irenelizihui commited on
Commit
9f8d51f
1 Parent(s): 5e3995c

Upload trainer_state.json

Browse files
Files changed (1) hide show
  1. trainer_state.json +2986 -0
trainer_state.json ADDED
@@ -0,0 +1,2986 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.9996687966007222,
5
+ "global_step": 247500,
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": 1.9959609341551486e-05,
13
+ "loss": 1.1369,
14
+ "step": 500
15
+ },
16
+ {
17
+ "epoch": 0.0,
18
+ "learning_rate": 1.9919218683102975e-05,
19
+ "loss": 1.1452,
20
+ "step": 1000
21
+ },
22
+ {
23
+ "epoch": 0.01,
24
+ "learning_rate": 1.987882802465446e-05,
25
+ "loss": 1.0938,
26
+ "step": 1500
27
+ },
28
+ {
29
+ "epoch": 0.01,
30
+ "learning_rate": 1.9838437366205944e-05,
31
+ "loss": 1.0933,
32
+ "step": 2000
33
+ },
34
+ {
35
+ "epoch": 0.01,
36
+ "learning_rate": 1.9798046707757432e-05,
37
+ "loss": 1.1217,
38
+ "step": 2500
39
+ },
40
+ {
41
+ "epoch": 0.01,
42
+ "learning_rate": 1.9757656049308917e-05,
43
+ "loss": 1.0401,
44
+ "step": 3000
45
+ },
46
+ {
47
+ "epoch": 0.01,
48
+ "learning_rate": 1.9717265390860402e-05,
49
+ "loss": 1.0169,
50
+ "step": 3500
51
+ },
52
+ {
53
+ "epoch": 0.02,
54
+ "learning_rate": 1.967687473241189e-05,
55
+ "loss": 1.1245,
56
+ "step": 4000
57
+ },
58
+ {
59
+ "epoch": 0.02,
60
+ "learning_rate": 1.9636484073963375e-05,
61
+ "loss": 1.0174,
62
+ "step": 4500
63
+ },
64
+ {
65
+ "epoch": 0.02,
66
+ "learning_rate": 1.959609341551486e-05,
67
+ "loss": 1.024,
68
+ "step": 5000
69
+ },
70
+ {
71
+ "epoch": 0.02,
72
+ "learning_rate": 1.9555702757066348e-05,
73
+ "loss": 1.0263,
74
+ "step": 5500
75
+ },
76
+ {
77
+ "epoch": 0.02,
78
+ "learning_rate": 1.9515312098617832e-05,
79
+ "loss": 1.042,
80
+ "step": 6000
81
+ },
82
+ {
83
+ "epoch": 0.03,
84
+ "learning_rate": 1.9474921440169317e-05,
85
+ "loss": 1.0445,
86
+ "step": 6500
87
+ },
88
+ {
89
+ "epoch": 0.03,
90
+ "learning_rate": 1.9434530781720805e-05,
91
+ "loss": 1.0388,
92
+ "step": 7000
93
+ },
94
+ {
95
+ "epoch": 0.03,
96
+ "learning_rate": 1.939414012327229e-05,
97
+ "loss": 1.0556,
98
+ "step": 7500
99
+ },
100
+ {
101
+ "epoch": 0.03,
102
+ "learning_rate": 1.9353749464823778e-05,
103
+ "loss": 1.0525,
104
+ "step": 8000
105
+ },
106
+ {
107
+ "epoch": 0.03,
108
+ "learning_rate": 1.9313358806375263e-05,
109
+ "loss": 1.09,
110
+ "step": 8500
111
+ },
112
+ {
113
+ "epoch": 0.04,
114
+ "learning_rate": 1.9272968147926748e-05,
115
+ "loss": 1.0217,
116
+ "step": 9000
117
+ },
118
+ {
119
+ "epoch": 0.04,
120
+ "learning_rate": 1.9232577489478236e-05,
121
+ "loss": 1.0252,
122
+ "step": 9500
123
+ },
124
+ {
125
+ "epoch": 0.04,
126
+ "learning_rate": 1.919218683102972e-05,
127
+ "loss": 1.0263,
128
+ "step": 10000
129
+ },
130
+ {
131
+ "epoch": 0.04,
132
+ "learning_rate": 1.915179617258121e-05,
133
+ "loss": 1.041,
134
+ "step": 10500
135
+ },
136
+ {
137
+ "epoch": 0.04,
138
+ "learning_rate": 1.9111405514132694e-05,
139
+ "loss": 1.0094,
140
+ "step": 11000
141
+ },
142
+ {
143
+ "epoch": 0.05,
144
+ "learning_rate": 1.907101485568418e-05,
145
+ "loss": 1.0667,
146
+ "step": 11500
147
+ },
148
+ {
149
+ "epoch": 0.05,
150
+ "learning_rate": 1.9030624197235667e-05,
151
+ "loss": 1.0105,
152
+ "step": 12000
153
+ },
154
+ {
155
+ "epoch": 0.05,
156
+ "learning_rate": 1.899023353878715e-05,
157
+ "loss": 1.0098,
158
+ "step": 12500
159
+ },
160
+ {
161
+ "epoch": 0.05,
162
+ "learning_rate": 1.8949842880338636e-05,
163
+ "loss": 1.0656,
164
+ "step": 13000
165
+ },
166
+ {
167
+ "epoch": 0.05,
168
+ "learning_rate": 1.8909452221890124e-05,
169
+ "loss": 1.0973,
170
+ "step": 13500
171
+ },
172
+ {
173
+ "epoch": 0.06,
174
+ "learning_rate": 1.886906156344161e-05,
175
+ "loss": 0.9736,
176
+ "step": 14000
177
+ },
178
+ {
179
+ "epoch": 0.06,
180
+ "learning_rate": 1.8828670904993094e-05,
181
+ "loss": 1.0193,
182
+ "step": 14500
183
+ },
184
+ {
185
+ "epoch": 0.06,
186
+ "learning_rate": 1.8788280246544582e-05,
187
+ "loss": 0.9827,
188
+ "step": 15000
189
+ },
190
+ {
191
+ "epoch": 0.06,
192
+ "learning_rate": 1.8747889588096067e-05,
193
+ "loss": 1.0219,
194
+ "step": 15500
195
+ },
196
+ {
197
+ "epoch": 0.06,
198
+ "learning_rate": 1.870749892964755e-05,
199
+ "loss": 1.0261,
200
+ "step": 16000
201
+ },
202
+ {
203
+ "epoch": 0.07,
204
+ "learning_rate": 1.866710827119904e-05,
205
+ "loss": 0.993,
206
+ "step": 16500
207
+ },
208
+ {
209
+ "epoch": 0.07,
210
+ "learning_rate": 1.8626717612750524e-05,
211
+ "loss": 1.0186,
212
+ "step": 17000
213
+ },
214
+ {
215
+ "epoch": 0.07,
216
+ "learning_rate": 1.858632695430201e-05,
217
+ "loss": 0.9749,
218
+ "step": 17500
219
+ },
220
+ {
221
+ "epoch": 0.07,
222
+ "learning_rate": 1.8545936295853497e-05,
223
+ "loss": 1.0013,
224
+ "step": 18000
225
+ },
226
+ {
227
+ "epoch": 0.07,
228
+ "learning_rate": 1.8505545637404982e-05,
229
+ "loss": 0.9743,
230
+ "step": 18500
231
+ },
232
+ {
233
+ "epoch": 0.08,
234
+ "learning_rate": 1.8465154978956467e-05,
235
+ "loss": 0.9796,
236
+ "step": 19000
237
+ },
238
+ {
239
+ "epoch": 0.08,
240
+ "learning_rate": 1.8424764320507955e-05,
241
+ "loss": 1.0136,
242
+ "step": 19500
243
+ },
244
+ {
245
+ "epoch": 0.08,
246
+ "learning_rate": 1.838437366205944e-05,
247
+ "loss": 1.0118,
248
+ "step": 20000
249
+ },
250
+ {
251
+ "epoch": 0.08,
252
+ "learning_rate": 1.8343983003610925e-05,
253
+ "loss": 1.0037,
254
+ "step": 20500
255
+ },
256
+ {
257
+ "epoch": 0.08,
258
+ "learning_rate": 1.8303592345162413e-05,
259
+ "loss": 0.9659,
260
+ "step": 21000
261
+ },
262
+ {
263
+ "epoch": 0.09,
264
+ "learning_rate": 1.8263201686713897e-05,
265
+ "loss": 1.0248,
266
+ "step": 21500
267
+ },
268
+ {
269
+ "epoch": 0.09,
270
+ "learning_rate": 1.8222811028265382e-05,
271
+ "loss": 1.0028,
272
+ "step": 22000
273
+ },
274
+ {
275
+ "epoch": 0.09,
276
+ "learning_rate": 1.818242036981687e-05,
277
+ "loss": 1.0514,
278
+ "step": 22500
279
+ },
280
+ {
281
+ "epoch": 0.09,
282
+ "learning_rate": 1.8142029711368355e-05,
283
+ "loss": 1.0166,
284
+ "step": 23000
285
+ },
286
+ {
287
+ "epoch": 0.09,
288
+ "learning_rate": 1.810163905291984e-05,
289
+ "loss": 0.9976,
290
+ "step": 23500
291
+ },
292
+ {
293
+ "epoch": 0.1,
294
+ "learning_rate": 1.8061248394471328e-05,
295
+ "loss": 1.0427,
296
+ "step": 24000
297
+ },
298
+ {
299
+ "epoch": 0.1,
300
+ "learning_rate": 1.8020857736022813e-05,
301
+ "loss": 1.0069,
302
+ "step": 24500
303
+ },
304
+ {
305
+ "epoch": 0.1,
306
+ "learning_rate": 1.79804670775743e-05,
307
+ "loss": 1.035,
308
+ "step": 25000
309
+ },
310
+ {
311
+ "epoch": 0.1,
312
+ "learning_rate": 1.7940076419125786e-05,
313
+ "loss": 0.9844,
314
+ "step": 25500
315
+ },
316
+ {
317
+ "epoch": 0.11,
318
+ "learning_rate": 1.7899685760677274e-05,
319
+ "loss": 0.9708,
320
+ "step": 26000
321
+ },
322
+ {
323
+ "epoch": 0.11,
324
+ "learning_rate": 1.785929510222876e-05,
325
+ "loss": 1.0022,
326
+ "step": 26500
327
+ },
328
+ {
329
+ "epoch": 0.11,
330
+ "learning_rate": 1.7818904443780243e-05,
331
+ "loss": 1.037,
332
+ "step": 27000
333
+ },
334
+ {
335
+ "epoch": 0.11,
336
+ "learning_rate": 1.777851378533173e-05,
337
+ "loss": 0.9876,
338
+ "step": 27500
339
+ },
340
+ {
341
+ "epoch": 0.11,
342
+ "learning_rate": 1.7738123126883216e-05,
343
+ "loss": 0.9781,
344
+ "step": 28000
345
+ },
346
+ {
347
+ "epoch": 0.12,
348
+ "learning_rate": 1.7697732468434705e-05,
349
+ "loss": 0.9913,
350
+ "step": 28500
351
+ },
352
+ {
353
+ "epoch": 0.12,
354
+ "learning_rate": 1.765734180998619e-05,
355
+ "loss": 1.0041,
356
+ "step": 29000
357
+ },
358
+ {
359
+ "epoch": 0.12,
360
+ "learning_rate": 1.7616951151537674e-05,
361
+ "loss": 1.0104,
362
+ "step": 29500
363
+ },
364
+ {
365
+ "epoch": 0.12,
366
+ "learning_rate": 1.757656049308916e-05,
367
+ "loss": 0.9863,
368
+ "step": 30000
369
+ },
370
+ {
371
+ "epoch": 0.12,
372
+ "learning_rate": 1.7536169834640647e-05,
373
+ "loss": 1.0076,
374
+ "step": 30500
375
+ },
376
+ {
377
+ "epoch": 0.13,
378
+ "learning_rate": 1.7495779176192132e-05,
379
+ "loss": 0.9887,
380
+ "step": 31000
381
+ },
382
+ {
383
+ "epoch": 0.13,
384
+ "learning_rate": 1.7455388517743617e-05,
385
+ "loss": 0.9595,
386
+ "step": 31500
387
+ },
388
+ {
389
+ "epoch": 0.13,
390
+ "learning_rate": 1.7414997859295105e-05,
391
+ "loss": 0.9512,
392
+ "step": 32000
393
+ },
394
+ {
395
+ "epoch": 0.13,
396
+ "learning_rate": 1.737460720084659e-05,
397
+ "loss": 0.9956,
398
+ "step": 32500
399
+ },
400
+ {
401
+ "epoch": 0.13,
402
+ "learning_rate": 1.7334216542398074e-05,
403
+ "loss": 0.9986,
404
+ "step": 33000
405
+ },
406
+ {
407
+ "epoch": 0.14,
408
+ "learning_rate": 1.7293825883949562e-05,
409
+ "loss": 0.9447,
410
+ "step": 33500
411
+ },
412
+ {
413
+ "epoch": 0.14,
414
+ "learning_rate": 1.7253435225501047e-05,
415
+ "loss": 0.9576,
416
+ "step": 34000
417
+ },
418
+ {
419
+ "epoch": 0.14,
420
+ "learning_rate": 1.7213044567052532e-05,
421
+ "loss": 0.9977,
422
+ "step": 34500
423
+ },
424
+ {
425
+ "epoch": 0.14,
426
+ "learning_rate": 1.717265390860402e-05,
427
+ "loss": 0.9411,
428
+ "step": 35000
429
+ },
430
+ {
431
+ "epoch": 0.14,
432
+ "learning_rate": 1.7132263250155505e-05,
433
+ "loss": 1.0055,
434
+ "step": 35500
435
+ },
436
+ {
437
+ "epoch": 0.15,
438
+ "learning_rate": 1.709187259170699e-05,
439
+ "loss": 1.0319,
440
+ "step": 36000
441
+ },
442
+ {
443
+ "epoch": 0.15,
444
+ "learning_rate": 1.7051481933258478e-05,
445
+ "loss": 0.9554,
446
+ "step": 36500
447
+ },
448
+ {
449
+ "epoch": 0.15,
450
+ "learning_rate": 1.7011091274809963e-05,
451
+ "loss": 0.9689,
452
+ "step": 37000
453
+ },
454
+ {
455
+ "epoch": 0.15,
456
+ "learning_rate": 1.6970700616361447e-05,
457
+ "loss": 0.9857,
458
+ "step": 37500
459
+ },
460
+ {
461
+ "epoch": 0.15,
462
+ "learning_rate": 1.6930309957912935e-05,
463
+ "loss": 0.9764,
464
+ "step": 38000
465
+ },
466
+ {
467
+ "epoch": 0.16,
468
+ "learning_rate": 1.688991929946442e-05,
469
+ "loss": 0.9616,
470
+ "step": 38500
471
+ },
472
+ {
473
+ "epoch": 0.16,
474
+ "learning_rate": 1.6849528641015905e-05,
475
+ "loss": 0.9978,
476
+ "step": 39000
477
+ },
478
+ {
479
+ "epoch": 0.16,
480
+ "learning_rate": 1.6809137982567393e-05,
481
+ "loss": 0.9387,
482
+ "step": 39500
483
+ },
484
+ {
485
+ "epoch": 0.16,
486
+ "learning_rate": 1.6768747324118878e-05,
487
+ "loss": 0.9403,
488
+ "step": 40000
489
+ },
490
+ {
491
+ "epoch": 0.16,
492
+ "learning_rate": 1.6728356665670366e-05,
493
+ "loss": 0.9615,
494
+ "step": 40500
495
+ },
496
+ {
497
+ "epoch": 0.17,
498
+ "learning_rate": 1.668796600722185e-05,
499
+ "loss": 0.9558,
500
+ "step": 41000
501
+ },
502
+ {
503
+ "epoch": 0.17,
504
+ "learning_rate": 1.6647575348773336e-05,
505
+ "loss": 0.9855,
506
+ "step": 41500
507
+ },
508
+ {
509
+ "epoch": 0.17,
510
+ "learning_rate": 1.6607184690324824e-05,
511
+ "loss": 1.0088,
512
+ "step": 42000
513
+ },
514
+ {
515
+ "epoch": 0.17,
516
+ "learning_rate": 1.656679403187631e-05,
517
+ "loss": 0.9751,
518
+ "step": 42500
519
+ },
520
+ {
521
+ "epoch": 0.17,
522
+ "learning_rate": 1.6526403373427797e-05,
523
+ "loss": 0.9547,
524
+ "step": 43000
525
+ },
526
+ {
527
+ "epoch": 0.18,
528
+ "learning_rate": 1.648601271497928e-05,
529
+ "loss": 0.9572,
530
+ "step": 43500
531
+ },
532
+ {
533
+ "epoch": 0.18,
534
+ "learning_rate": 1.6445622056530766e-05,
535
+ "loss": 0.9317,
536
+ "step": 44000
537
+ },
538
+ {
539
+ "epoch": 0.18,
540
+ "learning_rate": 1.6405231398082254e-05,
541
+ "loss": 0.9836,
542
+ "step": 44500
543
+ },
544
+ {
545
+ "epoch": 0.18,
546
+ "learning_rate": 1.636484073963374e-05,
547
+ "loss": 0.9795,
548
+ "step": 45000
549
+ },
550
+ {
551
+ "epoch": 0.18,
552
+ "learning_rate": 1.6324450081185227e-05,
553
+ "loss": 0.9789,
554
+ "step": 45500
555
+ },
556
+ {
557
+ "epoch": 0.19,
558
+ "learning_rate": 1.6284059422736712e-05,
559
+ "loss": 0.9628,
560
+ "step": 46000
561
+ },
562
+ {
563
+ "epoch": 0.19,
564
+ "learning_rate": 1.6243668764288197e-05,
565
+ "loss": 0.9666,
566
+ "step": 46500
567
+ },
568
+ {
569
+ "epoch": 0.19,
570
+ "learning_rate": 1.6203278105839685e-05,
571
+ "loss": 0.9821,
572
+ "step": 47000
573
+ },
574
+ {
575
+ "epoch": 0.19,
576
+ "learning_rate": 1.616288744739117e-05,
577
+ "loss": 0.9409,
578
+ "step": 47500
579
+ },
580
+ {
581
+ "epoch": 0.19,
582
+ "learning_rate": 1.6122496788942655e-05,
583
+ "loss": 1.0201,
584
+ "step": 48000
585
+ },
586
+ {
587
+ "epoch": 0.2,
588
+ "learning_rate": 1.6082106130494143e-05,
589
+ "loss": 0.9693,
590
+ "step": 48500
591
+ },
592
+ {
593
+ "epoch": 0.2,
594
+ "learning_rate": 1.6041715472045627e-05,
595
+ "loss": 0.9974,
596
+ "step": 49000
597
+ },
598
+ {
599
+ "epoch": 0.2,
600
+ "learning_rate": 1.6001324813597112e-05,
601
+ "loss": 0.9703,
602
+ "step": 49500
603
+ },
604
+ {
605
+ "epoch": 0.2,
606
+ "learning_rate": 1.5960934155148597e-05,
607
+ "loss": 0.9504,
608
+ "step": 50000
609
+ },
610
+ {
611
+ "epoch": 0.2,
612
+ "learning_rate": 1.5920543496700085e-05,
613
+ "loss": 0.9595,
614
+ "step": 50500
615
+ },
616
+ {
617
+ "epoch": 0.21,
618
+ "learning_rate": 1.588015283825157e-05,
619
+ "loss": 0.9734,
620
+ "step": 51000
621
+ },
622
+ {
623
+ "epoch": 0.21,
624
+ "learning_rate": 1.5839762179803055e-05,
625
+ "loss": 0.945,
626
+ "step": 51500
627
+ },
628
+ {
629
+ "epoch": 0.21,
630
+ "learning_rate": 1.5799371521354543e-05,
631
+ "loss": 0.9776,
632
+ "step": 52000
633
+ },
634
+ {
635
+ "epoch": 0.21,
636
+ "learning_rate": 1.5758980862906028e-05,
637
+ "loss": 0.9735,
638
+ "step": 52500
639
+ },
640
+ {
641
+ "epoch": 0.21,
642
+ "learning_rate": 1.5718590204457512e-05,
643
+ "loss": 1.007,
644
+ "step": 53000
645
+ },
646
+ {
647
+ "epoch": 0.22,
648
+ "learning_rate": 1.5678199546009e-05,
649
+ "loss": 0.9473,
650
+ "step": 53500
651
+ },
652
+ {
653
+ "epoch": 0.22,
654
+ "learning_rate": 1.5637808887560485e-05,
655
+ "loss": 0.9741,
656
+ "step": 54000
657
+ },
658
+ {
659
+ "epoch": 0.22,
660
+ "learning_rate": 1.559741822911197e-05,
661
+ "loss": 0.9218,
662
+ "step": 54500
663
+ },
664
+ {
665
+ "epoch": 0.22,
666
+ "learning_rate": 1.5557027570663458e-05,
667
+ "loss": 0.9297,
668
+ "step": 55000
669
+ },
670
+ {
671
+ "epoch": 0.22,
672
+ "learning_rate": 1.5516636912214943e-05,
673
+ "loss": 0.9443,
674
+ "step": 55500
675
+ },
676
+ {
677
+ "epoch": 0.23,
678
+ "learning_rate": 1.5476246253766428e-05,
679
+ "loss": 0.9553,
680
+ "step": 56000
681
+ },
682
+ {
683
+ "epoch": 0.23,
684
+ "learning_rate": 1.5435855595317916e-05,
685
+ "loss": 1.0098,
686
+ "step": 56500
687
+ },
688
+ {
689
+ "epoch": 0.23,
690
+ "learning_rate": 1.53954649368694e-05,
691
+ "loss": 0.9684,
692
+ "step": 57000
693
+ },
694
+ {
695
+ "epoch": 0.23,
696
+ "learning_rate": 1.535507427842089e-05,
697
+ "loss": 0.9874,
698
+ "step": 57500
699
+ },
700
+ {
701
+ "epoch": 0.23,
702
+ "learning_rate": 1.5314683619972374e-05,
703
+ "loss": 0.9602,
704
+ "step": 58000
705
+ },
706
+ {
707
+ "epoch": 0.24,
708
+ "learning_rate": 1.527429296152386e-05,
709
+ "loss": 0.9409,
710
+ "step": 58500
711
+ },
712
+ {
713
+ "epoch": 0.24,
714
+ "learning_rate": 1.5233902303075345e-05,
715
+ "loss": 0.9379,
716
+ "step": 59000
717
+ },
718
+ {
719
+ "epoch": 0.24,
720
+ "learning_rate": 1.5193511644626833e-05,
721
+ "loss": 0.9913,
722
+ "step": 59500
723
+ },
724
+ {
725
+ "epoch": 0.24,
726
+ "learning_rate": 1.5153120986178318e-05,
727
+ "loss": 0.9938,
728
+ "step": 60000
729
+ },
730
+ {
731
+ "epoch": 0.24,
732
+ "learning_rate": 1.5112730327729802e-05,
733
+ "loss": 0.9803,
734
+ "step": 60500
735
+ },
736
+ {
737
+ "epoch": 0.25,
738
+ "learning_rate": 1.507233966928129e-05,
739
+ "loss": 0.9494,
740
+ "step": 61000
741
+ },
742
+ {
743
+ "epoch": 0.25,
744
+ "learning_rate": 1.5031949010832775e-05,
745
+ "loss": 0.9691,
746
+ "step": 61500
747
+ },
748
+ {
749
+ "epoch": 0.25,
750
+ "learning_rate": 1.4991558352384262e-05,
751
+ "loss": 0.9712,
752
+ "step": 62000
753
+ },
754
+ {
755
+ "epoch": 0.25,
756
+ "learning_rate": 1.4951167693935748e-05,
757
+ "loss": 0.9205,
758
+ "step": 62500
759
+ },
760
+ {
761
+ "epoch": 0.25,
762
+ "learning_rate": 1.4910777035487233e-05,
763
+ "loss": 0.9914,
764
+ "step": 63000
765
+ },
766
+ {
767
+ "epoch": 0.26,
768
+ "learning_rate": 1.487038637703872e-05,
769
+ "loss": 0.9812,
770
+ "step": 63500
771
+ },
772
+ {
773
+ "epoch": 0.26,
774
+ "learning_rate": 1.4829995718590206e-05,
775
+ "loss": 0.9621,
776
+ "step": 64000
777
+ },
778
+ {
779
+ "epoch": 0.26,
780
+ "learning_rate": 1.4789605060141692e-05,
781
+ "loss": 0.964,
782
+ "step": 64500
783
+ },
784
+ {
785
+ "epoch": 0.26,
786
+ "learning_rate": 1.4749214401693177e-05,
787
+ "loss": 0.9682,
788
+ "step": 65000
789
+ },
790
+ {
791
+ "epoch": 0.26,
792
+ "learning_rate": 1.4708823743244664e-05,
793
+ "loss": 0.983,
794
+ "step": 65500
795
+ },
796
+ {
797
+ "epoch": 0.27,
798
+ "learning_rate": 1.466843308479615e-05,
799
+ "loss": 0.9768,
800
+ "step": 66000
801
+ },
802
+ {
803
+ "epoch": 0.27,
804
+ "learning_rate": 1.4628042426347635e-05,
805
+ "loss": 0.962,
806
+ "step": 66500
807
+ },
808
+ {
809
+ "epoch": 0.27,
810
+ "learning_rate": 1.4587651767899123e-05,
811
+ "loss": 1.0047,
812
+ "step": 67000
813
+ },
814
+ {
815
+ "epoch": 0.27,
816
+ "learning_rate": 1.4547261109450608e-05,
817
+ "loss": 0.9667,
818
+ "step": 67500
819
+ },
820
+ {
821
+ "epoch": 0.27,
822
+ "learning_rate": 1.4506870451002093e-05,
823
+ "loss": 1.0206,
824
+ "step": 68000
825
+ },
826
+ {
827
+ "epoch": 0.28,
828
+ "learning_rate": 1.446647979255358e-05,
829
+ "loss": 0.937,
830
+ "step": 68500
831
+ },
832
+ {
833
+ "epoch": 0.28,
834
+ "learning_rate": 1.4426089134105066e-05,
835
+ "loss": 0.9796,
836
+ "step": 69000
837
+ },
838
+ {
839
+ "epoch": 0.28,
840
+ "learning_rate": 1.438569847565655e-05,
841
+ "loss": 0.9328,
842
+ "step": 69500
843
+ },
844
+ {
845
+ "epoch": 0.28,
846
+ "learning_rate": 1.4345307817208037e-05,
847
+ "loss": 0.9244,
848
+ "step": 70000
849
+ },
850
+ {
851
+ "epoch": 0.28,
852
+ "learning_rate": 1.4304917158759523e-05,
853
+ "loss": 0.9437,
854
+ "step": 70500
855
+ },
856
+ {
857
+ "epoch": 0.29,
858
+ "learning_rate": 1.426452650031101e-05,
859
+ "loss": 0.9466,
860
+ "step": 71000
861
+ },
862
+ {
863
+ "epoch": 0.29,
864
+ "learning_rate": 1.4224135841862494e-05,
865
+ "loss": 0.9172,
866
+ "step": 71500
867
+ },
868
+ {
869
+ "epoch": 0.29,
870
+ "learning_rate": 1.4183745183413981e-05,
871
+ "loss": 0.9681,
872
+ "step": 72000
873
+ },
874
+ {
875
+ "epoch": 0.29,
876
+ "learning_rate": 1.4143354524965467e-05,
877
+ "loss": 0.9339,
878
+ "step": 72500
879
+ },
880
+ {
881
+ "epoch": 0.29,
882
+ "learning_rate": 1.4102963866516952e-05,
883
+ "loss": 0.9663,
884
+ "step": 73000
885
+ },
886
+ {
887
+ "epoch": 0.3,
888
+ "learning_rate": 1.406257320806844e-05,
889
+ "loss": 0.9324,
890
+ "step": 73500
891
+ },
892
+ {
893
+ "epoch": 0.3,
894
+ "learning_rate": 1.4022182549619925e-05,
895
+ "loss": 0.9784,
896
+ "step": 74000
897
+ },
898
+ {
899
+ "epoch": 0.3,
900
+ "learning_rate": 1.398179189117141e-05,
901
+ "loss": 0.9536,
902
+ "step": 74500
903
+ },
904
+ {
905
+ "epoch": 0.3,
906
+ "learning_rate": 1.3941401232722898e-05,
907
+ "loss": 0.9418,
908
+ "step": 75000
909
+ },
910
+ {
911
+ "epoch": 0.3,
912
+ "learning_rate": 1.3901010574274383e-05,
913
+ "loss": 0.9225,
914
+ "step": 75500
915
+ },
916
+ {
917
+ "epoch": 0.31,
918
+ "learning_rate": 1.3860619915825868e-05,
919
+ "loss": 0.9436,
920
+ "step": 76000
921
+ },
922
+ {
923
+ "epoch": 0.31,
924
+ "learning_rate": 1.3820229257377356e-05,
925
+ "loss": 0.9656,
926
+ "step": 76500
927
+ },
928
+ {
929
+ "epoch": 0.31,
930
+ "learning_rate": 1.377983859892884e-05,
931
+ "loss": 0.9451,
932
+ "step": 77000
933
+ },
934
+ {
935
+ "epoch": 0.31,
936
+ "learning_rate": 1.3739447940480325e-05,
937
+ "loss": 0.9152,
938
+ "step": 77500
939
+ },
940
+ {
941
+ "epoch": 0.32,
942
+ "learning_rate": 1.3699057282031813e-05,
943
+ "loss": 0.9567,
944
+ "step": 78000
945
+ },
946
+ {
947
+ "epoch": 0.32,
948
+ "learning_rate": 1.3658666623583298e-05,
949
+ "loss": 0.9142,
950
+ "step": 78500
951
+ },
952
+ {
953
+ "epoch": 0.32,
954
+ "learning_rate": 1.3618275965134785e-05,
955
+ "loss": 0.9347,
956
+ "step": 79000
957
+ },
958
+ {
959
+ "epoch": 0.32,
960
+ "learning_rate": 1.3577885306686271e-05,
961
+ "loss": 0.9835,
962
+ "step": 79500
963
+ },
964
+ {
965
+ "epoch": 0.32,
966
+ "learning_rate": 1.3537494648237758e-05,
967
+ "loss": 0.9171,
968
+ "step": 80000
969
+ },
970
+ {
971
+ "epoch": 0.33,
972
+ "learning_rate": 1.3497103989789242e-05,
973
+ "loss": 0.9485,
974
+ "step": 80500
975
+ },
976
+ {
977
+ "epoch": 0.33,
978
+ "learning_rate": 1.3456713331340729e-05,
979
+ "loss": 0.9879,
980
+ "step": 81000
981
+ },
982
+ {
983
+ "epoch": 0.33,
984
+ "learning_rate": 1.3416322672892215e-05,
985
+ "loss": 0.9326,
986
+ "step": 81500
987
+ },
988
+ {
989
+ "epoch": 0.33,
990
+ "learning_rate": 1.33759320144437e-05,
991
+ "loss": 0.9512,
992
+ "step": 82000
993
+ },
994
+ {
995
+ "epoch": 0.33,
996
+ "learning_rate": 1.3335541355995188e-05,
997
+ "loss": 0.9233,
998
+ "step": 82500
999
+ },
1000
+ {
1001
+ "epoch": 0.34,
1002
+ "learning_rate": 1.3295150697546673e-05,
1003
+ "loss": 0.9232,
1004
+ "step": 83000
1005
+ },
1006
+ {
1007
+ "epoch": 0.34,
1008
+ "learning_rate": 1.3254760039098158e-05,
1009
+ "loss": 0.9353,
1010
+ "step": 83500
1011
+ },
1012
+ {
1013
+ "epoch": 0.34,
1014
+ "learning_rate": 1.3214369380649646e-05,
1015
+ "loss": 0.9536,
1016
+ "step": 84000
1017
+ },
1018
+ {
1019
+ "epoch": 0.34,
1020
+ "learning_rate": 1.317397872220113e-05,
1021
+ "loss": 0.9547,
1022
+ "step": 84500
1023
+ },
1024
+ {
1025
+ "epoch": 0.34,
1026
+ "learning_rate": 1.3133588063752615e-05,
1027
+ "loss": 0.9323,
1028
+ "step": 85000
1029
+ },
1030
+ {
1031
+ "epoch": 0.35,
1032
+ "learning_rate": 1.3093197405304104e-05,
1033
+ "loss": 0.9414,
1034
+ "step": 85500
1035
+ },
1036
+ {
1037
+ "epoch": 0.35,
1038
+ "learning_rate": 1.3052806746855588e-05,
1039
+ "loss": 0.9694,
1040
+ "step": 86000
1041
+ },
1042
+ {
1043
+ "epoch": 0.35,
1044
+ "learning_rate": 1.3012416088407073e-05,
1045
+ "loss": 0.931,
1046
+ "step": 86500
1047
+ },
1048
+ {
1049
+ "epoch": 0.35,
1050
+ "learning_rate": 1.2972025429958561e-05,
1051
+ "loss": 0.949,
1052
+ "step": 87000
1053
+ },
1054
+ {
1055
+ "epoch": 0.35,
1056
+ "learning_rate": 1.2931634771510046e-05,
1057
+ "loss": 0.9511,
1058
+ "step": 87500
1059
+ },
1060
+ {
1061
+ "epoch": 0.36,
1062
+ "learning_rate": 1.2891244113061532e-05,
1063
+ "loss": 0.9378,
1064
+ "step": 88000
1065
+ },
1066
+ {
1067
+ "epoch": 0.36,
1068
+ "learning_rate": 1.2850853454613019e-05,
1069
+ "loss": 0.9475,
1070
+ "step": 88500
1071
+ },
1072
+ {
1073
+ "epoch": 0.36,
1074
+ "learning_rate": 1.2810462796164504e-05,
1075
+ "loss": 0.9869,
1076
+ "step": 89000
1077
+ },
1078
+ {
1079
+ "epoch": 0.36,
1080
+ "learning_rate": 1.277007213771599e-05,
1081
+ "loss": 0.9298,
1082
+ "step": 89500
1083
+ },
1084
+ {
1085
+ "epoch": 0.36,
1086
+ "learning_rate": 1.2729681479267475e-05,
1087
+ "loss": 0.9133,
1088
+ "step": 90000
1089
+ },
1090
+ {
1091
+ "epoch": 0.37,
1092
+ "learning_rate": 1.2689290820818963e-05,
1093
+ "loss": 0.9378,
1094
+ "step": 90500
1095
+ },
1096
+ {
1097
+ "epoch": 0.37,
1098
+ "learning_rate": 1.2648900162370448e-05,
1099
+ "loss": 0.9598,
1100
+ "step": 91000
1101
+ },
1102
+ {
1103
+ "epoch": 0.37,
1104
+ "learning_rate": 1.2608509503921933e-05,
1105
+ "loss": 0.9438,
1106
+ "step": 91500
1107
+ },
1108
+ {
1109
+ "epoch": 0.37,
1110
+ "learning_rate": 1.256811884547342e-05,
1111
+ "loss": 0.9571,
1112
+ "step": 92000
1113
+ },
1114
+ {
1115
+ "epoch": 0.37,
1116
+ "learning_rate": 1.2527728187024906e-05,
1117
+ "loss": 0.9755,
1118
+ "step": 92500
1119
+ },
1120
+ {
1121
+ "epoch": 0.38,
1122
+ "learning_rate": 1.248733752857639e-05,
1123
+ "loss": 0.9469,
1124
+ "step": 93000
1125
+ },
1126
+ {
1127
+ "epoch": 0.38,
1128
+ "learning_rate": 1.2446946870127878e-05,
1129
+ "loss": 0.9402,
1130
+ "step": 93500
1131
+ },
1132
+ {
1133
+ "epoch": 0.38,
1134
+ "learning_rate": 1.2406556211679363e-05,
1135
+ "loss": 0.9425,
1136
+ "step": 94000
1137
+ },
1138
+ {
1139
+ "epoch": 0.38,
1140
+ "learning_rate": 1.236616555323085e-05,
1141
+ "loss": 0.9816,
1142
+ "step": 94500
1143
+ },
1144
+ {
1145
+ "epoch": 0.38,
1146
+ "learning_rate": 1.2325774894782336e-05,
1147
+ "loss": 0.9738,
1148
+ "step": 95000
1149
+ },
1150
+ {
1151
+ "epoch": 0.39,
1152
+ "learning_rate": 1.2285384236333821e-05,
1153
+ "loss": 0.9631,
1154
+ "step": 95500
1155
+ },
1156
+ {
1157
+ "epoch": 0.39,
1158
+ "learning_rate": 1.2244993577885307e-05,
1159
+ "loss": 0.9766,
1160
+ "step": 96000
1161
+ },
1162
+ {
1163
+ "epoch": 0.39,
1164
+ "learning_rate": 1.2204602919436794e-05,
1165
+ "loss": 0.9972,
1166
+ "step": 96500
1167
+ },
1168
+ {
1169
+ "epoch": 0.39,
1170
+ "learning_rate": 1.216421226098828e-05,
1171
+ "loss": 0.9506,
1172
+ "step": 97000
1173
+ },
1174
+ {
1175
+ "epoch": 0.39,
1176
+ "learning_rate": 1.2123821602539765e-05,
1177
+ "loss": 0.9384,
1178
+ "step": 97500
1179
+ },
1180
+ {
1181
+ "epoch": 0.4,
1182
+ "learning_rate": 1.2083430944091252e-05,
1183
+ "loss": 0.9416,
1184
+ "step": 98000
1185
+ },
1186
+ {
1187
+ "epoch": 0.4,
1188
+ "learning_rate": 1.2043040285642738e-05,
1189
+ "loss": 0.9529,
1190
+ "step": 98500
1191
+ },
1192
+ {
1193
+ "epoch": 0.4,
1194
+ "learning_rate": 1.2002649627194223e-05,
1195
+ "loss": 0.9469,
1196
+ "step": 99000
1197
+ },
1198
+ {
1199
+ "epoch": 0.4,
1200
+ "learning_rate": 1.1962258968745711e-05,
1201
+ "loss": 0.8995,
1202
+ "step": 99500
1203
+ },
1204
+ {
1205
+ "epoch": 0.4,
1206
+ "learning_rate": 1.1921868310297196e-05,
1207
+ "loss": 0.9206,
1208
+ "step": 100000
1209
+ },
1210
+ {
1211
+ "epoch": 0.41,
1212
+ "learning_rate": 1.188147765184868e-05,
1213
+ "loss": 0.9306,
1214
+ "step": 100500
1215
+ },
1216
+ {
1217
+ "epoch": 0.41,
1218
+ "learning_rate": 1.1841086993400169e-05,
1219
+ "loss": 0.9977,
1220
+ "step": 101000
1221
+ },
1222
+ {
1223
+ "epoch": 0.41,
1224
+ "learning_rate": 1.1800696334951653e-05,
1225
+ "loss": 0.9301,
1226
+ "step": 101500
1227
+ },
1228
+ {
1229
+ "epoch": 0.41,
1230
+ "learning_rate": 1.1760305676503138e-05,
1231
+ "loss": 0.9425,
1232
+ "step": 102000
1233
+ },
1234
+ {
1235
+ "epoch": 0.41,
1236
+ "learning_rate": 1.1719915018054626e-05,
1237
+ "loss": 0.8925,
1238
+ "step": 102500
1239
+ },
1240
+ {
1241
+ "epoch": 0.42,
1242
+ "learning_rate": 1.1679524359606111e-05,
1243
+ "loss": 0.9098,
1244
+ "step": 103000
1245
+ },
1246
+ {
1247
+ "epoch": 0.42,
1248
+ "learning_rate": 1.1639133701157598e-05,
1249
+ "loss": 0.9129,
1250
+ "step": 103500
1251
+ },
1252
+ {
1253
+ "epoch": 0.42,
1254
+ "learning_rate": 1.1598743042709084e-05,
1255
+ "loss": 0.9247,
1256
+ "step": 104000
1257
+ },
1258
+ {
1259
+ "epoch": 0.42,
1260
+ "learning_rate": 1.1558352384260569e-05,
1261
+ "loss": 0.9092,
1262
+ "step": 104500
1263
+ },
1264
+ {
1265
+ "epoch": 0.42,
1266
+ "learning_rate": 1.1517961725812055e-05,
1267
+ "loss": 0.9261,
1268
+ "step": 105000
1269
+ },
1270
+ {
1271
+ "epoch": 0.43,
1272
+ "learning_rate": 1.1477571067363542e-05,
1273
+ "loss": 0.9342,
1274
+ "step": 105500
1275
+ },
1276
+ {
1277
+ "epoch": 0.43,
1278
+ "learning_rate": 1.1437180408915028e-05,
1279
+ "loss": 0.9485,
1280
+ "step": 106000
1281
+ },
1282
+ {
1283
+ "epoch": 0.43,
1284
+ "learning_rate": 1.1396789750466513e-05,
1285
+ "loss": 0.8406,
1286
+ "step": 106500
1287
+ },
1288
+ {
1289
+ "epoch": 0.43,
1290
+ "learning_rate": 1.1356399092018e-05,
1291
+ "loss": 0.9847,
1292
+ "step": 107000
1293
+ },
1294
+ {
1295
+ "epoch": 0.43,
1296
+ "learning_rate": 1.1316008433569486e-05,
1297
+ "loss": 0.9163,
1298
+ "step": 107500
1299
+ },
1300
+ {
1301
+ "epoch": 0.44,
1302
+ "learning_rate": 1.127561777512097e-05,
1303
+ "loss": 0.9396,
1304
+ "step": 108000
1305
+ },
1306
+ {
1307
+ "epoch": 0.44,
1308
+ "learning_rate": 1.1235227116672459e-05,
1309
+ "loss": 0.928,
1310
+ "step": 108500
1311
+ },
1312
+ {
1313
+ "epoch": 0.44,
1314
+ "learning_rate": 1.1194836458223944e-05,
1315
+ "loss": 0.9116,
1316
+ "step": 109000
1317
+ },
1318
+ {
1319
+ "epoch": 0.44,
1320
+ "learning_rate": 1.1154445799775428e-05,
1321
+ "loss": 0.9316,
1322
+ "step": 109500
1323
+ },
1324
+ {
1325
+ "epoch": 0.44,
1326
+ "learning_rate": 1.1114055141326913e-05,
1327
+ "loss": 0.915,
1328
+ "step": 110000
1329
+ },
1330
+ {
1331
+ "epoch": 0.45,
1332
+ "learning_rate": 1.1073664482878401e-05,
1333
+ "loss": 0.9313,
1334
+ "step": 110500
1335
+ },
1336
+ {
1337
+ "epoch": 0.45,
1338
+ "learning_rate": 1.1033273824429886e-05,
1339
+ "loss": 0.9005,
1340
+ "step": 111000
1341
+ },
1342
+ {
1343
+ "epoch": 0.45,
1344
+ "learning_rate": 1.0992883165981372e-05,
1345
+ "loss": 0.9014,
1346
+ "step": 111500
1347
+ },
1348
+ {
1349
+ "epoch": 0.45,
1350
+ "learning_rate": 1.0952492507532859e-05,
1351
+ "loss": 0.923,
1352
+ "step": 112000
1353
+ },
1354
+ {
1355
+ "epoch": 0.45,
1356
+ "learning_rate": 1.0912101849084344e-05,
1357
+ "loss": 0.9115,
1358
+ "step": 112500
1359
+ },
1360
+ {
1361
+ "epoch": 0.46,
1362
+ "learning_rate": 1.087171119063583e-05,
1363
+ "loss": 0.9482,
1364
+ "step": 113000
1365
+ },
1366
+ {
1367
+ "epoch": 0.46,
1368
+ "learning_rate": 1.0831320532187317e-05,
1369
+ "loss": 0.9236,
1370
+ "step": 113500
1371
+ },
1372
+ {
1373
+ "epoch": 0.46,
1374
+ "learning_rate": 1.0790929873738803e-05,
1375
+ "loss": 0.9172,
1376
+ "step": 114000
1377
+ },
1378
+ {
1379
+ "epoch": 0.46,
1380
+ "learning_rate": 1.0750539215290288e-05,
1381
+ "loss": 0.9374,
1382
+ "step": 114500
1383
+ },
1384
+ {
1385
+ "epoch": 0.46,
1386
+ "learning_rate": 1.0710148556841776e-05,
1387
+ "loss": 0.9542,
1388
+ "step": 115000
1389
+ },
1390
+ {
1391
+ "epoch": 0.47,
1392
+ "learning_rate": 1.066975789839326e-05,
1393
+ "loss": 0.9249,
1394
+ "step": 115500
1395
+ },
1396
+ {
1397
+ "epoch": 0.47,
1398
+ "learning_rate": 1.0629367239944745e-05,
1399
+ "loss": 0.93,
1400
+ "step": 116000
1401
+ },
1402
+ {
1403
+ "epoch": 0.47,
1404
+ "learning_rate": 1.0588976581496234e-05,
1405
+ "loss": 0.9265,
1406
+ "step": 116500
1407
+ },
1408
+ {
1409
+ "epoch": 0.47,
1410
+ "learning_rate": 1.0548585923047718e-05,
1411
+ "loss": 0.8939,
1412
+ "step": 117000
1413
+ },
1414
+ {
1415
+ "epoch": 0.47,
1416
+ "learning_rate": 1.0508195264599203e-05,
1417
+ "loss": 0.9213,
1418
+ "step": 117500
1419
+ },
1420
+ {
1421
+ "epoch": 0.48,
1422
+ "learning_rate": 1.0467804606150691e-05,
1423
+ "loss": 0.9554,
1424
+ "step": 118000
1425
+ },
1426
+ {
1427
+ "epoch": 0.48,
1428
+ "learning_rate": 1.0427413947702176e-05,
1429
+ "loss": 0.9253,
1430
+ "step": 118500
1431
+ },
1432
+ {
1433
+ "epoch": 0.48,
1434
+ "learning_rate": 1.0387023289253661e-05,
1435
+ "loss": 0.9161,
1436
+ "step": 119000
1437
+ },
1438
+ {
1439
+ "epoch": 0.48,
1440
+ "learning_rate": 1.0346632630805149e-05,
1441
+ "loss": 0.8829,
1442
+ "step": 119500
1443
+ },
1444
+ {
1445
+ "epoch": 0.48,
1446
+ "learning_rate": 1.0306241972356634e-05,
1447
+ "loss": 0.9496,
1448
+ "step": 120000
1449
+ },
1450
+ {
1451
+ "epoch": 0.49,
1452
+ "learning_rate": 1.026585131390812e-05,
1453
+ "loss": 0.8863,
1454
+ "step": 120500
1455
+ },
1456
+ {
1457
+ "epoch": 0.49,
1458
+ "learning_rate": 1.0225460655459607e-05,
1459
+ "loss": 0.911,
1460
+ "step": 121000
1461
+ },
1462
+ {
1463
+ "epoch": 0.49,
1464
+ "learning_rate": 1.0185069997011091e-05,
1465
+ "loss": 0.936,
1466
+ "step": 121500
1467
+ },
1468
+ {
1469
+ "epoch": 0.49,
1470
+ "learning_rate": 1.0144679338562578e-05,
1471
+ "loss": 0.9418,
1472
+ "step": 122000
1473
+ },
1474
+ {
1475
+ "epoch": 0.49,
1476
+ "learning_rate": 1.0104288680114064e-05,
1477
+ "loss": 0.9327,
1478
+ "step": 122500
1479
+ },
1480
+ {
1481
+ "epoch": 0.5,
1482
+ "learning_rate": 1.0063898021665551e-05,
1483
+ "loss": 0.9365,
1484
+ "step": 123000
1485
+ },
1486
+ {
1487
+ "epoch": 0.5,
1488
+ "learning_rate": 1.0023507363217036e-05,
1489
+ "loss": 0.9386,
1490
+ "step": 123500
1491
+ },
1492
+ {
1493
+ "epoch": 0.5,
1494
+ "learning_rate": 9.983116704768522e-06,
1495
+ "loss": 0.9122,
1496
+ "step": 124000
1497
+ },
1498
+ {
1499
+ "epoch": 0.5,
1500
+ "learning_rate": 9.942726046320009e-06,
1501
+ "loss": 0.9047,
1502
+ "step": 124500
1503
+ },
1504
+ {
1505
+ "epoch": 0.5,
1506
+ "learning_rate": 9.902335387871493e-06,
1507
+ "loss": 0.9056,
1508
+ "step": 125000
1509
+ },
1510
+ {
1511
+ "epoch": 0.51,
1512
+ "learning_rate": 9.86194472942298e-06,
1513
+ "loss": 0.9117,
1514
+ "step": 125500
1515
+ },
1516
+ {
1517
+ "epoch": 0.51,
1518
+ "learning_rate": 9.821554070974466e-06,
1519
+ "loss": 0.9186,
1520
+ "step": 126000
1521
+ },
1522
+ {
1523
+ "epoch": 0.51,
1524
+ "learning_rate": 9.781163412525951e-06,
1525
+ "loss": 0.9437,
1526
+ "step": 126500
1527
+ },
1528
+ {
1529
+ "epoch": 0.51,
1530
+ "learning_rate": 9.740772754077437e-06,
1531
+ "loss": 0.8771,
1532
+ "step": 127000
1533
+ },
1534
+ {
1535
+ "epoch": 0.51,
1536
+ "learning_rate": 9.700382095628924e-06,
1537
+ "loss": 0.9197,
1538
+ "step": 127500
1539
+ },
1540
+ {
1541
+ "epoch": 0.52,
1542
+ "learning_rate": 9.659991437180409e-06,
1543
+ "loss": 0.9057,
1544
+ "step": 128000
1545
+ },
1546
+ {
1547
+ "epoch": 0.52,
1548
+ "learning_rate": 9.619600778731895e-06,
1549
+ "loss": 0.946,
1550
+ "step": 128500
1551
+ },
1552
+ {
1553
+ "epoch": 0.52,
1554
+ "learning_rate": 9.579210120283382e-06,
1555
+ "loss": 0.884,
1556
+ "step": 129000
1557
+ },
1558
+ {
1559
+ "epoch": 0.52,
1560
+ "learning_rate": 9.538819461834868e-06,
1561
+ "loss": 0.9384,
1562
+ "step": 129500
1563
+ },
1564
+ {
1565
+ "epoch": 0.53,
1566
+ "learning_rate": 9.498428803386353e-06,
1567
+ "loss": 0.913,
1568
+ "step": 130000
1569
+ },
1570
+ {
1571
+ "epoch": 0.53,
1572
+ "learning_rate": 9.45803814493784e-06,
1573
+ "loss": 0.8897,
1574
+ "step": 130500
1575
+ },
1576
+ {
1577
+ "epoch": 0.53,
1578
+ "learning_rate": 9.417647486489326e-06,
1579
+ "loss": 0.9152,
1580
+ "step": 131000
1581
+ },
1582
+ {
1583
+ "epoch": 0.53,
1584
+ "learning_rate": 9.377256828040812e-06,
1585
+ "loss": 0.8772,
1586
+ "step": 131500
1587
+ },
1588
+ {
1589
+ "epoch": 0.53,
1590
+ "learning_rate": 9.336866169592299e-06,
1591
+ "loss": 0.9231,
1592
+ "step": 132000
1593
+ },
1594
+ {
1595
+ "epoch": 0.54,
1596
+ "learning_rate": 9.296475511143783e-06,
1597
+ "loss": 0.9081,
1598
+ "step": 132500
1599
+ },
1600
+ {
1601
+ "epoch": 0.54,
1602
+ "learning_rate": 9.25608485269527e-06,
1603
+ "loss": 0.9135,
1604
+ "step": 133000
1605
+ },
1606
+ {
1607
+ "epoch": 0.54,
1608
+ "learning_rate": 9.215694194246755e-06,
1609
+ "loss": 0.9331,
1610
+ "step": 133500
1611
+ },
1612
+ {
1613
+ "epoch": 0.54,
1614
+ "learning_rate": 9.175303535798241e-06,
1615
+ "loss": 0.95,
1616
+ "step": 134000
1617
+ },
1618
+ {
1619
+ "epoch": 0.54,
1620
+ "learning_rate": 9.134912877349728e-06,
1621
+ "loss": 0.9205,
1622
+ "step": 134500
1623
+ },
1624
+ {
1625
+ "epoch": 0.55,
1626
+ "learning_rate": 9.094522218901212e-06,
1627
+ "loss": 0.8951,
1628
+ "step": 135000
1629
+ },
1630
+ {
1631
+ "epoch": 0.55,
1632
+ "learning_rate": 9.054131560452699e-06,
1633
+ "loss": 0.9264,
1634
+ "step": 135500
1635
+ },
1636
+ {
1637
+ "epoch": 0.55,
1638
+ "learning_rate": 9.013740902004185e-06,
1639
+ "loss": 0.879,
1640
+ "step": 136000
1641
+ },
1642
+ {
1643
+ "epoch": 0.55,
1644
+ "learning_rate": 8.97335024355567e-06,
1645
+ "loss": 0.9176,
1646
+ "step": 136500
1647
+ },
1648
+ {
1649
+ "epoch": 0.55,
1650
+ "learning_rate": 8.932959585107157e-06,
1651
+ "loss": 0.905,
1652
+ "step": 137000
1653
+ },
1654
+ {
1655
+ "epoch": 0.56,
1656
+ "learning_rate": 8.892568926658643e-06,
1657
+ "loss": 0.9876,
1658
+ "step": 137500
1659
+ },
1660
+ {
1661
+ "epoch": 0.56,
1662
+ "learning_rate": 8.85217826821013e-06,
1663
+ "loss": 0.958,
1664
+ "step": 138000
1665
+ },
1666
+ {
1667
+ "epoch": 0.56,
1668
+ "learning_rate": 8.811787609761614e-06,
1669
+ "loss": 0.8806,
1670
+ "step": 138500
1671
+ },
1672
+ {
1673
+ "epoch": 0.56,
1674
+ "learning_rate": 8.7713969513131e-06,
1675
+ "loss": 0.9072,
1676
+ "step": 139000
1677
+ },
1678
+ {
1679
+ "epoch": 0.56,
1680
+ "learning_rate": 8.731006292864587e-06,
1681
+ "loss": 0.91,
1682
+ "step": 139500
1683
+ },
1684
+ {
1685
+ "epoch": 0.57,
1686
+ "learning_rate": 8.690615634416074e-06,
1687
+ "loss": 0.8987,
1688
+ "step": 140000
1689
+ },
1690
+ {
1691
+ "epoch": 0.57,
1692
+ "learning_rate": 8.65022497596756e-06,
1693
+ "loss": 0.921,
1694
+ "step": 140500
1695
+ },
1696
+ {
1697
+ "epoch": 0.57,
1698
+ "learning_rate": 8.609834317519045e-06,
1699
+ "loss": 0.955,
1700
+ "step": 141000
1701
+ },
1702
+ {
1703
+ "epoch": 0.57,
1704
+ "learning_rate": 8.569443659070531e-06,
1705
+ "loss": 0.9451,
1706
+ "step": 141500
1707
+ },
1708
+ {
1709
+ "epoch": 0.57,
1710
+ "learning_rate": 8.529053000622018e-06,
1711
+ "loss": 0.8754,
1712
+ "step": 142000
1713
+ },
1714
+ {
1715
+ "epoch": 0.58,
1716
+ "learning_rate": 8.488662342173503e-06,
1717
+ "loss": 0.9248,
1718
+ "step": 142500
1719
+ },
1720
+ {
1721
+ "epoch": 0.58,
1722
+ "learning_rate": 8.448271683724989e-06,
1723
+ "loss": 0.9433,
1724
+ "step": 143000
1725
+ },
1726
+ {
1727
+ "epoch": 0.58,
1728
+ "learning_rate": 8.407881025276474e-06,
1729
+ "loss": 0.9401,
1730
+ "step": 143500
1731
+ },
1732
+ {
1733
+ "epoch": 0.58,
1734
+ "learning_rate": 8.36749036682796e-06,
1735
+ "loss": 0.926,
1736
+ "step": 144000
1737
+ },
1738
+ {
1739
+ "epoch": 0.58,
1740
+ "learning_rate": 8.327099708379447e-06,
1741
+ "loss": 0.9239,
1742
+ "step": 144500
1743
+ },
1744
+ {
1745
+ "epoch": 0.59,
1746
+ "learning_rate": 8.286709049930931e-06,
1747
+ "loss": 0.9226,
1748
+ "step": 145000
1749
+ },
1750
+ {
1751
+ "epoch": 0.59,
1752
+ "learning_rate": 8.246318391482418e-06,
1753
+ "loss": 0.8886,
1754
+ "step": 145500
1755
+ },
1756
+ {
1757
+ "epoch": 0.59,
1758
+ "learning_rate": 8.205927733033904e-06,
1759
+ "loss": 0.9008,
1760
+ "step": 146000
1761
+ },
1762
+ {
1763
+ "epoch": 0.59,
1764
+ "learning_rate": 8.16553707458539e-06,
1765
+ "loss": 0.8982,
1766
+ "step": 146500
1767
+ },
1768
+ {
1769
+ "epoch": 0.59,
1770
+ "learning_rate": 8.125146416136877e-06,
1771
+ "loss": 0.9281,
1772
+ "step": 147000
1773
+ },
1774
+ {
1775
+ "epoch": 0.6,
1776
+ "learning_rate": 8.084755757688362e-06,
1777
+ "loss": 0.8951,
1778
+ "step": 147500
1779
+ },
1780
+ {
1781
+ "epoch": 0.6,
1782
+ "learning_rate": 8.044365099239849e-06,
1783
+ "loss": 0.8947,
1784
+ "step": 148000
1785
+ },
1786
+ {
1787
+ "epoch": 0.6,
1788
+ "learning_rate": 8.003974440791335e-06,
1789
+ "loss": 0.903,
1790
+ "step": 148500
1791
+ },
1792
+ {
1793
+ "epoch": 0.6,
1794
+ "learning_rate": 7.963583782342821e-06,
1795
+ "loss": 0.8872,
1796
+ "step": 149000
1797
+ },
1798
+ {
1799
+ "epoch": 0.6,
1800
+ "learning_rate": 7.923193123894306e-06,
1801
+ "loss": 0.923,
1802
+ "step": 149500
1803
+ },
1804
+ {
1805
+ "epoch": 0.61,
1806
+ "learning_rate": 7.882802465445793e-06,
1807
+ "loss": 0.8929,
1808
+ "step": 150000
1809
+ },
1810
+ {
1811
+ "epoch": 0.61,
1812
+ "learning_rate": 7.842411806997279e-06,
1813
+ "loss": 0.9127,
1814
+ "step": 150500
1815
+ },
1816
+ {
1817
+ "epoch": 0.61,
1818
+ "learning_rate": 7.802021148548764e-06,
1819
+ "loss": 0.8948,
1820
+ "step": 151000
1821
+ },
1822
+ {
1823
+ "epoch": 0.61,
1824
+ "learning_rate": 7.76163049010025e-06,
1825
+ "loss": 0.9375,
1826
+ "step": 151500
1827
+ },
1828
+ {
1829
+ "epoch": 0.61,
1830
+ "learning_rate": 7.721239831651737e-06,
1831
+ "loss": 0.9496,
1832
+ "step": 152000
1833
+ },
1834
+ {
1835
+ "epoch": 0.62,
1836
+ "learning_rate": 7.680849173203222e-06,
1837
+ "loss": 0.8727,
1838
+ "step": 152500
1839
+ },
1840
+ {
1841
+ "epoch": 0.62,
1842
+ "learning_rate": 7.640458514754708e-06,
1843
+ "loss": 0.8689,
1844
+ "step": 153000
1845
+ },
1846
+ {
1847
+ "epoch": 0.62,
1848
+ "learning_rate": 7.600067856306194e-06,
1849
+ "loss": 0.8999,
1850
+ "step": 153500
1851
+ },
1852
+ {
1853
+ "epoch": 0.62,
1854
+ "learning_rate": 7.55967719785768e-06,
1855
+ "loss": 0.9765,
1856
+ "step": 154000
1857
+ },
1858
+ {
1859
+ "epoch": 0.62,
1860
+ "learning_rate": 7.519286539409167e-06,
1861
+ "loss": 0.9241,
1862
+ "step": 154500
1863
+ },
1864
+ {
1865
+ "epoch": 0.63,
1866
+ "learning_rate": 7.478895880960651e-06,
1867
+ "loss": 0.8522,
1868
+ "step": 155000
1869
+ },
1870
+ {
1871
+ "epoch": 0.63,
1872
+ "learning_rate": 7.438505222512138e-06,
1873
+ "loss": 0.919,
1874
+ "step": 155500
1875
+ },
1876
+ {
1877
+ "epoch": 0.63,
1878
+ "learning_rate": 7.398114564063624e-06,
1879
+ "loss": 0.9277,
1880
+ "step": 156000
1881
+ },
1882
+ {
1883
+ "epoch": 0.63,
1884
+ "learning_rate": 7.35772390561511e-06,
1885
+ "loss": 0.9202,
1886
+ "step": 156500
1887
+ },
1888
+ {
1889
+ "epoch": 0.63,
1890
+ "learning_rate": 7.3173332471665955e-06,
1891
+ "loss": 0.8872,
1892
+ "step": 157000
1893
+ },
1894
+ {
1895
+ "epoch": 0.64,
1896
+ "learning_rate": 7.276942588718082e-06,
1897
+ "loss": 0.9106,
1898
+ "step": 157500
1899
+ },
1900
+ {
1901
+ "epoch": 0.64,
1902
+ "learning_rate": 7.236551930269568e-06,
1903
+ "loss": 0.9147,
1904
+ "step": 158000
1905
+ },
1906
+ {
1907
+ "epoch": 0.64,
1908
+ "learning_rate": 7.196161271821054e-06,
1909
+ "loss": 0.8906,
1910
+ "step": 158500
1911
+ },
1912
+ {
1913
+ "epoch": 0.64,
1914
+ "learning_rate": 7.1557706133725405e-06,
1915
+ "loss": 0.8656,
1916
+ "step": 159000
1917
+ },
1918
+ {
1919
+ "epoch": 0.64,
1920
+ "learning_rate": 7.115379954924025e-06,
1921
+ "loss": 0.8898,
1922
+ "step": 159500
1923
+ },
1924
+ {
1925
+ "epoch": 0.65,
1926
+ "learning_rate": 7.074989296475512e-06,
1927
+ "loss": 0.8818,
1928
+ "step": 160000
1929
+ },
1930
+ {
1931
+ "epoch": 0.65,
1932
+ "learning_rate": 7.034598638026998e-06,
1933
+ "loss": 0.9191,
1934
+ "step": 160500
1935
+ },
1936
+ {
1937
+ "epoch": 0.65,
1938
+ "learning_rate": 6.994207979578484e-06,
1939
+ "loss": 0.9219,
1940
+ "step": 161000
1941
+ },
1942
+ {
1943
+ "epoch": 0.65,
1944
+ "learning_rate": 6.9538173211299694e-06,
1945
+ "loss": 0.9275,
1946
+ "step": 161500
1947
+ },
1948
+ {
1949
+ "epoch": 0.65,
1950
+ "learning_rate": 6.913426662681456e-06,
1951
+ "loss": 0.9306,
1952
+ "step": 162000
1953
+ },
1954
+ {
1955
+ "epoch": 0.66,
1956
+ "learning_rate": 6.8730360042329415e-06,
1957
+ "loss": 0.903,
1958
+ "step": 162500
1959
+ },
1960
+ {
1961
+ "epoch": 0.66,
1962
+ "learning_rate": 6.832645345784428e-06,
1963
+ "loss": 0.914,
1964
+ "step": 163000
1965
+ },
1966
+ {
1967
+ "epoch": 0.66,
1968
+ "learning_rate": 6.792254687335913e-06,
1969
+ "loss": 0.9402,
1970
+ "step": 163500
1971
+ },
1972
+ {
1973
+ "epoch": 0.66,
1974
+ "learning_rate": 6.751864028887399e-06,
1975
+ "loss": 0.9246,
1976
+ "step": 164000
1977
+ },
1978
+ {
1979
+ "epoch": 0.66,
1980
+ "learning_rate": 6.711473370438886e-06,
1981
+ "loss": 0.8842,
1982
+ "step": 164500
1983
+ },
1984
+ {
1985
+ "epoch": 0.67,
1986
+ "learning_rate": 6.671082711990371e-06,
1987
+ "loss": 0.8796,
1988
+ "step": 165000
1989
+ },
1990
+ {
1991
+ "epoch": 0.67,
1992
+ "learning_rate": 6.630692053541857e-06,
1993
+ "loss": 0.9058,
1994
+ "step": 165500
1995
+ },
1996
+ {
1997
+ "epoch": 0.67,
1998
+ "learning_rate": 6.590301395093343e-06,
1999
+ "loss": 0.8838,
2000
+ "step": 166000
2001
+ },
2002
+ {
2003
+ "epoch": 0.67,
2004
+ "learning_rate": 6.549910736644829e-06,
2005
+ "loss": 0.9267,
2006
+ "step": 166500
2007
+ },
2008
+ {
2009
+ "epoch": 0.67,
2010
+ "learning_rate": 6.5095200781963154e-06,
2011
+ "loss": 0.8483,
2012
+ "step": 167000
2013
+ },
2014
+ {
2015
+ "epoch": 0.68,
2016
+ "learning_rate": 6.469129419747802e-06,
2017
+ "loss": 0.9276,
2018
+ "step": 167500
2019
+ },
2020
+ {
2021
+ "epoch": 0.68,
2022
+ "learning_rate": 6.428738761299287e-06,
2023
+ "loss": 0.9106,
2024
+ "step": 168000
2025
+ },
2026
+ {
2027
+ "epoch": 0.68,
2028
+ "learning_rate": 6.388348102850773e-06,
2029
+ "loss": 0.8863,
2030
+ "step": 168500
2031
+ },
2032
+ {
2033
+ "epoch": 0.68,
2034
+ "learning_rate": 6.34795744440226e-06,
2035
+ "loss": 0.9168,
2036
+ "step": 169000
2037
+ },
2038
+ {
2039
+ "epoch": 0.68,
2040
+ "learning_rate": 6.307566785953745e-06,
2041
+ "loss": 0.8849,
2042
+ "step": 169500
2043
+ },
2044
+ {
2045
+ "epoch": 0.69,
2046
+ "learning_rate": 6.267176127505231e-06,
2047
+ "loss": 0.8818,
2048
+ "step": 170000
2049
+ },
2050
+ {
2051
+ "epoch": 0.69,
2052
+ "learning_rate": 6.226785469056717e-06,
2053
+ "loss": 0.8809,
2054
+ "step": 170500
2055
+ },
2056
+ {
2057
+ "epoch": 0.69,
2058
+ "learning_rate": 6.186394810608203e-06,
2059
+ "loss": 0.9211,
2060
+ "step": 171000
2061
+ },
2062
+ {
2063
+ "epoch": 0.69,
2064
+ "learning_rate": 6.146004152159689e-06,
2065
+ "loss": 0.9225,
2066
+ "step": 171500
2067
+ },
2068
+ {
2069
+ "epoch": 0.69,
2070
+ "learning_rate": 6.105613493711176e-06,
2071
+ "loss": 0.9148,
2072
+ "step": 172000
2073
+ },
2074
+ {
2075
+ "epoch": 0.7,
2076
+ "learning_rate": 6.065222835262661e-06,
2077
+ "loss": 0.9156,
2078
+ "step": 172500
2079
+ },
2080
+ {
2081
+ "epoch": 0.7,
2082
+ "learning_rate": 6.024832176814147e-06,
2083
+ "loss": 0.9236,
2084
+ "step": 173000
2085
+ },
2086
+ {
2087
+ "epoch": 0.7,
2088
+ "learning_rate": 5.984441518365633e-06,
2089
+ "loss": 0.8661,
2090
+ "step": 173500
2091
+ },
2092
+ {
2093
+ "epoch": 0.7,
2094
+ "learning_rate": 5.944050859917119e-06,
2095
+ "loss": 0.9123,
2096
+ "step": 174000
2097
+ },
2098
+ {
2099
+ "epoch": 0.7,
2100
+ "learning_rate": 5.903660201468605e-06,
2101
+ "loss": 0.894,
2102
+ "step": 174500
2103
+ },
2104
+ {
2105
+ "epoch": 0.71,
2106
+ "learning_rate": 5.86326954302009e-06,
2107
+ "loss": 0.9081,
2108
+ "step": 175000
2109
+ },
2110
+ {
2111
+ "epoch": 0.71,
2112
+ "learning_rate": 5.822878884571577e-06,
2113
+ "loss": 0.8975,
2114
+ "step": 175500
2115
+ },
2116
+ {
2117
+ "epoch": 0.71,
2118
+ "learning_rate": 5.782488226123063e-06,
2119
+ "loss": 0.8747,
2120
+ "step": 176000
2121
+ },
2122
+ {
2123
+ "epoch": 0.71,
2124
+ "learning_rate": 5.742097567674548e-06,
2125
+ "loss": 0.9273,
2126
+ "step": 176500
2127
+ },
2128
+ {
2129
+ "epoch": 0.71,
2130
+ "learning_rate": 5.7017069092260345e-06,
2131
+ "loss": 0.8736,
2132
+ "step": 177000
2133
+ },
2134
+ {
2135
+ "epoch": 0.72,
2136
+ "learning_rate": 5.661316250777521e-06,
2137
+ "loss": 0.8794,
2138
+ "step": 177500
2139
+ },
2140
+ {
2141
+ "epoch": 0.72,
2142
+ "learning_rate": 5.6209255923290066e-06,
2143
+ "loss": 0.9084,
2144
+ "step": 178000
2145
+ },
2146
+ {
2147
+ "epoch": 0.72,
2148
+ "learning_rate": 5.580534933880492e-06,
2149
+ "loss": 0.8854,
2150
+ "step": 178500
2151
+ },
2152
+ {
2153
+ "epoch": 0.72,
2154
+ "learning_rate": 5.540144275431979e-06,
2155
+ "loss": 0.9172,
2156
+ "step": 179000
2157
+ },
2158
+ {
2159
+ "epoch": 0.73,
2160
+ "learning_rate": 5.499753616983464e-06,
2161
+ "loss": 0.9057,
2162
+ "step": 179500
2163
+ },
2164
+ {
2165
+ "epoch": 0.73,
2166
+ "learning_rate": 5.459362958534951e-06,
2167
+ "loss": 0.9074,
2168
+ "step": 180000
2169
+ },
2170
+ {
2171
+ "epoch": 0.73,
2172
+ "learning_rate": 5.418972300086437e-06,
2173
+ "loss": 0.9053,
2174
+ "step": 180500
2175
+ },
2176
+ {
2177
+ "epoch": 0.73,
2178
+ "learning_rate": 5.378581641637922e-06,
2179
+ "loss": 0.893,
2180
+ "step": 181000
2181
+ },
2182
+ {
2183
+ "epoch": 0.73,
2184
+ "learning_rate": 5.338190983189408e-06,
2185
+ "loss": 0.8762,
2186
+ "step": 181500
2187
+ },
2188
+ {
2189
+ "epoch": 0.74,
2190
+ "learning_rate": 5.297800324740895e-06,
2191
+ "loss": 0.8405,
2192
+ "step": 182000
2193
+ },
2194
+ {
2195
+ "epoch": 0.74,
2196
+ "learning_rate": 5.2574096662923805e-06,
2197
+ "loss": 0.8939,
2198
+ "step": 182500
2199
+ },
2200
+ {
2201
+ "epoch": 0.74,
2202
+ "learning_rate": 5.217019007843866e-06,
2203
+ "loss": 0.8593,
2204
+ "step": 183000
2205
+ },
2206
+ {
2207
+ "epoch": 0.74,
2208
+ "learning_rate": 5.176628349395352e-06,
2209
+ "loss": 0.9004,
2210
+ "step": 183500
2211
+ },
2212
+ {
2213
+ "epoch": 0.74,
2214
+ "learning_rate": 5.136237690946838e-06,
2215
+ "loss": 0.903,
2216
+ "step": 184000
2217
+ },
2218
+ {
2219
+ "epoch": 0.75,
2220
+ "learning_rate": 5.095847032498325e-06,
2221
+ "loss": 0.8813,
2222
+ "step": 184500
2223
+ },
2224
+ {
2225
+ "epoch": 0.75,
2226
+ "learning_rate": 5.055456374049809e-06,
2227
+ "loss": 0.8457,
2228
+ "step": 185000
2229
+ },
2230
+ {
2231
+ "epoch": 0.75,
2232
+ "learning_rate": 5.015065715601296e-06,
2233
+ "loss": 0.8782,
2234
+ "step": 185500
2235
+ },
2236
+ {
2237
+ "epoch": 0.75,
2238
+ "learning_rate": 4.9746750571527815e-06,
2239
+ "loss": 0.8841,
2240
+ "step": 186000
2241
+ },
2242
+ {
2243
+ "epoch": 0.75,
2244
+ "learning_rate": 4.934284398704268e-06,
2245
+ "loss": 0.8989,
2246
+ "step": 186500
2247
+ },
2248
+ {
2249
+ "epoch": 0.76,
2250
+ "learning_rate": 4.893893740255754e-06,
2251
+ "loss": 0.8968,
2252
+ "step": 187000
2253
+ },
2254
+ {
2255
+ "epoch": 0.76,
2256
+ "learning_rate": 4.85350308180724e-06,
2257
+ "loss": 0.845,
2258
+ "step": 187500
2259
+ },
2260
+ {
2261
+ "epoch": 0.76,
2262
+ "learning_rate": 4.8131124233587265e-06,
2263
+ "loss": 0.85,
2264
+ "step": 188000
2265
+ },
2266
+ {
2267
+ "epoch": 0.76,
2268
+ "learning_rate": 4.772721764910212e-06,
2269
+ "loss": 0.8625,
2270
+ "step": 188500
2271
+ },
2272
+ {
2273
+ "epoch": 0.76,
2274
+ "learning_rate": 4.732331106461698e-06,
2275
+ "loss": 0.8833,
2276
+ "step": 189000
2277
+ },
2278
+ {
2279
+ "epoch": 0.77,
2280
+ "learning_rate": 4.691940448013184e-06,
2281
+ "loss": 0.9184,
2282
+ "step": 189500
2283
+ },
2284
+ {
2285
+ "epoch": 0.77,
2286
+ "learning_rate": 4.65154978956467e-06,
2287
+ "loss": 0.889,
2288
+ "step": 190000
2289
+ },
2290
+ {
2291
+ "epoch": 0.77,
2292
+ "learning_rate": 4.611159131116155e-06,
2293
+ "loss": 0.8676,
2294
+ "step": 190500
2295
+ },
2296
+ {
2297
+ "epoch": 0.77,
2298
+ "learning_rate": 4.570768472667642e-06,
2299
+ "loss": 0.8962,
2300
+ "step": 191000
2301
+ },
2302
+ {
2303
+ "epoch": 0.77,
2304
+ "learning_rate": 4.530377814219128e-06,
2305
+ "loss": 0.8647,
2306
+ "step": 191500
2307
+ },
2308
+ {
2309
+ "epoch": 0.78,
2310
+ "learning_rate": 4.489987155770614e-06,
2311
+ "loss": 0.8796,
2312
+ "step": 192000
2313
+ },
2314
+ {
2315
+ "epoch": 0.78,
2316
+ "learning_rate": 4.4495964973220996e-06,
2317
+ "loss": 0.9004,
2318
+ "step": 192500
2319
+ },
2320
+ {
2321
+ "epoch": 0.78,
2322
+ "learning_rate": 4.409205838873586e-06,
2323
+ "loss": 0.8612,
2324
+ "step": 193000
2325
+ },
2326
+ {
2327
+ "epoch": 0.78,
2328
+ "learning_rate": 4.368815180425072e-06,
2329
+ "loss": 0.8612,
2330
+ "step": 193500
2331
+ },
2332
+ {
2333
+ "epoch": 0.78,
2334
+ "learning_rate": 4.328424521976557e-06,
2335
+ "loss": 0.9331,
2336
+ "step": 194000
2337
+ },
2338
+ {
2339
+ "epoch": 0.79,
2340
+ "learning_rate": 4.288033863528044e-06,
2341
+ "loss": 0.8905,
2342
+ "step": 194500
2343
+ },
2344
+ {
2345
+ "epoch": 0.79,
2346
+ "learning_rate": 4.247643205079529e-06,
2347
+ "loss": 0.9151,
2348
+ "step": 195000
2349
+ },
2350
+ {
2351
+ "epoch": 0.79,
2352
+ "learning_rate": 4.207252546631016e-06,
2353
+ "loss": 0.8583,
2354
+ "step": 195500
2355
+ },
2356
+ {
2357
+ "epoch": 0.79,
2358
+ "learning_rate": 4.166861888182501e-06,
2359
+ "loss": 0.9202,
2360
+ "step": 196000
2361
+ },
2362
+ {
2363
+ "epoch": 0.79,
2364
+ "learning_rate": 4.126471229733988e-06,
2365
+ "loss": 0.8658,
2366
+ "step": 196500
2367
+ },
2368
+ {
2369
+ "epoch": 0.8,
2370
+ "learning_rate": 4.0860805712854735e-06,
2371
+ "loss": 0.9142,
2372
+ "step": 197000
2373
+ },
2374
+ {
2375
+ "epoch": 0.8,
2376
+ "learning_rate": 4.045689912836959e-06,
2377
+ "loss": 0.8564,
2378
+ "step": 197500
2379
+ },
2380
+ {
2381
+ "epoch": 0.8,
2382
+ "learning_rate": 4.0052992543884456e-06,
2383
+ "loss": 0.8799,
2384
+ "step": 198000
2385
+ },
2386
+ {
2387
+ "epoch": 0.8,
2388
+ "learning_rate": 3.964908595939931e-06,
2389
+ "loss": 0.8971,
2390
+ "step": 198500
2391
+ },
2392
+ {
2393
+ "epoch": 0.8,
2394
+ "learning_rate": 3.924517937491417e-06,
2395
+ "loss": 0.8952,
2396
+ "step": 199000
2397
+ },
2398
+ {
2399
+ "epoch": 0.81,
2400
+ "learning_rate": 3.884127279042903e-06,
2401
+ "loss": 0.8958,
2402
+ "step": 199500
2403
+ },
2404
+ {
2405
+ "epoch": 0.81,
2406
+ "learning_rate": 3.84373662059439e-06,
2407
+ "loss": 0.8655,
2408
+ "step": 200000
2409
+ },
2410
+ {
2411
+ "epoch": 0.81,
2412
+ "learning_rate": 3.8033459621458753e-06,
2413
+ "loss": 0.9043,
2414
+ "step": 200500
2415
+ },
2416
+ {
2417
+ "epoch": 0.81,
2418
+ "learning_rate": 3.762955303697361e-06,
2419
+ "loss": 0.8537,
2420
+ "step": 201000
2421
+ },
2422
+ {
2423
+ "epoch": 0.81,
2424
+ "learning_rate": 3.7225646452488474e-06,
2425
+ "loss": 0.8716,
2426
+ "step": 201500
2427
+ },
2428
+ {
2429
+ "epoch": 0.82,
2430
+ "learning_rate": 3.682173986800333e-06,
2431
+ "loss": 0.9348,
2432
+ "step": 202000
2433
+ },
2434
+ {
2435
+ "epoch": 0.82,
2436
+ "learning_rate": 3.641783328351819e-06,
2437
+ "loss": 0.9048,
2438
+ "step": 202500
2439
+ },
2440
+ {
2441
+ "epoch": 0.82,
2442
+ "learning_rate": 3.601392669903305e-06,
2443
+ "loss": 0.8867,
2444
+ "step": 203000
2445
+ },
2446
+ {
2447
+ "epoch": 0.82,
2448
+ "learning_rate": 3.561002011454791e-06,
2449
+ "loss": 0.88,
2450
+ "step": 203500
2451
+ },
2452
+ {
2453
+ "epoch": 0.82,
2454
+ "learning_rate": 3.5206113530062767e-06,
2455
+ "loss": 0.8901,
2456
+ "step": 204000
2457
+ },
2458
+ {
2459
+ "epoch": 0.83,
2460
+ "learning_rate": 3.480220694557763e-06,
2461
+ "loss": 0.8875,
2462
+ "step": 204500
2463
+ },
2464
+ {
2465
+ "epoch": 0.83,
2466
+ "learning_rate": 3.4398300361092492e-06,
2467
+ "loss": 0.8641,
2468
+ "step": 205000
2469
+ },
2470
+ {
2471
+ "epoch": 0.83,
2472
+ "learning_rate": 3.399439377660735e-06,
2473
+ "loss": 0.8864,
2474
+ "step": 205500
2475
+ },
2476
+ {
2477
+ "epoch": 0.83,
2478
+ "learning_rate": 3.3590487192122205e-06,
2479
+ "loss": 0.8575,
2480
+ "step": 206000
2481
+ },
2482
+ {
2483
+ "epoch": 0.83,
2484
+ "learning_rate": 3.318658060763707e-06,
2485
+ "loss": 0.8565,
2486
+ "step": 206500
2487
+ },
2488
+ {
2489
+ "epoch": 0.84,
2490
+ "learning_rate": 3.278267402315193e-06,
2491
+ "loss": 0.8848,
2492
+ "step": 207000
2493
+ },
2494
+ {
2495
+ "epoch": 0.84,
2496
+ "learning_rate": 3.2378767438666786e-06,
2497
+ "loss": 0.879,
2498
+ "step": 207500
2499
+ },
2500
+ {
2501
+ "epoch": 0.84,
2502
+ "learning_rate": 3.197486085418165e-06,
2503
+ "loss": 0.8805,
2504
+ "step": 208000
2505
+ },
2506
+ {
2507
+ "epoch": 0.84,
2508
+ "learning_rate": 3.1570954269696507e-06,
2509
+ "loss": 0.8843,
2510
+ "step": 208500
2511
+ },
2512
+ {
2513
+ "epoch": 0.84,
2514
+ "learning_rate": 3.1167047685211367e-06,
2515
+ "loss": 0.8928,
2516
+ "step": 209000
2517
+ },
2518
+ {
2519
+ "epoch": 0.85,
2520
+ "learning_rate": 3.0763141100726227e-06,
2521
+ "loss": 0.8742,
2522
+ "step": 209500
2523
+ },
2524
+ {
2525
+ "epoch": 0.85,
2526
+ "learning_rate": 3.0359234516241088e-06,
2527
+ "loss": 0.8339,
2528
+ "step": 210000
2529
+ },
2530
+ {
2531
+ "epoch": 0.85,
2532
+ "learning_rate": 2.9955327931755944e-06,
2533
+ "loss": 0.8919,
2534
+ "step": 210500
2535
+ },
2536
+ {
2537
+ "epoch": 0.85,
2538
+ "learning_rate": 2.9551421347270804e-06,
2539
+ "loss": 0.8697,
2540
+ "step": 211000
2541
+ },
2542
+ {
2543
+ "epoch": 0.85,
2544
+ "learning_rate": 2.914751476278567e-06,
2545
+ "loss": 0.8697,
2546
+ "step": 211500
2547
+ },
2548
+ {
2549
+ "epoch": 0.86,
2550
+ "learning_rate": 2.8743608178300525e-06,
2551
+ "loss": 0.8874,
2552
+ "step": 212000
2553
+ },
2554
+ {
2555
+ "epoch": 0.86,
2556
+ "learning_rate": 2.833970159381538e-06,
2557
+ "loss": 0.8891,
2558
+ "step": 212500
2559
+ },
2560
+ {
2561
+ "epoch": 0.86,
2562
+ "learning_rate": 2.7935795009330246e-06,
2563
+ "loss": 0.9008,
2564
+ "step": 213000
2565
+ },
2566
+ {
2567
+ "epoch": 0.86,
2568
+ "learning_rate": 2.7531888424845106e-06,
2569
+ "loss": 0.9161,
2570
+ "step": 213500
2571
+ },
2572
+ {
2573
+ "epoch": 0.86,
2574
+ "learning_rate": 2.7127981840359962e-06,
2575
+ "loss": 0.9063,
2576
+ "step": 214000
2577
+ },
2578
+ {
2579
+ "epoch": 0.87,
2580
+ "learning_rate": 2.6724075255874827e-06,
2581
+ "loss": 0.8291,
2582
+ "step": 214500
2583
+ },
2584
+ {
2585
+ "epoch": 0.87,
2586
+ "learning_rate": 2.6320168671389683e-06,
2587
+ "loss": 0.8717,
2588
+ "step": 215000
2589
+ },
2590
+ {
2591
+ "epoch": 0.87,
2592
+ "learning_rate": 2.5916262086904543e-06,
2593
+ "loss": 0.8955,
2594
+ "step": 215500
2595
+ },
2596
+ {
2597
+ "epoch": 0.87,
2598
+ "learning_rate": 2.55123555024194e-06,
2599
+ "loss": 0.8767,
2600
+ "step": 216000
2601
+ },
2602
+ {
2603
+ "epoch": 0.87,
2604
+ "learning_rate": 2.5108448917934264e-06,
2605
+ "loss": 0.8314,
2606
+ "step": 216500
2607
+ },
2608
+ {
2609
+ "epoch": 0.88,
2610
+ "learning_rate": 2.470454233344912e-06,
2611
+ "loss": 0.8677,
2612
+ "step": 217000
2613
+ },
2614
+ {
2615
+ "epoch": 0.88,
2616
+ "learning_rate": 2.430063574896398e-06,
2617
+ "loss": 0.8947,
2618
+ "step": 217500
2619
+ },
2620
+ {
2621
+ "epoch": 0.88,
2622
+ "learning_rate": 2.389672916447884e-06,
2623
+ "loss": 0.8415,
2624
+ "step": 218000
2625
+ },
2626
+ {
2627
+ "epoch": 0.88,
2628
+ "learning_rate": 2.34928225799937e-06,
2629
+ "loss": 0.9029,
2630
+ "step": 218500
2631
+ },
2632
+ {
2633
+ "epoch": 0.88,
2634
+ "learning_rate": 2.3088915995508558e-06,
2635
+ "loss": 0.8588,
2636
+ "step": 219000
2637
+ },
2638
+ {
2639
+ "epoch": 0.89,
2640
+ "learning_rate": 2.2685009411023422e-06,
2641
+ "loss": 0.8956,
2642
+ "step": 219500
2643
+ },
2644
+ {
2645
+ "epoch": 0.89,
2646
+ "learning_rate": 2.2281102826538283e-06,
2647
+ "loss": 0.8821,
2648
+ "step": 220000
2649
+ },
2650
+ {
2651
+ "epoch": 0.89,
2652
+ "learning_rate": 2.187719624205314e-06,
2653
+ "loss": 0.8565,
2654
+ "step": 220500
2655
+ },
2656
+ {
2657
+ "epoch": 0.89,
2658
+ "learning_rate": 2.1473289657568e-06,
2659
+ "loss": 0.8276,
2660
+ "step": 221000
2661
+ },
2662
+ {
2663
+ "epoch": 0.89,
2664
+ "learning_rate": 2.106938307308286e-06,
2665
+ "loss": 0.8415,
2666
+ "step": 221500
2667
+ },
2668
+ {
2669
+ "epoch": 0.9,
2670
+ "learning_rate": 2.066547648859772e-06,
2671
+ "loss": 0.8428,
2672
+ "step": 222000
2673
+ },
2674
+ {
2675
+ "epoch": 0.9,
2676
+ "learning_rate": 2.026156990411258e-06,
2677
+ "loss": 0.8904,
2678
+ "step": 222500
2679
+ },
2680
+ {
2681
+ "epoch": 0.9,
2682
+ "learning_rate": 1.9857663319627436e-06,
2683
+ "loss": 0.8784,
2684
+ "step": 223000
2685
+ },
2686
+ {
2687
+ "epoch": 0.9,
2688
+ "learning_rate": 1.9453756735142297e-06,
2689
+ "loss": 0.8684,
2690
+ "step": 223500
2691
+ },
2692
+ {
2693
+ "epoch": 0.9,
2694
+ "learning_rate": 1.9049850150657157e-06,
2695
+ "loss": 0.9086,
2696
+ "step": 224000
2697
+ },
2698
+ {
2699
+ "epoch": 0.91,
2700
+ "learning_rate": 1.8645943566172017e-06,
2701
+ "loss": 0.8657,
2702
+ "step": 224500
2703
+ },
2704
+ {
2705
+ "epoch": 0.91,
2706
+ "learning_rate": 1.8242036981686878e-06,
2707
+ "loss": 0.8467,
2708
+ "step": 225000
2709
+ },
2710
+ {
2711
+ "epoch": 0.91,
2712
+ "learning_rate": 1.7838130397201736e-06,
2713
+ "loss": 0.8263,
2714
+ "step": 225500
2715
+ },
2716
+ {
2717
+ "epoch": 0.91,
2718
+ "learning_rate": 1.7434223812716597e-06,
2719
+ "loss": 0.8868,
2720
+ "step": 226000
2721
+ },
2722
+ {
2723
+ "epoch": 0.91,
2724
+ "learning_rate": 1.7030317228231455e-06,
2725
+ "loss": 0.8744,
2726
+ "step": 226500
2727
+ },
2728
+ {
2729
+ "epoch": 0.92,
2730
+ "learning_rate": 1.6626410643746315e-06,
2731
+ "loss": 0.8546,
2732
+ "step": 227000
2733
+ },
2734
+ {
2735
+ "epoch": 0.92,
2736
+ "learning_rate": 1.6222504059261176e-06,
2737
+ "loss": 0.8566,
2738
+ "step": 227500
2739
+ },
2740
+ {
2741
+ "epoch": 0.92,
2742
+ "learning_rate": 1.5818597474776034e-06,
2743
+ "loss": 0.9289,
2744
+ "step": 228000
2745
+ },
2746
+ {
2747
+ "epoch": 0.92,
2748
+ "learning_rate": 1.5414690890290896e-06,
2749
+ "loss": 0.8764,
2750
+ "step": 228500
2751
+ },
2752
+ {
2753
+ "epoch": 0.92,
2754
+ "learning_rate": 1.5010784305805752e-06,
2755
+ "loss": 0.8833,
2756
+ "step": 229000
2757
+ },
2758
+ {
2759
+ "epoch": 0.93,
2760
+ "learning_rate": 1.4606877721320615e-06,
2761
+ "loss": 0.8887,
2762
+ "step": 229500
2763
+ },
2764
+ {
2765
+ "epoch": 0.93,
2766
+ "learning_rate": 1.4202971136835475e-06,
2767
+ "loss": 0.8791,
2768
+ "step": 230000
2769
+ },
2770
+ {
2771
+ "epoch": 0.93,
2772
+ "learning_rate": 1.3799064552350334e-06,
2773
+ "loss": 0.8815,
2774
+ "step": 230500
2775
+ },
2776
+ {
2777
+ "epoch": 0.93,
2778
+ "learning_rate": 1.3395157967865194e-06,
2779
+ "loss": 0.8936,
2780
+ "step": 231000
2781
+ },
2782
+ {
2783
+ "epoch": 0.94,
2784
+ "learning_rate": 1.2991251383380052e-06,
2785
+ "loss": 0.8494,
2786
+ "step": 231500
2787
+ },
2788
+ {
2789
+ "epoch": 0.94,
2790
+ "learning_rate": 1.2587344798894913e-06,
2791
+ "loss": 0.8743,
2792
+ "step": 232000
2793
+ },
2794
+ {
2795
+ "epoch": 0.94,
2796
+ "learning_rate": 1.2183438214409773e-06,
2797
+ "loss": 0.888,
2798
+ "step": 232500
2799
+ },
2800
+ {
2801
+ "epoch": 0.94,
2802
+ "learning_rate": 1.1779531629924631e-06,
2803
+ "loss": 0.8898,
2804
+ "step": 233000
2805
+ },
2806
+ {
2807
+ "epoch": 0.94,
2808
+ "learning_rate": 1.1375625045439492e-06,
2809
+ "loss": 0.8638,
2810
+ "step": 233500
2811
+ },
2812
+ {
2813
+ "epoch": 0.95,
2814
+ "learning_rate": 1.0971718460954352e-06,
2815
+ "loss": 0.8727,
2816
+ "step": 234000
2817
+ },
2818
+ {
2819
+ "epoch": 0.95,
2820
+ "learning_rate": 1.056781187646921e-06,
2821
+ "loss": 0.853,
2822
+ "step": 234500
2823
+ },
2824
+ {
2825
+ "epoch": 0.95,
2826
+ "learning_rate": 1.016390529198407e-06,
2827
+ "loss": 0.8674,
2828
+ "step": 235000
2829
+ },
2830
+ {
2831
+ "epoch": 0.95,
2832
+ "learning_rate": 9.75999870749893e-07,
2833
+ "loss": 0.8506,
2834
+ "step": 235500
2835
+ },
2836
+ {
2837
+ "epoch": 0.95,
2838
+ "learning_rate": 9.35609212301379e-07,
2839
+ "loss": 0.8516,
2840
+ "step": 236000
2841
+ },
2842
+ {
2843
+ "epoch": 0.96,
2844
+ "learning_rate": 8.952185538528651e-07,
2845
+ "loss": 0.8636,
2846
+ "step": 236500
2847
+ },
2848
+ {
2849
+ "epoch": 0.96,
2850
+ "learning_rate": 8.54827895404351e-07,
2851
+ "loss": 0.8434,
2852
+ "step": 237000
2853
+ },
2854
+ {
2855
+ "epoch": 0.96,
2856
+ "learning_rate": 8.144372369558369e-07,
2857
+ "loss": 0.884,
2858
+ "step": 237500
2859
+ },
2860
+ {
2861
+ "epoch": 0.96,
2862
+ "learning_rate": 7.740465785073229e-07,
2863
+ "loss": 0.89,
2864
+ "step": 238000
2865
+ },
2866
+ {
2867
+ "epoch": 0.96,
2868
+ "learning_rate": 7.336559200588089e-07,
2869
+ "loss": 0.8915,
2870
+ "step": 238500
2871
+ },
2872
+ {
2873
+ "epoch": 0.97,
2874
+ "learning_rate": 6.932652616102948e-07,
2875
+ "loss": 0.8749,
2876
+ "step": 239000
2877
+ },
2878
+ {
2879
+ "epoch": 0.97,
2880
+ "learning_rate": 6.528746031617808e-07,
2881
+ "loss": 0.8963,
2882
+ "step": 239500
2883
+ },
2884
+ {
2885
+ "epoch": 0.97,
2886
+ "learning_rate": 6.124839447132667e-07,
2887
+ "loss": 0.8905,
2888
+ "step": 240000
2889
+ },
2890
+ {
2891
+ "epoch": 0.97,
2892
+ "learning_rate": 5.720932862647527e-07,
2893
+ "loss": 0.8566,
2894
+ "step": 240500
2895
+ },
2896
+ {
2897
+ "epoch": 0.97,
2898
+ "learning_rate": 5.317026278162387e-07,
2899
+ "loss": 0.8718,
2900
+ "step": 241000
2901
+ },
2902
+ {
2903
+ "epoch": 0.98,
2904
+ "learning_rate": 4.913119693677246e-07,
2905
+ "loss": 0.8383,
2906
+ "step": 241500
2907
+ },
2908
+ {
2909
+ "epoch": 0.98,
2910
+ "learning_rate": 4.5092131091921064e-07,
2911
+ "loss": 0.8678,
2912
+ "step": 242000
2913
+ },
2914
+ {
2915
+ "epoch": 0.98,
2916
+ "learning_rate": 4.1053065247069657e-07,
2917
+ "loss": 0.811,
2918
+ "step": 242500
2919
+ },
2920
+ {
2921
+ "epoch": 0.98,
2922
+ "learning_rate": 3.701399940221826e-07,
2923
+ "loss": 0.8551,
2924
+ "step": 243000
2925
+ },
2926
+ {
2927
+ "epoch": 0.98,
2928
+ "learning_rate": 3.2974933557366854e-07,
2929
+ "loss": 0.9153,
2930
+ "step": 243500
2931
+ },
2932
+ {
2933
+ "epoch": 0.99,
2934
+ "learning_rate": 2.8935867712515453e-07,
2935
+ "loss": 0.8721,
2936
+ "step": 244000
2937
+ },
2938
+ {
2939
+ "epoch": 0.99,
2940
+ "learning_rate": 2.489680186766405e-07,
2941
+ "loss": 0.8805,
2942
+ "step": 244500
2943
+ },
2944
+ {
2945
+ "epoch": 0.99,
2946
+ "learning_rate": 2.0857736022812644e-07,
2947
+ "loss": 0.8698,
2948
+ "step": 245000
2949
+ },
2950
+ {
2951
+ "epoch": 0.99,
2952
+ "learning_rate": 1.6818670177961243e-07,
2953
+ "loss": 0.8793,
2954
+ "step": 245500
2955
+ },
2956
+ {
2957
+ "epoch": 0.99,
2958
+ "learning_rate": 1.277960433310984e-07,
2959
+ "loss": 0.8455,
2960
+ "step": 246000
2961
+ },
2962
+ {
2963
+ "epoch": 1.0,
2964
+ "learning_rate": 8.740538488258437e-08,
2965
+ "loss": 0.884,
2966
+ "step": 246500
2967
+ },
2968
+ {
2969
+ "epoch": 1.0,
2970
+ "learning_rate": 4.7014726434070325e-08,
2971
+ "loss": 0.8582,
2972
+ "step": 247000
2973
+ },
2974
+ {
2975
+ "epoch": 1.0,
2976
+ "learning_rate": 6.624067985556301e-09,
2977
+ "loss": 0.8516,
2978
+ "step": 247500
2979
+ }
2980
+ ],
2981
+ "max_steps": 247582,
2982
+ "num_train_epochs": 1,
2983
+ "total_flos": 9502800674291712.0,
2984
+ "trial_name": null,
2985
+ "trial_params": null
2986
+ }