+ },
1762
+ {
1763
+ "epoch": 23.44,
1764
+ "learning_rate": 2.8869565217391304e-05,
1765
+ "loss": 0.8029,
1766
+ "step": 2930
1767
+ },
1768
+ {
1769
+ "epoch": 23.52,
1770
+ "learning_rate": 2.8782608695652174e-05,
1771
+ "loss": 0.7901,
1772
+ "step": 2940
1773
+ },
1774
+ {
1775
+ "epoch": 23.6,
1776
+ "learning_rate": 2.8695652173913044e-05,
1777
+ "loss": 0.8657,
1778
+ "step": 2950
1779
+ },
1780
+ {
1781
+ "epoch": 23.68,
1782
+ "learning_rate": 2.8608695652173918e-05,
1783
+ "loss": 0.8878,
1784
+ "step": 2960
1785
+ },
1786
+ {
1787
+ "epoch": 23.76,
1788
+ "learning_rate": 2.852173913043478e-05,
1789
+ "loss": 0.8956,
1790
+ "step": 2970
1791
+ },
1792
+ {
1793
+ "epoch": 23.84,
1794
+ "learning_rate": 2.8434782608695655e-05,
1795
+ "loss": 0.8472,
1796
+ "step": 2980
1797
+ },
1798
+ {
1799
+ "epoch": 23.92,
1800
+ "learning_rate": 2.8347826086956525e-05,
1801
+ "loss": 0.8308,
1802
+ "step": 2990
1803
+ },
1804
+ {
1805
+ "epoch": 24.0,
1806
+ "learning_rate": 2.826086956521739e-05,
1807
+ "loss": 0.8754,
1808
+ "step": 3000
1809
+ },
1810
+ {
1811
+ "epoch": 24.08,
1812
+ "learning_rate": 2.8173913043478262e-05,
1813
+ "loss": 0.7869,
1814
+ "step": 3010
1815
+ },
1816
+ {
1817
+ "epoch": 24.16,
1818
+ "learning_rate": 2.8086956521739133e-05,
1819
+ "loss": 0.7275,
1820
+ "step": 3020
1821
+ },
1822
+ {
1823
+ "epoch": 24.24,
1824
+ "learning_rate": 2.8000000000000003e-05,
1825
+ "loss": 0.7898,
1826
+ "step": 3030
1827
+ },
1828
+ {
1829
+ "epoch": 24.32,
1830
+ "learning_rate": 2.791304347826087e-05,
1831
+ "loss": 0.781,
1832
+ "step": 3040
1833
+ },
1834
+ {
1835
+ "epoch": 24.4,
1836
+ "learning_rate": 2.782608695652174e-05,
1837
+ "loss": 0.8289,
1838
+ "step": 3050
1839
+ },
1840
+ {
1841
+ "epoch": 24.48,
1842
+ "learning_rate": 2.773913043478261e-05,
1843
+ "loss": 0.8367,
1844
+ "step": 3060
1845
+ },
1846
+ {
1847
+ "epoch": 24.56,
1848
+ "learning_rate": 2.7652173913043477e-05,
1849
+ "loss": 0.8201,
1850
+ "step": 3070
1851
+ },
1852
+ {
1853
+ "epoch": 24.64,
1854
+ "learning_rate": 2.7565217391304347e-05,
1855
+ "loss": 0.8339,
1856
+ "step": 3080
1857
+ },
1858
+ {
1859
+ "epoch": 24.72,
1860
+ "learning_rate": 2.747826086956522e-05,
1861
+ "loss": 0.8857,
1862
+ "step": 3090
1863
+ },
1864
+ {
1865
+ "epoch": 24.8,
1866
+ "learning_rate": 2.7391304347826085e-05,
1867
+ "loss": 0.8688,
1868
+ "step": 3100
1869
+ },
1870
+ {
1871
+ "epoch": 24.88,
1872
+ "learning_rate": 2.7304347826086958e-05,
1873
+ "loss": 0.8792,
1874
+ "step": 3110
1875
+ },
1876
+ {
1877
+ "epoch": 24.96,
1878
+ "learning_rate": 2.721739130434783e-05,
1879
+ "loss": 0.8835,
1880
+ "step": 3120
1881
+ },
1882
+ {
1883
+ "epoch": 25.04,
1884
+ "learning_rate": 2.71304347826087e-05,
1885
+ "loss": 0.8183,
1886
+ "step": 3130
1887
+ },
1888
+ {
1889
+ "epoch": 25.12,
1890
+ "learning_rate": 2.7043478260869566e-05,
1891
+ "loss": 0.772,
1892
+ "step": 3140
1893
+ },
1894
+ {
1895
+ "epoch": 25.2,
1896
+ "learning_rate": 2.6956521739130436e-05,
1897
+ "loss": 0.7913,
1898
+ "step": 3150
1899
+ },
1900
+ {
1901
+ "epoch": 25.28,
1902
+ "learning_rate": 2.6869565217391306e-05,
1903
+ "loss": 0.782,
1904
+ "step": 3160
1905
+ },
1906
+ {
1907
+ "epoch": 25.36,
1908
+ "learning_rate": 2.6782608695652173e-05,
1909
+ "loss": 0.7693,
1910
+ "step": 3170
1911
+ },
1912
+ {
1913
+ "epoch": 25.44,
1914
+ "learning_rate": 2.6695652173913043e-05,
1915
+ "loss": 0.814,
1916
+ "step": 3180
1917
+ },
1918
+ {
1919
+ "epoch": 25.52,
1920
+ "learning_rate": 2.6608695652173913e-05,
1921
+ "loss": 0.771,
1922
+ "step": 3190
1923
+ },
1924
+ {
1925
+ "epoch": 25.6,
1926
+ "learning_rate": 2.6521739130434787e-05,
1927
+ "loss": 0.8702,
1928
+ "step": 3200
1929
+ },
1930
+ {
1931
+ "epoch": 25.68,
1932
+ "learning_rate": 2.643478260869565e-05,
1933
+ "loss": 0.8653,
1934
+ "step": 3210
1935
+ },
1936
+ {
1937
+ "epoch": 25.76,
1938
+ "learning_rate": 2.6347826086956524e-05,
1939
+ "loss": 0.8006,
1940
+ "step": 3220
1941
+ },
1942
+ {
1943
+ "epoch": 25.84,
1944
+ "learning_rate": 2.6260869565217394e-05,
1945
+ "loss": 0.8638,
1946
+ "step": 3230
1947
+ },
1948
+ {
1949
+ "epoch": 25.92,
1950
+ "learning_rate": 2.617391304347826e-05,
1951
+ "loss": 0.8425,
1952
+ "step": 3240
1953
+ },
1954
+ {
1955
+ "epoch": 26.0,
1956
+ "learning_rate": 2.608695652173913e-05,
1957
+ "loss": 0.8959,
1958
+ "step": 3250
1959
+ },
1960
+ {
1961
+ "epoch": 26.08,
1962
+ "learning_rate": 2.6000000000000002e-05,
1963
+ "loss": 0.7395,
1964
+ "step": 3260
1965
+ },
1966
+ {
1967
+ "epoch": 26.16,
1968
+ "learning_rate": 2.5913043478260872e-05,
1969
+ "loss": 0.7737,
1970
+ "step": 3270
1971
+ },
1972
+ {
1973
+ "epoch": 26.24,
1974
+ "learning_rate": 2.582608695652174e-05,
1975
+ "loss": 0.7703,
1976
+ "step": 3280
1977
+ },
1978
+ {
1979
+ "epoch": 26.32,
1980
+ "learning_rate": 2.573913043478261e-05,
1981
+ "loss": 0.7901,
1982
+ "step": 3290
1983
+ },
1984
+ {
1985
+ "epoch": 26.4,
1986
+ "learning_rate": 2.5652173913043483e-05,
1987
+ "loss": 0.8004,
1988
+ "step": 3300
1989
+ },
1990
+ {
1991
+ "epoch": 26.48,
1992
+ "learning_rate": 2.5565217391304346e-05,
1993
+ "loss": 0.8087,
1994
+ "step": 3310
1995
+ },
1996
+ {
1997
+ "epoch": 26.56,
1998
+ "learning_rate": 2.5478260869565217e-05,
1999
+ "loss": 0.8025,
2000
+ "step": 3320
2001
+ },
2002
+ {
2003
+ "epoch": 26.64,
2004
+ "learning_rate": 2.539130434782609e-05,
2005
+ "loss": 0.8545,
2006
+ "step": 3330
2007
+ },
2008
+ {
2009
+ "epoch": 26.72,
2010
+ "learning_rate": 2.530434782608696e-05,
2011
+ "loss": 0.8215,
2012
+ "step": 3340
2013
+ },
2014
+ {
2015
+ "epoch": 26.8,
2016
+ "learning_rate": 2.5217391304347827e-05,
2017
+ "loss": 0.7817,
2018
+ "step": 3350
2019
+ },
2020
+ {
2021
+ "epoch": 26.88,
2022
+ "learning_rate": 2.5130434782608698e-05,
2023
+ "loss": 0.8443,
2024
+ "step": 3360
2025
+ },
2026
+ {
2027
+ "epoch": 26.96,
2028
+ "learning_rate": 2.5043478260869568e-05,
2029
+ "loss": 0.8381,
2030
+ "step": 3370
2031
+ },
2032
+ {
2033
+ "epoch": 27.04,
2034
+ "learning_rate": 2.4956521739130438e-05,
2035
+ "loss": 0.7931,
2036
+ "step": 3380
2037
+ },
2038
+ {
2039
+ "epoch": 27.12,
2040
+ "learning_rate": 2.4869565217391305e-05,
2041
+ "loss": 0.7931,
2042
+ "step": 3390
2043
+ },
2044
+ {
2045
+ "epoch": 27.2,
2046
+ "learning_rate": 2.4782608695652175e-05,
2047
+ "loss": 0.729,
2048
+ "step": 3400
2049
+ },
2050
+ {
2051
+ "epoch": 27.28,
2052
+ "learning_rate": 2.4695652173913045e-05,
2053
+ "loss": 0.8028,
2054
+ "step": 3410
2055
+ },
2056
+ {
2057
+ "epoch": 27.36,
2058
+ "learning_rate": 2.4608695652173912e-05,
2059
+ "loss": 0.7529,
2060
+ "step": 3420
2061
+ },
2062
+ {
2063
+ "epoch": 27.44,
2064
+ "learning_rate": 2.4521739130434786e-05,
2065
+ "loss": 0.825,
2066
+ "step": 3430
2067
+ },
2068
+ {
2069
+ "epoch": 27.52,
2070
+ "learning_rate": 2.4434782608695653e-05,
2071
+ "loss": 0.785,
2072
+ "step": 3440
2073
+ },
2074
+ {
2075
+ "epoch": 27.6,
2076
+ "learning_rate": 2.4347826086956523e-05,
2077
+ "loss": 0.8408,
2078
+ "step": 3450
2079
+ },
2080
+ {
2081
+ "epoch": 27.68,
2082
+ "learning_rate": 2.4260869565217393e-05,
2083
+ "loss": 0.8812,
2084
+ "step": 3460
2085
+ },
2086
+ {
2087
+ "epoch": 27.76,
2088
+ "learning_rate": 2.417391304347826e-05,
2089
+ "loss": 0.8036,
2090
+ "step": 3470
2091
+ },
2092
+ {
2093
+ "epoch": 27.84,
2094
+ "learning_rate": 2.408695652173913e-05,
2095
+ "loss": 0.8844,
2096
+ "step": 3480
2097
+ },
2098
+ {
2099
+ "epoch": 27.92,
2100
+ "learning_rate": 2.4e-05,
2101
+ "loss": 0.8191,
2102
+ "step": 3490
2103
+ },
2104
+ {
2105
+ "epoch": 28.0,
2106
+ "learning_rate": 2.391304347826087e-05,
2107
+ "loss": 0.8419,
2108
+ "step": 3500
2109
+ },
2110
+ {
2111
+ "epoch": 28.08,
2112
+ "learning_rate": 2.382608695652174e-05,
2113
+ "loss": 0.7546,
2114
+ "step": 3510
2115
+ },
2116
+ {
2117
+ "epoch": 28.16,
2118
+ "learning_rate": 2.373913043478261e-05,
2119
+ "loss": 0.74,
2120
+ "step": 3520
2121
+ },
2122
+ {
2123
+ "epoch": 28.24,
2124
+ "learning_rate": 2.3652173913043478e-05,
2125
+ "loss": 0.7936,
2126
+ "step": 3530
2127
+ },
2128
+ {
2129
+ "epoch": 28.32,
2130
+ "learning_rate": 2.356521739130435e-05,
2131
+ "loss": 0.7558,
2132
+ "step": 3540
2133
+ },
2134
+ {
2135
+ "epoch": 28.4,
2136
+ "learning_rate": 2.347826086956522e-05,
2137
+ "loss": 0.8113,
2138
+ "step": 3550
2139
+ },
2140
+ {
2141
+ "epoch": 28.48,
2142
+ "learning_rate": 2.339130434782609e-05,
2143
+ "loss": 0.793,
2144
+ "step": 3560
2145
+ },
2146
+ {
2147
+ "epoch": 28.56,
2148
+ "learning_rate": 2.330434782608696e-05,
2149
+ "loss": 0.7783,
2150
+ "step": 3570
2151
+ },
2152
+ {
2153
+ "epoch": 28.64,
2154
+ "learning_rate": 2.3217391304347826e-05,
2155
+ "loss": 0.825,
2156
+ "step": 3580
2157
+ },
2158
+ {
2159
+ "epoch": 28.72,
2160
+ "learning_rate": 2.31304347826087e-05,
2161
+ "loss": 0.7734,
2162
+ "step": 3590
2163
+ },
2164
+ {
2165
+ "epoch": 28.8,
2166
+ "learning_rate": 2.3043478260869567e-05,
2167
+ "loss": 0.7998,
2168
+ "step": 3600
2169
+ },
2170
+ {
2171
+ "epoch": 28.88,
2172
+ "learning_rate": 2.2956521739130433e-05,
2173
+ "loss": 0.8553,
2174
+ "step": 3610
2175
+ },
2176
+ {
2177
+ "epoch": 28.96,
2178
+ "learning_rate": 2.2869565217391307e-05,
2179
+ "loss": 0.8441,
2180
+ "step": 3620
2181
+ },
2182
+ {
2183
+ "epoch": 29.04,
2184
+ "learning_rate": 2.2782608695652174e-05,
2185
+ "loss": 0.7547,
2186
+ "step": 3630
2187
+ },
2188
+ {
2189
+ "epoch": 29.12,
2190
+ "learning_rate": 2.2695652173913044e-05,
2191
+ "loss": 0.7616,
2192
+ "step": 3640
2193
+ },
2194
+ {
2195
+ "epoch": 29.2,
2196
+ "learning_rate": 2.2608695652173914e-05,
2197
+ "loss": 0.7356,
2198
+ "step": 3650
2199
+ },
2200
+ {
2201
+ "epoch": 29.28,
2202
+ "learning_rate": 2.252173913043478e-05,
2203
+ "loss": 0.739,
2204
+ "step": 3660
2205
+ },
2206
+ {
2207
+ "epoch": 29.36,
2208
+ "learning_rate": 2.2434782608695655e-05,
2209
+ "loss": 0.7591,
2210
+ "step": 3670
2211
+ },
2212
+ {
2213
+ "epoch": 29.44,
2214
+ "learning_rate": 2.2347826086956522e-05,
2215
+ "loss": 0.7885,
2216
+ "step": 3680
2217
+ },
2218
+ {
2219
+ "epoch": 29.52,
2220
+ "learning_rate": 2.2260869565217392e-05,
2221
+ "loss": 0.8358,
2222
+ "step": 3690
2223
+ },
2224
+ {
2225
+ "epoch": 29.6,
2226
+ "learning_rate": 2.2173913043478262e-05,
2227
+ "loss": 0.8253,
2228
+ "step": 3700
2229
+ },
2230
+ {
2231
+ "epoch": 29.68,
2232
+ "learning_rate": 2.2086956521739133e-05,
2233
+ "loss": 0.8004,
2234
+ "step": 3710
2235
+ },
2236
+ {
2237
+ "epoch": 29.76,
2238
+ "learning_rate": 2.2000000000000003e-05,
2239
+ "loss": 0.7941,
2240
+ "step": 3720
2241
+ },
2242
+ {
2243
+ "epoch": 29.84,
2244
+ "learning_rate": 2.191304347826087e-05,
2245
+ "loss": 0.7904,
2246
+ "step": 3730
2247
+ },
2248
+ {
2249
+ "epoch": 29.92,
2250
+ "learning_rate": 2.182608695652174e-05,
2251
+ "loss": 0.8109,
2252
+ "step": 3740
2253
+ },
2254
+ {
2255
+ "epoch": 30.0,
2256
+ "learning_rate": 2.173913043478261e-05,
2257
+ "loss": 0.8528,
2258
+ "step": 3750
2259
+ },
2260
+ {
2261
+ "epoch": 30.08,
2262
+ "learning_rate": 2.165217391304348e-05,
2263
+ "loss": 0.7457,
2264
+ "step": 3760
2265
+ },
2266
+ {
2267
+ "epoch": 30.16,
2268
+ "learning_rate": 2.1565217391304347e-05,
2269
+ "loss": 0.755,
2270
+ "step": 3770
2271
+ },
2272
+ {
2273
+ "epoch": 30.24,
2274
+ "learning_rate": 2.1478260869565218e-05,
2275
+ "loss": 0.7449,
2276
+ "step": 3780
2277
+ },
2278
+ {
2279
+ "epoch": 30.32,
2280
+ "learning_rate": 2.1391304347826088e-05,
2281
+ "loss": 0.7583,
2282
+ "step": 3790
2283
+ },
2284
+ {
2285
+ "epoch": 30.4,
2286
+ "learning_rate": 2.1304347826086958e-05,
2287
+ "loss": 0.7352,
2288
+ "step": 3800
2289
+ },
2290
+ {
2291
+ "epoch": 30.48,
2292
+ "learning_rate": 2.1217391304347828e-05,
2293
+ "loss": 0.7401,
2294
+ "step": 3810
2295
+ },
2296
+ {
2297
+ "epoch": 30.56,
2298
+ "learning_rate": 2.1130434782608695e-05,
2299
+ "loss": 0.7649,
2300
+ "step": 3820
2301
+ },
2302
+ {
2303
+ "epoch": 30.64,
2304
+ "learning_rate": 2.104347826086957e-05,
2305
+ "loss": 0.8294,
2306
+ "step": 3830
2307
+ },
2308
+ {
2309
+ "epoch": 30.72,
2310
+ "learning_rate": 2.0956521739130436e-05,
2311
+ "loss": 0.7539,
2312
+ "step": 3840
2313
+ },
2314
+ {
2315
+ "epoch": 30.8,
2316
+ "learning_rate": 2.0869565217391303e-05,
2317
+ "loss": 0.8142,
2318
+ "step": 3850
2319
+ },
2320
+ {
2321
+ "epoch": 30.88,
2322
+ "learning_rate": 2.0782608695652176e-05,
2323
+ "loss": 0.875,
2324
+ "step": 3860
2325
+ },
2326
+ {
2327
+ "epoch": 30.96,
2328
+ "learning_rate": 2.0695652173913043e-05,
2329
+ "loss": 0.7937,
2330
+ "step": 3870
2331
+ },
2332
+ {
2333
+ "epoch": 31.04,
2334
+ "learning_rate": 2.0608695652173913e-05,
2335
+ "loss": 0.7895,
2336
+ "step": 3880
2337
+ },
2338
+ {
2339
+ "epoch": 31.12,
2340
+ "learning_rate": 2.0521739130434784e-05,
2341
+ "loss": 0.7183,
2342
+ "step": 3890
2343
+ },
2344
+ {
2345
+ "epoch": 31.2,
2346
+ "learning_rate": 2.0434782608695654e-05,
2347
+ "loss": 0.7453,
2348
+ "step": 3900
2349
+ },
2350
+ {
2351
+ "epoch": 31.28,
2352
+ "learning_rate": 2.0347826086956524e-05,
2353
+ "loss": 0.741,
2354
+ "step": 3910
2355
+ },
2356
+ {
2357
+ "epoch": 31.36,
2358
+ "learning_rate": 2.026086956521739e-05,
2359
+ "loss": 0.7427,
2360
+ "step": 3920
2361
+ },
2362
+ {
2363
+ "epoch": 31.44,
2364
+ "learning_rate": 2.017391304347826e-05,
2365
+ "loss": 0.7211,
2366
+ "step": 3930
2367
+ },
2368
+ {
2369
+ "epoch": 31.52,
2370
+ "learning_rate": 2.008695652173913e-05,
2371
+ "loss": 0.7671,
2372
+ "step": 3940
2373
+ },
2374
+ {
2375
+ "epoch": 31.6,
2376
+ "learning_rate": 2e-05,
2377
+ "loss": 0.7455,
2378
+ "step": 3950
2379
+ },
2380
+ {
2381
+ "epoch": 31.68,
2382
+ "learning_rate": 1.9913043478260872e-05,
2383
+ "loss": 0.7925,
2384
+ "step": 3960
2385
+ },
2386
+ {
2387
+ "epoch": 31.76,
2388
+ "learning_rate": 1.982608695652174e-05,
2389
+ "loss": 0.8201,
2390
+ "step": 3970
2391
+ },
2392
+ {
2393
+ "epoch": 31.84,
2394
+ "learning_rate": 1.973913043478261e-05,
2395
+ "loss": 0.8209,
2396
+ "step": 3980
2397
+ },
2398
+ {
2399
+ "epoch": 31.92,
2400
+ "learning_rate": 1.965217391304348e-05,
2401
+ "loss": 0.8644,
2402
+ "step": 3990
2403
+ },
2404
+ {
2405
+ "epoch": 32.0,
2406
+ "learning_rate": 1.956521739130435e-05,
2407
+ "loss": 0.7728,
2408
+ "step": 4000
2409
+ },
2410
+ {
2411
+ "epoch": 32.08,
2412
+ "learning_rate": 1.9478260869565216e-05,
2413
+ "loss": 0.7284,
2414
+ "step": 4010
2415
+ },
2416
+ {
2417
+ "epoch": 32.16,
2418
+ "learning_rate": 1.939130434782609e-05,
2419
+ "loss": 0.7171,
2420
+ "step": 4020
2421
+ },
2422
+ {
2423
+ "epoch": 32.24,
2424
+ "learning_rate": 1.9304347826086957e-05,
2425
+ "loss": 0.7435,
2426
+ "step": 4030
2427
+ },
2428
+ {
2429
+ "epoch": 32.32,
2430
+ "learning_rate": 1.9217391304347827e-05,
2431
+ "loss": 0.736,
2432
+ "step": 4040
2433
+ },
2434
+ {
2435
+ "epoch": 32.4,
2436
+ "learning_rate": 1.9130434782608697e-05,
2437
+ "loss": 0.7511,
2438
+ "step": 4050
2439
+ },
2440
+ {
2441
+ "epoch": 32.48,
2442
+ "learning_rate": 1.9043478260869564e-05,
2443
+ "loss": 0.8181,
2444
+ "step": 4060
2445
+ },
2446
+ {
2447
+ "epoch": 32.56,
2448
+ "learning_rate": 1.8956521739130438e-05,
2449
+ "loss": 0.7707,
2450
+ "step": 4070
2451
+ },
2452
+ {
2453
+ "epoch": 32.64,
2454
+ "learning_rate": 1.8869565217391305e-05,
2455
+ "loss": 0.7859,
2456
+ "step": 4080
2457
+ },
2458
+ {
2459
+ "epoch": 32.72,
2460
+ "learning_rate": 1.8782608695652175e-05,
2461
+ "loss": 0.7545,
2462
+ "step": 4090
2463
+ },
2464
+ {
2465
+ "epoch": 32.8,
2466
+ "learning_rate": 1.8695652173913045e-05,
2467
+ "loss": 0.773,
2468
+ "step": 4100
2469
+ },
2470
+ {
2471
+ "epoch": 32.88,
2472
+ "learning_rate": 1.8608695652173912e-05,
2473
+ "loss": 0.8116,
2474
+ "step": 4110
2475
+ },
2476
+ {
2477
+ "epoch": 32.96,
2478
+ "learning_rate": 1.8521739130434786e-05,
2479
+ "loss": 0.8306,
2480
+ "step": 4120
2481
+ },
2482
+ {
2483
+ "epoch": 33.04,
2484
+ "learning_rate": 1.8434782608695653e-05,
2485
+ "loss": 0.715,
2486
+ "step": 4130
2487
+ },
2488
+ {
2489
+ "epoch": 33.12,
2490
+ "learning_rate": 1.8347826086956523e-05,
2491
+ "loss": 0.7313,
2492
+ "step": 4140
2493
+ },
2494
+ {
2495
+ "epoch": 33.2,
2496
+ "learning_rate": 1.8260869565217393e-05,
2497
+ "loss": 0.7405,
2498
+ "step": 4150
2499
+ },
2500
+ {
2501
+ "epoch": 33.28,
2502
+ "learning_rate": 1.817391304347826e-05,
2503
+ "loss": 0.7378,
2504
+ "step": 4160
2505
+ },
2506
+ {
2507
+ "epoch": 33.36,
2508
+ "learning_rate": 1.808695652173913e-05,
2509
+ "loss": 0.7845,
2510
+ "step": 4170
2511
+ },
2512
+ {
2513
+ "epoch": 33.44,
2514
+ "learning_rate": 1.8e-05,
2515
+ "loss": 0.7204,
2516
+ "step": 4180
2517
+ },
2518
+ {
2519
+ "epoch": 33.52,
2520
+ "learning_rate": 1.791304347826087e-05,
2521
+ "loss": 0.7728,
2522
+ "step": 4190
2523
+ },
2524
+ {
2525
+ "epoch": 33.6,
2526
+ "learning_rate": 1.782608695652174e-05,
2527
+ "loss": 0.8212,
2528
+ "step": 4200
2529
+ },
2530
+ {
2531
+ "epoch": 33.68,
2532
+ "learning_rate": 1.7739130434782608e-05,
2533
+ "loss": 0.7131,
2534
+ "step": 4210
2535
+ },
2536
+ {
2537
+ "epoch": 33.76,
2538
+ "learning_rate": 1.7652173913043478e-05,
2539
+ "loss": 0.7555,
2540
+ "step": 4220
2541
+ },
2542
+ {
2543
+ "epoch": 33.84,
2544
+ "learning_rate": 1.756521739130435e-05,
2545
+ "loss": 0.8008,
2546
+ "step": 4230
2547
+ },
2548
+ {
2549
+ "epoch": 33.92,
2550
+ "learning_rate": 1.747826086956522e-05,
2551
+ "loss": 0.8415,
2552
+ "step": 4240
2553
+ },
2554
+ {
2555
+ "epoch": 34.0,
2556
+ "learning_rate": 1.739130434782609e-05,
2557
+ "loss": 0.7872,
2558
+ "step": 4250
2559
+ },
2560
+ {
2561
+ "epoch": 34.08,
2562
+ "learning_rate": 1.730434782608696e-05,
2563
+ "loss": 0.7312,
2564
+ "step": 4260
2565
+ },
2566
+ {
2567
+ "epoch": 34.16,
2568
+ "learning_rate": 1.7217391304347826e-05,
2569
+ "loss": 0.7106,
2570
+ "step": 4270
2571
+ },
2572
+ {
2573
+ "epoch": 34.24,
2574
+ "learning_rate": 1.7130434782608696e-05,
2575
+ "loss": 0.6981,
2576
+ "step": 4280
2577
+ },
2578
+ {
2579
+ "epoch": 34.32,
2580
+ "learning_rate": 1.7043478260869566e-05,
2581
+ "loss": 0.6903,
2582
+ "step": 4290
2583
+ },
2584
+ {
2585
+ "epoch": 34.4,
2586
+ "learning_rate": 1.6956521739130433e-05,
2587
+ "loss": 0.7726,
2588
+ "step": 4300
2589
+ },
2590
+ {
2591
+ "epoch": 34.48,
2592
+ "learning_rate": 1.6869565217391307e-05,
2593
+ "loss": 0.776,
2594
+ "step": 4310
2595
+ },
2596
+ {
2597
+ "epoch": 34.56,
2598
+ "learning_rate": 1.6782608695652174e-05,
2599
+ "loss": 0.7745,
2600
+ "step": 4320
2601
+ },
2602
+ {
2603
+ "epoch": 34.64,
2604
+ "learning_rate": 1.6695652173913044e-05,
2605
+ "loss": 0.7843,
2606
+ "step": 4330
2607
+ },
2608
+ {
2609
+ "epoch": 34.72,
2610
+ "learning_rate": 1.6608695652173914e-05,
2611
+ "loss": 0.7705,
2612
+ "step": 4340
2613
+ },
2614
+ {
2615
+ "epoch": 34.8,
2616
+ "learning_rate": 1.652173913043478e-05,
2617
+ "loss": 0.8103,
2618
+ "step": 4350
2619
+ },
2620
+ {
2621
+ "epoch": 34.88,
2622
+ "learning_rate": 1.6434782608695655e-05,
2623
+ "loss": 0.793,
2624
+ "step": 4360
2625
+ },
2626
+ {
2627
+ "epoch": 34.96,
2628
+ "learning_rate": 1.634782608695652e-05,
2629
+ "loss": 0.7965,
2630
+ "step": 4370
2631
+ },
2632
+ {
2633
+ "epoch": 35.04,
2634
+ "learning_rate": 1.6260869565217392e-05,
2635
+ "loss": 0.7574,
2636
+ "step": 4380
2637
+ },
2638
+ {
2639
+ "epoch": 35.12,
2640
+ "learning_rate": 1.6173913043478262e-05,
2641
+ "loss": 0.6871,
2642
+ "step": 4390
2643
+ },
2644
+ {
2645
+ "epoch": 35.2,
2646
+ "learning_rate": 1.608695652173913e-05,
2647
+ "loss": 0.7258,
2648
+ "step": 4400
2649
+ },
2650
+ {
2651
+ "epoch": 35.28,
2652
+ "learning_rate": 1.6000000000000003e-05,
2653
+ "loss": 0.7213,
2654
+ "step": 4410
2655
+ },
2656
+ {
2657
+ "epoch": 35.36,
2658
+ "learning_rate": 1.591304347826087e-05,
2659
+ "loss": 0.704,
2660
+ "step": 4420
2661
+ },
2662
+ {
2663
+ "epoch": 35.44,
2664
+ "learning_rate": 1.582608695652174e-05,
2665
+ "loss": 0.7534,
2666
+ "step": 4430
2667
+ },
2668
+ {
2669
+ "epoch": 35.52,
2670
+ "learning_rate": 1.573913043478261e-05,
2671
+ "loss": 0.7666,
2672
+ "step": 4440
2673
+ },
2674
+ {
2675
+ "epoch": 35.6,
2676
+ "learning_rate": 1.565217391304348e-05,
2677
+ "loss": 0.7837,
2678
+ "step": 4450
2679
+ },
2680
+ {
2681
+ "epoch": 35.68,
2682
+ "learning_rate": 1.5565217391304347e-05,
2683
+ "loss": 0.7544,
2684
+ "step": 4460
2685
+ },
2686
+ {
2687
+ "epoch": 35.76,
2688
+ "learning_rate": 1.5478260869565217e-05,
2689
+ "loss": 0.7645,
2690
+ "step": 4470
2691
+ },
2692
+ {
2693
+ "epoch": 35.84,
2694
+ "learning_rate": 1.5391304347826088e-05,
2695
+ "loss": 0.7887,
2696
+ "step": 4480
2697
+ },
2698
+ {
2699
+ "epoch": 35.92,
2700
+ "learning_rate": 1.5304347826086958e-05,
2701
+ "loss": 0.7201,
2702
+ "step": 4490
2703
+ },
2704
+ {
2705
+ "epoch": 36.0,
2706
+ "learning_rate": 1.5217391304347828e-05,
2707
+ "loss": 0.8047,
2708
+ "step": 4500
2709
+ },
2710
+ {
2711
+ "epoch": 36.08,
2712
+ "learning_rate": 1.5130434782608697e-05,
2713
+ "loss": 0.682,
2714
+ "step": 4510
2715
+ },
2716
+ {
2717
+ "epoch": 36.16,
2718
+ "learning_rate": 1.5043478260869565e-05,
2719
+ "loss": 0.7343,
2720
+ "step": 4520
2721
+ },
2722
+ {
2723
+ "epoch": 36.24,
2724
+ "learning_rate": 1.4956521739130436e-05,
2725
+ "loss": 0.7002,
2726
+ "step": 4530
2727
+ },
2728
+ {
2729
+ "epoch": 36.32,
2730
+ "learning_rate": 1.4869565217391304e-05,
2731
+ "loss": 0.7418,
2732
+ "step": 4540
2733
+ },
2734
+ {
2735
+ "epoch": 36.4,
2736
+ "learning_rate": 1.4782608695652176e-05,
2737
+ "loss": 0.7528,
2738
+ "step": 4550
2739
+ },
2740
+ {
2741
+ "epoch": 36.48,
2742
+ "learning_rate": 1.4695652173913045e-05,
2743
+ "loss": 0.7683,
2744
+ "step": 4560
2745
+ },
2746
+ {
2747
+ "epoch": 36.56,
2748
+ "learning_rate": 1.4608695652173915e-05,
2749
+ "loss": 0.7687,
2750
+ "step": 4570
2751
+ },
2752
+ {
2753
+ "epoch": 36.64,
2754
+ "learning_rate": 1.4521739130434783e-05,
2755
+ "loss": 0.7404,
2756
+ "step": 4580
2757
+ },
2758
+ {
2759
+ "epoch": 36.72,
2760
+ "learning_rate": 1.4434782608695652e-05,
2761
+ "loss": 0.7402,
2762
+ "step": 4590
2763
+ },
2764
+ {
2765
+ "epoch": 36.8,
2766
+ "learning_rate": 1.4347826086956522e-05,
2767
+ "loss": 0.7542,
2768
+ "step": 4600
2769
+ },
2770
+ {
2771
+ "epoch": 36.88,
2772
+ "learning_rate": 1.426086956521739e-05,
2773
+ "loss": 0.8203,
2774
+ "step": 4610
2775
+ },
2776
+ {
2777
+ "epoch": 36.96,
2778
+ "learning_rate": 1.4173913043478263e-05,
2779
+ "loss": 0.7896,
2780
+ "step": 4620
2781
+ },
2782
+ {
2783
+ "epoch": 37.04,
2784
+ "learning_rate": 1.4086956521739131e-05,
2785
+ "loss": 0.7888,
2786
+ "step": 4630
2787
+ },
2788
+ {
2789
+ "epoch": 37.12,
2790
+ "learning_rate": 1.4000000000000001e-05,
2791
+ "loss": 0.691,
2792
+ "step": 4640
2793
+ },
2794
+ {
2795
+ "epoch": 37.2,
2796
+ "learning_rate": 1.391304347826087e-05,
2797
+ "loss": 0.7158,
2798
+ "step": 4650
2799
+ },
2800
+ {
2801
+ "epoch": 37.28,
2802
+ "learning_rate": 1.3826086956521739e-05,
2803
+ "loss": 0.6983,
2804
+ "step": 4660
2805
+ },
2806
+ {
2807
+ "epoch": 37.36,
2808
+ "learning_rate": 1.373913043478261e-05,
2809
+ "loss": 0.7424,
2810
+ "step": 4670
2811
+ },
2812
+ {
2813
+ "epoch": 37.44,
2814
+ "learning_rate": 1.3652173913043479e-05,
2815
+ "loss": 0.7273,
2816
+ "step": 4680
2817
+ },
2818
+ {
2819
+ "epoch": 37.52,
2820
+ "learning_rate": 1.356521739130435e-05,
2821
+ "loss": 0.7105,
2822
+ "step": 4690
2823
+ },
2824
+ {
2825
+ "epoch": 37.6,
2826
+ "learning_rate": 1.3478260869565218e-05,
2827
+ "loss": 0.765,
2828
+ "step": 4700
2829
+ },
2830
+ {
2831
+ "epoch": 37.68,
2832
+ "learning_rate": 1.3391304347826086e-05,
2833
+ "loss": 0.7455,
2834
+ "step": 4710
2835
+ },
2836
+ {
2837
+ "epoch": 37.76,
2838
+ "learning_rate": 1.3304347826086957e-05,
2839
+ "loss": 0.7378,
2840
+ "step": 4720
2841
+ },
2842
+ {
2843
+ "epoch": 37.84,
2844
+ "learning_rate": 1.3217391304347825e-05,
2845
+ "loss": 0.7797,
2846
+ "step": 4730
2847
+ },
2848
+ {
2849
+ "epoch": 37.92,
2850
+ "learning_rate": 1.3130434782608697e-05,
2851
+ "loss": 0.7881,
2852
+ "step": 4740
2853
+ },
2854
+ {
2855
+ "epoch": 38.0,
2856
+ "learning_rate": 1.3043478260869566e-05,
2857
+ "loss": 0.7675,
2858
+ "step": 4750
2859
+ },
2860
+ {
2861
+ "epoch": 38.08,
2862
+ "learning_rate": 1.2956521739130436e-05,
2863
+ "loss": 0.7251,
2864
+ "step": 4760
2865
+ },
2866
+ {
2867
+ "epoch": 38.16,
2868
+ "learning_rate": 1.2869565217391305e-05,
2869
+ "loss": 0.736,
2870
+ "step": 4770
2871
+ },
2872
+ {
2873
+ "epoch": 38.24,
2874
+ "learning_rate": 1.2782608695652173e-05,
2875
+ "loss": 0.6723,
2876
+ "step": 4780
2877
+ },
2878
+ {
2879
+ "epoch": 38.32,
2880
+ "learning_rate": 1.2695652173913045e-05,
2881
+ "loss": 0.736,
2882
+ "step": 4790
2883
+ },
2884
+ {
2885
+ "epoch": 38.4,
2886
+ "learning_rate": 1.2608695652173914e-05,
2887
+ "loss": 0.7543,
2888
+ "step": 4800
2889
+ },
2890
+ {
2891
+ "epoch": 38.48,
2892
+ "learning_rate": 1.2521739130434784e-05,
2893
+ "loss": 0.708,
2894
+ "step": 4810
2895
+ },
2896
+ {
2897
+ "epoch": 38.56,
2898
+ "learning_rate": 1.2434782608695652e-05,
2899
+ "loss": 0.7561,
2900
+ "step": 4820
2901
+ },
2902
+ {
2903
+ "epoch": 38.64,
2904
+ "learning_rate": 1.2347826086956523e-05,
2905
+ "loss": 0.7343,
2906
+ "step": 4830
2907
+ },
2908
+ {
2909
+ "epoch": 38.72,
2910
+ "learning_rate": 1.2260869565217393e-05,
2911
+ "loss": 0.748,
2912
+ "step": 4840
2913
+ },
2914
+ {
2915
+ "epoch": 38.8,
2916
+ "learning_rate": 1.2173913043478261e-05,
2917
+ "loss": 0.7882,
2918
+ "step": 4850
2919
+ },
2920
+ {
2921
+ "epoch": 38.88,
2922
+ "learning_rate": 1.208695652173913e-05,
2923
+ "loss": 0.7107,
2924
+ "step": 4860
2925
+ },
2926
+ {
2927
+ "epoch": 38.96,
2928
+ "learning_rate": 1.2e-05,
2929
+ "loss": 0.7572,
2930
+ "step": 4870
2931
+ },
2932
+ {
2933
+ "epoch": 39.04,
2934
+ "learning_rate": 1.191304347826087e-05,
2935
+ "loss": 0.7749,
2936
+ "step": 4880
2937
+ },
2938
+ {
2939
+ "epoch": 39.12,
2940
+ "learning_rate": 1.1826086956521739e-05,
2941
+ "loss": 0.7238,
2942
+ "step": 4890
2943
+ },
2944
+ {
2945
+ "epoch": 39.2,
2946
+ "learning_rate": 1.173913043478261e-05,
2947
+ "loss": 0.7204,
2948
+ "step": 4900
2949
+ },
2950
+ {
2951
+ "epoch": 39.28,
2952
+ "learning_rate": 1.165217391304348e-05,
2953
+ "loss": 0.7088,
2954
+ "step": 4910
2955
+ },
2956
+ {
2957
+ "epoch": 39.36,
2958
+ "learning_rate": 1.156521739130435e-05,
2959
+ "loss": 0.7424,
2960
+ "step": 4920
2961
+ },
2962
+ {
2963
+ "epoch": 39.44,
2964
+ "learning_rate": 1.1478260869565217e-05,
2965
+ "loss": 0.7483,
2966
+ "step": 4930
2967
+ },
2968
+ {
2969
+ "epoch": 39.52,
2970
+ "learning_rate": 1.1391304347826087e-05,
2971
+ "loss": 0.7048,
2972
+ "step": 4940
2973
+ },
2974
+ {
2975
+ "epoch": 39.6,
2976
+ "learning_rate": 1.1304347826086957e-05,
2977
+ "loss": 0.7583,
2978
+ "step": 4950
2979
+ },
2980
+ {
2981
+ "epoch": 39.68,
2982
+ "learning_rate": 1.1217391304347827e-05,
2983
+ "loss": 0.6905,
2984
+ "step": 4960
2985
+ },
2986
+ {
2987
+ "epoch": 39.76,
2988
+ "learning_rate": 1.1130434782608696e-05,
2989
+ "loss": 0.6934,
2990
+ "step": 4970
2991
+ },
2992
+ {
2993
+ "epoch": 39.84,
2994
+ "learning_rate": 1.1043478260869566e-05,
2995
+ "loss": 0.7766,
2996
+ "step": 4980
2997
+ },
2998
+ {
2999
+ "epoch": 39.92,
3000
+ "learning_rate": 1.0956521739130435e-05,
3001
+ "loss": 0.7716,
3002
+ "step": 4990
3003
+ },
3004
+ {
3005
+ "epoch": 40.0,
3006
+ "learning_rate": 1.0869565217391305e-05,
3007
+ "loss": 0.7482,
3008
+ "step": 5000
3009
+ },
3010
+ {
3011
+ "epoch": 40.08,
3012
+ "learning_rate": 1.0782608695652174e-05,
3013
+ "loss": 0.6505,
3014
+ "step": 5010
3015
+ },
3016
+ {
3017
+ "epoch": 40.16,
3018
+ "learning_rate": 1.0695652173913044e-05,
3019
+ "loss": 0.6988,
3020
+ "step": 5020
3021
+ },
3022
+ {
3023
+ "epoch": 40.24,
3024
+ "learning_rate": 1.0608695652173914e-05,
3025
+ "loss": 0.7083,
3026
+ "step": 5030
3027
+ },
3028
+ {
3029
+ "epoch": 40.32,
3030
+ "learning_rate": 1.0521739130434784e-05,
3031
+ "loss": 0.7353,
3032
+ "step": 5040
3033
+ },
3034
+ {
3035
+ "epoch": 40.4,
3036
+ "learning_rate": 1.0434782608695651e-05,
3037
+ "loss": 0.7568,
3038
+ "step": 5050
3039
+ },
3040
+ {
3041
+ "epoch": 40.48,
3042
+ "learning_rate": 1.0347826086956522e-05,
3043
+ "loss": 0.7481,
3044
+ "step": 5060
3045
+ },
3046
+ {
3047
+ "epoch": 40.56,
3048
+ "learning_rate": 1.0260869565217392e-05,
3049
+ "loss": 0.7494,
3050
+ "step": 5070
3051
+ },
3052
+ {
3053
+ "epoch": 40.64,
3054
+ "learning_rate": 1.0173913043478262e-05,
3055
+ "loss": 0.7494,
3056
+ "step": 5080
3057
+ },
3058
+ {
3059
+ "epoch": 40.72,
3060
+ "learning_rate": 1.008695652173913e-05,
3061
+ "loss": 0.7494,
3062
+ "step": 5090
3063
+ },
3064
+ {
3065
+ "epoch": 40.8,
3066
+ "learning_rate": 1e-05,
3067
+ "loss": 0.7479,
3068
+ "step": 5100
3069
+ },
3070
+ {
3071
+ "epoch": 40.88,
3072
+ "learning_rate": 9.91304347826087e-06,
3073
+ "loss": 0.7476,
3074
+ "step": 5110
3075
+ },
3076
+ {
3077
+ "epoch": 40.96,
3078
+ "learning_rate": 9.82608695652174e-06,
3079
+ "loss": 0.7544,
3080
+ "step": 5120
3081
+ },
3082
+ {
3083
+ "epoch": 41.04,
3084
+ "learning_rate": 9.739130434782608e-06,
3085
+ "loss": 0.6957,
3086
+ "step": 5130
3087
+ },
3088
+ {
3089
+ "epoch": 41.12,
3090
+ "learning_rate": 9.652173913043478e-06,
3091
+ "loss": 0.7071,
3092
+ "step": 5140
3093
+ },
3094
+ {
3095
+ "epoch": 41.2,
3096
+ "learning_rate": 9.565217391304349e-06,
3097
+ "loss": 0.7036,
3098
+ "step": 5150
3099
+ },
3100
+ {
3101
+ "epoch": 41.28,
3102
+ "learning_rate": 9.478260869565219e-06,
3103
+ "loss": 0.6897,
3104
+ "step": 5160
3105
+ },
3106
+ {
3107
+ "epoch": 41.36,
3108
+ "learning_rate": 9.391304347826087e-06,
3109
+ "loss": 0.7168,
3110
+ "step": 5170
3111
+ },
3112
+ {
3113
+ "epoch": 41.44,
3114
+ "learning_rate": 9.304347826086956e-06,
3115
+ "loss": 0.6823,
3116
+ "step": 5180
3117
+ },
3118
+ {
3119
+ "epoch": 41.52,
3120
+ "learning_rate": 9.217391304347826e-06,
3121
+ "loss": 0.7238,
3122
+ "step": 5190
3123
+ },
3124
+ {
3125
+ "epoch": 41.6,
3126
+ "learning_rate": 9.130434782608697e-06,
3127
+ "loss": 0.7366,
3128
+ "step": 5200
3129
+ },
3130
+ {
3131
+ "epoch": 41.68,
3132
+ "learning_rate": 9.043478260869565e-06,
3133
+ "loss": 0.7688,
3134
+ "step": 5210
3135
+ },
3136
+ {
3137
+ "epoch": 41.76,
3138
+ "learning_rate": 8.956521739130435e-06,
3139
+ "loss": 0.7606,
3140
+ "step": 5220
3141
+ },
3142
+ {
3143
+ "epoch": 41.84,
3144
+ "learning_rate": 8.869565217391304e-06,
3145
+ "loss": 0.7199,
3146
+ "step": 5230
3147
+ },
3148
+ {
3149
+ "epoch": 41.92,
3150
+ "learning_rate": 8.782608695652174e-06,
3151
+ "loss": 0.7336,
3152
+ "step": 5240
3153
+ },
3154
+ {
3155
+ "epoch": 42.0,
3156
+ "learning_rate": 8.695652173913044e-06,
3157
+ "loss": 0.7061,
3158
+ "step": 5250
3159
+ },
3160
+ {
3161
+ "epoch": 42.08,
3162
+ "learning_rate": 8.608695652173913e-06,
3163
+ "loss": 0.7108,
3164
+ "step": 5260
3165
+ },
3166
+ {
3167
+ "epoch": 42.16,
3168
+ "learning_rate": 8.521739130434783e-06,
3169
+ "loss": 0.6611,
3170
+ "step": 5270
3171
+ },
3172
+ {
3173
+ "epoch": 42.24,
3174
+ "learning_rate": 8.434782608695653e-06,
3175
+ "loss": 0.7001,
3176
+ "step": 5280
3177
+ },
3178
+ {
3179
+ "epoch": 42.32,
3180
+ "learning_rate": 8.347826086956522e-06,
3181
+ "loss": 0.7304,
3182
+ "step": 5290
3183
+ },
3184
+ {
3185
+ "epoch": 42.4,
3186
+ "learning_rate": 8.26086956521739e-06,
3187
+ "loss": 0.7143,
3188
+ "step": 5300
3189
+ },
3190
+ {
3191
+ "epoch": 42.48,
3192
+ "learning_rate": 8.17391304347826e-06,
3193
+ "loss": 0.6987,
3194
+ "step": 5310
3195
+ },
3196
+ {
3197
+ "epoch": 42.56,
3198
+ "learning_rate": 8.086956521739131e-06,
3199
+ "loss": 0.7078,
3200
+ "step": 5320
3201
+ },
3202
+ {
3203
+ "epoch": 42.64,
3204
+ "learning_rate": 8.000000000000001e-06,
3205
+ "loss": 0.7553,
3206
+ "step": 5330
3207
+ },
3208
+ {
3209
+ "epoch": 42.72,
3210
+ "learning_rate": 7.91304347826087e-06,
3211
+ "loss": 0.7014,
3212
+ "step": 5340
3213
+ },
3214
+ {
3215
+ "epoch": 42.8,
3216
+ "learning_rate": 7.82608695652174e-06,
3217
+ "loss": 0.7483,
3218
+ "step": 5350
3219
+ },
3220
+ {
3221
+ "epoch": 42.88,
3222
+ "learning_rate": 7.739130434782609e-06,
3223
+ "loss": 0.7629,
3224
+ "step": 5360
3225
+ },
3226
+ {
3227
+ "epoch": 42.96,
3228
+ "learning_rate": 7.652173913043479e-06,
3229
+ "loss": 0.7117,
3230
+ "step": 5370
3231
+ },
3232
+ {
3233
+ "epoch": 43.04,
3234
+ "learning_rate": 7.565217391304348e-06,
3235
+ "loss": 0.7447,
3236
+ "step": 5380
3237
+ },
3238
+ {
3239
+ "epoch": 43.12,
3240
+ "learning_rate": 7.478260869565218e-06,
3241
+ "loss": 0.6614,
3242
+ "step": 5390
3243
+ },
3244
+ {
3245
+ "epoch": 43.2,
3246
+ "learning_rate": 7.391304347826088e-06,
3247
+ "loss": 0.7417,
3248
+ "step": 5400
3249
+ },
3250
+ {
3251
+ "epoch": 43.28,
3252
+ "learning_rate": 7.304347826086957e-06,
3253
+ "loss": 0.7217,
3254
+ "step": 5410
3255
+ },
3256
+ {
3257
+ "epoch": 43.36,
3258
+ "learning_rate": 7.217391304347826e-06,
3259
+ "loss": 0.704,
3260
+ "step": 5420
3261
+ },
3262
+ {
3263
+ "epoch": 43.44,
3264
+ "learning_rate": 7.130434782608695e-06,
3265
+ "loss": 0.704,
3266
+ "step": 5430
3267
+ },
3268
+ {
3269
+ "epoch": 43.52,
3270
+ "learning_rate": 7.043478260869566e-06,
3271
+ "loss": 0.7896,
3272
+ "step": 5440
3273
+ },
3274
+ {
3275
+ "epoch": 43.6,
3276
+ "learning_rate": 6.956521739130435e-06,
3277
+ "loss": 0.7481,
3278
+ "step": 5450
3279
+ },
3280
+ {
3281
+ "epoch": 43.68,
3282
+ "learning_rate": 6.869565217391305e-06,
3283
+ "loss": 0.7105,
3284
+ "step": 5460
3285
+ },
3286
+ {
3287
+ "epoch": 43.76,
3288
+ "learning_rate": 6.782608695652175e-06,
3289
+ "loss": 0.6717,
3290
+ "step": 5470
3291
+ },
3292
+ {
3293
+ "epoch": 43.84,
3294
+ "learning_rate": 6.695652173913043e-06,
3295
+ "loss": 0.6871,
3296
+ "step": 5480
3297
+ },
3298
+ {
3299
+ "epoch": 43.92,
3300
+ "learning_rate": 6.608695652173913e-06,
3301
+ "loss": 0.7229,
3302
+ "step": 5490
3303
+ },
3304
+ {
3305
+ "epoch": 44.0,
3306
+ "learning_rate": 6.521739130434783e-06,
3307
+ "loss": 0.7324,
3308
+ "step": 5500
3309
+ },
3310
+ {
3311
+ "epoch": 44.08,
3312
+ "learning_rate": 6.434782608695652e-06,
3313
+ "loss": 0.6597,
3314
+ "step": 5510
3315
+ },
3316
+ {
3317
+ "epoch": 44.16,
3318
+ "learning_rate": 6.3478260869565225e-06,
3319
+ "loss": 0.6771,
3320
+ "step": 5520
3321
+ },
3322
+ {
3323
+ "epoch": 44.24,
3324
+ "learning_rate": 6.260869565217392e-06,
3325
+ "loss": 0.7184,
3326
+ "step": 5530
3327
+ },
3328
+ {
3329
+ "epoch": 44.32,
3330
+ "learning_rate": 6.173913043478261e-06,
3331
+ "loss": 0.6816,
3332
+ "step": 5540
3333
+ },
3334
+ {
3335
+ "epoch": 44.4,
3336
+ "learning_rate": 6.086956521739131e-06,
3337
+ "loss": 0.7171,
3338
+ "step": 5550
3339
+ },
3340
+ {
3341
+ "epoch": 44.48,
3342
+ "learning_rate": 6e-06,
3343
+ "loss": 0.6723,
3344
+ "step": 5560
3345
+ },
3346
+ {
3347
+ "epoch": 44.56,
3348
+ "learning_rate": 5.9130434782608696e-06,
3349
+ "loss": 0.6991,
3350
+ "step": 5570
3351
+ },
3352
+ {
3353
+ "epoch": 44.64,
3354
+ "learning_rate": 5.82608695652174e-06,
3355
+ "loss": 0.7538,
3356
+ "step": 5580
3357
+ },
3358
+ {
3359
+ "epoch": 44.72,
3360
+ "learning_rate": 5.739130434782608e-06,
3361
+ "loss": 0.7405,
3362
+ "step": 5590
3363
+ },
3364
+ {
3365
+ "epoch": 44.8,
3366
+ "learning_rate": 5.652173913043479e-06,
3367
+ "loss": 0.6698,
3368
+ "step": 5600
3369
+ },
3370
+ {
3371
+ "epoch": 44.88,
3372
+ "learning_rate": 5.565217391304348e-06,
3373
+ "loss": 0.7276,
3374
+ "step": 5610
3375
+ },
3376
+ {
3377
+ "epoch": 44.96,
3378
+ "learning_rate": 5.478260869565217e-06,
3379
+ "loss": 0.7446,
3380
+ "step": 5620
3381
+ },
3382
+ {
3383
+ "epoch": 45.04,
3384
+ "learning_rate": 5.391304347826087e-06,
3385
+ "loss": 0.7269,
3386
+ "step": 5630
3387
+ },
3388
+ {
3389
+ "epoch": 45.12,
3390
+ "learning_rate": 5.304347826086957e-06,
3391
+ "loss": 0.6781,
3392
+ "step": 5640
3393
+ },
3394
+ {
3395
+ "epoch": 45.2,
3396
+ "learning_rate": 5.217391304347826e-06,
3397
+ "loss": 0.7352,
3398
+ "step": 5650
3399
+ },
3400
+ {
3401
+ "epoch": 45.28,
3402
+ "learning_rate": 5.130434782608696e-06,
3403
+ "loss": 0.7599,
3404
+ "step": 5660
3405
+ },
3406
+ {
3407
+ "epoch": 45.36,
3408
+ "learning_rate": 5.043478260869565e-06,
3409
+ "loss": 0.7081,
3410
+ "step": 5670
3411
+ },
3412
+ {
3413
+ "epoch": 45.44,
3414
+ "learning_rate": 4.956521739130435e-06,
3415
+ "loss": 0.7287,
3416
+ "step": 5680
3417
+ },
3418
+ {
3419
+ "epoch": 45.52,
3420
+ "learning_rate": 4.869565217391304e-06,
3421
+ "loss": 0.6988,
3422
+ "step": 5690
3423
+ },
3424
+ {
3425
+ "epoch": 45.6,
3426
+ "learning_rate": 4.782608695652174e-06,
3427
+ "loss": 0.716,
3428
+ "step": 5700
3429
+ },
3430
+ {
3431
+ "epoch": 45.68,
3432
+ "learning_rate": 4.695652173913044e-06,
3433
+ "loss": 0.6744,
3434
+ "step": 5710
3435
+ },
3436
+ {
3437
+ "epoch": 45.76,
3438
+ "learning_rate": 4.608695652173913e-06,
3439
+ "loss": 0.6928,
3440
+ "step": 5720
3441
+ },
3442
+ {
3443
+ "epoch": 45.84,
3444
+ "learning_rate": 4.5217391304347826e-06,
3445
+ "loss": 0.721,
3446
+ "step": 5730
3447
+ },
3448
+ {
3449
+ "epoch": 45.92,
3450
+ "learning_rate": 4.434782608695652e-06,
3451
+ "loss": 0.7276,
3452
+ "step": 5740
3453
+ },
3454
+ {
3455
+ "epoch": 46.0,
3456
+ "learning_rate": 4.347826086956522e-06,
3457
+ "loss": 0.7224,
3458
+ "step": 5750
3459
+ },
3460
+ {
3461
+ "epoch": 46.08,
3462
+ "learning_rate": 4.260869565217392e-06,
3463
+ "loss": 0.7101,
3464
+ "step": 5760
3465
+ },
3466
+ {
3467
+ "epoch": 46.16,
3468
+ "learning_rate": 4.173913043478261e-06,
3469
+ "loss": 0.6956,
3470
+ "step": 5770
3471
+ },
3472
+ {
3473
+ "epoch": 46.24,
3474
+ "learning_rate": 4.08695652173913e-06,
3475
+ "loss": 0.734,
3476
+ "step": 5780
3477
+ },
3478
+ {
3479
+ "epoch": 46.32,
3480
+ "learning_rate": 4.000000000000001e-06,
3481
+ "loss": 0.6655,
3482
+ "step": 5790
3483
+ },
3484
+ {
3485
+ "epoch": 46.4,
3486
+ "learning_rate": 3.91304347826087e-06,
3487
+ "loss": 0.6528,
3488
+ "step": 5800
3489
+ },
3490
+ {
3491
+ "epoch": 46.48,
3492
+ "learning_rate": 3.8260869565217395e-06,
3493
+ "loss": 0.6876,
3494
+ "step": 5810
3495
+ },
3496
+ {
3497
+ "epoch": 46.56,
3498
+ "learning_rate": 3.739130434782609e-06,
3499
+ "loss": 0.6792,
3500
+ "step": 5820
3501
+ },
3502
+ {
3503
+ "epoch": 46.64,
3504
+ "learning_rate": 3.6521739130434787e-06,
3505
+ "loss": 0.7622,
3506
+ "step": 5830
3507
+ },
3508
+ {
3509
+ "epoch": 46.72,
3510
+ "learning_rate": 3.5652173913043477e-06,
3511
+ "loss": 0.6852,
3512
+ "step": 5840
3513
+ },
3514
+ {
3515
+ "epoch": 46.8,
3516
+ "learning_rate": 3.4782608695652175e-06,
3517
+ "loss": 0.6863,
3518
+ "step": 5850
3519
+ },
3520
+ {
3521
+ "epoch": 46.88,
3522
+ "learning_rate": 3.3913043478260873e-06,
3523
+ "loss": 0.7497,
3524
+ "step": 5860
3525
+ },
3526
+ {
3527
+ "epoch": 46.96,
3528
+ "learning_rate": 3.3043478260869563e-06,
3529
+ "loss": 0.7246,
3530
+ "step": 5870
3531
+ },
3532
+ {
3533
+ "epoch": 47.04,
3534
+ "learning_rate": 3.217391304347826e-06,
3535
+ "loss": 0.7235,
3536
+ "step": 5880
3537
+ },
3538
+ {
3539
+ "epoch": 47.12,
3540
+ "learning_rate": 3.130434782608696e-06,
3541
+ "loss": 0.7068,
3542
+ "step": 5890
3543
+ },
3544
+ {
3545
+ "epoch": 47.2,
3546
+ "learning_rate": 3.0434782608695654e-06,
3547
+ "loss": 0.7105,
3548
+ "step": 5900
3549
+ },
3550
+ {
3551
+ "epoch": 47.28,
3552
+ "learning_rate": 2.9565217391304348e-06,
3553
+ "loss": 0.6939,
3554
+ "step": 5910
3555
+ },
3556
+ {
3557
+ "epoch": 47.36,
3558
+ "learning_rate": 2.869565217391304e-06,
3559
+ "loss": 0.6867,
3560
+ "step": 5920
3561
+ },
3562
+ {
3563
+ "epoch": 47.44,
3564
+ "learning_rate": 2.782608695652174e-06,
3565
+ "loss": 0.6587,
3566
+ "step": 5930
3567
+ },
3568
+ {
3569
+ "epoch": 47.52,
3570
+ "learning_rate": 2.6956521739130434e-06,
3571
+ "loss": 0.6914,
3572
+ "step": 5940
3573
+ },
3574
+ {
3575
+ "epoch": 47.6,
3576
+ "learning_rate": 2.608695652173913e-06,
3577
+ "loss": 0.6501,
3578
+ "step": 5950
3579
+ },
3580
+ {
3581
+ "epoch": 47.68,
3582
+ "learning_rate": 2.5217391304347826e-06,
3583
+ "loss": 0.7491,
3584
+ "step": 5960
3585
+ },
3586
+ {
3587
+ "epoch": 47.76,
3588
+ "learning_rate": 2.434782608695652e-06,
3589
+ "loss": 0.7428,
3590
+ "step": 5970
3591
+ },
3592
+ {
3593
+ "epoch": 47.84,
3594
+ "learning_rate": 2.347826086956522e-06,
3595
+ "loss": 0.716,
3596
+ "step": 5980
3597
+ },
3598
+ {
3599
+ "epoch": 47.92,
3600
+ "learning_rate": 2.2608695652173913e-06,
3601
+ "loss": 0.6706,
3602
+ "step": 5990
3603
+ },
3604
+ {
3605
+ "epoch": 48.0,
3606
+ "learning_rate": 2.173913043478261e-06,
3607
+ "loss": 0.6691,
3608
+ "step": 6000
3609
+ }
3610
+ ],
3611
+ "max_steps": 6250,
3612
+ "num_train_epochs": 50,
3613
+ "total_flos": 8948999081401344.0,
3614
+ "trial_name": null,
3615
+ "trial_params": null
3616
+ }