DunnBC22 commited on
Commit
b10d4f1
·
1 Parent(s): cea5767

All Dunn!!!

Browse files
all_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 2.0,
3
+ "total_flos": 6.728145843607406e+16,
4
+ "train_loss": 0.7582937260732105,
5
+ "train_runtime": 12885.6962,
6
+ "train_samples_per_second": 65.589,
7
+ "train_steps_per_second": 2.05
8
+ }
runs/Jun21_14-46-56_3cb495f9eec1/events.out.tfevents.1687372638.3cb495f9eec1.488.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d837b6bc9be21f9d4164157bda99cd290ee4bcc03c09650f22c0643eaa052f67
3
+ size 936
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 2.0,
3
+ "total_flos": 6.728145843607406e+16,
4
+ "train_loss": 0.7582937260732105,
5
+ "train_runtime": 12885.6962,
6
+ "train_samples_per_second": 65.589,
7
+ "train_steps_per_second": 2.05
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,3199 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.999886413115499,
5
+ "global_step": 26410,
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": 1.9999242711094284e-05,
13
+ "loss": 1.3163,
14
+ "step": 1
15
+ },
16
+ {
17
+ "epoch": 0.0,
18
+ "learning_rate": 1.9963650132525562e-05,
19
+ "loss": 1.0093,
20
+ "step": 50
21
+ },
22
+ {
23
+ "epoch": 0.01,
24
+ "learning_rate": 1.99265429761454e-05,
25
+ "loss": 0.8939,
26
+ "step": 100
27
+ },
28
+ {
29
+ "epoch": 0.01,
30
+ "learning_rate": 1.9888678530859524e-05,
31
+ "loss": 0.8368,
32
+ "step": 150
33
+ },
34
+ {
35
+ "epoch": 0.02,
36
+ "learning_rate": 1.985081408557365e-05,
37
+ "loss": 0.8794,
38
+ "step": 200
39
+ },
40
+ {
41
+ "epoch": 0.02,
42
+ "learning_rate": 1.9812949640287772e-05,
43
+ "loss": 0.843,
44
+ "step": 250
45
+ },
46
+ {
47
+ "epoch": 0.02,
48
+ "learning_rate": 1.9775085195001894e-05,
49
+ "loss": 0.8488,
50
+ "step": 300
51
+ },
52
+ {
53
+ "epoch": 0.03,
54
+ "learning_rate": 1.9737220749716017e-05,
55
+ "loss": 0.8465,
56
+ "step": 350
57
+ },
58
+ {
59
+ "epoch": 0.03,
60
+ "learning_rate": 1.9699356304430143e-05,
61
+ "loss": 0.8591,
62
+ "step": 400
63
+ },
64
+ {
65
+ "epoch": 0.03,
66
+ "learning_rate": 1.9661491859144265e-05,
67
+ "loss": 0.8978,
68
+ "step": 450
69
+ },
70
+ {
71
+ "epoch": 0.04,
72
+ "learning_rate": 1.9623627413858387e-05,
73
+ "loss": 0.8035,
74
+ "step": 500
75
+ },
76
+ {
77
+ "epoch": 0.04,
78
+ "learning_rate": 1.9585762968572513e-05,
79
+ "loss": 0.8153,
80
+ "step": 550
81
+ },
82
+ {
83
+ "epoch": 0.05,
84
+ "learning_rate": 1.9547898523286635e-05,
85
+ "loss": 0.8079,
86
+ "step": 600
87
+ },
88
+ {
89
+ "epoch": 0.05,
90
+ "learning_rate": 1.9510034078000758e-05,
91
+ "loss": 0.8092,
92
+ "step": 650
93
+ },
94
+ {
95
+ "epoch": 0.05,
96
+ "learning_rate": 1.947216963271488e-05,
97
+ "loss": 0.8294,
98
+ "step": 700
99
+ },
100
+ {
101
+ "epoch": 0.06,
102
+ "learning_rate": 1.9434305187429006e-05,
103
+ "loss": 0.8099,
104
+ "step": 750
105
+ },
106
+ {
107
+ "epoch": 0.06,
108
+ "learning_rate": 1.939644074214313e-05,
109
+ "loss": 0.85,
110
+ "step": 800
111
+ },
112
+ {
113
+ "epoch": 0.06,
114
+ "learning_rate": 1.935857629685725e-05,
115
+ "loss": 0.8509,
116
+ "step": 850
117
+ },
118
+ {
119
+ "epoch": 0.07,
120
+ "learning_rate": 1.9320711851571376e-05,
121
+ "loss": 0.8229,
122
+ "step": 900
123
+ },
124
+ {
125
+ "epoch": 0.07,
126
+ "learning_rate": 1.92828474062855e-05,
127
+ "loss": 0.8251,
128
+ "step": 950
129
+ },
130
+ {
131
+ "epoch": 0.08,
132
+ "learning_rate": 1.9244982960999625e-05,
133
+ "loss": 0.8116,
134
+ "step": 1000
135
+ },
136
+ {
137
+ "epoch": 0.08,
138
+ "learning_rate": 1.9207118515713747e-05,
139
+ "loss": 0.8277,
140
+ "step": 1050
141
+ },
142
+ {
143
+ "epoch": 0.08,
144
+ "learning_rate": 1.916925407042787e-05,
145
+ "loss": 0.8004,
146
+ "step": 1100
147
+ },
148
+ {
149
+ "epoch": 0.09,
150
+ "learning_rate": 1.9131389625141995e-05,
151
+ "loss": 0.8347,
152
+ "step": 1150
153
+ },
154
+ {
155
+ "epoch": 0.09,
156
+ "learning_rate": 1.9093525179856118e-05,
157
+ "loss": 0.8103,
158
+ "step": 1200
159
+ },
160
+ {
161
+ "epoch": 0.09,
162
+ "learning_rate": 1.905566073457024e-05,
163
+ "loss": 0.8566,
164
+ "step": 1250
165
+ },
166
+ {
167
+ "epoch": 0.1,
168
+ "learning_rate": 1.9017796289284362e-05,
169
+ "loss": 0.7921,
170
+ "step": 1300
171
+ },
172
+ {
173
+ "epoch": 0.1,
174
+ "learning_rate": 1.8979931843998488e-05,
175
+ "loss": 0.7959,
176
+ "step": 1350
177
+ },
178
+ {
179
+ "epoch": 0.11,
180
+ "learning_rate": 1.894206739871261e-05,
181
+ "loss": 0.7827,
182
+ "step": 1400
183
+ },
184
+ {
185
+ "epoch": 0.11,
186
+ "learning_rate": 1.8904202953426733e-05,
187
+ "loss": 0.785,
188
+ "step": 1450
189
+ },
190
+ {
191
+ "epoch": 0.11,
192
+ "learning_rate": 1.886633850814086e-05,
193
+ "loss": 0.8061,
194
+ "step": 1500
195
+ },
196
+ {
197
+ "epoch": 0.12,
198
+ "learning_rate": 1.882847406285498e-05,
199
+ "loss": 0.7717,
200
+ "step": 1550
201
+ },
202
+ {
203
+ "epoch": 0.12,
204
+ "learning_rate": 1.8790609617569103e-05,
205
+ "loss": 0.8362,
206
+ "step": 1600
207
+ },
208
+ {
209
+ "epoch": 0.12,
210
+ "learning_rate": 1.8752745172283226e-05,
211
+ "loss": 0.7975,
212
+ "step": 1650
213
+ },
214
+ {
215
+ "epoch": 0.13,
216
+ "learning_rate": 1.871488072699735e-05,
217
+ "loss": 0.8123,
218
+ "step": 1700
219
+ },
220
+ {
221
+ "epoch": 0.13,
222
+ "learning_rate": 1.8677016281711474e-05,
223
+ "loss": 0.8526,
224
+ "step": 1750
225
+ },
226
+ {
227
+ "epoch": 0.14,
228
+ "learning_rate": 1.8639151836425596e-05,
229
+ "loss": 0.808,
230
+ "step": 1800
231
+ },
232
+ {
233
+ "epoch": 0.14,
234
+ "learning_rate": 1.8601287391139722e-05,
235
+ "loss": 0.7989,
236
+ "step": 1850
237
+ },
238
+ {
239
+ "epoch": 0.14,
240
+ "learning_rate": 1.8563422945853844e-05,
241
+ "loss": 0.7755,
242
+ "step": 1900
243
+ },
244
+ {
245
+ "epoch": 0.15,
246
+ "learning_rate": 1.8525558500567967e-05,
247
+ "loss": 0.8467,
248
+ "step": 1950
249
+ },
250
+ {
251
+ "epoch": 0.15,
252
+ "learning_rate": 1.8487694055282093e-05,
253
+ "loss": 0.8221,
254
+ "step": 2000
255
+ },
256
+ {
257
+ "epoch": 0.16,
258
+ "learning_rate": 1.8449829609996215e-05,
259
+ "loss": 0.7851,
260
+ "step": 2050
261
+ },
262
+ {
263
+ "epoch": 0.16,
264
+ "learning_rate": 1.841196516471034e-05,
265
+ "loss": 0.8159,
266
+ "step": 2100
267
+ },
268
+ {
269
+ "epoch": 0.16,
270
+ "learning_rate": 1.837410071942446e-05,
271
+ "loss": 0.8027,
272
+ "step": 2150
273
+ },
274
+ {
275
+ "epoch": 0.17,
276
+ "learning_rate": 1.8336236274138586e-05,
277
+ "loss": 0.7585,
278
+ "step": 2200
279
+ },
280
+ {
281
+ "epoch": 0.17,
282
+ "learning_rate": 1.8298371828852708e-05,
283
+ "loss": 0.8287,
284
+ "step": 2250
285
+ },
286
+ {
287
+ "epoch": 0.17,
288
+ "learning_rate": 1.8260507383566834e-05,
289
+ "loss": 0.8127,
290
+ "step": 2300
291
+ },
292
+ {
293
+ "epoch": 0.18,
294
+ "learning_rate": 1.8222642938280956e-05,
295
+ "loss": 0.8015,
296
+ "step": 2350
297
+ },
298
+ {
299
+ "epoch": 0.18,
300
+ "learning_rate": 1.818477849299508e-05,
301
+ "loss": 0.8228,
302
+ "step": 2400
303
+ },
304
+ {
305
+ "epoch": 0.19,
306
+ "learning_rate": 1.8146914047709204e-05,
307
+ "loss": 0.7865,
308
+ "step": 2450
309
+ },
310
+ {
311
+ "epoch": 0.19,
312
+ "learning_rate": 1.8109049602423327e-05,
313
+ "loss": 0.8093,
314
+ "step": 2500
315
+ },
316
+ {
317
+ "epoch": 0.19,
318
+ "learning_rate": 1.807118515713745e-05,
319
+ "loss": 0.7812,
320
+ "step": 2550
321
+ },
322
+ {
323
+ "epoch": 0.2,
324
+ "learning_rate": 1.803332071185157e-05,
325
+ "loss": 0.805,
326
+ "step": 2600
327
+ },
328
+ {
329
+ "epoch": 0.2,
330
+ "learning_rate": 1.7995456266565697e-05,
331
+ "loss": 0.7844,
332
+ "step": 2650
333
+ },
334
+ {
335
+ "epoch": 0.2,
336
+ "learning_rate": 1.795759182127982e-05,
337
+ "loss": 0.7976,
338
+ "step": 2700
339
+ },
340
+ {
341
+ "epoch": 0.21,
342
+ "learning_rate": 1.7919727375993942e-05,
343
+ "loss": 0.7683,
344
+ "step": 2750
345
+ },
346
+ {
347
+ "epoch": 0.21,
348
+ "learning_rate": 1.7881862930708068e-05,
349
+ "loss": 0.8157,
350
+ "step": 2800
351
+ },
352
+ {
353
+ "epoch": 0.22,
354
+ "learning_rate": 1.784399848542219e-05,
355
+ "loss": 0.7801,
356
+ "step": 2850
357
+ },
358
+ {
359
+ "epoch": 0.22,
360
+ "learning_rate": 1.7806134040136312e-05,
361
+ "loss": 0.7817,
362
+ "step": 2900
363
+ },
364
+ {
365
+ "epoch": 0.22,
366
+ "learning_rate": 1.7768269594850435e-05,
367
+ "loss": 0.8286,
368
+ "step": 2950
369
+ },
370
+ {
371
+ "epoch": 0.23,
372
+ "learning_rate": 1.773040514956456e-05,
373
+ "loss": 0.7994,
374
+ "step": 3000
375
+ },
376
+ {
377
+ "epoch": 0.23,
378
+ "learning_rate": 1.7692540704278683e-05,
379
+ "loss": 0.758,
380
+ "step": 3050
381
+ },
382
+ {
383
+ "epoch": 0.23,
384
+ "learning_rate": 1.7654676258992805e-05,
385
+ "loss": 0.8116,
386
+ "step": 3100
387
+ },
388
+ {
389
+ "epoch": 0.24,
390
+ "learning_rate": 1.761681181370693e-05,
391
+ "loss": 0.7762,
392
+ "step": 3150
393
+ },
394
+ {
395
+ "epoch": 0.24,
396
+ "learning_rate": 1.7578947368421054e-05,
397
+ "loss": 0.7603,
398
+ "step": 3200
399
+ },
400
+ {
401
+ "epoch": 0.25,
402
+ "learning_rate": 1.7541082923135176e-05,
403
+ "loss": 0.825,
404
+ "step": 3250
405
+ },
406
+ {
407
+ "epoch": 0.25,
408
+ "learning_rate": 1.7503218477849302e-05,
409
+ "loss": 0.7649,
410
+ "step": 3300
411
+ },
412
+ {
413
+ "epoch": 0.25,
414
+ "learning_rate": 1.7465354032563424e-05,
415
+ "loss": 0.7886,
416
+ "step": 3350
417
+ },
418
+ {
419
+ "epoch": 0.26,
420
+ "learning_rate": 1.742748958727755e-05,
421
+ "loss": 0.796,
422
+ "step": 3400
423
+ },
424
+ {
425
+ "epoch": 0.26,
426
+ "learning_rate": 1.7389625141991672e-05,
427
+ "loss": 0.7712,
428
+ "step": 3450
429
+ },
430
+ {
431
+ "epoch": 0.27,
432
+ "learning_rate": 1.7351760696705795e-05,
433
+ "loss": 0.8061,
434
+ "step": 3500
435
+ },
436
+ {
437
+ "epoch": 0.27,
438
+ "learning_rate": 1.7313896251419917e-05,
439
+ "loss": 0.7691,
440
+ "step": 3550
441
+ },
442
+ {
443
+ "epoch": 0.27,
444
+ "learning_rate": 1.7276031806134043e-05,
445
+ "loss": 0.8038,
446
+ "step": 3600
447
+ },
448
+ {
449
+ "epoch": 0.28,
450
+ "learning_rate": 1.7238167360848165e-05,
451
+ "loss": 0.7679,
452
+ "step": 3650
453
+ },
454
+ {
455
+ "epoch": 0.28,
456
+ "learning_rate": 1.7200302915562288e-05,
457
+ "loss": 0.7909,
458
+ "step": 3700
459
+ },
460
+ {
461
+ "epoch": 0.28,
462
+ "learning_rate": 1.7162438470276413e-05,
463
+ "loss": 0.8001,
464
+ "step": 3750
465
+ },
466
+ {
467
+ "epoch": 0.29,
468
+ "learning_rate": 1.7124574024990536e-05,
469
+ "loss": 0.7369,
470
+ "step": 3800
471
+ },
472
+ {
473
+ "epoch": 0.29,
474
+ "learning_rate": 1.7086709579704658e-05,
475
+ "loss": 0.7716,
476
+ "step": 3850
477
+ },
478
+ {
479
+ "epoch": 0.3,
480
+ "learning_rate": 1.704884513441878e-05,
481
+ "loss": 0.8003,
482
+ "step": 3900
483
+ },
484
+ {
485
+ "epoch": 0.3,
486
+ "learning_rate": 1.7010980689132906e-05,
487
+ "loss": 0.7973,
488
+ "step": 3950
489
+ },
490
+ {
491
+ "epoch": 0.3,
492
+ "learning_rate": 1.697311624384703e-05,
493
+ "loss": 0.7745,
494
+ "step": 4000
495
+ },
496
+ {
497
+ "epoch": 0.31,
498
+ "learning_rate": 1.693525179856115e-05,
499
+ "loss": 0.7266,
500
+ "step": 4050
501
+ },
502
+ {
503
+ "epoch": 0.31,
504
+ "learning_rate": 1.6897387353275277e-05,
505
+ "loss": 0.7917,
506
+ "step": 4100
507
+ },
508
+ {
509
+ "epoch": 0.31,
510
+ "learning_rate": 1.68595229079894e-05,
511
+ "loss": 0.7393,
512
+ "step": 4150
513
+ },
514
+ {
515
+ "epoch": 0.32,
516
+ "learning_rate": 1.682165846270352e-05,
517
+ "loss": 0.7822,
518
+ "step": 4200
519
+ },
520
+ {
521
+ "epoch": 0.32,
522
+ "learning_rate": 1.6783794017417647e-05,
523
+ "loss": 0.7919,
524
+ "step": 4250
525
+ },
526
+ {
527
+ "epoch": 0.33,
528
+ "learning_rate": 1.674592957213177e-05,
529
+ "loss": 0.7943,
530
+ "step": 4300
531
+ },
532
+ {
533
+ "epoch": 0.33,
534
+ "learning_rate": 1.6708065126845892e-05,
535
+ "loss": 0.7731,
536
+ "step": 4350
537
+ },
538
+ {
539
+ "epoch": 0.33,
540
+ "learning_rate": 1.6670200681560014e-05,
541
+ "loss": 0.7623,
542
+ "step": 4400
543
+ },
544
+ {
545
+ "epoch": 0.34,
546
+ "learning_rate": 1.663233623627414e-05,
547
+ "loss": 0.7891,
548
+ "step": 4450
549
+ },
550
+ {
551
+ "epoch": 0.34,
552
+ "learning_rate": 1.6594471790988263e-05,
553
+ "loss": 0.7769,
554
+ "step": 4500
555
+ },
556
+ {
557
+ "epoch": 0.34,
558
+ "learning_rate": 1.655660734570239e-05,
559
+ "loss": 0.7958,
560
+ "step": 4550
561
+ },
562
+ {
563
+ "epoch": 0.35,
564
+ "learning_rate": 1.651874290041651e-05,
565
+ "loss": 0.7555,
566
+ "step": 4600
567
+ },
568
+ {
569
+ "epoch": 0.35,
570
+ "learning_rate": 1.6480878455130633e-05,
571
+ "loss": 0.7993,
572
+ "step": 4650
573
+ },
574
+ {
575
+ "epoch": 0.36,
576
+ "learning_rate": 1.644301400984476e-05,
577
+ "loss": 0.7916,
578
+ "step": 4700
579
+ },
580
+ {
581
+ "epoch": 0.36,
582
+ "learning_rate": 1.640514956455888e-05,
583
+ "loss": 0.8101,
584
+ "step": 4750
585
+ },
586
+ {
587
+ "epoch": 0.36,
588
+ "learning_rate": 1.6367285119273004e-05,
589
+ "loss": 0.8066,
590
+ "step": 4800
591
+ },
592
+ {
593
+ "epoch": 0.37,
594
+ "learning_rate": 1.6329420673987126e-05,
595
+ "loss": 0.761,
596
+ "step": 4850
597
+ },
598
+ {
599
+ "epoch": 0.37,
600
+ "learning_rate": 1.6291556228701252e-05,
601
+ "loss": 0.7919,
602
+ "step": 4900
603
+ },
604
+ {
605
+ "epoch": 0.37,
606
+ "learning_rate": 1.6253691783415374e-05,
607
+ "loss": 0.7484,
608
+ "step": 4950
609
+ },
610
+ {
611
+ "epoch": 0.38,
612
+ "learning_rate": 1.6215827338129497e-05,
613
+ "loss": 0.809,
614
+ "step": 5000
615
+ },
616
+ {
617
+ "epoch": 0.38,
618
+ "learning_rate": 1.6177962892843622e-05,
619
+ "loss": 0.7683,
620
+ "step": 5050
621
+ },
622
+ {
623
+ "epoch": 0.39,
624
+ "learning_rate": 1.6140098447557745e-05,
625
+ "loss": 0.7622,
626
+ "step": 5100
627
+ },
628
+ {
629
+ "epoch": 0.39,
630
+ "learning_rate": 1.6102234002271867e-05,
631
+ "loss": 0.7618,
632
+ "step": 5150
633
+ },
634
+ {
635
+ "epoch": 0.39,
636
+ "learning_rate": 1.6064369556985993e-05,
637
+ "loss": 0.7687,
638
+ "step": 5200
639
+ },
640
+ {
641
+ "epoch": 0.4,
642
+ "learning_rate": 1.6026505111700115e-05,
643
+ "loss": 0.7598,
644
+ "step": 5250
645
+ },
646
+ {
647
+ "epoch": 0.4,
648
+ "learning_rate": 1.5988640666414238e-05,
649
+ "loss": 0.7245,
650
+ "step": 5300
651
+ },
652
+ {
653
+ "epoch": 0.41,
654
+ "learning_rate": 1.595077622112836e-05,
655
+ "loss": 0.7751,
656
+ "step": 5350
657
+ },
658
+ {
659
+ "epoch": 0.41,
660
+ "learning_rate": 1.5912911775842486e-05,
661
+ "loss": 0.7903,
662
+ "step": 5400
663
+ },
664
+ {
665
+ "epoch": 0.41,
666
+ "learning_rate": 1.5875047330556608e-05,
667
+ "loss": 0.7399,
668
+ "step": 5450
669
+ },
670
+ {
671
+ "epoch": 0.42,
672
+ "learning_rate": 1.583718288527073e-05,
673
+ "loss": 0.7529,
674
+ "step": 5500
675
+ },
676
+ {
677
+ "epoch": 0.42,
678
+ "learning_rate": 1.5799318439984856e-05,
679
+ "loss": 0.7777,
680
+ "step": 5550
681
+ },
682
+ {
683
+ "epoch": 0.42,
684
+ "learning_rate": 1.576145399469898e-05,
685
+ "loss": 0.7411,
686
+ "step": 5600
687
+ },
688
+ {
689
+ "epoch": 0.43,
690
+ "learning_rate": 1.5723589549413105e-05,
691
+ "loss": 0.7805,
692
+ "step": 5650
693
+ },
694
+ {
695
+ "epoch": 0.43,
696
+ "learning_rate": 1.5685725104127224e-05,
697
+ "loss": 0.7585,
698
+ "step": 5700
699
+ },
700
+ {
701
+ "epoch": 0.44,
702
+ "learning_rate": 1.564786065884135e-05,
703
+ "loss": 0.7908,
704
+ "step": 5750
705
+ },
706
+ {
707
+ "epoch": 0.44,
708
+ "learning_rate": 1.560999621355547e-05,
709
+ "loss": 0.7302,
710
+ "step": 5800
711
+ },
712
+ {
713
+ "epoch": 0.44,
714
+ "learning_rate": 1.5572131768269597e-05,
715
+ "loss": 0.7492,
716
+ "step": 5850
717
+ },
718
+ {
719
+ "epoch": 0.45,
720
+ "learning_rate": 1.553426732298372e-05,
721
+ "loss": 0.8124,
722
+ "step": 5900
723
+ },
724
+ {
725
+ "epoch": 0.45,
726
+ "learning_rate": 1.5496402877697842e-05,
727
+ "loss": 0.7609,
728
+ "step": 5950
729
+ },
730
+ {
731
+ "epoch": 0.45,
732
+ "learning_rate": 1.5458538432411968e-05,
733
+ "loss": 0.7383,
734
+ "step": 6000
735
+ },
736
+ {
737
+ "epoch": 0.46,
738
+ "learning_rate": 1.542067398712609e-05,
739
+ "loss": 0.7585,
740
+ "step": 6050
741
+ },
742
+ {
743
+ "epoch": 0.46,
744
+ "learning_rate": 1.538356683074593e-05,
745
+ "loss": 0.7852,
746
+ "step": 6100
747
+ },
748
+ {
749
+ "epoch": 0.47,
750
+ "learning_rate": 1.5345702385460056e-05,
751
+ "loss": 0.7737,
752
+ "step": 6150
753
+ },
754
+ {
755
+ "epoch": 0.47,
756
+ "learning_rate": 1.5307837940174178e-05,
757
+ "loss": 0.7881,
758
+ "step": 6200
759
+ },
760
+ {
761
+ "epoch": 0.47,
762
+ "learning_rate": 1.52699734948883e-05,
763
+ "loss": 0.7716,
764
+ "step": 6250
765
+ },
766
+ {
767
+ "epoch": 0.48,
768
+ "learning_rate": 1.5232109049602425e-05,
769
+ "loss": 0.7478,
770
+ "step": 6300
771
+ },
772
+ {
773
+ "epoch": 0.48,
774
+ "learning_rate": 1.5194244604316549e-05,
775
+ "loss": 0.7681,
776
+ "step": 6350
777
+ },
778
+ {
779
+ "epoch": 0.48,
780
+ "learning_rate": 1.5156380159030673e-05,
781
+ "loss": 0.7703,
782
+ "step": 6400
783
+ },
784
+ {
785
+ "epoch": 0.49,
786
+ "learning_rate": 1.5118515713744795e-05,
787
+ "loss": 0.8031,
788
+ "step": 6450
789
+ },
790
+ {
791
+ "epoch": 0.49,
792
+ "learning_rate": 1.508065126845892e-05,
793
+ "loss": 0.7812,
794
+ "step": 6500
795
+ },
796
+ {
797
+ "epoch": 0.5,
798
+ "learning_rate": 1.5042786823173042e-05,
799
+ "loss": 0.8151,
800
+ "step": 6550
801
+ },
802
+ {
803
+ "epoch": 0.5,
804
+ "learning_rate": 1.5004922377887166e-05,
805
+ "loss": 0.8384,
806
+ "step": 6600
807
+ },
808
+ {
809
+ "epoch": 0.5,
810
+ "learning_rate": 1.4967057932601288e-05,
811
+ "loss": 0.7861,
812
+ "step": 6650
813
+ },
814
+ {
815
+ "epoch": 0.51,
816
+ "learning_rate": 1.4929193487315412e-05,
817
+ "loss": 0.7473,
818
+ "step": 6700
819
+ },
820
+ {
821
+ "epoch": 0.51,
822
+ "learning_rate": 1.4891329042029536e-05,
823
+ "loss": 0.7741,
824
+ "step": 6750
825
+ },
826
+ {
827
+ "epoch": 0.51,
828
+ "learning_rate": 1.4853464596743659e-05,
829
+ "loss": 0.799,
830
+ "step": 6800
831
+ },
832
+ {
833
+ "epoch": 0.52,
834
+ "learning_rate": 1.4815600151457783e-05,
835
+ "loss": 0.7929,
836
+ "step": 6850
837
+ },
838
+ {
839
+ "epoch": 0.52,
840
+ "learning_rate": 1.4777735706171905e-05,
841
+ "loss": 0.7594,
842
+ "step": 6900
843
+ },
844
+ {
845
+ "epoch": 0.53,
846
+ "learning_rate": 1.4739871260886029e-05,
847
+ "loss": 0.736,
848
+ "step": 6950
849
+ },
850
+ {
851
+ "epoch": 0.53,
852
+ "learning_rate": 1.4702006815600151e-05,
853
+ "loss": 0.7608,
854
+ "step": 7000
855
+ },
856
+ {
857
+ "epoch": 0.53,
858
+ "learning_rate": 1.4664142370314276e-05,
859
+ "loss": 0.7927,
860
+ "step": 7050
861
+ },
862
+ {
863
+ "epoch": 0.54,
864
+ "learning_rate": 1.46262779250284e-05,
865
+ "loss": 0.7753,
866
+ "step": 7100
867
+ },
868
+ {
869
+ "epoch": 0.54,
870
+ "learning_rate": 1.4588413479742522e-05,
871
+ "loss": 0.7476,
872
+ "step": 7150
873
+ },
874
+ {
875
+ "epoch": 0.55,
876
+ "learning_rate": 1.4550549034456646e-05,
877
+ "loss": 0.7907,
878
+ "step": 7200
879
+ },
880
+ {
881
+ "epoch": 0.55,
882
+ "learning_rate": 1.4512684589170768e-05,
883
+ "loss": 0.7874,
884
+ "step": 7250
885
+ },
886
+ {
887
+ "epoch": 0.55,
888
+ "learning_rate": 1.4474820143884894e-05,
889
+ "loss": 0.7626,
890
+ "step": 7300
891
+ },
892
+ {
893
+ "epoch": 0.56,
894
+ "learning_rate": 1.4436955698599018e-05,
895
+ "loss": 0.7527,
896
+ "step": 7350
897
+ },
898
+ {
899
+ "epoch": 0.56,
900
+ "learning_rate": 1.439909125331314e-05,
901
+ "loss": 0.7451,
902
+ "step": 7400
903
+ },
904
+ {
905
+ "epoch": 0.56,
906
+ "learning_rate": 1.4361226808027265e-05,
907
+ "loss": 0.7615,
908
+ "step": 7450
909
+ },
910
+ {
911
+ "epoch": 0.57,
912
+ "learning_rate": 1.4323362362741387e-05,
913
+ "loss": 0.7365,
914
+ "step": 7500
915
+ },
916
+ {
917
+ "epoch": 0.57,
918
+ "learning_rate": 1.4285497917455511e-05,
919
+ "loss": 0.7627,
920
+ "step": 7550
921
+ },
922
+ {
923
+ "epoch": 0.58,
924
+ "learning_rate": 1.4247633472169634e-05,
925
+ "loss": 0.7769,
926
+ "step": 7600
927
+ },
928
+ {
929
+ "epoch": 0.58,
930
+ "learning_rate": 1.4209769026883758e-05,
931
+ "loss": 0.7253,
932
+ "step": 7650
933
+ },
934
+ {
935
+ "epoch": 0.58,
936
+ "learning_rate": 1.4171904581597882e-05,
937
+ "loss": 0.7422,
938
+ "step": 7700
939
+ },
940
+ {
941
+ "epoch": 0.59,
942
+ "learning_rate": 1.4134040136312004e-05,
943
+ "loss": 0.7459,
944
+ "step": 7750
945
+ },
946
+ {
947
+ "epoch": 0.59,
948
+ "learning_rate": 1.4096175691026128e-05,
949
+ "loss": 0.7355,
950
+ "step": 7800
951
+ },
952
+ {
953
+ "epoch": 0.59,
954
+ "learning_rate": 1.405831124574025e-05,
955
+ "loss": 0.7658,
956
+ "step": 7850
957
+ },
958
+ {
959
+ "epoch": 0.6,
960
+ "learning_rate": 1.4020446800454375e-05,
961
+ "loss": 0.8133,
962
+ "step": 7900
963
+ },
964
+ {
965
+ "epoch": 0.6,
966
+ "learning_rate": 1.3982582355168497e-05,
967
+ "loss": 0.7574,
968
+ "step": 7950
969
+ },
970
+ {
971
+ "epoch": 0.61,
972
+ "learning_rate": 1.3944717909882621e-05,
973
+ "loss": 0.7755,
974
+ "step": 8000
975
+ },
976
+ {
977
+ "epoch": 0.61,
978
+ "learning_rate": 1.3906853464596745e-05,
979
+ "loss": 0.7434,
980
+ "step": 8050
981
+ },
982
+ {
983
+ "epoch": 0.61,
984
+ "learning_rate": 1.3868989019310868e-05,
985
+ "loss": 0.7565,
986
+ "step": 8100
987
+ },
988
+ {
989
+ "epoch": 0.62,
990
+ "learning_rate": 1.3831124574024992e-05,
991
+ "loss": 0.7705,
992
+ "step": 8150
993
+ },
994
+ {
995
+ "epoch": 0.62,
996
+ "learning_rate": 1.3793260128739114e-05,
997
+ "loss": 0.7699,
998
+ "step": 8200
999
+ },
1000
+ {
1001
+ "epoch": 0.62,
1002
+ "learning_rate": 1.3755395683453238e-05,
1003
+ "loss": 0.7834,
1004
+ "step": 8250
1005
+ },
1006
+ {
1007
+ "epoch": 0.63,
1008
+ "learning_rate": 1.3718288527073081e-05,
1009
+ "loss": 0.7788,
1010
+ "step": 8300
1011
+ },
1012
+ {
1013
+ "epoch": 0.63,
1014
+ "learning_rate": 1.3680424081787202e-05,
1015
+ "loss": 0.7329,
1016
+ "step": 8350
1017
+ },
1018
+ {
1019
+ "epoch": 0.64,
1020
+ "learning_rate": 1.3642559636501328e-05,
1021
+ "loss": 0.7774,
1022
+ "step": 8400
1023
+ },
1024
+ {
1025
+ "epoch": 0.64,
1026
+ "learning_rate": 1.3604695191215448e-05,
1027
+ "loss": 0.7427,
1028
+ "step": 8450
1029
+ },
1030
+ {
1031
+ "epoch": 0.64,
1032
+ "learning_rate": 1.3566830745929574e-05,
1033
+ "loss": 0.7482,
1034
+ "step": 8500
1035
+ },
1036
+ {
1037
+ "epoch": 0.65,
1038
+ "learning_rate": 1.3528966300643698e-05,
1039
+ "loss": 0.7888,
1040
+ "step": 8550
1041
+ },
1042
+ {
1043
+ "epoch": 0.65,
1044
+ "learning_rate": 1.349110185535782e-05,
1045
+ "loss": 0.7535,
1046
+ "step": 8600
1047
+ },
1048
+ {
1049
+ "epoch": 0.66,
1050
+ "learning_rate": 1.3453237410071945e-05,
1051
+ "loss": 0.7456,
1052
+ "step": 8650
1053
+ },
1054
+ {
1055
+ "epoch": 0.66,
1056
+ "learning_rate": 1.3415372964786067e-05,
1057
+ "loss": 0.7964,
1058
+ "step": 8700
1059
+ },
1060
+ {
1061
+ "epoch": 0.66,
1062
+ "learning_rate": 1.3377508519500191e-05,
1063
+ "loss": 0.7913,
1064
+ "step": 8750
1065
+ },
1066
+ {
1067
+ "epoch": 0.67,
1068
+ "learning_rate": 1.3339644074214313e-05,
1069
+ "loss": 0.7745,
1070
+ "step": 8800
1071
+ },
1072
+ {
1073
+ "epoch": 0.67,
1074
+ "learning_rate": 1.3301779628928438e-05,
1075
+ "loss": 0.7595,
1076
+ "step": 8850
1077
+ },
1078
+ {
1079
+ "epoch": 0.67,
1080
+ "learning_rate": 1.3263915183642562e-05,
1081
+ "loss": 0.7117,
1082
+ "step": 8900
1083
+ },
1084
+ {
1085
+ "epoch": 0.68,
1086
+ "learning_rate": 1.3226050738356684e-05,
1087
+ "loss": 0.8142,
1088
+ "step": 8950
1089
+ },
1090
+ {
1091
+ "epoch": 0.68,
1092
+ "learning_rate": 1.3188186293070808e-05,
1093
+ "loss": 0.7539,
1094
+ "step": 9000
1095
+ },
1096
+ {
1097
+ "epoch": 0.69,
1098
+ "learning_rate": 1.315032184778493e-05,
1099
+ "loss": 0.7135,
1100
+ "step": 9050
1101
+ },
1102
+ {
1103
+ "epoch": 0.69,
1104
+ "learning_rate": 1.3112457402499055e-05,
1105
+ "loss": 0.7883,
1106
+ "step": 9100
1107
+ },
1108
+ {
1109
+ "epoch": 0.69,
1110
+ "learning_rate": 1.3074592957213177e-05,
1111
+ "loss": 0.7335,
1112
+ "step": 9150
1113
+ },
1114
+ {
1115
+ "epoch": 0.7,
1116
+ "learning_rate": 1.3036728511927301e-05,
1117
+ "loss": 0.771,
1118
+ "step": 9200
1119
+ },
1120
+ {
1121
+ "epoch": 0.7,
1122
+ "learning_rate": 1.2998864066641425e-05,
1123
+ "loss": 0.7341,
1124
+ "step": 9250
1125
+ },
1126
+ {
1127
+ "epoch": 0.7,
1128
+ "learning_rate": 1.2960999621355547e-05,
1129
+ "loss": 0.7177,
1130
+ "step": 9300
1131
+ },
1132
+ {
1133
+ "epoch": 0.71,
1134
+ "learning_rate": 1.2923135176069672e-05,
1135
+ "loss": 0.7784,
1136
+ "step": 9350
1137
+ },
1138
+ {
1139
+ "epoch": 0.71,
1140
+ "learning_rate": 1.2885270730783794e-05,
1141
+ "loss": 0.7906,
1142
+ "step": 9400
1143
+ },
1144
+ {
1145
+ "epoch": 0.72,
1146
+ "learning_rate": 1.2847406285497918e-05,
1147
+ "loss": 0.7762,
1148
+ "step": 9450
1149
+ },
1150
+ {
1151
+ "epoch": 0.72,
1152
+ "learning_rate": 1.280954184021204e-05,
1153
+ "loss": 0.7482,
1154
+ "step": 9500
1155
+ },
1156
+ {
1157
+ "epoch": 0.72,
1158
+ "learning_rate": 1.2771677394926164e-05,
1159
+ "loss": 0.7069,
1160
+ "step": 9550
1161
+ },
1162
+ {
1163
+ "epoch": 0.73,
1164
+ "learning_rate": 1.2734570238546007e-05,
1165
+ "loss": 0.761,
1166
+ "step": 9600
1167
+ },
1168
+ {
1169
+ "epoch": 0.73,
1170
+ "learning_rate": 1.269670579326013e-05,
1171
+ "loss": 0.7375,
1172
+ "step": 9650
1173
+ },
1174
+ {
1175
+ "epoch": 0.73,
1176
+ "learning_rate": 1.2658841347974254e-05,
1177
+ "loss": 0.7937,
1178
+ "step": 9700
1179
+ },
1180
+ {
1181
+ "epoch": 0.74,
1182
+ "learning_rate": 1.2620976902688378e-05,
1183
+ "loss": 0.7983,
1184
+ "step": 9750
1185
+ },
1186
+ {
1187
+ "epoch": 0.74,
1188
+ "learning_rate": 1.25831124574025e-05,
1189
+ "loss": 0.7339,
1190
+ "step": 9800
1191
+ },
1192
+ {
1193
+ "epoch": 0.75,
1194
+ "learning_rate": 1.2545248012116624e-05,
1195
+ "loss": 0.7671,
1196
+ "step": 9850
1197
+ },
1198
+ {
1199
+ "epoch": 0.75,
1200
+ "learning_rate": 1.2507383566830747e-05,
1201
+ "loss": 0.7311,
1202
+ "step": 9900
1203
+ },
1204
+ {
1205
+ "epoch": 0.75,
1206
+ "learning_rate": 1.2469519121544871e-05,
1207
+ "loss": 0.7652,
1208
+ "step": 9950
1209
+ },
1210
+ {
1211
+ "epoch": 0.76,
1212
+ "learning_rate": 1.2431654676258993e-05,
1213
+ "loss": 0.7549,
1214
+ "step": 10000
1215
+ },
1216
+ {
1217
+ "epoch": 0.76,
1218
+ "learning_rate": 1.2393790230973117e-05,
1219
+ "loss": 0.7802,
1220
+ "step": 10050
1221
+ },
1222
+ {
1223
+ "epoch": 0.76,
1224
+ "learning_rate": 1.2355925785687241e-05,
1225
+ "loss": 0.7573,
1226
+ "step": 10100
1227
+ },
1228
+ {
1229
+ "epoch": 0.77,
1230
+ "learning_rate": 1.2318061340401364e-05,
1231
+ "loss": 0.695,
1232
+ "step": 10150
1233
+ },
1234
+ {
1235
+ "epoch": 0.77,
1236
+ "learning_rate": 1.2280196895115488e-05,
1237
+ "loss": 0.7214,
1238
+ "step": 10200
1239
+ },
1240
+ {
1241
+ "epoch": 0.78,
1242
+ "learning_rate": 1.224233244982961e-05,
1243
+ "loss": 0.772,
1244
+ "step": 10250
1245
+ },
1246
+ {
1247
+ "epoch": 0.78,
1248
+ "learning_rate": 1.2204468004543734e-05,
1249
+ "loss": 0.7855,
1250
+ "step": 10300
1251
+ },
1252
+ {
1253
+ "epoch": 0.78,
1254
+ "learning_rate": 1.2166603559257857e-05,
1255
+ "loss": 0.7345,
1256
+ "step": 10350
1257
+ },
1258
+ {
1259
+ "epoch": 0.79,
1260
+ "learning_rate": 1.212873911397198e-05,
1261
+ "loss": 0.7416,
1262
+ "step": 10400
1263
+ },
1264
+ {
1265
+ "epoch": 0.79,
1266
+ "learning_rate": 1.2090874668686105e-05,
1267
+ "loss": 0.7723,
1268
+ "step": 10450
1269
+ },
1270
+ {
1271
+ "epoch": 0.8,
1272
+ "learning_rate": 1.2053010223400227e-05,
1273
+ "loss": 0.7446,
1274
+ "step": 10500
1275
+ },
1276
+ {
1277
+ "epoch": 0.8,
1278
+ "learning_rate": 1.2015145778114351e-05,
1279
+ "loss": 0.7657,
1280
+ "step": 10550
1281
+ },
1282
+ {
1283
+ "epoch": 0.8,
1284
+ "learning_rate": 1.1977281332828474e-05,
1285
+ "loss": 0.7944,
1286
+ "step": 10600
1287
+ },
1288
+ {
1289
+ "epoch": 0.81,
1290
+ "learning_rate": 1.1939416887542598e-05,
1291
+ "loss": 0.7598,
1292
+ "step": 10650
1293
+ },
1294
+ {
1295
+ "epoch": 0.81,
1296
+ "learning_rate": 1.1901552442256724e-05,
1297
+ "loss": 0.7212,
1298
+ "step": 10700
1299
+ },
1300
+ {
1301
+ "epoch": 0.81,
1302
+ "learning_rate": 1.1863687996970846e-05,
1303
+ "loss": 0.7564,
1304
+ "step": 10750
1305
+ },
1306
+ {
1307
+ "epoch": 0.82,
1308
+ "learning_rate": 1.182582355168497e-05,
1309
+ "loss": 0.7552,
1310
+ "step": 10800
1311
+ },
1312
+ {
1313
+ "epoch": 0.82,
1314
+ "learning_rate": 1.1787959106399092e-05,
1315
+ "loss": 0.7582,
1316
+ "step": 10850
1317
+ },
1318
+ {
1319
+ "epoch": 0.83,
1320
+ "learning_rate": 1.1750094661113217e-05,
1321
+ "loss": 0.7506,
1322
+ "step": 10900
1323
+ },
1324
+ {
1325
+ "epoch": 0.83,
1326
+ "learning_rate": 1.1712230215827339e-05,
1327
+ "loss": 0.7738,
1328
+ "step": 10950
1329
+ },
1330
+ {
1331
+ "epoch": 0.83,
1332
+ "learning_rate": 1.1674365770541463e-05,
1333
+ "loss": 0.7295,
1334
+ "step": 11000
1335
+ },
1336
+ {
1337
+ "epoch": 0.84,
1338
+ "learning_rate": 1.1636501325255587e-05,
1339
+ "loss": 0.7627,
1340
+ "step": 11050
1341
+ },
1342
+ {
1343
+ "epoch": 0.84,
1344
+ "learning_rate": 1.159863687996971e-05,
1345
+ "loss": 0.7918,
1346
+ "step": 11100
1347
+ },
1348
+ {
1349
+ "epoch": 0.84,
1350
+ "learning_rate": 1.1560772434683833e-05,
1351
+ "loss": 0.7235,
1352
+ "step": 11150
1353
+ },
1354
+ {
1355
+ "epoch": 0.85,
1356
+ "learning_rate": 1.1522907989397956e-05,
1357
+ "loss": 0.7988,
1358
+ "step": 11200
1359
+ },
1360
+ {
1361
+ "epoch": 0.85,
1362
+ "learning_rate": 1.148504354411208e-05,
1363
+ "loss": 0.7646,
1364
+ "step": 11250
1365
+ },
1366
+ {
1367
+ "epoch": 0.86,
1368
+ "learning_rate": 1.1447179098826202e-05,
1369
+ "loss": 0.7794,
1370
+ "step": 11300
1371
+ },
1372
+ {
1373
+ "epoch": 0.86,
1374
+ "learning_rate": 1.1409314653540326e-05,
1375
+ "loss": 0.79,
1376
+ "step": 11350
1377
+ },
1378
+ {
1379
+ "epoch": 0.86,
1380
+ "learning_rate": 1.137145020825445e-05,
1381
+ "loss": 0.7291,
1382
+ "step": 11400
1383
+ },
1384
+ {
1385
+ "epoch": 0.87,
1386
+ "learning_rate": 1.1333585762968573e-05,
1387
+ "loss": 0.7682,
1388
+ "step": 11450
1389
+ },
1390
+ {
1391
+ "epoch": 0.87,
1392
+ "learning_rate": 1.1295721317682697e-05,
1393
+ "loss": 0.8078,
1394
+ "step": 11500
1395
+ },
1396
+ {
1397
+ "epoch": 0.87,
1398
+ "learning_rate": 1.125785687239682e-05,
1399
+ "loss": 0.8196,
1400
+ "step": 11550
1401
+ },
1402
+ {
1403
+ "epoch": 0.88,
1404
+ "learning_rate": 1.1219992427110943e-05,
1405
+ "loss": 0.7173,
1406
+ "step": 11600
1407
+ },
1408
+ {
1409
+ "epoch": 0.88,
1410
+ "learning_rate": 1.1182127981825066e-05,
1411
+ "loss": 0.7908,
1412
+ "step": 11650
1413
+ },
1414
+ {
1415
+ "epoch": 0.89,
1416
+ "learning_rate": 1.114426353653919e-05,
1417
+ "loss": 0.7784,
1418
+ "step": 11700
1419
+ },
1420
+ {
1421
+ "epoch": 0.89,
1422
+ "learning_rate": 1.1106399091253316e-05,
1423
+ "loss": 0.7317,
1424
+ "step": 11750
1425
+ },
1426
+ {
1427
+ "epoch": 0.89,
1428
+ "learning_rate": 1.1068534645967436e-05,
1429
+ "loss": 0.7497,
1430
+ "step": 11800
1431
+ },
1432
+ {
1433
+ "epoch": 0.9,
1434
+ "learning_rate": 1.1030670200681562e-05,
1435
+ "loss": 0.7504,
1436
+ "step": 11850
1437
+ },
1438
+ {
1439
+ "epoch": 0.9,
1440
+ "learning_rate": 1.0992805755395683e-05,
1441
+ "loss": 0.7962,
1442
+ "step": 11900
1443
+ },
1444
+ {
1445
+ "epoch": 0.9,
1446
+ "learning_rate": 1.0954941310109809e-05,
1447
+ "loss": 0.7615,
1448
+ "step": 11950
1449
+ },
1450
+ {
1451
+ "epoch": 0.91,
1452
+ "learning_rate": 1.0917076864823933e-05,
1453
+ "loss": 0.7485,
1454
+ "step": 12000
1455
+ },
1456
+ {
1457
+ "epoch": 0.91,
1458
+ "learning_rate": 1.0879969708443772e-05,
1459
+ "loss": 0.7088,
1460
+ "step": 12050
1461
+ },
1462
+ {
1463
+ "epoch": 0.92,
1464
+ "learning_rate": 1.0842105263157896e-05,
1465
+ "loss": 0.7479,
1466
+ "step": 12100
1467
+ },
1468
+ {
1469
+ "epoch": 0.92,
1470
+ "learning_rate": 1.0804240817872019e-05,
1471
+ "loss": 0.7571,
1472
+ "step": 12150
1473
+ },
1474
+ {
1475
+ "epoch": 0.92,
1476
+ "learning_rate": 1.0766376372586143e-05,
1477
+ "loss": 0.7576,
1478
+ "step": 12200
1479
+ },
1480
+ {
1481
+ "epoch": 0.93,
1482
+ "learning_rate": 1.0728511927300267e-05,
1483
+ "loss": 0.7267,
1484
+ "step": 12250
1485
+ },
1486
+ {
1487
+ "epoch": 0.93,
1488
+ "learning_rate": 1.069064748201439e-05,
1489
+ "loss": 0.7521,
1490
+ "step": 12300
1491
+ },
1492
+ {
1493
+ "epoch": 0.94,
1494
+ "learning_rate": 1.0652783036728513e-05,
1495
+ "loss": 0.7616,
1496
+ "step": 12350
1497
+ },
1498
+ {
1499
+ "epoch": 0.94,
1500
+ "learning_rate": 1.0614918591442636e-05,
1501
+ "loss": 0.7232,
1502
+ "step": 12400
1503
+ },
1504
+ {
1505
+ "epoch": 0.94,
1506
+ "learning_rate": 1.057705414615676e-05,
1507
+ "loss": 0.797,
1508
+ "step": 12450
1509
+ },
1510
+ {
1511
+ "epoch": 0.95,
1512
+ "learning_rate": 1.0539189700870882e-05,
1513
+ "loss": 0.7575,
1514
+ "step": 12500
1515
+ },
1516
+ {
1517
+ "epoch": 0.95,
1518
+ "learning_rate": 1.0501325255585006e-05,
1519
+ "loss": 0.7384,
1520
+ "step": 12550
1521
+ },
1522
+ {
1523
+ "epoch": 0.95,
1524
+ "learning_rate": 1.046346081029913e-05,
1525
+ "loss": 0.7523,
1526
+ "step": 12600
1527
+ },
1528
+ {
1529
+ "epoch": 0.96,
1530
+ "learning_rate": 1.0425596365013253e-05,
1531
+ "loss": 0.7271,
1532
+ "step": 12650
1533
+ },
1534
+ {
1535
+ "epoch": 0.96,
1536
+ "learning_rate": 1.0387731919727377e-05,
1537
+ "loss": 0.7746,
1538
+ "step": 12700
1539
+ },
1540
+ {
1541
+ "epoch": 0.97,
1542
+ "learning_rate": 1.03498674744415e-05,
1543
+ "loss": 0.7463,
1544
+ "step": 12750
1545
+ },
1546
+ {
1547
+ "epoch": 0.97,
1548
+ "learning_rate": 1.0312003029155623e-05,
1549
+ "loss": 0.7234,
1550
+ "step": 12800
1551
+ },
1552
+ {
1553
+ "epoch": 0.97,
1554
+ "learning_rate": 1.0274138583869746e-05,
1555
+ "loss": 0.7462,
1556
+ "step": 12850
1557
+ },
1558
+ {
1559
+ "epoch": 0.98,
1560
+ "learning_rate": 1.023627413858387e-05,
1561
+ "loss": 0.7391,
1562
+ "step": 12900
1563
+ },
1564
+ {
1565
+ "epoch": 0.98,
1566
+ "learning_rate": 1.0198409693297995e-05,
1567
+ "loss": 0.7761,
1568
+ "step": 12950
1569
+ },
1570
+ {
1571
+ "epoch": 0.98,
1572
+ "learning_rate": 1.0160545248012116e-05,
1573
+ "loss": 0.7849,
1574
+ "step": 13000
1575
+ },
1576
+ {
1577
+ "epoch": 0.99,
1578
+ "learning_rate": 1.0122680802726242e-05,
1579
+ "loss": 0.7391,
1580
+ "step": 13050
1581
+ },
1582
+ {
1583
+ "epoch": 0.99,
1584
+ "learning_rate": 1.0084816357440364e-05,
1585
+ "loss": 0.7203,
1586
+ "step": 13100
1587
+ },
1588
+ {
1589
+ "epoch": 1.0,
1590
+ "learning_rate": 1.0046951912154488e-05,
1591
+ "loss": 0.7439,
1592
+ "step": 13150
1593
+ },
1594
+ {
1595
+ "epoch": 1.0,
1596
+ "learning_rate": 1.0009087466868612e-05,
1597
+ "loss": 0.7551,
1598
+ "step": 13200
1599
+ },
1600
+ {
1601
+ "epoch": 1.0,
1602
+ "learning_rate": 9.971223021582735e-06,
1603
+ "loss": 0.6935,
1604
+ "step": 13250
1605
+ },
1606
+ {
1607
+ "epoch": 1.01,
1608
+ "learning_rate": 9.933358576296857e-06,
1609
+ "loss": 0.7541,
1610
+ "step": 13300
1611
+ },
1612
+ {
1613
+ "epoch": 1.01,
1614
+ "learning_rate": 9.895494131010981e-06,
1615
+ "loss": 0.7828,
1616
+ "step": 13350
1617
+ },
1618
+ {
1619
+ "epoch": 1.01,
1620
+ "learning_rate": 9.857629685725105e-06,
1621
+ "loss": 0.7204,
1622
+ "step": 13400
1623
+ },
1624
+ {
1625
+ "epoch": 1.02,
1626
+ "learning_rate": 9.819765240439228e-06,
1627
+ "loss": 0.7374,
1628
+ "step": 13450
1629
+ },
1630
+ {
1631
+ "epoch": 1.02,
1632
+ "learning_rate": 9.781900795153352e-06,
1633
+ "loss": 0.7477,
1634
+ "step": 13500
1635
+ },
1636
+ {
1637
+ "epoch": 1.03,
1638
+ "learning_rate": 9.744036349867474e-06,
1639
+ "loss": 0.7516,
1640
+ "step": 13550
1641
+ },
1642
+ {
1643
+ "epoch": 1.03,
1644
+ "learning_rate": 9.706171904581598e-06,
1645
+ "loss": 0.727,
1646
+ "step": 13600
1647
+ },
1648
+ {
1649
+ "epoch": 1.03,
1650
+ "learning_rate": 9.668307459295722e-06,
1651
+ "loss": 0.7343,
1652
+ "step": 13650
1653
+ },
1654
+ {
1655
+ "epoch": 1.04,
1656
+ "learning_rate": 9.630443014009846e-06,
1657
+ "loss": 0.7739,
1658
+ "step": 13700
1659
+ },
1660
+ {
1661
+ "epoch": 1.04,
1662
+ "learning_rate": 9.592578568723969e-06,
1663
+ "loss": 0.7447,
1664
+ "step": 13750
1665
+ },
1666
+ {
1667
+ "epoch": 1.04,
1668
+ "learning_rate": 9.554714123438093e-06,
1669
+ "loss": 0.7607,
1670
+ "step": 13800
1671
+ },
1672
+ {
1673
+ "epoch": 1.05,
1674
+ "learning_rate": 9.516849678152215e-06,
1675
+ "loss": 0.79,
1676
+ "step": 13850
1677
+ },
1678
+ {
1679
+ "epoch": 1.05,
1680
+ "learning_rate": 9.47898523286634e-06,
1681
+ "loss": 0.7209,
1682
+ "step": 13900
1683
+ },
1684
+ {
1685
+ "epoch": 1.06,
1686
+ "learning_rate": 9.441120787580462e-06,
1687
+ "loss": 0.7745,
1688
+ "step": 13950
1689
+ },
1690
+ {
1691
+ "epoch": 1.06,
1692
+ "learning_rate": 9.403256342294586e-06,
1693
+ "loss": 0.7179,
1694
+ "step": 14000
1695
+ },
1696
+ {
1697
+ "epoch": 1.06,
1698
+ "learning_rate": 9.36539189700871e-06,
1699
+ "loss": 0.774,
1700
+ "step": 14050
1701
+ },
1702
+ {
1703
+ "epoch": 1.07,
1704
+ "learning_rate": 9.327527451722832e-06,
1705
+ "loss": 0.7592,
1706
+ "step": 14100
1707
+ },
1708
+ {
1709
+ "epoch": 1.07,
1710
+ "learning_rate": 9.289663006436956e-06,
1711
+ "loss": 0.7598,
1712
+ "step": 14150
1713
+ },
1714
+ {
1715
+ "epoch": 1.08,
1716
+ "learning_rate": 9.25179856115108e-06,
1717
+ "loss": 0.7665,
1718
+ "step": 14200
1719
+ },
1720
+ {
1721
+ "epoch": 1.08,
1722
+ "learning_rate": 9.213934115865203e-06,
1723
+ "loss": 0.7529,
1724
+ "step": 14250
1725
+ },
1726
+ {
1727
+ "epoch": 1.08,
1728
+ "learning_rate": 9.176069670579327e-06,
1729
+ "loss": 0.6873,
1730
+ "step": 14300
1731
+ },
1732
+ {
1733
+ "epoch": 1.09,
1734
+ "learning_rate": 9.138205225293451e-06,
1735
+ "loss": 0.7574,
1736
+ "step": 14350
1737
+ },
1738
+ {
1739
+ "epoch": 1.09,
1740
+ "learning_rate": 9.100340780007573e-06,
1741
+ "loss": 0.7215,
1742
+ "step": 14400
1743
+ },
1744
+ {
1745
+ "epoch": 1.09,
1746
+ "learning_rate": 9.062476334721697e-06,
1747
+ "loss": 0.7411,
1748
+ "step": 14450
1749
+ },
1750
+ {
1751
+ "epoch": 1.1,
1752
+ "learning_rate": 9.02461188943582e-06,
1753
+ "loss": 0.7627,
1754
+ "step": 14500
1755
+ },
1756
+ {
1757
+ "epoch": 1.1,
1758
+ "learning_rate": 8.986747444149944e-06,
1759
+ "loss": 0.7313,
1760
+ "step": 14550
1761
+ },
1762
+ {
1763
+ "epoch": 1.11,
1764
+ "learning_rate": 8.948882998864066e-06,
1765
+ "loss": 0.765,
1766
+ "step": 14600
1767
+ },
1768
+ {
1769
+ "epoch": 1.11,
1770
+ "learning_rate": 8.911018553578192e-06,
1771
+ "loss": 0.7696,
1772
+ "step": 14650
1773
+ },
1774
+ {
1775
+ "epoch": 1.11,
1776
+ "learning_rate": 8.873154108292314e-06,
1777
+ "loss": 0.7305,
1778
+ "step": 14700
1779
+ },
1780
+ {
1781
+ "epoch": 1.12,
1782
+ "learning_rate": 8.835289663006439e-06,
1783
+ "loss": 0.7653,
1784
+ "step": 14750
1785
+ },
1786
+ {
1787
+ "epoch": 1.12,
1788
+ "learning_rate": 8.797425217720561e-06,
1789
+ "loss": 0.7444,
1790
+ "step": 14800
1791
+ },
1792
+ {
1793
+ "epoch": 1.12,
1794
+ "learning_rate": 8.759560772434685e-06,
1795
+ "loss": 0.7858,
1796
+ "step": 14850
1797
+ },
1798
+ {
1799
+ "epoch": 1.13,
1800
+ "learning_rate": 8.721696327148807e-06,
1801
+ "loss": 0.7634,
1802
+ "step": 14900
1803
+ },
1804
+ {
1805
+ "epoch": 1.13,
1806
+ "learning_rate": 8.683831881862931e-06,
1807
+ "loss": 0.7471,
1808
+ "step": 14950
1809
+ },
1810
+ {
1811
+ "epoch": 1.14,
1812
+ "learning_rate": 8.645967436577056e-06,
1813
+ "loss": 0.7065,
1814
+ "step": 15000
1815
+ },
1816
+ {
1817
+ "epoch": 1.14,
1818
+ "learning_rate": 8.608102991291178e-06,
1819
+ "loss": 0.7659,
1820
+ "step": 15050
1821
+ },
1822
+ {
1823
+ "epoch": 1.14,
1824
+ "learning_rate": 8.570238546005302e-06,
1825
+ "loss": 0.7085,
1826
+ "step": 15100
1827
+ },
1828
+ {
1829
+ "epoch": 1.15,
1830
+ "learning_rate": 8.532374100719424e-06,
1831
+ "loss": 0.7312,
1832
+ "step": 15150
1833
+ },
1834
+ {
1835
+ "epoch": 1.15,
1836
+ "learning_rate": 8.494509655433548e-06,
1837
+ "loss": 0.814,
1838
+ "step": 15200
1839
+ },
1840
+ {
1841
+ "epoch": 1.15,
1842
+ "learning_rate": 8.45664521014767e-06,
1843
+ "loss": 0.7194,
1844
+ "step": 15250
1845
+ },
1846
+ {
1847
+ "epoch": 1.16,
1848
+ "learning_rate": 8.418780764861797e-06,
1849
+ "loss": 0.7266,
1850
+ "step": 15300
1851
+ },
1852
+ {
1853
+ "epoch": 1.16,
1854
+ "learning_rate": 8.380916319575919e-06,
1855
+ "loss": 0.7345,
1856
+ "step": 15350
1857
+ },
1858
+ {
1859
+ "epoch": 1.17,
1860
+ "learning_rate": 8.343051874290043e-06,
1861
+ "loss": 0.7121,
1862
+ "step": 15400
1863
+ },
1864
+ {
1865
+ "epoch": 1.17,
1866
+ "learning_rate": 8.305187429004165e-06,
1867
+ "loss": 0.7716,
1868
+ "step": 15450
1869
+ },
1870
+ {
1871
+ "epoch": 1.17,
1872
+ "learning_rate": 8.26732298371829e-06,
1873
+ "loss": 0.7601,
1874
+ "step": 15500
1875
+ },
1876
+ {
1877
+ "epoch": 1.18,
1878
+ "learning_rate": 8.229458538432412e-06,
1879
+ "loss": 0.736,
1880
+ "step": 15550
1881
+ },
1882
+ {
1883
+ "epoch": 1.18,
1884
+ "learning_rate": 8.191594093146536e-06,
1885
+ "loss": 0.7659,
1886
+ "step": 15600
1887
+ },
1888
+ {
1889
+ "epoch": 1.19,
1890
+ "learning_rate": 8.15372964786066e-06,
1891
+ "loss": 0.7504,
1892
+ "step": 15650
1893
+ },
1894
+ {
1895
+ "epoch": 1.19,
1896
+ "learning_rate": 8.115865202574782e-06,
1897
+ "loss": 0.7367,
1898
+ "step": 15700
1899
+ },
1900
+ {
1901
+ "epoch": 1.19,
1902
+ "learning_rate": 8.078000757288907e-06,
1903
+ "loss": 0.739,
1904
+ "step": 15750
1905
+ },
1906
+ {
1907
+ "epoch": 1.2,
1908
+ "learning_rate": 8.040136312003029e-06,
1909
+ "loss": 0.7421,
1910
+ "step": 15800
1911
+ },
1912
+ {
1913
+ "epoch": 1.2,
1914
+ "learning_rate": 8.002271866717153e-06,
1915
+ "loss": 0.7181,
1916
+ "step": 15850
1917
+ },
1918
+ {
1919
+ "epoch": 1.2,
1920
+ "learning_rate": 7.964407421431277e-06,
1921
+ "loss": 0.7699,
1922
+ "step": 15900
1923
+ },
1924
+ {
1925
+ "epoch": 1.21,
1926
+ "learning_rate": 7.926542976145401e-06,
1927
+ "loss": 0.7725,
1928
+ "step": 15950
1929
+ },
1930
+ {
1931
+ "epoch": 1.21,
1932
+ "learning_rate": 7.888678530859524e-06,
1933
+ "loss": 0.7658,
1934
+ "step": 16000
1935
+ },
1936
+ {
1937
+ "epoch": 1.22,
1938
+ "learning_rate": 7.850814085573648e-06,
1939
+ "loss": 0.6961,
1940
+ "step": 16050
1941
+ },
1942
+ {
1943
+ "epoch": 1.22,
1944
+ "learning_rate": 7.813706929193487e-06,
1945
+ "loss": 0.7598,
1946
+ "step": 16100
1947
+ },
1948
+ {
1949
+ "epoch": 1.22,
1950
+ "learning_rate": 7.775842483907611e-06,
1951
+ "loss": 0.7684,
1952
+ "step": 16150
1953
+ },
1954
+ {
1955
+ "epoch": 1.23,
1956
+ "learning_rate": 7.737978038621735e-06,
1957
+ "loss": 0.7258,
1958
+ "step": 16200
1959
+ },
1960
+ {
1961
+ "epoch": 1.23,
1962
+ "learning_rate": 7.700113593335858e-06,
1963
+ "loss": 0.7602,
1964
+ "step": 16250
1965
+ },
1966
+ {
1967
+ "epoch": 1.23,
1968
+ "learning_rate": 7.662249148049982e-06,
1969
+ "loss": 0.775,
1970
+ "step": 16300
1971
+ },
1972
+ {
1973
+ "epoch": 1.24,
1974
+ "learning_rate": 7.624384702764105e-06,
1975
+ "loss": 0.7311,
1976
+ "step": 16350
1977
+ },
1978
+ {
1979
+ "epoch": 1.24,
1980
+ "learning_rate": 7.586520257478228e-06,
1981
+ "loss": 0.7252,
1982
+ "step": 16400
1983
+ },
1984
+ {
1985
+ "epoch": 1.25,
1986
+ "learning_rate": 7.548655812192352e-06,
1987
+ "loss": 0.7078,
1988
+ "step": 16450
1989
+ },
1990
+ {
1991
+ "epoch": 1.25,
1992
+ "learning_rate": 7.5107913669064756e-06,
1993
+ "loss": 0.7607,
1994
+ "step": 16500
1995
+ },
1996
+ {
1997
+ "epoch": 1.25,
1998
+ "learning_rate": 7.472926921620599e-06,
1999
+ "loss": 0.7462,
2000
+ "step": 16550
2001
+ },
2002
+ {
2003
+ "epoch": 1.26,
2004
+ "learning_rate": 7.435062476334722e-06,
2005
+ "loss": 0.7574,
2006
+ "step": 16600
2007
+ },
2008
+ {
2009
+ "epoch": 1.26,
2010
+ "learning_rate": 7.397198031048845e-06,
2011
+ "loss": 0.7223,
2012
+ "step": 16650
2013
+ },
2014
+ {
2015
+ "epoch": 1.26,
2016
+ "learning_rate": 7.3593335857629685e-06,
2017
+ "loss": 0.7306,
2018
+ "step": 16700
2019
+ },
2020
+ {
2021
+ "epoch": 1.27,
2022
+ "learning_rate": 7.321469140477092e-06,
2023
+ "loss": 0.7638,
2024
+ "step": 16750
2025
+ },
2026
+ {
2027
+ "epoch": 1.27,
2028
+ "learning_rate": 7.283604695191217e-06,
2029
+ "loss": 0.7236,
2030
+ "step": 16800
2031
+ },
2032
+ {
2033
+ "epoch": 1.28,
2034
+ "learning_rate": 7.24574024990534e-06,
2035
+ "loss": 0.7381,
2036
+ "step": 16850
2037
+ },
2038
+ {
2039
+ "epoch": 1.28,
2040
+ "learning_rate": 7.207875804619463e-06,
2041
+ "loss": 0.7257,
2042
+ "step": 16900
2043
+ },
2044
+ {
2045
+ "epoch": 1.28,
2046
+ "learning_rate": 7.170011359333586e-06,
2047
+ "loss": 0.7516,
2048
+ "step": 16950
2049
+ },
2050
+ {
2051
+ "epoch": 1.29,
2052
+ "learning_rate": 7.132904202953427e-06,
2053
+ "loss": 0.7189,
2054
+ "step": 17000
2055
+ },
2056
+ {
2057
+ "epoch": 1.29,
2058
+ "learning_rate": 7.095039757667551e-06,
2059
+ "loss": 0.7672,
2060
+ "step": 17050
2061
+ },
2062
+ {
2063
+ "epoch": 1.29,
2064
+ "learning_rate": 7.057175312381674e-06,
2065
+ "loss": 0.7289,
2066
+ "step": 17100
2067
+ },
2068
+ {
2069
+ "epoch": 1.3,
2070
+ "learning_rate": 7.019310867095797e-06,
2071
+ "loss": 0.7194,
2072
+ "step": 17150
2073
+ },
2074
+ {
2075
+ "epoch": 1.3,
2076
+ "learning_rate": 6.9814464218099205e-06,
2077
+ "loss": 0.6933,
2078
+ "step": 17200
2079
+ },
2080
+ {
2081
+ "epoch": 1.31,
2082
+ "learning_rate": 6.943581976524045e-06,
2083
+ "loss": 0.8152,
2084
+ "step": 17250
2085
+ },
2086
+ {
2087
+ "epoch": 1.31,
2088
+ "learning_rate": 6.905717531238168e-06,
2089
+ "loss": 0.76,
2090
+ "step": 17300
2091
+ },
2092
+ {
2093
+ "epoch": 1.31,
2094
+ "learning_rate": 6.867853085952292e-06,
2095
+ "loss": 0.7271,
2096
+ "step": 17350
2097
+ },
2098
+ {
2099
+ "epoch": 1.32,
2100
+ "learning_rate": 6.829988640666415e-06,
2101
+ "loss": 0.7494,
2102
+ "step": 17400
2103
+ },
2104
+ {
2105
+ "epoch": 1.32,
2106
+ "learning_rate": 6.792124195380538e-06,
2107
+ "loss": 0.7738,
2108
+ "step": 17450
2109
+ },
2110
+ {
2111
+ "epoch": 1.33,
2112
+ "learning_rate": 6.754259750094662e-06,
2113
+ "loss": 0.7071,
2114
+ "step": 17500
2115
+ },
2116
+ {
2117
+ "epoch": 1.33,
2118
+ "learning_rate": 6.716395304808785e-06,
2119
+ "loss": 0.7339,
2120
+ "step": 17550
2121
+ },
2122
+ {
2123
+ "epoch": 1.33,
2124
+ "learning_rate": 6.678530859522908e-06,
2125
+ "loss": 0.7506,
2126
+ "step": 17600
2127
+ },
2128
+ {
2129
+ "epoch": 1.34,
2130
+ "learning_rate": 6.640666414237032e-06,
2131
+ "loss": 0.7121,
2132
+ "step": 17650
2133
+ },
2134
+ {
2135
+ "epoch": 1.34,
2136
+ "learning_rate": 6.602801968951155e-06,
2137
+ "loss": 0.7018,
2138
+ "step": 17700
2139
+ },
2140
+ {
2141
+ "epoch": 1.34,
2142
+ "learning_rate": 6.564937523665279e-06,
2143
+ "loss": 0.7677,
2144
+ "step": 17750
2145
+ },
2146
+ {
2147
+ "epoch": 1.35,
2148
+ "learning_rate": 6.527073078379403e-06,
2149
+ "loss": 0.6925,
2150
+ "step": 17800
2151
+ },
2152
+ {
2153
+ "epoch": 1.35,
2154
+ "learning_rate": 6.489208633093526e-06,
2155
+ "loss": 0.7205,
2156
+ "step": 17850
2157
+ },
2158
+ {
2159
+ "epoch": 1.36,
2160
+ "learning_rate": 6.451344187807649e-06,
2161
+ "loss": 0.7163,
2162
+ "step": 17900
2163
+ },
2164
+ {
2165
+ "epoch": 1.36,
2166
+ "learning_rate": 6.413479742521772e-06,
2167
+ "loss": 0.752,
2168
+ "step": 17950
2169
+ },
2170
+ {
2171
+ "epoch": 1.36,
2172
+ "learning_rate": 6.3756152972358965e-06,
2173
+ "loss": 0.7706,
2174
+ "step": 18000
2175
+ },
2176
+ {
2177
+ "epoch": 1.37,
2178
+ "learning_rate": 6.33775085195002e-06,
2179
+ "loss": 0.7556,
2180
+ "step": 18050
2181
+ },
2182
+ {
2183
+ "epoch": 1.37,
2184
+ "learning_rate": 6.299886406664143e-06,
2185
+ "loss": 0.7626,
2186
+ "step": 18100
2187
+ },
2188
+ {
2189
+ "epoch": 1.37,
2190
+ "learning_rate": 6.262021961378266e-06,
2191
+ "loss": 0.7447,
2192
+ "step": 18150
2193
+ },
2194
+ {
2195
+ "epoch": 1.38,
2196
+ "learning_rate": 6.224157516092389e-06,
2197
+ "loss": 0.741,
2198
+ "step": 18200
2199
+ },
2200
+ {
2201
+ "epoch": 1.38,
2202
+ "learning_rate": 6.186293070806513e-06,
2203
+ "loss": 0.7567,
2204
+ "step": 18250
2205
+ },
2206
+ {
2207
+ "epoch": 1.39,
2208
+ "learning_rate": 6.1484286255206375e-06,
2209
+ "loss": 0.7364,
2210
+ "step": 18300
2211
+ },
2212
+ {
2213
+ "epoch": 1.39,
2214
+ "learning_rate": 6.110564180234761e-06,
2215
+ "loss": 0.739,
2216
+ "step": 18350
2217
+ },
2218
+ {
2219
+ "epoch": 1.39,
2220
+ "learning_rate": 6.072699734948884e-06,
2221
+ "loss": 0.7309,
2222
+ "step": 18400
2223
+ },
2224
+ {
2225
+ "epoch": 1.4,
2226
+ "learning_rate": 6.034835289663007e-06,
2227
+ "loss": 0.7385,
2228
+ "step": 18450
2229
+ },
2230
+ {
2231
+ "epoch": 1.4,
2232
+ "learning_rate": 5.9969708443771305e-06,
2233
+ "loss": 0.7588,
2234
+ "step": 18500
2235
+ },
2236
+ {
2237
+ "epoch": 1.4,
2238
+ "learning_rate": 5.959106399091254e-06,
2239
+ "loss": 0.7442,
2240
+ "step": 18550
2241
+ },
2242
+ {
2243
+ "epoch": 1.41,
2244
+ "learning_rate": 5.921241953805377e-06,
2245
+ "loss": 0.7871,
2246
+ "step": 18600
2247
+ },
2248
+ {
2249
+ "epoch": 1.41,
2250
+ "learning_rate": 5.883377508519501e-06,
2251
+ "loss": 0.7817,
2252
+ "step": 18650
2253
+ },
2254
+ {
2255
+ "epoch": 1.42,
2256
+ "learning_rate": 5.845513063233624e-06,
2257
+ "loss": 0.751,
2258
+ "step": 18700
2259
+ },
2260
+ {
2261
+ "epoch": 1.42,
2262
+ "learning_rate": 5.8076486179477474e-06,
2263
+ "loss": 0.7379,
2264
+ "step": 18750
2265
+ },
2266
+ {
2267
+ "epoch": 1.42,
2268
+ "learning_rate": 5.769784172661871e-06,
2269
+ "loss": 0.7151,
2270
+ "step": 18800
2271
+ },
2272
+ {
2273
+ "epoch": 1.43,
2274
+ "learning_rate": 5.731919727375994e-06,
2275
+ "loss": 0.7989,
2276
+ "step": 18850
2277
+ },
2278
+ {
2279
+ "epoch": 1.43,
2280
+ "learning_rate": 5.694055282090117e-06,
2281
+ "loss": 0.752,
2282
+ "step": 18900
2283
+ },
2284
+ {
2285
+ "epoch": 1.43,
2286
+ "learning_rate": 5.656190836804242e-06,
2287
+ "loss": 0.755,
2288
+ "step": 18950
2289
+ },
2290
+ {
2291
+ "epoch": 1.44,
2292
+ "learning_rate": 5.618326391518365e-06,
2293
+ "loss": 0.7355,
2294
+ "step": 19000
2295
+ },
2296
+ {
2297
+ "epoch": 1.44,
2298
+ "learning_rate": 5.5804619462324885e-06,
2299
+ "loss": 0.668,
2300
+ "step": 19050
2301
+ },
2302
+ {
2303
+ "epoch": 1.45,
2304
+ "learning_rate": 5.542597500946612e-06,
2305
+ "loss": 0.763,
2306
+ "step": 19100
2307
+ },
2308
+ {
2309
+ "epoch": 1.45,
2310
+ "learning_rate": 5.504733055660735e-06,
2311
+ "loss": 0.7369,
2312
+ "step": 19150
2313
+ },
2314
+ {
2315
+ "epoch": 1.45,
2316
+ "learning_rate": 5.466868610374858e-06,
2317
+ "loss": 0.7389,
2318
+ "step": 19200
2319
+ },
2320
+ {
2321
+ "epoch": 1.46,
2322
+ "learning_rate": 5.429004165088982e-06,
2323
+ "loss": 0.7435,
2324
+ "step": 19250
2325
+ },
2326
+ {
2327
+ "epoch": 1.46,
2328
+ "learning_rate": 5.3911397198031055e-06,
2329
+ "loss": 0.7103,
2330
+ "step": 19300
2331
+ },
2332
+ {
2333
+ "epoch": 1.47,
2334
+ "learning_rate": 5.353275274517229e-06,
2335
+ "loss": 0.7,
2336
+ "step": 19350
2337
+ },
2338
+ {
2339
+ "epoch": 1.47,
2340
+ "learning_rate": 5.315410829231352e-06,
2341
+ "loss": 0.7269,
2342
+ "step": 19400
2343
+ },
2344
+ {
2345
+ "epoch": 1.47,
2346
+ "learning_rate": 5.277546383945475e-06,
2347
+ "loss": 0.713,
2348
+ "step": 19450
2349
+ },
2350
+ {
2351
+ "epoch": 1.48,
2352
+ "learning_rate": 5.2396819386595984e-06,
2353
+ "loss": 0.7442,
2354
+ "step": 19500
2355
+ },
2356
+ {
2357
+ "epoch": 1.48,
2358
+ "learning_rate": 5.201817493373722e-06,
2359
+ "loss": 0.731,
2360
+ "step": 19550
2361
+ },
2362
+ {
2363
+ "epoch": 1.48,
2364
+ "learning_rate": 5.163953048087847e-06,
2365
+ "loss": 0.7389,
2366
+ "step": 19600
2367
+ },
2368
+ {
2369
+ "epoch": 1.49,
2370
+ "learning_rate": 5.12608860280197e-06,
2371
+ "loss": 0.7348,
2372
+ "step": 19650
2373
+ },
2374
+ {
2375
+ "epoch": 1.49,
2376
+ "learning_rate": 5.088224157516093e-06,
2377
+ "loss": 0.7607,
2378
+ "step": 19700
2379
+ },
2380
+ {
2381
+ "epoch": 1.5,
2382
+ "learning_rate": 5.050359712230216e-06,
2383
+ "loss": 0.7151,
2384
+ "step": 19750
2385
+ },
2386
+ {
2387
+ "epoch": 1.5,
2388
+ "learning_rate": 5.0124952669443395e-06,
2389
+ "loss": 0.7002,
2390
+ "step": 19800
2391
+ },
2392
+ {
2393
+ "epoch": 1.5,
2394
+ "learning_rate": 4.974630821658464e-06,
2395
+ "loss": 0.7734,
2396
+ "step": 19850
2397
+ },
2398
+ {
2399
+ "epoch": 1.51,
2400
+ "learning_rate": 4.936766376372587e-06,
2401
+ "loss": 0.7506,
2402
+ "step": 19900
2403
+ },
2404
+ {
2405
+ "epoch": 1.51,
2406
+ "learning_rate": 4.89890193108671e-06,
2407
+ "loss": 0.7369,
2408
+ "step": 19950
2409
+ },
2410
+ {
2411
+ "epoch": 1.51,
2412
+ "learning_rate": 4.861037485800833e-06,
2413
+ "loss": 0.7118,
2414
+ "step": 20000
2415
+ },
2416
+ {
2417
+ "epoch": 1.52,
2418
+ "learning_rate": 4.8231730405149565e-06,
2419
+ "loss": 0.7152,
2420
+ "step": 20050
2421
+ },
2422
+ {
2423
+ "epoch": 1.52,
2424
+ "learning_rate": 4.78530859522908e-06,
2425
+ "loss": 0.7366,
2426
+ "step": 20100
2427
+ },
2428
+ {
2429
+ "epoch": 1.53,
2430
+ "learning_rate": 4.747444149943204e-06,
2431
+ "loss": 0.7086,
2432
+ "step": 20150
2433
+ },
2434
+ {
2435
+ "epoch": 1.53,
2436
+ "learning_rate": 4.709579704657327e-06,
2437
+ "loss": 0.7306,
2438
+ "step": 20200
2439
+ },
2440
+ {
2441
+ "epoch": 1.53,
2442
+ "learning_rate": 4.67171525937145e-06,
2443
+ "loss": 0.7266,
2444
+ "step": 20250
2445
+ },
2446
+ {
2447
+ "epoch": 1.54,
2448
+ "learning_rate": 4.633850814085574e-06,
2449
+ "loss": 0.7241,
2450
+ "step": 20300
2451
+ },
2452
+ {
2453
+ "epoch": 1.54,
2454
+ "learning_rate": 4.596743657705415e-06,
2455
+ "loss": 0.708,
2456
+ "step": 20350
2457
+ },
2458
+ {
2459
+ "epoch": 1.54,
2460
+ "learning_rate": 4.558879212419539e-06,
2461
+ "loss": 0.7544,
2462
+ "step": 20400
2463
+ },
2464
+ {
2465
+ "epoch": 1.55,
2466
+ "learning_rate": 4.521014767133662e-06,
2467
+ "loss": 0.739,
2468
+ "step": 20450
2469
+ },
2470
+ {
2471
+ "epoch": 1.55,
2472
+ "learning_rate": 4.483150321847785e-06,
2473
+ "loss": 0.7492,
2474
+ "step": 20500
2475
+ },
2476
+ {
2477
+ "epoch": 1.56,
2478
+ "learning_rate": 4.4452858765619086e-06,
2479
+ "loss": 0.7323,
2480
+ "step": 20550
2481
+ },
2482
+ {
2483
+ "epoch": 1.56,
2484
+ "learning_rate": 4.407421431276032e-06,
2485
+ "loss": 0.7431,
2486
+ "step": 20600
2487
+ },
2488
+ {
2489
+ "epoch": 1.56,
2490
+ "learning_rate": 4.369556985990155e-06,
2491
+ "loss": 0.7302,
2492
+ "step": 20650
2493
+ },
2494
+ {
2495
+ "epoch": 1.57,
2496
+ "learning_rate": 4.331692540704279e-06,
2497
+ "loss": 0.7112,
2498
+ "step": 20700
2499
+ },
2500
+ {
2501
+ "epoch": 1.57,
2502
+ "learning_rate": 4.293828095418402e-06,
2503
+ "loss": 0.7359,
2504
+ "step": 20750
2505
+ },
2506
+ {
2507
+ "epoch": 1.58,
2508
+ "learning_rate": 4.2559636501325256e-06,
2509
+ "loss": 0.7274,
2510
+ "step": 20800
2511
+ },
2512
+ {
2513
+ "epoch": 1.58,
2514
+ "learning_rate": 4.21809920484665e-06,
2515
+ "loss": 0.7307,
2516
+ "step": 20850
2517
+ },
2518
+ {
2519
+ "epoch": 1.58,
2520
+ "learning_rate": 4.180234759560773e-06,
2521
+ "loss": 0.7339,
2522
+ "step": 20900
2523
+ },
2524
+ {
2525
+ "epoch": 1.59,
2526
+ "learning_rate": 4.142370314274896e-06,
2527
+ "loss": 0.7743,
2528
+ "step": 20950
2529
+ },
2530
+ {
2531
+ "epoch": 1.59,
2532
+ "learning_rate": 4.104505868989019e-06,
2533
+ "loss": 0.7424,
2534
+ "step": 21000
2535
+ },
2536
+ {
2537
+ "epoch": 1.59,
2538
+ "learning_rate": 4.066641423703143e-06,
2539
+ "loss": 0.7152,
2540
+ "step": 21050
2541
+ },
2542
+ {
2543
+ "epoch": 1.6,
2544
+ "learning_rate": 4.028776978417267e-06,
2545
+ "loss": 0.707,
2546
+ "step": 21100
2547
+ },
2548
+ {
2549
+ "epoch": 1.6,
2550
+ "learning_rate": 3.99091253313139e-06,
2551
+ "loss": 0.7538,
2552
+ "step": 21150
2553
+ },
2554
+ {
2555
+ "epoch": 1.61,
2556
+ "learning_rate": 3.953048087845514e-06,
2557
+ "loss": 0.7293,
2558
+ "step": 21200
2559
+ },
2560
+ {
2561
+ "epoch": 1.61,
2562
+ "learning_rate": 3.915183642559637e-06,
2563
+ "loss": 0.7524,
2564
+ "step": 21250
2565
+ },
2566
+ {
2567
+ "epoch": 1.61,
2568
+ "learning_rate": 3.87731919727376e-06,
2569
+ "loss": 0.7938,
2570
+ "step": 21300
2571
+ },
2572
+ {
2573
+ "epoch": 1.62,
2574
+ "learning_rate": 3.839454751987884e-06,
2575
+ "loss": 0.7266,
2576
+ "step": 21350
2577
+ },
2578
+ {
2579
+ "epoch": 1.62,
2580
+ "learning_rate": 3.8015903067020073e-06,
2581
+ "loss": 0.7566,
2582
+ "step": 21400
2583
+ },
2584
+ {
2585
+ "epoch": 1.62,
2586
+ "learning_rate": 3.7637258614161305e-06,
2587
+ "loss": 0.713,
2588
+ "step": 21450
2589
+ },
2590
+ {
2591
+ "epoch": 1.63,
2592
+ "learning_rate": 3.725861416130254e-06,
2593
+ "loss": 0.7546,
2594
+ "step": 21500
2595
+ },
2596
+ {
2597
+ "epoch": 1.63,
2598
+ "learning_rate": 3.6879969708443774e-06,
2599
+ "loss": 0.6955,
2600
+ "step": 21550
2601
+ },
2602
+ {
2603
+ "epoch": 1.64,
2604
+ "learning_rate": 3.6501325255585006e-06,
2605
+ "loss": 0.7084,
2606
+ "step": 21600
2607
+ },
2608
+ {
2609
+ "epoch": 1.64,
2610
+ "learning_rate": 3.6122680802726247e-06,
2611
+ "loss": 0.7447,
2612
+ "step": 21650
2613
+ },
2614
+ {
2615
+ "epoch": 1.64,
2616
+ "learning_rate": 3.574403634986748e-06,
2617
+ "loss": 0.7173,
2618
+ "step": 21700
2619
+ },
2620
+ {
2621
+ "epoch": 1.65,
2622
+ "learning_rate": 3.536539189700871e-06,
2623
+ "loss": 0.7324,
2624
+ "step": 21750
2625
+ },
2626
+ {
2627
+ "epoch": 1.65,
2628
+ "learning_rate": 3.498674744414995e-06,
2629
+ "loss": 0.7561,
2630
+ "step": 21800
2631
+ },
2632
+ {
2633
+ "epoch": 1.65,
2634
+ "learning_rate": 3.460810299129118e-06,
2635
+ "loss": 0.7505,
2636
+ "step": 21850
2637
+ },
2638
+ {
2639
+ "epoch": 1.66,
2640
+ "learning_rate": 3.4229458538432413e-06,
2641
+ "loss": 0.7337,
2642
+ "step": 21900
2643
+ },
2644
+ {
2645
+ "epoch": 1.66,
2646
+ "learning_rate": 3.3850814085573645e-06,
2647
+ "loss": 0.7753,
2648
+ "step": 21950
2649
+ },
2650
+ {
2651
+ "epoch": 1.67,
2652
+ "learning_rate": 3.3472169632714886e-06,
2653
+ "loss": 0.7703,
2654
+ "step": 22000
2655
+ },
2656
+ {
2657
+ "epoch": 1.67,
2658
+ "learning_rate": 3.309352517985612e-06,
2659
+ "loss": 0.7384,
2660
+ "step": 22050
2661
+ },
2662
+ {
2663
+ "epoch": 1.67,
2664
+ "learning_rate": 3.271488072699735e-06,
2665
+ "loss": 0.7759,
2666
+ "step": 22100
2667
+ },
2668
+ {
2669
+ "epoch": 1.68,
2670
+ "learning_rate": 3.2336236274138587e-06,
2671
+ "loss": 0.7446,
2672
+ "step": 22150
2673
+ },
2674
+ {
2675
+ "epoch": 1.68,
2676
+ "learning_rate": 3.195759182127982e-06,
2677
+ "loss": 0.7437,
2678
+ "step": 22200
2679
+ },
2680
+ {
2681
+ "epoch": 1.68,
2682
+ "learning_rate": 3.157894736842105e-06,
2683
+ "loss": 0.7817,
2684
+ "step": 22250
2685
+ },
2686
+ {
2687
+ "epoch": 1.69,
2688
+ "learning_rate": 3.1200302915562292e-06,
2689
+ "loss": 0.7396,
2690
+ "step": 22300
2691
+ },
2692
+ {
2693
+ "epoch": 1.69,
2694
+ "learning_rate": 3.0821658462703525e-06,
2695
+ "loss": 0.743,
2696
+ "step": 22350
2697
+ },
2698
+ {
2699
+ "epoch": 1.7,
2700
+ "learning_rate": 3.0443014009844757e-06,
2701
+ "loss": 0.7597,
2702
+ "step": 22400
2703
+ },
2704
+ {
2705
+ "epoch": 1.7,
2706
+ "learning_rate": 3.0064369556985994e-06,
2707
+ "loss": 0.6844,
2708
+ "step": 22450
2709
+ },
2710
+ {
2711
+ "epoch": 1.7,
2712
+ "learning_rate": 2.9685725104127226e-06,
2713
+ "loss": 0.7737,
2714
+ "step": 22500
2715
+ },
2716
+ {
2717
+ "epoch": 1.71,
2718
+ "learning_rate": 2.930708065126846e-06,
2719
+ "loss": 0.7024,
2720
+ "step": 22550
2721
+ },
2722
+ {
2723
+ "epoch": 1.71,
2724
+ "learning_rate": 2.89284361984097e-06,
2725
+ "loss": 0.7136,
2726
+ "step": 22600
2727
+ },
2728
+ {
2729
+ "epoch": 1.72,
2730
+ "learning_rate": 2.854979174555093e-06,
2731
+ "loss": 0.7482,
2732
+ "step": 22650
2733
+ },
2734
+ {
2735
+ "epoch": 1.72,
2736
+ "learning_rate": 2.8171147292692164e-06,
2737
+ "loss": 0.7608,
2738
+ "step": 22700
2739
+ },
2740
+ {
2741
+ "epoch": 1.72,
2742
+ "learning_rate": 2.7792502839833396e-06,
2743
+ "loss": 0.7442,
2744
+ "step": 22750
2745
+ },
2746
+ {
2747
+ "epoch": 1.73,
2748
+ "learning_rate": 2.7413858386974632e-06,
2749
+ "loss": 0.6687,
2750
+ "step": 22800
2751
+ },
2752
+ {
2753
+ "epoch": 1.73,
2754
+ "learning_rate": 2.703521393411587e-06,
2755
+ "loss": 0.7594,
2756
+ "step": 22850
2757
+ },
2758
+ {
2759
+ "epoch": 1.73,
2760
+ "learning_rate": 2.66565694812571e-06,
2761
+ "loss": 0.7042,
2762
+ "step": 22900
2763
+ },
2764
+ {
2765
+ "epoch": 1.74,
2766
+ "learning_rate": 2.6277925028398338e-06,
2767
+ "loss": 0.7556,
2768
+ "step": 22950
2769
+ },
2770
+ {
2771
+ "epoch": 1.74,
2772
+ "learning_rate": 2.589928057553957e-06,
2773
+ "loss": 0.7357,
2774
+ "step": 23000
2775
+ },
2776
+ {
2777
+ "epoch": 1.75,
2778
+ "learning_rate": 2.5520636122680802e-06,
2779
+ "loss": 0.7292,
2780
+ "step": 23050
2781
+ },
2782
+ {
2783
+ "epoch": 1.75,
2784
+ "learning_rate": 2.5141991669822043e-06,
2785
+ "loss": 0.6899,
2786
+ "step": 23100
2787
+ },
2788
+ {
2789
+ "epoch": 1.75,
2790
+ "learning_rate": 2.4763347216963275e-06,
2791
+ "loss": 0.729,
2792
+ "step": 23150
2793
+ },
2794
+ {
2795
+ "epoch": 1.76,
2796
+ "learning_rate": 2.4384702764104508e-06,
2797
+ "loss": 0.7473,
2798
+ "step": 23200
2799
+ },
2800
+ {
2801
+ "epoch": 1.76,
2802
+ "learning_rate": 2.400605831124574e-06,
2803
+ "loss": 0.7264,
2804
+ "step": 23250
2805
+ },
2806
+ {
2807
+ "epoch": 1.76,
2808
+ "learning_rate": 2.3627413858386977e-06,
2809
+ "loss": 0.7314,
2810
+ "step": 23300
2811
+ },
2812
+ {
2813
+ "epoch": 1.77,
2814
+ "learning_rate": 2.3256342294585385e-06,
2815
+ "loss": 0.6819,
2816
+ "step": 23350
2817
+ },
2818
+ {
2819
+ "epoch": 1.77,
2820
+ "learning_rate": 2.287769784172662e-06,
2821
+ "loss": 0.7637,
2822
+ "step": 23400
2823
+ },
2824
+ {
2825
+ "epoch": 1.78,
2826
+ "learning_rate": 2.2499053388867854e-06,
2827
+ "loss": 0.727,
2828
+ "step": 23450
2829
+ },
2830
+ {
2831
+ "epoch": 1.78,
2832
+ "learning_rate": 2.2120408936009086e-06,
2833
+ "loss": 0.7044,
2834
+ "step": 23500
2835
+ },
2836
+ {
2837
+ "epoch": 1.78,
2838
+ "learning_rate": 2.1741764483150323e-06,
2839
+ "loss": 0.6926,
2840
+ "step": 23550
2841
+ },
2842
+ {
2843
+ "epoch": 1.79,
2844
+ "learning_rate": 2.136312003029156e-06,
2845
+ "loss": 0.7406,
2846
+ "step": 23600
2847
+ },
2848
+ {
2849
+ "epoch": 1.79,
2850
+ "learning_rate": 2.098447557743279e-06,
2851
+ "loss": 0.7572,
2852
+ "step": 23650
2853
+ },
2854
+ {
2855
+ "epoch": 1.79,
2856
+ "learning_rate": 2.060583112457403e-06,
2857
+ "loss": 0.7239,
2858
+ "step": 23700
2859
+ },
2860
+ {
2861
+ "epoch": 1.8,
2862
+ "learning_rate": 2.022718667171526e-06,
2863
+ "loss": 0.7353,
2864
+ "step": 23750
2865
+ },
2866
+ {
2867
+ "epoch": 1.8,
2868
+ "learning_rate": 1.9848542218856497e-06,
2869
+ "loss": 0.7287,
2870
+ "step": 23800
2871
+ },
2872
+ {
2873
+ "epoch": 1.81,
2874
+ "learning_rate": 1.946989776599773e-06,
2875
+ "loss": 0.7461,
2876
+ "step": 23850
2877
+ },
2878
+ {
2879
+ "epoch": 1.81,
2880
+ "learning_rate": 1.909125331313896e-06,
2881
+ "loss": 0.7489,
2882
+ "step": 23900
2883
+ },
2884
+ {
2885
+ "epoch": 1.81,
2886
+ "learning_rate": 1.8712608860280198e-06,
2887
+ "loss": 0.7156,
2888
+ "step": 23950
2889
+ },
2890
+ {
2891
+ "epoch": 1.82,
2892
+ "learning_rate": 1.8333964407421435e-06,
2893
+ "loss": 0.6765,
2894
+ "step": 24000
2895
+ },
2896
+ {
2897
+ "epoch": 1.82,
2898
+ "learning_rate": 1.7955319954562667e-06,
2899
+ "loss": 0.7496,
2900
+ "step": 24050
2901
+ },
2902
+ {
2903
+ "epoch": 1.82,
2904
+ "learning_rate": 1.7576675501703901e-06,
2905
+ "loss": 0.7269,
2906
+ "step": 24100
2907
+ },
2908
+ {
2909
+ "epoch": 1.83,
2910
+ "learning_rate": 1.7198031048845134e-06,
2911
+ "loss": 0.728,
2912
+ "step": 24150
2913
+ },
2914
+ {
2915
+ "epoch": 1.83,
2916
+ "learning_rate": 1.681938659598637e-06,
2917
+ "loss": 0.7914,
2918
+ "step": 24200
2919
+ },
2920
+ {
2921
+ "epoch": 1.84,
2922
+ "learning_rate": 1.6440742143127605e-06,
2923
+ "loss": 0.7419,
2924
+ "step": 24250
2925
+ },
2926
+ {
2927
+ "epoch": 1.84,
2928
+ "learning_rate": 1.6062097690268837e-06,
2929
+ "loss": 0.7509,
2930
+ "step": 24300
2931
+ },
2932
+ {
2933
+ "epoch": 1.84,
2934
+ "learning_rate": 1.5683453237410074e-06,
2935
+ "loss": 0.7372,
2936
+ "step": 24350
2937
+ },
2938
+ {
2939
+ "epoch": 1.85,
2940
+ "learning_rate": 1.5304808784551308e-06,
2941
+ "loss": 0.7159,
2942
+ "step": 24400
2943
+ },
2944
+ {
2945
+ "epoch": 1.85,
2946
+ "learning_rate": 1.4926164331692542e-06,
2947
+ "loss": 0.7272,
2948
+ "step": 24450
2949
+ },
2950
+ {
2951
+ "epoch": 1.86,
2952
+ "learning_rate": 1.4547519878833777e-06,
2953
+ "loss": 0.735,
2954
+ "step": 24500
2955
+ },
2956
+ {
2957
+ "epoch": 1.86,
2958
+ "learning_rate": 1.416887542597501e-06,
2959
+ "loss": 0.715,
2960
+ "step": 24550
2961
+ },
2962
+ {
2963
+ "epoch": 1.86,
2964
+ "learning_rate": 1.3790230973116246e-06,
2965
+ "loss": 0.7507,
2966
+ "step": 24600
2967
+ },
2968
+ {
2969
+ "epoch": 1.87,
2970
+ "learning_rate": 1.341158652025748e-06,
2971
+ "loss": 0.7723,
2972
+ "step": 24650
2973
+ },
2974
+ {
2975
+ "epoch": 1.87,
2976
+ "learning_rate": 1.3032942067398712e-06,
2977
+ "loss": 0.7658,
2978
+ "step": 24700
2979
+ },
2980
+ {
2981
+ "epoch": 1.87,
2982
+ "learning_rate": 1.2654297614539949e-06,
2983
+ "loss": 0.7727,
2984
+ "step": 24750
2985
+ },
2986
+ {
2987
+ "epoch": 1.88,
2988
+ "learning_rate": 1.2275653161681181e-06,
2989
+ "loss": 0.748,
2990
+ "step": 24800
2991
+ },
2992
+ {
2993
+ "epoch": 1.88,
2994
+ "learning_rate": 1.1897008708822416e-06,
2995
+ "loss": 0.7135,
2996
+ "step": 24850
2997
+ },
2998
+ {
2999
+ "epoch": 1.89,
3000
+ "learning_rate": 1.1518364255963652e-06,
3001
+ "loss": 0.76,
3002
+ "step": 24900
3003
+ },
3004
+ {
3005
+ "epoch": 1.89,
3006
+ "learning_rate": 1.1139719803104887e-06,
3007
+ "loss": 0.7439,
3008
+ "step": 24950
3009
+ },
3010
+ {
3011
+ "epoch": 1.89,
3012
+ "learning_rate": 1.0761075350246119e-06,
3013
+ "loss": 0.7496,
3014
+ "step": 25000
3015
+ },
3016
+ {
3017
+ "epoch": 1.9,
3018
+ "learning_rate": 1.0382430897387353e-06,
3019
+ "loss": 0.7631,
3020
+ "step": 25050
3021
+ },
3022
+ {
3023
+ "epoch": 1.9,
3024
+ "learning_rate": 1.000378644452859e-06,
3025
+ "loss": 0.7033,
3026
+ "step": 25100
3027
+ },
3028
+ {
3029
+ "epoch": 1.9,
3030
+ "learning_rate": 9.625141991669822e-07,
3031
+ "loss": 0.717,
3032
+ "step": 25150
3033
+ },
3034
+ {
3035
+ "epoch": 1.91,
3036
+ "learning_rate": 9.246497538811057e-07,
3037
+ "loss": 0.7329,
3038
+ "step": 25200
3039
+ },
3040
+ {
3041
+ "epoch": 1.91,
3042
+ "learning_rate": 8.867853085952291e-07,
3043
+ "loss": 0.7065,
3044
+ "step": 25250
3045
+ },
3046
+ {
3047
+ "epoch": 1.92,
3048
+ "learning_rate": 8.489208633093526e-07,
3049
+ "loss": 0.702,
3050
+ "step": 25300
3051
+ },
3052
+ {
3053
+ "epoch": 1.92,
3054
+ "learning_rate": 8.110564180234761e-07,
3055
+ "loss": 0.7161,
3056
+ "step": 25350
3057
+ },
3058
+ {
3059
+ "epoch": 1.92,
3060
+ "learning_rate": 7.731919727375994e-07,
3061
+ "loss": 0.725,
3062
+ "step": 25400
3063
+ },
3064
+ {
3065
+ "epoch": 1.93,
3066
+ "learning_rate": 7.353275274517229e-07,
3067
+ "loss": 0.7256,
3068
+ "step": 25450
3069
+ },
3070
+ {
3071
+ "epoch": 1.93,
3072
+ "learning_rate": 6.974630821658464e-07,
3073
+ "loss": 0.7395,
3074
+ "step": 25500
3075
+ },
3076
+ {
3077
+ "epoch": 1.93,
3078
+ "learning_rate": 6.595986368799697e-07,
3079
+ "loss": 0.794,
3080
+ "step": 25550
3081
+ },
3082
+ {
3083
+ "epoch": 1.94,
3084
+ "learning_rate": 6.217341915940932e-07,
3085
+ "loss": 0.7349,
3086
+ "step": 25600
3087
+ },
3088
+ {
3089
+ "epoch": 1.94,
3090
+ "learning_rate": 5.838697463082166e-07,
3091
+ "loss": 0.7394,
3092
+ "step": 25650
3093
+ },
3094
+ {
3095
+ "epoch": 1.95,
3096
+ "learning_rate": 5.460053010223401e-07,
3097
+ "loss": 0.7345,
3098
+ "step": 25700
3099
+ },
3100
+ {
3101
+ "epoch": 1.95,
3102
+ "learning_rate": 5.081408557364635e-07,
3103
+ "loss": 0.6888,
3104
+ "step": 25750
3105
+ },
3106
+ {
3107
+ "epoch": 1.95,
3108
+ "learning_rate": 4.702764104505869e-07,
3109
+ "loss": 0.7347,
3110
+ "step": 25800
3111
+ },
3112
+ {
3113
+ "epoch": 1.96,
3114
+ "learning_rate": 4.324119651647104e-07,
3115
+ "loss": 0.7246,
3116
+ "step": 25850
3117
+ },
3118
+ {
3119
+ "epoch": 1.96,
3120
+ "learning_rate": 3.945475198788338e-07,
3121
+ "loss": 0.7438,
3122
+ "step": 25900
3123
+ },
3124
+ {
3125
+ "epoch": 1.97,
3126
+ "learning_rate": 3.5668307459295723e-07,
3127
+ "loss": 0.7108,
3128
+ "step": 25950
3129
+ },
3130
+ {
3131
+ "epoch": 1.97,
3132
+ "learning_rate": 3.1881862930708067e-07,
3133
+ "loss": 0.7208,
3134
+ "step": 26000
3135
+ },
3136
+ {
3137
+ "epoch": 1.97,
3138
+ "learning_rate": 2.809541840212041e-07,
3139
+ "loss": 0.7633,
3140
+ "step": 26050
3141
+ },
3142
+ {
3143
+ "epoch": 1.98,
3144
+ "learning_rate": 2.4308973873532756e-07,
3145
+ "loss": 0.7707,
3146
+ "step": 26100
3147
+ },
3148
+ {
3149
+ "epoch": 1.98,
3150
+ "learning_rate": 2.05225293449451e-07,
3151
+ "loss": 0.7524,
3152
+ "step": 26150
3153
+ },
3154
+ {
3155
+ "epoch": 1.98,
3156
+ "learning_rate": 1.6736084816357444e-07,
3157
+ "loss": 0.707,
3158
+ "step": 26200
3159
+ },
3160
+ {
3161
+ "epoch": 1.99,
3162
+ "learning_rate": 1.2949640287769786e-07,
3163
+ "loss": 0.7751,
3164
+ "step": 26250
3165
+ },
3166
+ {
3167
+ "epoch": 1.99,
3168
+ "learning_rate": 9.163195759182128e-08,
3169
+ "loss": 0.7672,
3170
+ "step": 26300
3171
+ },
3172
+ {
3173
+ "epoch": 2.0,
3174
+ "learning_rate": 5.376751230594472e-08,
3175
+ "loss": 0.6998,
3176
+ "step": 26350
3177
+ },
3178
+ {
3179
+ "epoch": 2.0,
3180
+ "learning_rate": 1.5903067020068158e-08,
3181
+ "loss": 0.7069,
3182
+ "step": 26400
3183
+ },
3184
+ {
3185
+ "epoch": 2.0,
3186
+ "step": 26410,
3187
+ "total_flos": 6.728145843607406e+16,
3188
+ "train_loss": 0.7582937260732105,
3189
+ "train_runtime": 12885.6962,
3190
+ "train_samples_per_second": 65.589,
3191
+ "train_steps_per_second": 2.05
3192
+ }
3193
+ ],
3194
+ "max_steps": 26410,
3195
+ "num_train_epochs": 2,
3196
+ "total_flos": 6.728145843607406e+16,
3197
+ "trial_name": null,
3198
+ "trial_params": null
3199
+ }