Upload trainer_state.json
Browse files- trainer_state.json +1405 -0
trainer_state.json
ADDED
@@ -0,0 +1,1405 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.9962100703844072,
|
5 |
+
"global_step": 460,
|
6 |
+
"is_hyper_param_search": false,
|
7 |
+
"is_local_process_zero": true,
|
8 |
+
"is_world_process_zero": true,
|
9 |
+
"log_history": [
|
10 |
+
{
|
11 |
+
"epoch": 0.01,
|
12 |
+
"learning_rate": 0.00014285714285714284,
|
13 |
+
"loss": 1.8415,
|
14 |
+
"step": 2
|
15 |
+
},
|
16 |
+
{
|
17 |
+
"epoch": 0.02,
|
18 |
+
"learning_rate": 0.0002857142857142857,
|
19 |
+
"loss": 1.9512,
|
20 |
+
"step": 4
|
21 |
+
},
|
22 |
+
{
|
23 |
+
"epoch": 0.03,
|
24 |
+
"learning_rate": 0.00042857142857142855,
|
25 |
+
"loss": 2.0021,
|
26 |
+
"step": 6
|
27 |
+
},
|
28 |
+
{
|
29 |
+
"epoch": 0.03,
|
30 |
+
"learning_rate": 0.0005714285714285714,
|
31 |
+
"loss": 2.0064,
|
32 |
+
"step": 8
|
33 |
+
},
|
34 |
+
{
|
35 |
+
"epoch": 0.04,
|
36 |
+
"learning_rate": 0.0007142857142857143,
|
37 |
+
"loss": 1.9143,
|
38 |
+
"step": 10
|
39 |
+
},
|
40 |
+
{
|
41 |
+
"epoch": 0.05,
|
42 |
+
"learning_rate": 0.0008571428571428571,
|
43 |
+
"loss": 1.8679,
|
44 |
+
"step": 12
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.06,
|
48 |
+
"learning_rate": 0.001,
|
49 |
+
"loss": 1.9015,
|
50 |
+
"step": 14
|
51 |
+
},
|
52 |
+
{
|
53 |
+
"epoch": 0.07,
|
54 |
+
"learning_rate": 0.0009999503838747563,
|
55 |
+
"loss": 1.9679,
|
56 |
+
"step": 16
|
57 |
+
},
|
58 |
+
{
|
59 |
+
"epoch": 0.08,
|
60 |
+
"learning_rate": 0.000999801545346065,
|
61 |
+
"loss": 1.96,
|
62 |
+
"step": 18
|
63 |
+
},
|
64 |
+
{
|
65 |
+
"epoch": 0.09,
|
66 |
+
"learning_rate": 0.0009995535139530904,
|
67 |
+
"loss": 2.019,
|
68 |
+
"step": 20
|
69 |
+
},
|
70 |
+
{
|
71 |
+
"epoch": 0.1,
|
72 |
+
"learning_rate": 0.000999206338921259,
|
73 |
+
"loss": 1.9169,
|
74 |
+
"step": 22
|
75 |
+
},
|
76 |
+
{
|
77 |
+
"epoch": 0.1,
|
78 |
+
"learning_rate": 0.00099876008915249,
|
79 |
+
"loss": 2.002,
|
80 |
+
"step": 24
|
81 |
+
},
|
82 |
+
{
|
83 |
+
"epoch": 0.11,
|
84 |
+
"learning_rate": 0.0009982148532115217,
|
85 |
+
"loss": 1.966,
|
86 |
+
"step": 26
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.12,
|
90 |
+
"learning_rate": 0.0009975707393083326,
|
91 |
+
"loss": 1.9922,
|
92 |
+
"step": 28
|
93 |
+
},
|
94 |
+
{
|
95 |
+
"epoch": 0.13,
|
96 |
+
"learning_rate": 0.0009968278752766672,
|
97 |
+
"loss": 1.8925,
|
98 |
+
"step": 30
|
99 |
+
},
|
100 |
+
{
|
101 |
+
"epoch": 0.14,
|
102 |
+
"learning_rate": 0.0009959864085486647,
|
103 |
+
"loss": 1.9948,
|
104 |
+
"step": 32
|
105 |
+
},
|
106 |
+
{
|
107 |
+
"epoch": 0.15,
|
108 |
+
"learning_rate": 0.0009950465061255996,
|
109 |
+
"loss": 2.0015,
|
110 |
+
"step": 34
|
111 |
+
},
|
112 |
+
{
|
113 |
+
"epoch": 0.16,
|
114 |
+
"learning_rate": 0.0009940083545447369,
|
115 |
+
"loss": 1.9682,
|
116 |
+
"step": 36
|
117 |
+
},
|
118 |
+
{
|
119 |
+
"epoch": 0.16,
|
120 |
+
"learning_rate": 0.0009928721598423125,
|
121 |
+
"loss": 1.9633,
|
122 |
+
"step": 38
|
123 |
+
},
|
124 |
+
{
|
125 |
+
"epoch": 0.17,
|
126 |
+
"learning_rate": 0.0009916381475126405,
|
127 |
+
"loss": 2.0015,
|
128 |
+
"step": 40
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.18,
|
132 |
+
"learning_rate": 0.0009903065624633626,
|
133 |
+
"loss": 1.8838,
|
134 |
+
"step": 42
|
135 |
+
},
|
136 |
+
{
|
137 |
+
"epoch": 0.19,
|
138 |
+
"learning_rate": 0.000988877668966841,
|
139 |
+
"loss": 1.966,
|
140 |
+
"step": 44
|
141 |
+
},
|
142 |
+
{
|
143 |
+
"epoch": 0.2,
|
144 |
+
"learning_rate": 0.0009873517506077101,
|
145 |
+
"loss": 1.8983,
|
146 |
+
"step": 46
|
147 |
+
},
|
148 |
+
{
|
149 |
+
"epoch": 0.21,
|
150 |
+
"learning_rate": 0.000985729110226596,
|
151 |
+
"loss": 1.9624,
|
152 |
+
"step": 48
|
153 |
+
},
|
154 |
+
{
|
155 |
+
"epoch": 0.22,
|
156 |
+
"learning_rate": 0.0009840100698600119,
|
157 |
+
"loss": 1.9347,
|
158 |
+
"step": 50
|
159 |
+
},
|
160 |
+
{
|
161 |
+
"epoch": 0.23,
|
162 |
+
"learning_rate": 0.0009821949706764462,
|
163 |
+
"loss": 1.9835,
|
164 |
+
"step": 52
|
165 |
+
},
|
166 |
+
{
|
167 |
+
"epoch": 0.23,
|
168 |
+
"learning_rate": 0.000980284172908653,
|
169 |
+
"loss": 2.0104,
|
170 |
+
"step": 54
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 0.24,
|
174 |
+
"learning_rate": 0.0009782780557821575,
|
175 |
+
"loss": 1.885,
|
176 |
+
"step": 56
|
177 |
+
},
|
178 |
+
{
|
179 |
+
"epoch": 0.25,
|
180 |
+
"learning_rate": 0.0009761770174399942,
|
181 |
+
"loss": 1.964,
|
182 |
+
"step": 58
|
183 |
+
},
|
184 |
+
{
|
185 |
+
"epoch": 0.26,
|
186 |
+
"learning_rate": 0.000973981474863689,
|
187 |
+
"loss": 1.932,
|
188 |
+
"step": 60
|
189 |
+
},
|
190 |
+
{
|
191 |
+
"epoch": 0.27,
|
192 |
+
"learning_rate": 0.0009716918637905041,
|
193 |
+
"loss": 1.9233,
|
194 |
+
"step": 62
|
195 |
+
},
|
196 |
+
{
|
197 |
+
"epoch": 0.28,
|
198 |
+
"learning_rate": 0.0009693086386269581,
|
199 |
+
"loss": 1.8307,
|
200 |
+
"step": 64
|
201 |
+
},
|
202 |
+
{
|
203 |
+
"epoch": 0.29,
|
204 |
+
"learning_rate": 0.000966832272358644,
|
205 |
+
"loss": 1.9262,
|
206 |
+
"step": 66
|
207 |
+
},
|
208 |
+
{
|
209 |
+
"epoch": 0.29,
|
210 |
+
"learning_rate": 0.0009642632564563575,
|
211 |
+
"loss": 1.9522,
|
212 |
+
"step": 68
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 0.3,
|
216 |
+
"learning_rate": 0.0009616021007785576,
|
217 |
+
"loss": 1.9949,
|
218 |
+
"step": 70
|
219 |
+
},
|
220 |
+
{
|
221 |
+
"epoch": 0.31,
|
222 |
+
"learning_rate": 0.0009588493334701778,
|
223 |
+
"loss": 1.9875,
|
224 |
+
"step": 72
|
225 |
+
},
|
226 |
+
{
|
227 |
+
"epoch": 0.32,
|
228 |
+
"learning_rate": 0.0009560055008578085,
|
229 |
+
"loss": 2.0109,
|
230 |
+
"step": 74
|
231 |
+
},
|
232 |
+
{
|
233 |
+
"epoch": 0.33,
|
234 |
+
"learning_rate": 0.0009530711673412697,
|
235 |
+
"loss": 1.9399,
|
236 |
+
"step": 76
|
237 |
+
},
|
238 |
+
{
|
239 |
+
"epoch": 0.34,
|
240 |
+
"learning_rate": 0.0009500469152815987,
|
241 |
+
"loss": 1.9279,
|
242 |
+
"step": 78
|
243 |
+
},
|
244 |
+
{
|
245 |
+
"epoch": 0.35,
|
246 |
+
"learning_rate": 0.0009469333448854713,
|
247 |
+
"loss": 1.9739,
|
248 |
+
"step": 80
|
249 |
+
},
|
250 |
+
{
|
251 |
+
"epoch": 0.36,
|
252 |
+
"learning_rate": 0.0009437310740860822,
|
253 |
+
"loss": 1.9563,
|
254 |
+
"step": 82
|
255 |
+
},
|
256 |
+
{
|
257 |
+
"epoch": 0.36,
|
258 |
+
"learning_rate": 0.0009404407384205079,
|
259 |
+
"loss": 1.9654,
|
260 |
+
"step": 84
|
261 |
+
},
|
262 |
+
{
|
263 |
+
"epoch": 0.37,
|
264 |
+
"learning_rate": 0.0009370629909035739,
|
265 |
+
"loss": 2.0422,
|
266 |
+
"step": 86
|
267 |
+
},
|
268 |
+
{
|
269 |
+
"epoch": 0.38,
|
270 |
+
"learning_rate": 0.0009335985018982559,
|
271 |
+
"loss": 1.9137,
|
272 |
+
"step": 88
|
273 |
+
},
|
274 |
+
{
|
275 |
+
"epoch": 0.39,
|
276 |
+
"learning_rate": 0.0009300479589826354,
|
277 |
+
"loss": 1.9886,
|
278 |
+
"step": 90
|
279 |
+
},
|
280 |
+
{
|
281 |
+
"epoch": 0.4,
|
282 |
+
"learning_rate": 0.0009264120668134404,
|
283 |
+
"loss": 1.9525,
|
284 |
+
"step": 92
|
285 |
+
},
|
286 |
+
{
|
287 |
+
"epoch": 0.41,
|
288 |
+
"learning_rate": 0.0009226915469861956,
|
289 |
+
"loss": 1.9942,
|
290 |
+
"step": 94
|
291 |
+
},
|
292 |
+
{
|
293 |
+
"epoch": 0.42,
|
294 |
+
"learning_rate": 0.0009188871378920122,
|
295 |
+
"loss": 1.9555,
|
296 |
+
"step": 96
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 0.42,
|
300 |
+
"learning_rate": 0.0009149995945710424,
|
301 |
+
"loss": 1.9369,
|
302 |
+
"step": 98
|
303 |
+
},
|
304 |
+
{
|
305 |
+
"epoch": 0.43,
|
306 |
+
"learning_rate": 0.0009110296885626314,
|
307 |
+
"loss": 1.9038,
|
308 |
+
"step": 100
|
309 |
+
},
|
310 |
+
{
|
311 |
+
"epoch": 0.44,
|
312 |
+
"learning_rate": 0.0009069782077521942,
|
313 |
+
"loss": 1.9615,
|
314 |
+
"step": 102
|
315 |
+
},
|
316 |
+
{
|
317 |
+
"epoch": 0.45,
|
318 |
+
"learning_rate": 0.0009028459562148479,
|
319 |
+
"loss": 2.0167,
|
320 |
+
"step": 104
|
321 |
+
},
|
322 |
+
{
|
323 |
+
"epoch": 0.46,
|
324 |
+
"learning_rate": 0.0008986337540558318,
|
325 |
+
"loss": 1.8892,
|
326 |
+
"step": 106
|
327 |
+
},
|
328 |
+
{
|
329 |
+
"epoch": 0.47,
|
330 |
+
"learning_rate": 0.0008943424372477454,
|
331 |
+
"loss": 1.9468,
|
332 |
+
"step": 108
|
333 |
+
},
|
334 |
+
{
|
335 |
+
"epoch": 0.48,
|
336 |
+
"learning_rate": 0.0008899728574646376,
|
337 |
+
"loss": 1.9827,
|
338 |
+
"step": 110
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 0.49,
|
342 |
+
"learning_rate": 0.0008855258819129796,
|
343 |
+
"loss": 1.8949,
|
344 |
+
"step": 112
|
345 |
+
},
|
346 |
+
{
|
347 |
+
"epoch": 0.49,
|
348 |
+
"learning_rate": 0.0008810023931595549,
|
349 |
+
"loss": 1.9341,
|
350 |
+
"step": 114
|
351 |
+
},
|
352 |
+
{
|
353 |
+
"epoch": 0.5,
|
354 |
+
"learning_rate": 0.0008764032889563016,
|
355 |
+
"loss": 1.9982,
|
356 |
+
"step": 116
|
357 |
+
},
|
358 |
+
{
|
359 |
+
"epoch": 0.51,
|
360 |
+
"learning_rate": 0.0008717294820621407,
|
361 |
+
"loss": 1.982,
|
362 |
+
"step": 118
|
363 |
+
},
|
364 |
+
{
|
365 |
+
"epoch": 0.52,
|
366 |
+
"learning_rate": 0.0008669819000618246,
|
367 |
+
"loss": 1.8524,
|
368 |
+
"step": 120
|
369 |
+
},
|
370 |
+
{
|
371 |
+
"epoch": 0.53,
|
372 |
+
"learning_rate": 0.0008621614851818462,
|
373 |
+
"loss": 1.891,
|
374 |
+
"step": 122
|
375 |
+
},
|
376 |
+
{
|
377 |
+
"epoch": 0.54,
|
378 |
+
"learning_rate": 0.000857269194103439,
|
379 |
+
"loss": 1.9186,
|
380 |
+
"step": 124
|
381 |
+
},
|
382 |
+
{
|
383 |
+
"epoch": 0.55,
|
384 |
+
"learning_rate": 0.0008523059977727103,
|
385 |
+
"loss": 1.8855,
|
386 |
+
"step": 126
|
387 |
+
},
|
388 |
+
{
|
389 |
+
"epoch": 0.55,
|
390 |
+
"learning_rate": 0.0008472728812079436,
|
391 |
+
"loss": 1.9325,
|
392 |
+
"step": 128
|
393 |
+
},
|
394 |
+
{
|
395 |
+
"epoch": 0.56,
|
396 |
+
"learning_rate": 0.0008421708433041057,
|
397 |
+
"loss": 1.9798,
|
398 |
+
"step": 130
|
399 |
+
},
|
400 |
+
{
|
401 |
+
"epoch": 0.57,
|
402 |
+
"learning_rate": 0.0008370008966346037,
|
403 |
+
"loss": 1.9994,
|
404 |
+
"step": 132
|
405 |
+
},
|
406 |
+
{
|
407 |
+
"epoch": 0.58,
|
408 |
+
"learning_rate": 0.0008317640672503231,
|
409 |
+
"loss": 1.9706,
|
410 |
+
"step": 134
|
411 |
+
},
|
412 |
+
{
|
413 |
+
"epoch": 0.59,
|
414 |
+
"learning_rate": 0.0008264613944759942,
|
415 |
+
"loss": 1.9126,
|
416 |
+
"step": 136
|
417 |
+
},
|
418 |
+
{
|
419 |
+
"epoch": 0.6,
|
420 |
+
"learning_rate": 0.0008210939307039234,
|
421 |
+
"loss": 1.921,
|
422 |
+
"step": 138
|
423 |
+
},
|
424 |
+
{
|
425 |
+
"epoch": 0.61,
|
426 |
+
"learning_rate": 0.0008156627411851295,
|
427 |
+
"loss": 1.8735,
|
428 |
+
"step": 140
|
429 |
+
},
|
430 |
+
{
|
431 |
+
"epoch": 0.62,
|
432 |
+
"learning_rate": 0.00081016890381793,
|
433 |
+
"loss": 1.8251,
|
434 |
+
"step": 142
|
435 |
+
},
|
436 |
+
{
|
437 |
+
"epoch": 0.62,
|
438 |
+
"learning_rate": 0.0008046135089340164,
|
439 |
+
"loss": 1.9081,
|
440 |
+
"step": 144
|
441 |
+
},
|
442 |
+
{
|
443 |
+
"epoch": 0.63,
|
444 |
+
"learning_rate": 0.0007989976590820622,
|
445 |
+
"loss": 1.9156,
|
446 |
+
"step": 146
|
447 |
+
},
|
448 |
+
{
|
449 |
+
"epoch": 0.64,
|
450 |
+
"learning_rate": 0.0007933224688089058,
|
451 |
+
"loss": 1.8971,
|
452 |
+
"step": 148
|
453 |
+
},
|
454 |
+
{
|
455 |
+
"epoch": 0.65,
|
456 |
+
"learning_rate": 0.0007875890644383525,
|
457 |
+
"loss": 1.968,
|
458 |
+
"step": 150
|
459 |
+
},
|
460 |
+
{
|
461 |
+
"epoch": 0.66,
|
462 |
+
"learning_rate": 0.0007817985838476398,
|
463 |
+
"loss": 1.9105,
|
464 |
+
"step": 152
|
465 |
+
},
|
466 |
+
{
|
467 |
+
"epoch": 0.67,
|
468 |
+
"learning_rate": 0.0007759521762416084,
|
469 |
+
"loss": 1.9104,
|
470 |
+
"step": 154
|
471 |
+
},
|
472 |
+
{
|
473 |
+
"epoch": 0.68,
|
474 |
+
"learning_rate": 0.0007700510019246265,
|
475 |
+
"loss": 1.9086,
|
476 |
+
"step": 156
|
477 |
+
},
|
478 |
+
{
|
479 |
+
"epoch": 0.68,
|
480 |
+
"learning_rate": 0.0007640962320703099,
|
481 |
+
"loss": 1.8637,
|
482 |
+
"step": 158
|
483 |
+
},
|
484 |
+
{
|
485 |
+
"epoch": 0.69,
|
486 |
+
"learning_rate": 0.0007580890484890863,
|
487 |
+
"loss": 2.0117,
|
488 |
+
"step": 160
|
489 |
+
},
|
490 |
+
{
|
491 |
+
"epoch": 0.7,
|
492 |
+
"learning_rate": 0.0007520306433936473,
|
493 |
+
"loss": 1.9229,
|
494 |
+
"step": 162
|
495 |
+
},
|
496 |
+
{
|
497 |
+
"epoch": 0.71,
|
498 |
+
"learning_rate": 0.0007459222191623368,
|
499 |
+
"loss": 1.9669,
|
500 |
+
"step": 164
|
501 |
+
},
|
502 |
+
{
|
503 |
+
"epoch": 0.72,
|
504 |
+
"learning_rate": 0.0007397649881005219,
|
505 |
+
"loss": 1.9291,
|
506 |
+
"step": 166
|
507 |
+
},
|
508 |
+
{
|
509 |
+
"epoch": 0.73,
|
510 |
+
"learning_rate": 0.0007335601721999923,
|
511 |
+
"loss": 1.9108,
|
512 |
+
"step": 168
|
513 |
+
},
|
514 |
+
{
|
515 |
+
"epoch": 0.74,
|
516 |
+
"learning_rate": 0.0007273090028964396,
|
517 |
+
"loss": 1.9027,
|
518 |
+
"step": 170
|
519 |
+
},
|
520 |
+
{
|
521 |
+
"epoch": 0.74,
|
522 |
+
"learning_rate": 0.0007210127208250598,
|
523 |
+
"loss": 1.8513,
|
524 |
+
"step": 172
|
525 |
+
},
|
526 |
+
{
|
527 |
+
"epoch": 0.75,
|
528 |
+
"learning_rate": 0.0007146725755743328,
|
529 |
+
"loss": 1.9157,
|
530 |
+
"step": 174
|
531 |
+
},
|
532 |
+
{
|
533 |
+
"epoch": 0.76,
|
534 |
+
"learning_rate": 0.0007082898254380215,
|
535 |
+
"loss": 1.9181,
|
536 |
+
"step": 176
|
537 |
+
},
|
538 |
+
{
|
539 |
+
"epoch": 0.77,
|
540 |
+
"learning_rate": 0.0007018657371654464,
|
541 |
+
"loss": 1.8859,
|
542 |
+
"step": 178
|
543 |
+
},
|
544 |
+
{
|
545 |
+
"epoch": 0.78,
|
546 |
+
"learning_rate": 0.000695401585710081,
|
547 |
+
"loss": 1.8927,
|
548 |
+
"step": 180
|
549 |
+
},
|
550 |
+
{
|
551 |
+
"epoch": 0.79,
|
552 |
+
"learning_rate": 0.0006888986539765181,
|
553 |
+
"loss": 1.908,
|
554 |
+
"step": 182
|
555 |
+
},
|
556 |
+
{
|
557 |
+
"epoch": 0.8,
|
558 |
+
"learning_rate": 0.0006823582325658588,
|
559 |
+
"loss": 2.0016,
|
560 |
+
"step": 184
|
561 |
+
},
|
562 |
+
{
|
563 |
+
"epoch": 0.81,
|
564 |
+
"learning_rate": 0.0006757816195195749,
|
565 |
+
"loss": 2.0238,
|
566 |
+
"step": 186
|
567 |
+
},
|
568 |
+
{
|
569 |
+
"epoch": 0.81,
|
570 |
+
"learning_rate": 0.0006691701200618925,
|
571 |
+
"loss": 1.9548,
|
572 |
+
"step": 188
|
573 |
+
},
|
574 |
+
{
|
575 |
+
"epoch": 0.82,
|
576 |
+
"learning_rate": 0.0006625250463407523,
|
577 |
+
"loss": 1.9494,
|
578 |
+
"step": 190
|
579 |
+
},
|
580 |
+
{
|
581 |
+
"epoch": 0.83,
|
582 |
+
"learning_rate": 0.0006558477171673941,
|
583 |
+
"loss": 1.8334,
|
584 |
+
"step": 192
|
585 |
+
},
|
586 |
+
{
|
587 |
+
"epoch": 0.84,
|
588 |
+
"learning_rate": 0.0006491394577546203,
|
589 |
+
"loss": 1.9507,
|
590 |
+
"step": 194
|
591 |
+
},
|
592 |
+
{
|
593 |
+
"epoch": 0.85,
|
594 |
+
"learning_rate": 0.0006424015994537876,
|
595 |
+
"loss": 1.9014,
|
596 |
+
"step": 196
|
597 |
+
},
|
598 |
+
{
|
599 |
+
"epoch": 0.86,
|
600 |
+
"learning_rate": 0.0006356354794905814,
|
601 |
+
"loss": 1.9175,
|
602 |
+
"step": 198
|
603 |
+
},
|
604 |
+
{
|
605 |
+
"epoch": 0.87,
|
606 |
+
"learning_rate": 0.0006288424406996237,
|
607 |
+
"loss": 1.9221,
|
608 |
+
"step": 200
|
609 |
+
},
|
610 |
+
{
|
611 |
+
"epoch": 0.87,
|
612 |
+
"learning_rate": 0.0006220238312579681,
|
613 |
+
"loss": 1.9813,
|
614 |
+
"step": 202
|
615 |
+
},
|
616 |
+
{
|
617 |
+
"epoch": 0.88,
|
618 |
+
"learning_rate": 0.000615181004417535,
|
619 |
+
"loss": 1.9541,
|
620 |
+
"step": 204
|
621 |
+
},
|
622 |
+
{
|
623 |
+
"epoch": 0.89,
|
624 |
+
"learning_rate": 0.0006083153182365383,
|
625 |
+
"loss": 1.8112,
|
626 |
+
"step": 206
|
627 |
+
},
|
628 |
+
{
|
629 |
+
"epoch": 0.9,
|
630 |
+
"learning_rate": 0.00060142813530996,
|
631 |
+
"loss": 1.902,
|
632 |
+
"step": 208
|
633 |
+
},
|
634 |
+
{
|
635 |
+
"epoch": 0.91,
|
636 |
+
"learning_rate": 0.0005945208224991225,
|
637 |
+
"loss": 1.9269,
|
638 |
+
"step": 210
|
639 |
+
},
|
640 |
+
{
|
641 |
+
"epoch": 0.92,
|
642 |
+
"learning_rate": 0.0005875947506604162,
|
643 |
+
"loss": 1.949,
|
644 |
+
"step": 212
|
645 |
+
},
|
646 |
+
{
|
647 |
+
"epoch": 0.93,
|
648 |
+
"learning_rate": 0.000580651294373232,
|
649 |
+
"loss": 1.9206,
|
650 |
+
"step": 214
|
651 |
+
},
|
652 |
+
{
|
653 |
+
"epoch": 0.94,
|
654 |
+
"learning_rate": 0.0005736918316671572,
|
655 |
+
"loss": 1.8971,
|
656 |
+
"step": 216
|
657 |
+
},
|
658 |
+
{
|
659 |
+
"epoch": 0.94,
|
660 |
+
"learning_rate": 0.0005667177437484845,
|
661 |
+
"loss": 1.9171,
|
662 |
+
"step": 218
|
663 |
+
},
|
664 |
+
{
|
665 |
+
"epoch": 0.95,
|
666 |
+
"learning_rate": 0.0005597304147260927,
|
667 |
+
"loss": 1.8511,
|
668 |
+
"step": 220
|
669 |
+
},
|
670 |
+
{
|
671 |
+
"epoch": 0.96,
|
672 |
+
"learning_rate": 0.0005527312313367492,
|
673 |
+
"loss": 1.8867,
|
674 |
+
"step": 222
|
675 |
+
},
|
676 |
+
{
|
677 |
+
"epoch": 0.97,
|
678 |
+
"learning_rate": 0.0005457215826698928,
|
679 |
+
"loss": 1.9239,
|
680 |
+
"step": 224
|
681 |
+
},
|
682 |
+
{
|
683 |
+
"epoch": 0.98,
|
684 |
+
"learning_rate": 0.0005387028598919481,
|
685 |
+
"loss": 1.8722,
|
686 |
+
"step": 226
|
687 |
+
},
|
688 |
+
{
|
689 |
+
"epoch": 0.99,
|
690 |
+
"learning_rate": 0.0005316764559702285,
|
691 |
+
"loss": 1.8539,
|
692 |
+
"step": 228
|
693 |
+
},
|
694 |
+
{
|
695 |
+
"epoch": 1.0,
|
696 |
+
"learning_rate": 0.0005246437653964822,
|
697 |
+
"loss": 1.9282,
|
698 |
+
"step": 230
|
699 |
+
},
|
700 |
+
{
|
701 |
+
"epoch": 1.01,
|
702 |
+
"learning_rate": 0.0005176061839101343,
|
703 |
+
"loss": 2.6089,
|
704 |
+
"step": 232
|
705 |
+
},
|
706 |
+
{
|
707 |
+
"epoch": 1.02,
|
708 |
+
"learning_rate": 0.0005105651082212828,
|
709 |
+
"loss": 1.8528,
|
710 |
+
"step": 234
|
711 |
+
},
|
712 |
+
{
|
713 |
+
"epoch": 1.03,
|
714 |
+
"learning_rate": 0.0005035219357335002,
|
715 |
+
"loss": 1.916,
|
716 |
+
"step": 236
|
717 |
+
},
|
718 |
+
{
|
719 |
+
"epoch": 1.03,
|
720 |
+
"learning_rate": 0.0004964780642664998,
|
721 |
+
"loss": 1.9557,
|
722 |
+
"step": 238
|
723 |
+
},
|
724 |
+
{
|
725 |
+
"epoch": 1.04,
|
726 |
+
"learning_rate": 0.0004894348917787173,
|
727 |
+
"loss": 1.8235,
|
728 |
+
"step": 240
|
729 |
+
},
|
730 |
+
{
|
731 |
+
"epoch": 1.05,
|
732 |
+
"learning_rate": 0.0004823938160898657,
|
733 |
+
"loss": 1.9001,
|
734 |
+
"step": 242
|
735 |
+
},
|
736 |
+
{
|
737 |
+
"epoch": 1.06,
|
738 |
+
"learning_rate": 0.0004753562346035178,
|
739 |
+
"loss": 1.8572,
|
740 |
+
"step": 244
|
741 |
+
},
|
742 |
+
{
|
743 |
+
"epoch": 1.07,
|
744 |
+
"learning_rate": 0.0004683235440297717,
|
745 |
+
"loss": 1.8058,
|
746 |
+
"step": 246
|
747 |
+
},
|
748 |
+
{
|
749 |
+
"epoch": 1.08,
|
750 |
+
"learning_rate": 0.00046129714010805206,
|
751 |
+
"loss": 1.8738,
|
752 |
+
"step": 248
|
753 |
+
},
|
754 |
+
{
|
755 |
+
"epoch": 1.09,
|
756 |
+
"learning_rate": 0.00045427841733010723,
|
757 |
+
"loss": 1.7835,
|
758 |
+
"step": 250
|
759 |
+
},
|
760 |
+
{
|
761 |
+
"epoch": 1.1,
|
762 |
+
"learning_rate": 0.00044726876866325085,
|
763 |
+
"loss": 1.8462,
|
764 |
+
"step": 252
|
765 |
+
},
|
766 |
+
{
|
767 |
+
"epoch": 1.1,
|
768 |
+
"learning_rate": 0.0004402695852739074,
|
769 |
+
"loss": 1.8028,
|
770 |
+
"step": 254
|
771 |
+
},
|
772 |
+
{
|
773 |
+
"epoch": 1.11,
|
774 |
+
"learning_rate": 0.0004332822562515155,
|
775 |
+
"loss": 1.8254,
|
776 |
+
"step": 256
|
777 |
+
},
|
778 |
+
{
|
779 |
+
"epoch": 1.12,
|
780 |
+
"learning_rate": 0.00042630816833284286,
|
781 |
+
"loss": 1.846,
|
782 |
+
"step": 258
|
783 |
+
},
|
784 |
+
{
|
785 |
+
"epoch": 1.13,
|
786 |
+
"learning_rate": 0.0004193487056267679,
|
787 |
+
"loss": 1.7496,
|
788 |
+
"step": 260
|
789 |
+
},
|
790 |
+
{
|
791 |
+
"epoch": 1.14,
|
792 |
+
"learning_rate": 0.0004124052493395838,
|
793 |
+
"loss": 1.7965,
|
794 |
+
"step": 262
|
795 |
+
},
|
796 |
+
{
|
797 |
+
"epoch": 1.15,
|
798 |
+
"learning_rate": 0.0004054791775008775,
|
799 |
+
"loss": 1.7148,
|
800 |
+
"step": 264
|
801 |
+
},
|
802 |
+
{
|
803 |
+
"epoch": 1.16,
|
804 |
+
"learning_rate": 0.0003985718646900401,
|
805 |
+
"loss": 1.8289,
|
806 |
+
"step": 266
|
807 |
+
},
|
808 |
+
{
|
809 |
+
"epoch": 1.16,
|
810 |
+
"learning_rate": 0.00039168468176346176,
|
811 |
+
"loss": 1.795,
|
812 |
+
"step": 268
|
813 |
+
},
|
814 |
+
{
|
815 |
+
"epoch": 1.17,
|
816 |
+
"learning_rate": 0.00038481899558246514,
|
817 |
+
"loss": 1.8347,
|
818 |
+
"step": 270
|
819 |
+
},
|
820 |
+
{
|
821 |
+
"epoch": 1.18,
|
822 |
+
"learning_rate": 0.00037797616874203193,
|
823 |
+
"loss": 1.8465,
|
824 |
+
"step": 272
|
825 |
+
},
|
826 |
+
{
|
827 |
+
"epoch": 1.19,
|
828 |
+
"learning_rate": 0.0003711575593003763,
|
829 |
+
"loss": 1.8523,
|
830 |
+
"step": 274
|
831 |
+
},
|
832 |
+
{
|
833 |
+
"epoch": 1.2,
|
834 |
+
"learning_rate": 0.00036436452050941866,
|
835 |
+
"loss": 1.7976,
|
836 |
+
"step": 276
|
837 |
+
},
|
838 |
+
{
|
839 |
+
"epoch": 1.21,
|
840 |
+
"learning_rate": 0.0003575984005462124,
|
841 |
+
"loss": 1.8306,
|
842 |
+
"step": 278
|
843 |
+
},
|
844 |
+
{
|
845 |
+
"epoch": 1.22,
|
846 |
+
"learning_rate": 0.0003508605422453799,
|
847 |
+
"loss": 1.8349,
|
848 |
+
"step": 280
|
849 |
+
},
|
850 |
+
{
|
851 |
+
"epoch": 1.23,
|
852 |
+
"learning_rate": 0.0003441522828326061,
|
853 |
+
"loss": 1.8249,
|
854 |
+
"step": 282
|
855 |
+
},
|
856 |
+
{
|
857 |
+
"epoch": 1.23,
|
858 |
+
"learning_rate": 0.00033747495365924786,
|
859 |
+
"loss": 1.822,
|
860 |
+
"step": 284
|
861 |
+
},
|
862 |
+
{
|
863 |
+
"epoch": 1.24,
|
864 |
+
"learning_rate": 0.00033082987993810755,
|
865 |
+
"loss": 1.7984,
|
866 |
+
"step": 286
|
867 |
+
},
|
868 |
+
{
|
869 |
+
"epoch": 1.25,
|
870 |
+
"learning_rate": 0.00032421838048042514,
|
871 |
+
"loss": 1.8706,
|
872 |
+
"step": 288
|
873 |
+
},
|
874 |
+
{
|
875 |
+
"epoch": 1.26,
|
876 |
+
"learning_rate": 0.0003176417674341412,
|
877 |
+
"loss": 1.8664,
|
878 |
+
"step": 290
|
879 |
+
},
|
880 |
+
{
|
881 |
+
"epoch": 1.27,
|
882 |
+
"learning_rate": 0.000311101346023482,
|
883 |
+
"loss": 1.7938,
|
884 |
+
"step": 292
|
885 |
+
},
|
886 |
+
{
|
887 |
+
"epoch": 1.28,
|
888 |
+
"learning_rate": 0.000304598414289919,
|
889 |
+
"loss": 1.8195,
|
890 |
+
"step": 294
|
891 |
+
},
|
892 |
+
{
|
893 |
+
"epoch": 1.29,
|
894 |
+
"learning_rate": 0.0002981342628345535,
|
895 |
+
"loss": 1.8177,
|
896 |
+
"step": 296
|
897 |
+
},
|
898 |
+
{
|
899 |
+
"epoch": 1.29,
|
900 |
+
"learning_rate": 0.0002917101745619787,
|
901 |
+
"loss": 1.7852,
|
902 |
+
"step": 298
|
903 |
+
},
|
904 |
+
{
|
905 |
+
"epoch": 1.3,
|
906 |
+
"learning_rate": 0.00028532742442566734,
|
907 |
+
"loss": 1.8172,
|
908 |
+
"step": 300
|
909 |
+
},
|
910 |
+
{
|
911 |
+
"epoch": 1.31,
|
912 |
+
"learning_rate": 0.00027898727917494,
|
913 |
+
"loss": 1.8626,
|
914 |
+
"step": 302
|
915 |
+
},
|
916 |
+
{
|
917 |
+
"epoch": 1.32,
|
918 |
+
"learning_rate": 0.00027269099710356055,
|
919 |
+
"loss": 1.8716,
|
920 |
+
"step": 304
|
921 |
+
},
|
922 |
+
{
|
923 |
+
"epoch": 1.33,
|
924 |
+
"learning_rate": 0.0002664398278000076,
|
925 |
+
"loss": 1.7199,
|
926 |
+
"step": 306
|
927 |
+
},
|
928 |
+
{
|
929 |
+
"epoch": 1.34,
|
930 |
+
"learning_rate": 0.0002602350118994782,
|
931 |
+
"loss": 1.8076,
|
932 |
+
"step": 308
|
933 |
+
},
|
934 |
+
{
|
935 |
+
"epoch": 1.35,
|
936 |
+
"learning_rate": 0.0002540777808376632,
|
937 |
+
"loss": 1.8665,
|
938 |
+
"step": 310
|
939 |
+
},
|
940 |
+
{
|
941 |
+
"epoch": 1.36,
|
942 |
+
"learning_rate": 0.0002479693566063527,
|
943 |
+
"loss": 1.9296,
|
944 |
+
"step": 312
|
945 |
+
},
|
946 |
+
{
|
947 |
+
"epoch": 1.36,
|
948 |
+
"learning_rate": 0.0002419109515109137,
|
949 |
+
"loss": 1.8659,
|
950 |
+
"step": 314
|
951 |
+
},
|
952 |
+
{
|
953 |
+
"epoch": 1.37,
|
954 |
+
"learning_rate": 0.0002359037679296901,
|
955 |
+
"loss": 1.7887,
|
956 |
+
"step": 316
|
957 |
+
},
|
958 |
+
{
|
959 |
+
"epoch": 1.38,
|
960 |
+
"learning_rate": 0.00022994899807537357,
|
961 |
+
"loss": 1.6962,
|
962 |
+
"step": 318
|
963 |
+
},
|
964 |
+
{
|
965 |
+
"epoch": 1.39,
|
966 |
+
"learning_rate": 0.0002240478237583915,
|
967 |
+
"loss": 1.8846,
|
968 |
+
"step": 320
|
969 |
+
},
|
970 |
+
{
|
971 |
+
"epoch": 1.4,
|
972 |
+
"learning_rate": 0.00021820141615236022,
|
973 |
+
"loss": 1.8557,
|
974 |
+
"step": 322
|
975 |
+
},
|
976 |
+
{
|
977 |
+
"epoch": 1.41,
|
978 |
+
"learning_rate": 0.00021241093556164758,
|
979 |
+
"loss": 1.8232,
|
980 |
+
"step": 324
|
981 |
+
},
|
982 |
+
{
|
983 |
+
"epoch": 1.42,
|
984 |
+
"learning_rate": 0.00020667753119109423,
|
985 |
+
"loss": 1.8227,
|
986 |
+
"step": 326
|
987 |
+
},
|
988 |
+
{
|
989 |
+
"epoch": 1.42,
|
990 |
+
"learning_rate": 0.0002010023409179378,
|
991 |
+
"loss": 1.9507,
|
992 |
+
"step": 328
|
993 |
+
},
|
994 |
+
{
|
995 |
+
"epoch": 1.43,
|
996 |
+
"learning_rate": 0.0001953864910659835,
|
997 |
+
"loss": 1.7959,
|
998 |
+
"step": 330
|
999 |
+
},
|
1000 |
+
{
|
1001 |
+
"epoch": 1.44,
|
1002 |
+
"learning_rate": 0.00018983109618207018,
|
1003 |
+
"loss": 1.777,
|
1004 |
+
"step": 332
|
1005 |
+
},
|
1006 |
+
{
|
1007 |
+
"epoch": 1.45,
|
1008 |
+
"learning_rate": 0.00018433725881487057,
|
1009 |
+
"loss": 1.8493,
|
1010 |
+
"step": 334
|
1011 |
+
},
|
1012 |
+
{
|
1013 |
+
"epoch": 1.46,
|
1014 |
+
"learning_rate": 0.00017890606929607666,
|
1015 |
+
"loss": 1.8469,
|
1016 |
+
"step": 336
|
1017 |
+
},
|
1018 |
+
{
|
1019 |
+
"epoch": 1.47,
|
1020 |
+
"learning_rate": 0.00017353860552400585,
|
1021 |
+
"loss": 1.926,
|
1022 |
+
"step": 338
|
1023 |
+
},
|
1024 |
+
{
|
1025 |
+
"epoch": 1.48,
|
1026 |
+
"learning_rate": 0.00016823593274967703,
|
1027 |
+
"loss": 1.8596,
|
1028 |
+
"step": 340
|
1029 |
+
},
|
1030 |
+
{
|
1031 |
+
"epoch": 1.49,
|
1032 |
+
"learning_rate": 0.00016299910336539642,
|
1033 |
+
"loss": 1.9387,
|
1034 |
+
"step": 342
|
1035 |
+
},
|
1036 |
+
{
|
1037 |
+
"epoch": 1.49,
|
1038 |
+
"learning_rate": 0.0001578291566958942,
|
1039 |
+
"loss": 1.8431,
|
1040 |
+
"step": 344
|
1041 |
+
},
|
1042 |
+
{
|
1043 |
+
"epoch": 1.5,
|
1044 |
+
"learning_rate": 0.00015272711879205648,
|
1045 |
+
"loss": 1.7386,
|
1046 |
+
"step": 346
|
1047 |
+
},
|
1048 |
+
{
|
1049 |
+
"epoch": 1.51,
|
1050 |
+
"learning_rate": 0.00014769400222728975,
|
1051 |
+
"loss": 1.8143,
|
1052 |
+
"step": 348
|
1053 |
+
},
|
1054 |
+
{
|
1055 |
+
"epoch": 1.52,
|
1056 |
+
"learning_rate": 0.00014273080589656125,
|
1057 |
+
"loss": 1.8075,
|
1058 |
+
"step": 350
|
1059 |
+
},
|
1060 |
+
{
|
1061 |
+
"epoch": 1.53,
|
1062 |
+
"learning_rate": 0.00013783851481815403,
|
1063 |
+
"loss": 1.8118,
|
1064 |
+
"step": 352
|
1065 |
+
},
|
1066 |
+
{
|
1067 |
+
"epoch": 1.54,
|
1068 |
+
"learning_rate": 0.00013301809993817543,
|
1069 |
+
"loss": 1.8411,
|
1070 |
+
"step": 354
|
1071 |
+
},
|
1072 |
+
{
|
1073 |
+
"epoch": 1.55,
|
1074 |
+
"learning_rate": 0.00012827051793785942,
|
1075 |
+
"loss": 1.8274,
|
1076 |
+
"step": 356
|
1077 |
+
},
|
1078 |
+
{
|
1079 |
+
"epoch": 1.55,
|
1080 |
+
"learning_rate": 0.00012359671104369846,
|
1081 |
+
"loss": 1.912,
|
1082 |
+
"step": 358
|
1083 |
+
},
|
1084 |
+
{
|
1085 |
+
"epoch": 1.56,
|
1086 |
+
"learning_rate": 0.00011899760684044514,
|
1087 |
+
"loss": 1.7547,
|
1088 |
+
"step": 360
|
1089 |
+
},
|
1090 |
+
{
|
1091 |
+
"epoch": 1.57,
|
1092 |
+
"learning_rate": 0.00011447411808702052,
|
1093 |
+
"loss": 1.9167,
|
1094 |
+
"step": 362
|
1095 |
+
},
|
1096 |
+
{
|
1097 |
+
"epoch": 1.58,
|
1098 |
+
"learning_rate": 0.00011002714253536233,
|
1099 |
+
"loss": 1.7957,
|
1100 |
+
"step": 364
|
1101 |
+
},
|
1102 |
+
{
|
1103 |
+
"epoch": 1.59,
|
1104 |
+
"learning_rate": 0.00010565756275225474,
|
1105 |
+
"loss": 1.8266,
|
1106 |
+
"step": 366
|
1107 |
+
},
|
1108 |
+
{
|
1109 |
+
"epoch": 1.6,
|
1110 |
+
"learning_rate": 0.00010136624594416826,
|
1111 |
+
"loss": 1.8285,
|
1112 |
+
"step": 368
|
1113 |
+
},
|
1114 |
+
{
|
1115 |
+
"epoch": 1.61,
|
1116 |
+
"learning_rate": 9.715404378515225e-05,
|
1117 |
+
"loss": 1.8656,
|
1118 |
+
"step": 370
|
1119 |
+
},
|
1120 |
+
{
|
1121 |
+
"epoch": 1.62,
|
1122 |
+
"learning_rate": 9.302179224780604e-05,
|
1123 |
+
"loss": 1.8691,
|
1124 |
+
"step": 372
|
1125 |
+
},
|
1126 |
+
{
|
1127 |
+
"epoch": 1.62,
|
1128 |
+
"learning_rate": 8.897031143736866e-05,
|
1129 |
+
"loss": 1.7859,
|
1130 |
+
"step": 374
|
1131 |
+
},
|
1132 |
+
{
|
1133 |
+
"epoch": 1.63,
|
1134 |
+
"learning_rate": 8.50004054289577e-05,
|
1135 |
+
"loss": 1.822,
|
1136 |
+
"step": 376
|
1137 |
+
},
|
1138 |
+
{
|
1139 |
+
"epoch": 1.64,
|
1140 |
+
"learning_rate": 8.11128621079878e-05,
|
1141 |
+
"loss": 1.9343,
|
1142 |
+
"step": 378
|
1143 |
+
},
|
1144 |
+
{
|
1145 |
+
"epoch": 1.65,
|
1146 |
+
"learning_rate": 7.73084530138044e-05,
|
1147 |
+
"loss": 1.8374,
|
1148 |
+
"step": 380
|
1149 |
+
},
|
1150 |
+
{
|
1151 |
+
"epoch": 1.66,
|
1152 |
+
"learning_rate": 7.358793318655971e-05,
|
1153 |
+
"loss": 1.8533,
|
1154 |
+
"step": 382
|
1155 |
+
},
|
1156 |
+
{
|
1157 |
+
"epoch": 1.67,
|
1158 |
+
"learning_rate": 6.995204101736469e-05,
|
1159 |
+
"loss": 1.9424,
|
1160 |
+
"step": 384
|
1161 |
+
},
|
1162 |
+
{
|
1163 |
+
"epoch": 1.68,
|
1164 |
+
"learning_rate": 6.640149810174423e-05,
|
1165 |
+
"loss": 1.8204,
|
1166 |
+
"step": 386
|
1167 |
+
},
|
1168 |
+
{
|
1169 |
+
"epoch": 1.68,
|
1170 |
+
"learning_rate": 6.293700909642619e-05,
|
1171 |
+
"loss": 1.9254,
|
1172 |
+
"step": 388
|
1173 |
+
},
|
1174 |
+
{
|
1175 |
+
"epoch": 1.69,
|
1176 |
+
"learning_rate": 5.9559261579492316e-05,
|
1177 |
+
"loss": 1.795,
|
1178 |
+
"step": 390
|
1179 |
+
},
|
1180 |
+
{
|
1181 |
+
"epoch": 1.7,
|
1182 |
+
"learning_rate": 5.6268925913917855e-05,
|
1183 |
+
"loss": 1.8497,
|
1184 |
+
"step": 392
|
1185 |
+
},
|
1186 |
+
{
|
1187 |
+
"epoch": 1.71,
|
1188 |
+
"learning_rate": 5.30666551145288e-05,
|
1189 |
+
"loss": 1.8333,
|
1190 |
+
"step": 394
|
1191 |
+
},
|
1192 |
+
{
|
1193 |
+
"epoch": 1.72,
|
1194 |
+
"learning_rate": 4.995308471840132e-05,
|
1195 |
+
"loss": 1.783,
|
1196 |
+
"step": 396
|
1197 |
+
},
|
1198 |
+
{
|
1199 |
+
"epoch": 1.73,
|
1200 |
+
"learning_rate": 4.692883265873027e-05,
|
1201 |
+
"loss": 1.9231,
|
1202 |
+
"step": 398
|
1203 |
+
},
|
1204 |
+
{
|
1205 |
+
"epoch": 1.74,
|
1206 |
+
"learning_rate": 4.399449914219167e-05,
|
1207 |
+
"loss": 1.83,
|
1208 |
+
"step": 400
|
1209 |
+
},
|
1210 |
+
{
|
1211 |
+
"epoch": 1.74,
|
1212 |
+
"learning_rate": 4.1150666529822245e-05,
|
1213 |
+
"loss": 1.773,
|
1214 |
+
"step": 402
|
1215 |
+
},
|
1216 |
+
{
|
1217 |
+
"epoch": 1.75,
|
1218 |
+
"learning_rate": 3.8397899221442545e-05,
|
1219 |
+
"loss": 1.8058,
|
1220 |
+
"step": 404
|
1221 |
+
},
|
1222 |
+
{
|
1223 |
+
"epoch": 1.76,
|
1224 |
+
"learning_rate": 3.5736743543642534e-05,
|
1225 |
+
"loss": 1.8111,
|
1226 |
+
"step": 406
|
1227 |
+
},
|
1228 |
+
{
|
1229 |
+
"epoch": 1.77,
|
1230 |
+
"learning_rate": 3.316772764135606e-05,
|
1231 |
+
"loss": 1.7742,
|
1232 |
+
"step": 408
|
1233 |
+
},
|
1234 |
+
{
|
1235 |
+
"epoch": 1.78,
|
1236 |
+
"learning_rate": 3.069136137304201e-05,
|
1237 |
+
"loss": 1.8001,
|
1238 |
+
"step": 410
|
1239 |
+
},
|
1240 |
+
{
|
1241 |
+
"epoch": 1.79,
|
1242 |
+
"learning_rate": 2.830813620949596e-05,
|
1243 |
+
"loss": 1.7632,
|
1244 |
+
"step": 412
|
1245 |
+
},
|
1246 |
+
{
|
1247 |
+
"epoch": 1.8,
|
1248 |
+
"learning_rate": 2.6018525136310922e-05,
|
1249 |
+
"loss": 1.8808,
|
1250 |
+
"step": 414
|
1251 |
+
},
|
1252 |
+
{
|
1253 |
+
"epoch": 1.81,
|
1254 |
+
"learning_rate": 2.3822982560005814e-05,
|
1255 |
+
"loss": 1.8566,
|
1256 |
+
"step": 416
|
1257 |
+
},
|
1258 |
+
{
|
1259 |
+
"epoch": 1.81,
|
1260 |
+
"learning_rate": 2.1721944217842515e-05,
|
1261 |
+
"loss": 1.7656,
|
1262 |
+
"step": 418
|
1263 |
+
},
|
1264 |
+
{
|
1265 |
+
"epoch": 1.82,
|
1266 |
+
"learning_rate": 1.9715827091347003e-05,
|
1267 |
+
"loss": 1.9143,
|
1268 |
+
"step": 420
|
1269 |
+
},
|
1270 |
+
{
|
1271 |
+
"epoch": 1.83,
|
1272 |
+
"learning_rate": 1.7805029323553768e-05,
|
1273 |
+
"loss": 1.9037,
|
1274 |
+
"step": 422
|
1275 |
+
},
|
1276 |
+
{
|
1277 |
+
"epoch": 1.84,
|
1278 |
+
"learning_rate": 1.5989930139988284e-05,
|
1279 |
+
"loss": 1.786,
|
1280 |
+
"step": 424
|
1281 |
+
},
|
1282 |
+
{
|
1283 |
+
"epoch": 1.85,
|
1284 |
+
"learning_rate": 1.4270889773404128e-05,
|
1285 |
+
"loss": 1.9072,
|
1286 |
+
"step": 426
|
1287 |
+
},
|
1288 |
+
{
|
1289 |
+
"epoch": 1.86,
|
1290 |
+
"learning_rate": 1.2648249392289924e-05,
|
1291 |
+
"loss": 1.7677,
|
1292 |
+
"step": 428
|
1293 |
+
},
|
1294 |
+
{
|
1295 |
+
"epoch": 1.87,
|
1296 |
+
"learning_rate": 1.1122331033159072e-05,
|
1297 |
+
"loss": 1.9154,
|
1298 |
+
"step": 430
|
1299 |
+
},
|
1300 |
+
{
|
1301 |
+
"epoch": 1.87,
|
1302 |
+
"learning_rate": 9.693437536637318e-06,
|
1303 |
+
"loss": 1.8175,
|
1304 |
+
"step": 432
|
1305 |
+
},
|
1306 |
+
{
|
1307 |
+
"epoch": 1.88,
|
1308 |
+
"learning_rate": 8.361852487359379e-06,
|
1309 |
+
"loss": 1.7268,
|
1310 |
+
"step": 434
|
1311 |
+
},
|
1312 |
+
{
|
1313 |
+
"epoch": 1.89,
|
1314 |
+
"learning_rate": 7.127840157687582e-06,
|
1315 |
+
"loss": 1.8565,
|
1316 |
+
"step": 436
|
1317 |
+
},
|
1318 |
+
{
|
1319 |
+
"epoch": 1.9,
|
1320 |
+
"learning_rate": 5.99164545526304e-06,
|
1321 |
+
"loss": 1.9316,
|
1322 |
+
"step": 438
|
1323 |
+
},
|
1324 |
+
{
|
1325 |
+
"epoch": 1.91,
|
1326 |
+
"learning_rate": 4.953493874400472e-06,
|
1327 |
+
"loss": 1.792,
|
1328 |
+
"step": 440
|
1329 |
+
},
|
1330 |
+
{
|
1331 |
+
"epoch": 1.92,
|
1332 |
+
"learning_rate": 4.013591451335286e-06,
|
1333 |
+
"loss": 1.8061,
|
1334 |
+
"step": 442
|
1335 |
+
},
|
1336 |
+
{
|
1337 |
+
"epoch": 1.93,
|
1338 |
+
"learning_rate": 3.1721247233328366e-06,
|
1339 |
+
"loss": 1.8494,
|
1340 |
+
"step": 444
|
1341 |
+
},
|
1342 |
+
{
|
1343 |
+
"epoch": 1.94,
|
1344 |
+
"learning_rate": 2.4292606916674344e-06,
|
1345 |
+
"loss": 1.8836,
|
1346 |
+
"step": 446
|
1347 |
+
},
|
1348 |
+
{
|
1349 |
+
"epoch": 1.94,
|
1350 |
+
"learning_rate": 1.7851467884782979e-06,
|
1351 |
+
"loss": 1.7925,
|
1352 |
+
"step": 448
|
1353 |
+
},
|
1354 |
+
{
|
1355 |
+
"epoch": 1.95,
|
1356 |
+
"learning_rate": 1.2399108475099063e-06,
|
1357 |
+
"loss": 1.8539,
|
1358 |
+
"step": 450
|
1359 |
+
},
|
1360 |
+
{
|
1361 |
+
"epoch": 1.96,
|
1362 |
+
"learning_rate": 7.936610787411258e-07,
|
1363 |
+
"loss": 1.8058,
|
1364 |
+
"step": 452
|
1365 |
+
},
|
1366 |
+
{
|
1367 |
+
"epoch": 1.97,
|
1368 |
+
"learning_rate": 4.4648604690966656e-07,
|
1369 |
+
"loss": 1.8023,
|
1370 |
+
"step": 454
|
1371 |
+
},
|
1372 |
+
{
|
1373 |
+
"epoch": 1.98,
|
1374 |
+
"learning_rate": 1.984546539349763e-07,
|
1375 |
+
"loss": 1.777,
|
1376 |
+
"step": 456
|
1377 |
+
},
|
1378 |
+
{
|
1379 |
+
"epoch": 1.99,
|
1380 |
+
"learning_rate": 4.9616125243623354e-08,
|
1381 |
+
"loss": 1.8195,
|
1382 |
+
"step": 458
|
1383 |
+
},
|
1384 |
+
{
|
1385 |
+
"epoch": 2.0,
|
1386 |
+
"learning_rate": 0.0,
|
1387 |
+
"loss": 1.8233,
|
1388 |
+
"step": 460
|
1389 |
+
},
|
1390 |
+
{
|
1391 |
+
"epoch": 2.0,
|
1392 |
+
"step": 460,
|
1393 |
+
"total_flos": 4.03948633545769e+16,
|
1394 |
+
"train_loss": 1.8878583851067916,
|
1395 |
+
"train_runtime": 22914.0021,
|
1396 |
+
"train_samples_per_second": 5.159,
|
1397 |
+
"train_steps_per_second": 0.02
|
1398 |
+
}
|
1399 |
+
],
|
1400 |
+
"max_steps": 460,
|
1401 |
+
"num_train_epochs": 2,
|
1402 |
+
"total_flos": 4.03948633545769e+16,
|
1403 |
+
"trial_name": null,
|
1404 |
+
"trial_params": null
|
1405 |
+
}
|