Update log_history.json
Browse files- log_history.json +1542 -0
log_history.json
ADDED
@@ -0,0 +1,1542 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
[
|
2 |
+
{
|
3 |
+
"loss": 8.221125,
|
4 |
+
"learning_rate": 4.977302440441604e-05,
|
5 |
+
"epoch": 0.004539511911679256,
|
6 |
+
"total_flos": 1534078999360512,
|
7 |
+
"step": 500
|
8 |
+
},
|
9 |
+
{
|
10 |
+
"loss": 7.613228515625,
|
11 |
+
"learning_rate": 4.9546048808832074e-05,
|
12 |
+
"epoch": 0.009079023823358512,
|
13 |
+
"total_flos": 3066979584402432,
|
14 |
+
"step": 1000
|
15 |
+
},
|
16 |
+
{
|
17 |
+
"loss": 7.3869140625,
|
18 |
+
"learning_rate": 4.931907321324811e-05,
|
19 |
+
"epoch": 0.01361853573503777,
|
20 |
+
"total_flos": 4595094364354560,
|
21 |
+
"step": 1500
|
22 |
+
},
|
23 |
+
{
|
24 |
+
"loss": 7.200509765625,
|
25 |
+
"learning_rate": 4.909209761766415e-05,
|
26 |
+
"epoch": 0.018158047646717024,
|
27 |
+
"total_flos": 6127634210319360,
|
28 |
+
"step": 2000
|
29 |
+
},
|
30 |
+
{
|
31 |
+
"loss": 7.10742578125,
|
32 |
+
"learning_rate": 4.886512202208019e-05,
|
33 |
+
"epoch": 0.02269755955839628,
|
34 |
+
"total_flos": 7662338490746880,
|
35 |
+
"step": 2500
|
36 |
+
},
|
37 |
+
{
|
38 |
+
"loss": 6.98366796875,
|
39 |
+
"learning_rate": 4.8638146426496224e-05,
|
40 |
+
"epoch": 0.02723707147007554,
|
41 |
+
"total_flos": 9193050592054272,
|
42 |
+
"step": 3000
|
43 |
+
},
|
44 |
+
{
|
45 |
+
"loss": 6.89792578125,
|
46 |
+
"learning_rate": 4.841117083091226e-05,
|
47 |
+
"epoch": 0.03177658338175479,
|
48 |
+
"total_flos": 10724748713505792,
|
49 |
+
"step": 3500
|
50 |
+
},
|
51 |
+
{
|
52 |
+
"loss": 6.84346875,
|
53 |
+
"learning_rate": 4.81841952353283e-05,
|
54 |
+
"epoch": 0.03631609529343405,
|
55 |
+
"total_flos": 12256855672578048,
|
56 |
+
"step": 4000
|
57 |
+
},
|
58 |
+
{
|
59 |
+
"loss": 6.70800390625,
|
60 |
+
"learning_rate": 4.795721963974434e-05,
|
61 |
+
"epoch": 0.0408556072051133,
|
62 |
+
"total_flos": 13787471576798208,
|
63 |
+
"step": 4500
|
64 |
+
},
|
65 |
+
{
|
66 |
+
"loss": 6.61594140625,
|
67 |
+
"learning_rate": 4.7730244044160374e-05,
|
68 |
+
"epoch": 0.04539511911679256,
|
69 |
+
"total_flos": 15318303924464640,
|
70 |
+
"step": 5000
|
71 |
+
},
|
72 |
+
{
|
73 |
+
"loss": 6.4896953125,
|
74 |
+
"learning_rate": 4.750326844857641e-05,
|
75 |
+
"epoch": 0.049934631028471815,
|
76 |
+
"total_flos": 16850747573342208,
|
77 |
+
"step": 5500
|
78 |
+
},
|
79 |
+
{
|
80 |
+
"loss": 6.3819609375,
|
81 |
+
"learning_rate": 4.7276292852992446e-05,
|
82 |
+
"epoch": 0.05447414294015108,
|
83 |
+
"total_flos": 18381291329746944,
|
84 |
+
"step": 6000
|
85 |
+
},
|
86 |
+
{
|
87 |
+
"loss": 6.242921875,
|
88 |
+
"learning_rate": 4.7049317257408485e-05,
|
89 |
+
"epoch": 0.059013654851830334,
|
90 |
+
"total_flos": 19910728819648512,
|
91 |
+
"step": 6500
|
92 |
+
},
|
93 |
+
{
|
94 |
+
"loss": 6.1153828125,
|
95 |
+
"learning_rate": 4.682234166182452e-05,
|
96 |
+
"epoch": 0.06355316676350958,
|
97 |
+
"total_flos": 21443990143767552,
|
98 |
+
"step": 7000
|
99 |
+
},
|
100 |
+
{
|
101 |
+
"loss": 6.03621875,
|
102 |
+
"learning_rate": 4.6595366066240556e-05,
|
103 |
+
"epoch": 0.06809267867518884,
|
104 |
+
"total_flos": 22974990836336640,
|
105 |
+
"step": 7500
|
106 |
+
},
|
107 |
+
{
|
108 |
+
"loss": 5.920453125,
|
109 |
+
"learning_rate": 4.6368390470656596e-05,
|
110 |
+
"epoch": 0.0726321905868681,
|
111 |
+
"total_flos": 24509574870405120,
|
112 |
+
"step": 8000
|
113 |
+
},
|
114 |
+
{
|
115 |
+
"loss": 5.7900625,
|
116 |
+
"learning_rate": 4.6141414875072635e-05,
|
117 |
+
"epoch": 0.07717170249854735,
|
118 |
+
"total_flos": 26041994470010880,
|
119 |
+
"step": 8500
|
120 |
+
},
|
121 |
+
{
|
122 |
+
"loss": 5.6957890625,
|
123 |
+
"learning_rate": 4.591443927948867e-05,
|
124 |
+
"epoch": 0.0817112144102266,
|
125 |
+
"total_flos": 27574606463791104,
|
126 |
+
"step": 9000
|
127 |
+
},
|
128 |
+
{
|
129 |
+
"loss": 5.611875,
|
130 |
+
"learning_rate": 4.5687463683904706e-05,
|
131 |
+
"epoch": 0.08625072632190586,
|
132 |
+
"total_flos": 29103514869712896,
|
133 |
+
"step": 9500
|
134 |
+
},
|
135 |
+
{
|
136 |
+
"loss": 5.5246484375,
|
137 |
+
"learning_rate": 4.5460488088320745e-05,
|
138 |
+
"epoch": 0.09079023823358512,
|
139 |
+
"total_flos": 30636247109852160,
|
140 |
+
"step": 10000
|
141 |
+
},
|
142 |
+
{
|
143 |
+
"loss": 5.417921875,
|
144 |
+
"learning_rate": 4.5233512492736785e-05,
|
145 |
+
"epoch": 0.09532975014526437,
|
146 |
+
"total_flos": 32167800935672832,
|
147 |
+
"step": 10500
|
148 |
+
},
|
149 |
+
{
|
150 |
+
"loss": 5.356875,
|
151 |
+
"learning_rate": 4.500653689715282e-05,
|
152 |
+
"epoch": 0.09986926205694363,
|
153 |
+
"total_flos": 33702120427751424,
|
154 |
+
"step": 11000
|
155 |
+
},
|
156 |
+
{
|
157 |
+
"loss": 5.248140625,
|
158 |
+
"learning_rate": 4.4779561301568856e-05,
|
159 |
+
"epoch": 0.1044087739686229,
|
160 |
+
"total_flos": 35233578056484864,
|
161 |
+
"step": 11500
|
162 |
+
},
|
163 |
+
{
|
164 |
+
"loss": 5.219984375,
|
165 |
+
"learning_rate": 4.4552585705984895e-05,
|
166 |
+
"epoch": 0.10894828588030216,
|
167 |
+
"total_flos": 36765228079392768,
|
168 |
+
"step": 12000
|
169 |
+
},
|
170 |
+
{
|
171 |
+
"loss": 5.129609375,
|
172 |
+
"learning_rate": 4.4325610110400935e-05,
|
173 |
+
"epoch": 0.11348779779198141,
|
174 |
+
"total_flos": 38297816023901184,
|
175 |
+
"step": 12500
|
176 |
+
},
|
177 |
+
{
|
178 |
+
"loss": 5.08884375,
|
179 |
+
"learning_rate": 4.409863451481697e-05,
|
180 |
+
"epoch": 0.11802730970366067,
|
181 |
+
"total_flos": 39827782597782528,
|
182 |
+
"step": 13000
|
183 |
+
},
|
184 |
+
{
|
185 |
+
"loss": 5.06709375,
|
186 |
+
"learning_rate": 4.3871658919233006e-05,
|
187 |
+
"epoch": 0.12256682161533992,
|
188 |
+
"total_flos": 41360995823357952,
|
189 |
+
"step": 13500
|
190 |
+
},
|
191 |
+
{
|
192 |
+
"loss": 4.992921875,
|
193 |
+
"learning_rate": 4.3644683323649045e-05,
|
194 |
+
"epoch": 0.12710633352701917,
|
195 |
+
"total_flos": 42891299087044608,
|
196 |
+
"step": 14000
|
197 |
+
},
|
198 |
+
{
|
199 |
+
"loss": 4.955390625,
|
200 |
+
"learning_rate": 4.3417707728065085e-05,
|
201 |
+
"epoch": 0.13164584543869842,
|
202 |
+
"total_flos": 44422756715778048,
|
203 |
+
"step": 14500
|
204 |
+
},
|
205 |
+
{
|
206 |
+
"loss": 4.915578125,
|
207 |
+
"learning_rate": 4.319073213248112e-05,
|
208 |
+
"epoch": 0.13618535735037768,
|
209 |
+
"total_flos": 45955392758830080,
|
210 |
+
"step": 15000
|
211 |
+
},
|
212 |
+
{
|
213 |
+
"loss": 4.83159375,
|
214 |
+
"learning_rate": 4.2963756536897156e-05,
|
215 |
+
"epoch": 0.14072486926205693,
|
216 |
+
"total_flos": 47487668062804992,
|
217 |
+
"step": 15500
|
218 |
+
},
|
219 |
+
{
|
220 |
+
"loss": 4.842171875,
|
221 |
+
"learning_rate": 4.2736780941313195e-05,
|
222 |
+
"epoch": 0.1452643811737362,
|
223 |
+
"total_flos": 49018356114840576,
|
224 |
+
"step": 16000
|
225 |
+
},
|
226 |
+
{
|
227 |
+
"loss": 4.789875,
|
228 |
+
"learning_rate": 4.250980534572923e-05,
|
229 |
+
"epoch": 0.14980389308541545,
|
230 |
+
"total_flos": 50547168323675136,
|
231 |
+
"step": 16500
|
232 |
+
},
|
233 |
+
{
|
234 |
+
"loss": 4.773703125,
|
235 |
+
"learning_rate": 4.228282975014527e-05,
|
236 |
+
"epoch": 0.1543434049970947,
|
237 |
+
"total_flos": 52076846306294784,
|
238 |
+
"step": 17000
|
239 |
+
},
|
240 |
+
{
|
241 |
+
"loss": 4.713765625,
|
242 |
+
"learning_rate": 4.20558541545613e-05,
|
243 |
+
"epoch": 0.15888291690877396,
|
244 |
+
"total_flos": 53608448230659072,
|
245 |
+
"step": 17500
|
246 |
+
},
|
247 |
+
{
|
248 |
+
"loss": 4.704203125,
|
249 |
+
"learning_rate": 4.182887855897734e-05,
|
250 |
+
"epoch": 0.1634224288204532,
|
251 |
+
"total_flos": 55139160331966464,
|
252 |
+
"step": 18000
|
253 |
+
},
|
254 |
+
{
|
255 |
+
"loss": 4.666015625,
|
256 |
+
"learning_rate": 4.160190296339338e-05,
|
257 |
+
"epoch": 0.16796194073213247,
|
258 |
+
"total_flos": 56672445705357312,
|
259 |
+
"step": 18500
|
260 |
+
},
|
261 |
+
{
|
262 |
+
"loss": 4.6343125,
|
263 |
+
"learning_rate": 4.137492736780941e-05,
|
264 |
+
"epoch": 0.17250145264381173,
|
265 |
+
"total_flos": 58205875374379008,
|
266 |
+
"step": 19000
|
267 |
+
},
|
268 |
+
{
|
269 |
+
"loss": 4.595625,
|
270 |
+
"learning_rate": 4.114795177222545e-05,
|
271 |
+
"epoch": 0.17704096455549098,
|
272 |
+
"total_flos": 59740627753350144,
|
273 |
+
"step": 19500
|
274 |
+
},
|
275 |
+
{
|
276 |
+
"loss": 4.544984375,
|
277 |
+
"learning_rate": 4.092097617664149e-05,
|
278 |
+
"epoch": 0.18158047646717024,
|
279 |
+
"total_flos": 61268598237671424,
|
280 |
+
"step": 20000
|
281 |
+
},
|
282 |
+
{
|
283 |
+
"loss": 4.53390625,
|
284 |
+
"learning_rate": 4.069400058105753e-05,
|
285 |
+
"epoch": 0.1861199883788495,
|
286 |
+
"total_flos": 62799502733153280,
|
287 |
+
"step": 20500
|
288 |
+
},
|
289 |
+
{
|
290 |
+
"loss": 4.511140625,
|
291 |
+
"learning_rate": 4.046702498547356e-05,
|
292 |
+
"epoch": 0.19065950029052875,
|
293 |
+
"total_flos": 64333870323775488,
|
294 |
+
"step": 21000
|
295 |
+
},
|
296 |
+
{
|
297 |
+
"loss": 4.49559375,
|
298 |
+
"learning_rate": 4.02400493898896e-05,
|
299 |
+
"epoch": 0.195199012202208,
|
300 |
+
"total_flos": 65865183656878080,
|
301 |
+
"step": 21500
|
302 |
+
},
|
303 |
+
{
|
304 |
+
"loss": 4.464828125,
|
305 |
+
"learning_rate": 4.001307379430564e-05,
|
306 |
+
"epoch": 0.19973852411388726,
|
307 |
+
"total_flos": 67398517128812544,
|
308 |
+
"step": 22000
|
309 |
+
},
|
310 |
+
{
|
311 |
+
"loss": 4.4189375,
|
312 |
+
"learning_rate": 3.978609819872168e-05,
|
313 |
+
"epoch": 0.20427803602556652,
|
314 |
+
"total_flos": 68930359545894912,
|
315 |
+
"step": 22500
|
316 |
+
},
|
317 |
+
{
|
318 |
+
"loss": 4.41446875,
|
319 |
+
"learning_rate": 3.955912260313771e-05,
|
320 |
+
"epoch": 0.2088175479372458,
|
321 |
+
"total_flos": 70460855203756032,
|
322 |
+
"step": 23000
|
323 |
+
},
|
324 |
+
{
|
325 |
+
"loss": 4.362109375,
|
326 |
+
"learning_rate": 3.933214700755375e-05,
|
327 |
+
"epoch": 0.21335705984892506,
|
328 |
+
"total_flos": 71991879945596928,
|
329 |
+
"step": 23500
|
330 |
+
},
|
331 |
+
{
|
332 |
+
"loss": 4.349375,
|
333 |
+
"learning_rate": 3.910517141196979e-05,
|
334 |
+
"epoch": 0.2178965717606043,
|
335 |
+
"total_flos": 73525550107336704,
|
336 |
+
"step": 24000
|
337 |
+
},
|
338 |
+
{
|
339 |
+
"loss": 4.3429375,
|
340 |
+
"learning_rate": 3.887819581638583e-05,
|
341 |
+
"epoch": 0.22243608367228357,
|
342 |
+
"total_flos": 75056911538982912,
|
343 |
+
"step": 24500
|
344 |
+
},
|
345 |
+
{
|
346 |
+
"loss": 4.29459375,
|
347 |
+
"learning_rate": 3.865122022080186e-05,
|
348 |
+
"epoch": 0.22697559558396282,
|
349 |
+
"total_flos": 76589595680578560,
|
350 |
+
"step": 25000
|
351 |
+
},
|
352 |
+
{
|
353 |
+
"loss": 4.31003125,
|
354 |
+
"learning_rate": 3.84242446252179e-05,
|
355 |
+
"epoch": 0.23151510749564208,
|
356 |
+
"total_flos": 78121606442563584,
|
357 |
+
"step": 25500
|
358 |
+
},
|
359 |
+
{
|
360 |
+
"loss": 4.278125,
|
361 |
+
"learning_rate": 3.819726902963394e-05,
|
362 |
+
"epoch": 0.23605461940732134,
|
363 |
+
"total_flos": 79651645164260352,
|
364 |
+
"step": 26000
|
365 |
+
},
|
366 |
+
{
|
367 |
+
"loss": 4.2620625,
|
368 |
+
"learning_rate": 3.797029343404998e-05,
|
369 |
+
"epoch": 0.2405941313190006,
|
370 |
+
"total_flos": 81182621807557632,
|
371 |
+
"step": 26500
|
372 |
+
},
|
373 |
+
{
|
374 |
+
"loss": 4.2123125,
|
375 |
+
"learning_rate": 3.774331783846601e-05,
|
376 |
+
"epoch": 0.24513364323067985,
|
377 |
+
"total_flos": 82714921160804352,
|
378 |
+
"step": 27000
|
379 |
+
},
|
380 |
+
{
|
381 |
+
"loss": 4.2054375,
|
382 |
+
"learning_rate": 3.751634224288205e-05,
|
383 |
+
"epoch": 0.2496731551423591,
|
384 |
+
"total_flos": 84246883824245760,
|
385 |
+
"step": 27500
|
386 |
+
},
|
387 |
+
{
|
388 |
+
"loss": 4.176625,
|
389 |
+
"learning_rate": 3.728936664729809e-05,
|
390 |
+
"epoch": 0.25421266705403833,
|
391 |
+
"total_flos": 85777163038660608,
|
392 |
+
"step": 28000
|
393 |
+
},
|
394 |
+
{
|
395 |
+
"loss": 4.17928125,
|
396 |
+
"learning_rate": 3.706239105171412e-05,
|
397 |
+
"epoch": 0.2587521789657176,
|
398 |
+
"total_flos": 87308476371763200,
|
399 |
+
"step": 28500
|
400 |
+
},
|
401 |
+
{
|
402 |
+
"loss": 4.144,
|
403 |
+
"learning_rate": 3.683541545613016e-05,
|
404 |
+
"epoch": 0.26329169087739684,
|
405 |
+
"total_flos": 88841857942241280,
|
406 |
+
"step": 29000
|
407 |
+
},
|
408 |
+
{
|
409 |
+
"loss": 4.12915625,
|
410 |
+
"learning_rate": 3.660843986054619e-05,
|
411 |
+
"epoch": 0.26783120278907613,
|
412 |
+
"total_flos": 90372185255199744,
|
413 |
+
"step": 29500
|
414 |
+
},
|
415 |
+
{
|
416 |
+
"loss": 4.1188125,
|
417 |
+
"learning_rate": 3.638146426496223e-05,
|
418 |
+
"epoch": 0.27237071470075536,
|
419 |
+
"total_flos": 91901069611849728,
|
420 |
+
"step": 30000
|
421 |
+
},
|
422 |
+
{
|
423 |
+
"loss": 4.1068125,
|
424 |
+
"learning_rate": 3.615448866937827e-05,
|
425 |
+
"epoch": 0.27691022661243464,
|
426 |
+
"total_flos": 93430819742284800,
|
427 |
+
"step": 30500
|
428 |
+
},
|
429 |
+
{
|
430 |
+
"loss": 4.08590625,
|
431 |
+
"learning_rate": 3.59275130737943e-05,
|
432 |
+
"epoch": 0.28144973852411387,
|
433 |
+
"total_flos": 94959535754032128,
|
434 |
+
"step": 31000
|
435 |
+
},
|
436 |
+
{
|
437 |
+
"loss": 4.0940625,
|
438 |
+
"learning_rate": 3.570053747821034e-05,
|
439 |
+
"epoch": 0.28598925043579315,
|
440 |
+
"total_flos": 96491859156550656,
|
441 |
+
"step": 31500
|
442 |
+
},
|
443 |
+
{
|
444 |
+
"loss": 4.04140625,
|
445 |
+
"learning_rate": 3.547356188262638e-05,
|
446 |
+
"epoch": 0.2905287623474724,
|
447 |
+
"total_flos": 98024879987951616,
|
448 |
+
"step": 32000
|
449 |
+
},
|
450 |
+
{
|
451 |
+
"loss": 4.045,
|
452 |
+
"learning_rate": 3.524658628704242e-05,
|
453 |
+
"epoch": 0.29506827425915166,
|
454 |
+
"total_flos": 99554076985135104,
|
455 |
+
"step": 32500
|
456 |
+
},
|
457 |
+
{
|
458 |
+
"loss": 4.0135,
|
459 |
+
"learning_rate": 3.501961069145845e-05,
|
460 |
+
"epoch": 0.2996077861708309,
|
461 |
+
"total_flos": 101083803066298368,
|
462 |
+
"step": 33000
|
463 |
+
},
|
464 |
+
{
|
465 |
+
"loss": 4.00115625,
|
466 |
+
"learning_rate": 3.479263509587449e-05,
|
467 |
+
"epoch": 0.3041472980825102,
|
468 |
+
"total_flos": 102618507346725888,
|
469 |
+
"step": 33500
|
470 |
+
},
|
471 |
+
{
|
472 |
+
"loss": 3.9580625,
|
473 |
+
"learning_rate": 3.456565950029053e-05,
|
474 |
+
"epoch": 0.3086868099941894,
|
475 |
+
"total_flos": 104149532088566784,
|
476 |
+
"step": 34000
|
477 |
+
},
|
478 |
+
{
|
479 |
+
"loss": 3.9779375,
|
480 |
+
"learning_rate": 3.433868390470657e-05,
|
481 |
+
"epoch": 0.3132263219058687,
|
482 |
+
"total_flos": 105683202250306560,
|
483 |
+
"step": 34500
|
484 |
+
},
|
485 |
+
{
|
486 |
+
"loss": 3.9616875,
|
487 |
+
"learning_rate": 3.41117083091226e-05,
|
488 |
+
"epoch": 0.3177658338175479,
|
489 |
+
"total_flos": 107214082696516608,
|
490 |
+
"step": 35000
|
491 |
+
},
|
492 |
+
{
|
493 |
+
"loss": 3.93290625,
|
494 |
+
"learning_rate": 3.388473271353864e-05,
|
495 |
+
"epoch": 0.3223053457292272,
|
496 |
+
"total_flos": 108744746699280384,
|
497 |
+
"step": 35500
|
498 |
+
},
|
499 |
+
{
|
500 |
+
"loss": 3.93778125,
|
501 |
+
"learning_rate": 3.365775711795468e-05,
|
502 |
+
"epoch": 0.3268448576409064,
|
503 |
+
"total_flos": 110276949855439872,
|
504 |
+
"step": 36000
|
505 |
+
},
|
506 |
+
{
|
507 |
+
"loss": 3.9205625,
|
508 |
+
"learning_rate": 3.343078152237072e-05,
|
509 |
+
"epoch": 0.3313843695525857,
|
510 |
+
"total_flos": 111804607699227648,
|
511 |
+
"step": 36500
|
512 |
+
},
|
513 |
+
{
|
514 |
+
"loss": 3.88321875,
|
515 |
+
"learning_rate": 3.320380592678675e-05,
|
516 |
+
"epoch": 0.33592388146426494,
|
517 |
+
"total_flos": 113335848884514816,
|
518 |
+
"step": 37000
|
519 |
+
},
|
520 |
+
{
|
521 |
+
"loss": 3.888125,
|
522 |
+
"learning_rate": 3.297683033120279e-05,
|
523 |
+
"epoch": 0.3404633933759442,
|
524 |
+
"total_flos": 114868653272469504,
|
525 |
+
"step": 37500
|
526 |
+
},
|
527 |
+
{
|
528 |
+
"loss": 3.89365625,
|
529 |
+
"learning_rate": 3.274985473561883e-05,
|
530 |
+
"epoch": 0.34500290528762345,
|
531 |
+
"total_flos": 116403814489061376,
|
532 |
+
"step": 38000
|
533 |
+
},
|
534 |
+
{
|
535 |
+
"loss": 3.86290625,
|
536 |
+
"learning_rate": 3.252287914003487e-05,
|
537 |
+
"epoch": 0.34954241719930274,
|
538 |
+
"total_flos": 117934117752748032,
|
539 |
+
"step": 38500
|
540 |
+
},
|
541 |
+
{
|
542 |
+
"loss": 3.84853125,
|
543 |
+
"learning_rate": 3.22959035444509e-05,
|
544 |
+
"epoch": 0.35408192911098196,
|
545 |
+
"total_flos": 119463771686095872,
|
546 |
+
"step": 39000
|
547 |
+
},
|
548 |
+
{
|
549 |
+
"loss": 3.8525,
|
550 |
+
"learning_rate": 3.206892794886694e-05,
|
551 |
+
"epoch": 0.35862144102266125,
|
552 |
+
"total_flos": 120998981001231360,
|
553 |
+
"step": 39500
|
554 |
+
},
|
555 |
+
{
|
556 |
+
"loss": 3.8159375,
|
557 |
+
"learning_rate": 3.1841952353282975e-05,
|
558 |
+
"epoch": 0.3631609529343405,
|
559 |
+
"total_flos": 122533011902048256,
|
560 |
+
"step": 40000
|
561 |
+
},
|
562 |
+
{
|
563 |
+
"loss": 3.81184375,
|
564 |
+
"learning_rate": 3.1614976757699014e-05,
|
565 |
+
"epoch": 0.36770046484601976,
|
566 |
+
"total_flos": 124066008684177408,
|
567 |
+
"step": 40500
|
568 |
+
},
|
569 |
+
{
|
570 |
+
"loss": 3.80665625,
|
571 |
+
"learning_rate": 3.1388001162115046e-05,
|
572 |
+
"epoch": 0.372239976757699,
|
573 |
+
"total_flos": 125597610608541696,
|
574 |
+
"step": 41000
|
575 |
+
},
|
576 |
+
{
|
577 |
+
"loss": 3.78921875,
|
578 |
+
"learning_rate": 3.1161025566531086e-05,
|
579 |
+
"epoch": 0.37677948866937827,
|
580 |
+
"total_flos": 127130006158875648,
|
581 |
+
"step": 41500
|
582 |
+
},
|
583 |
+
{
|
584 |
+
"loss": 3.7755,
|
585 |
+
"learning_rate": 3.0934049970947125e-05,
|
586 |
+
"epoch": 0.3813190005810575,
|
587 |
+
"total_flos": 128661632132511744,
|
588 |
+
"step": 42000
|
589 |
+
},
|
590 |
+
{
|
591 |
+
"loss": 3.76021875,
|
592 |
+
"learning_rate": 3.0707074375363164e-05,
|
593 |
+
"epoch": 0.3858585124927368,
|
594 |
+
"total_flos": 130193210007604224,
|
595 |
+
"step": 42500
|
596 |
+
},
|
597 |
+
{
|
598 |
+
"loss": 3.75678125,
|
599 |
+
"learning_rate": 3.04800987797792e-05,
|
600 |
+
"epoch": 0.390398024404416,
|
601 |
+
"total_flos": 131723801862552576,
|
602 |
+
"step": 43000
|
603 |
+
},
|
604 |
+
{
|
605 |
+
"loss": 3.74840625,
|
606 |
+
"learning_rate": 3.0253123184195235e-05,
|
607 |
+
"epoch": 0.3949375363160953,
|
608 |
+
"total_flos": 133253768436433920,
|
609 |
+
"step": 43500
|
610 |
+
},
|
611 |
+
{
|
612 |
+
"loss": 3.7486875,
|
613 |
+
"learning_rate": 3.0026147588611275e-05,
|
614 |
+
"epoch": 0.3994770482277745,
|
615 |
+
"total_flos": 134787727189435392,
|
616 |
+
"step": 44000
|
617 |
+
},
|
618 |
+
{
|
619 |
+
"loss": 3.7429375,
|
620 |
+
"learning_rate": 2.9799171993027314e-05,
|
621 |
+
"epoch": 0.4040165601394538,
|
622 |
+
"total_flos": 136321589745349632,
|
623 |
+
"step": 44500
|
624 |
+
},
|
625 |
+
{
|
626 |
+
"loss": 3.70103125,
|
627 |
+
"learning_rate": 2.9572196397443346e-05,
|
628 |
+
"epoch": 0.40855607205113303,
|
629 |
+
"total_flos": 137853696704421888,
|
630 |
+
"step": 45000
|
631 |
+
},
|
632 |
+
{
|
633 |
+
"loss": 3.7165625,
|
634 |
+
"learning_rate": 2.9345220801859385e-05,
|
635 |
+
"epoch": 0.4130955839628123,
|
636 |
+
"total_flos": 139387583309607936,
|
637 |
+
"step": 45500
|
638 |
+
},
|
639 |
+
{
|
640 |
+
"loss": 3.6774375,
|
641 |
+
"learning_rate": 2.9118245206275425e-05,
|
642 |
+
"epoch": 0.4176350958744916,
|
643 |
+
"total_flos": 140920796535183360,
|
644 |
+
"step": 46000
|
645 |
+
},
|
646 |
+
{
|
647 |
+
"loss": 3.6928125,
|
648 |
+
"learning_rate": 2.8891269610691457e-05,
|
649 |
+
"epoch": 0.42217460778617083,
|
650 |
+
"total_flos": 142455452717067264,
|
651 |
+
"step": 46500
|
652 |
+
},
|
653 |
+
{
|
654 |
+
"loss": 3.6948125,
|
655 |
+
"learning_rate": 2.8664294015107496e-05,
|
656 |
+
"epoch": 0.4267141196978501,
|
657 |
+
"total_flos": 143988545696283648,
|
658 |
+
"step": 47000
|
659 |
+
},
|
660 |
+
{
|
661 |
+
"loss": 3.65459375,
|
662 |
+
"learning_rate": 2.8437318419523535e-05,
|
663 |
+
"epoch": 0.43125363160952934,
|
664 |
+
"total_flos": 145519618536668160,
|
665 |
+
"step": 47500
|
666 |
+
},
|
667 |
+
{
|
668 |
+
"loss": 3.66340625,
|
669 |
+
"learning_rate": 2.821034282393957e-05,
|
670 |
+
"epoch": 0.4357931435212086,
|
671 |
+
"total_flos": 147054851901075456,
|
672 |
+
"step": 48000
|
673 |
+
},
|
674 |
+
{
|
675 |
+
"loss": 3.65359375,
|
676 |
+
"learning_rate": 2.7983367228355607e-05,
|
677 |
+
"epoch": 0.44033265543288785,
|
678 |
+
"total_flos": 148586742416701440,
|
679 |
+
"step": 48500
|
680 |
+
},
|
681 |
+
{
|
682 |
+
"loss": 3.64646875,
|
683 |
+
"learning_rate": 2.7756391632771643e-05,
|
684 |
+
"epoch": 0.44487216734456714,
|
685 |
+
"total_flos": 150117430468737024,
|
686 |
+
"step": 49000
|
687 |
+
},
|
688 |
+
{
|
689 |
+
"loss": 3.63471875,
|
690 |
+
"learning_rate": 2.7529416037187682e-05,
|
691 |
+
"epoch": 0.44941167925624637,
|
692 |
+
"total_flos": 151648912146742272,
|
693 |
+
"step": 49500
|
694 |
+
},
|
695 |
+
{
|
696 |
+
"loss": 3.60778125,
|
697 |
+
"learning_rate": 2.730244044160372e-05,
|
698 |
+
"epoch": 0.45395119116792565,
|
699 |
+
"total_flos": 153182077273774080,
|
700 |
+
"step": 50000
|
701 |
+
},
|
702 |
+
{
|
703 |
+
"loss": 3.5908125,
|
704 |
+
"learning_rate": 2.7075464846019754e-05,
|
705 |
+
"epoch": 0.4584907030796049,
|
706 |
+
"total_flos": 154711081876783104,
|
707 |
+
"step": 50500
|
708 |
+
},
|
709 |
+
{
|
710 |
+
"loss": 3.6209375,
|
711 |
+
"learning_rate": 2.6848489250435793e-05,
|
712 |
+
"epoch": 0.46303021499128416,
|
713 |
+
"total_flos": 156241240844838912,
|
714 |
+
"step": 51000
|
715 |
+
},
|
716 |
+
{
|
717 |
+
"loss": 3.596,
|
718 |
+
"learning_rate": 2.6621513654851832e-05,
|
719 |
+
"epoch": 0.4675697269029634,
|
720 |
+
"total_flos": 157773083261921280,
|
721 |
+
"step": 51500
|
722 |
+
},
|
723 |
+
{
|
724 |
+
"loss": 3.57665625,
|
725 |
+
"learning_rate": 2.639453805926787e-05,
|
726 |
+
"epoch": 0.4721092388146427,
|
727 |
+
"total_flos": 159301246140417024,
|
728 |
+
"step": 52000
|
729 |
+
},
|
730 |
+
{
|
731 |
+
"loss": 3.5551875,
|
732 |
+
"learning_rate": 2.6167562463683904e-05,
|
733 |
+
"epoch": 0.4766487507263219,
|
734 |
+
"total_flos": 160831573453375488,
|
735 |
+
"step": 52500
|
736 |
+
},
|
737 |
+
{
|
738 |
+
"loss": 3.57975,
|
739 |
+
"learning_rate": 2.5940586868099943e-05,
|
740 |
+
"epoch": 0.4811882626380012,
|
741 |
+
"total_flos": 162364834777494528,
|
742 |
+
"step": 53000
|
743 |
+
},
|
744 |
+
{
|
745 |
+
"loss": 3.55321875,
|
746 |
+
"learning_rate": 2.5713611272515982e-05,
|
747 |
+
"epoch": 0.4857277745496804,
|
748 |
+
"total_flos": 163897711313264640,
|
749 |
+
"step": 53500
|
750 |
+
},
|
751 |
+
{
|
752 |
+
"loss": 3.5406875,
|
753 |
+
"learning_rate": 2.5486635676932018e-05,
|
754 |
+
"epoch": 0.4902672864613597,
|
755 |
+
"total_flos": 165430708095393792,
|
756 |
+
"step": 54000
|
757 |
+
},
|
758 |
+
{
|
759 |
+
"loss": 3.5404375,
|
760 |
+
"learning_rate": 2.5259660081348054e-05,
|
761 |
+
"epoch": 0.4948067983730389,
|
762 |
+
"total_flos": 166962550512476160,
|
763 |
+
"step": 54500
|
764 |
+
},
|
765 |
+
{
|
766 |
+
"loss": 3.5504375,
|
767 |
+
"learning_rate": 2.503268448576409e-05,
|
768 |
+
"epoch": 0.4993463102847182,
|
769 |
+
"total_flos": 168496485216205824,
|
770 |
+
"step": 55000
|
771 |
+
},
|
772 |
+
{
|
773 |
+
"loss": 3.541,
|
774 |
+
"learning_rate": 2.480570889018013e-05,
|
775 |
+
"epoch": 0.5038858221963974,
|
776 |
+
"total_flos": 170029049111442432,
|
777 |
+
"step": 55500
|
778 |
+
},
|
779 |
+
{
|
780 |
+
"loss": 3.5195625,
|
781 |
+
"learning_rate": 2.4578733294596164e-05,
|
782 |
+
"epoch": 0.5084253341080767,
|
783 |
+
"total_flos": 171563681244054528,
|
784 |
+
"step": 56000
|
785 |
+
},
|
786 |
+
{
|
787 |
+
"loss": 3.51209375,
|
788 |
+
"learning_rate": 2.4351757699012204e-05,
|
789 |
+
"epoch": 0.512964846019756,
|
790 |
+
"total_flos": 173096076794388480,
|
791 |
+
"step": 56500
|
792 |
+
},
|
793 |
+
{
|
794 |
+
"loss": 3.509,
|
795 |
+
"learning_rate": 2.412478210342824e-05,
|
796 |
+
"epoch": 0.5175043579314352,
|
797 |
+
"total_flos": 174627462275306496,
|
798 |
+
"step": 57000
|
799 |
+
},
|
800 |
+
{
|
801 |
+
"loss": 3.51578125,
|
802 |
+
"learning_rate": 2.389780650784428e-05,
|
803 |
+
"epoch": 0.5220438698431145,
|
804 |
+
"total_flos": 176158102228798464,
|
805 |
+
"step": 57500
|
806 |
+
},
|
807 |
+
{
|
808 |
+
"loss": 3.48325,
|
809 |
+
"learning_rate": 2.3670830912260314e-05,
|
810 |
+
"epoch": 0.5265833817547937,
|
811 |
+
"total_flos": 177692157178887168,
|
812 |
+
"step": 58000
|
813 |
+
},
|
814 |
+
{
|
815 |
+
"loss": 3.474625,
|
816 |
+
"learning_rate": 2.3443855316676354e-05,
|
817 |
+
"epoch": 0.531122893666473,
|
818 |
+
"total_flos": 179225418503006208,
|
819 |
+
"step": 58500
|
820 |
+
},
|
821 |
+
{
|
822 |
+
"loss": 3.4898125,
|
823 |
+
"learning_rate": 2.321687972109239e-05,
|
824 |
+
"epoch": 0.5356624055781523,
|
825 |
+
"total_flos": 180754567401646080,
|
826 |
+
"step": 59000
|
827 |
+
},
|
828 |
+
{
|
829 |
+
"loss": 3.4839375,
|
830 |
+
"learning_rate": 2.298990412550843e-05,
|
831 |
+
"epoch": 0.5402019174898315,
|
832 |
+
"total_flos": 182287227493969920,
|
833 |
+
"step": 59500
|
834 |
+
},
|
835 |
+
{
|
836 |
+
"loss": 3.4523125,
|
837 |
+
"learning_rate": 2.2762928529924464e-05,
|
838 |
+
"epoch": 0.5447414294015107,
|
839 |
+
"total_flos": 183817578856200192,
|
840 |
+
"step": 60000
|
841 |
+
},
|
842 |
+
{
|
843 |
+
"loss": 3.467875,
|
844 |
+
"learning_rate": 2.25359529343405e-05,
|
845 |
+
"epoch": 0.54928094131319,
|
846 |
+
"total_flos": 185350335145611264,
|
847 |
+
"step": 60500
|
848 |
+
},
|
849 |
+
{
|
850 |
+
"loss": 3.4614375,
|
851 |
+
"learning_rate": 2.2308977338756536e-05,
|
852 |
+
"epoch": 0.5538204532248693,
|
853 |
+
"total_flos": 186880734606385152,
|
854 |
+
"step": 61000
|
855 |
+
},
|
856 |
+
{
|
857 |
+
"loss": 3.4434375,
|
858 |
+
"learning_rate": 2.2082001743172575e-05,
|
859 |
+
"epoch": 0.5583599651365485,
|
860 |
+
"total_flos": 188413827585601536,
|
861 |
+
"step": 61500
|
862 |
+
},
|
863 |
+
{
|
864 |
+
"loss": 3.440875,
|
865 |
+
"learning_rate": 2.185502614758861e-05,
|
866 |
+
"epoch": 0.5628994770482277,
|
867 |
+
"total_flos": 189945429509965824,
|
868 |
+
"step": 62000
|
869 |
+
},
|
870 |
+
{
|
871 |
+
"loss": 3.4280625,
|
872 |
+
"learning_rate": 2.162805055200465e-05,
|
873 |
+
"epoch": 0.5674389889599071,
|
874 |
+
"total_flos": 191479243967336448,
|
875 |
+
"step": 62500
|
876 |
+
},
|
877 |
+
{
|
878 |
+
"loss": 3.41775,
|
879 |
+
"learning_rate": 2.1401074956420686e-05,
|
880 |
+
"epoch": 0.5719785008715863,
|
881 |
+
"total_flos": 193010629448254464,
|
882 |
+
"step": 63000
|
883 |
+
},
|
884 |
+
{
|
885 |
+
"loss": 3.4230625,
|
886 |
+
"learning_rate": 2.1174099360836725e-05,
|
887 |
+
"epoch": 0.5765180127832655,
|
888 |
+
"total_flos": 194545213482322944,
|
889 |
+
"step": 63500
|
890 |
+
},
|
891 |
+
{
|
892 |
+
"loss": 3.3981875,
|
893 |
+
"learning_rate": 2.094712376525276e-05,
|
894 |
+
"epoch": 0.5810575246949448,
|
895 |
+
"total_flos": 196078571003529216,
|
896 |
+
"step": 64000
|
897 |
+
},
|
898 |
+
{
|
899 |
+
"loss": 3.41275,
|
900 |
+
"learning_rate": 2.07201481696688e-05,
|
901 |
+
"epoch": 0.5855970366066241,
|
902 |
+
"total_flos": 197608970464303104,
|
903 |
+
"step": 64500
|
904 |
+
},
|
905 |
+
{
|
906 |
+
"loss": 3.398,
|
907 |
+
"learning_rate": 2.0493172574084836e-05,
|
908 |
+
"epoch": 0.5901365485183033,
|
909 |
+
"total_flos": 199141414113180672,
|
910 |
+
"step": 65000
|
911 |
+
},
|
912 |
+
{
|
913 |
+
"loss": 3.399375,
|
914 |
+
"learning_rate": 2.026619697850087e-05,
|
915 |
+
"epoch": 0.5946760604299826,
|
916 |
+
"total_flos": 200673304628806656,
|
917 |
+
"step": 65500
|
918 |
+
},
|
919 |
+
{
|
920 |
+
"loss": 3.4090625,
|
921 |
+
"learning_rate": 2.0039221382916907e-05,
|
922 |
+
"epoch": 0.5992155723416618,
|
923 |
+
"total_flos": 202205628031325184,
|
924 |
+
"step": 66000
|
925 |
+
},
|
926 |
+
{
|
927 |
+
"loss": 3.3909375,
|
928 |
+
"learning_rate": 1.9812245787332947e-05,
|
929 |
+
"epoch": 0.6037550842533411,
|
930 |
+
"total_flos": 203738961503259648,
|
931 |
+
"step": 66500
|
932 |
+
},
|
933 |
+
{
|
934 |
+
"loss": 3.3735625,
|
935 |
+
"learning_rate": 1.9585270191748982e-05,
|
936 |
+
"epoch": 0.6082945961650204,
|
937 |
+
"total_flos": 205271789940486144,
|
938 |
+
"step": 67000
|
939 |
+
},
|
940 |
+
{
|
941 |
+
"loss": 3.36775,
|
942 |
+
"learning_rate": 1.935829459616502e-05,
|
943 |
+
"epoch": 0.6128341080766996,
|
944 |
+
"total_flos": 206805123412420608,
|
945 |
+
"step": 67500
|
946 |
+
},
|
947 |
+
{
|
948 |
+
"loss": 3.3604375,
|
949 |
+
"learning_rate": 1.9131319000581057e-05,
|
950 |
+
"epoch": 0.6173736199883788,
|
951 |
+
"total_flos": 208335643119553536,
|
952 |
+
"step": 68000
|
953 |
+
},
|
954 |
+
{
|
955 |
+
"loss": 3.36825,
|
956 |
+
"learning_rate": 1.8904343404997097e-05,
|
957 |
+
"epoch": 0.6219131319000581,
|
958 |
+
"total_flos": 209869096837847040,
|
959 |
+
"step": 68500
|
960 |
+
},
|
961 |
+
{
|
962 |
+
"loss": 3.340125,
|
963 |
+
"learning_rate": 1.8677367809413132e-05,
|
964 |
+
"epoch": 0.6264526438117374,
|
965 |
+
"total_flos": 211403344182110208,
|
966 |
+
"step": 69000
|
967 |
+
},
|
968 |
+
{
|
969 |
+
"loss": 3.33675,
|
970 |
+
"learning_rate": 1.845039221382917e-05,
|
971 |
+
"epoch": 0.6309921557234166,
|
972 |
+
"total_flos": 212935787830987776,
|
973 |
+
"step": 69500
|
974 |
+
},
|
975 |
+
{
|
976 |
+
"loss": 3.344125,
|
977 |
+
"learning_rate": 1.8223416618245207e-05,
|
978 |
+
"epoch": 0.6355316676350958,
|
979 |
+
"total_flos": 214466932819187712,
|
980 |
+
"step": 70000
|
981 |
+
},
|
982 |
+
{
|
983 |
+
"loss": 3.336,
|
984 |
+
"learning_rate": 1.7996441022661247e-05,
|
985 |
+
"epoch": 0.6400711795467752,
|
986 |
+
"total_flos": 215998222103018496,
|
987 |
+
"step": 70500
|
988 |
+
},
|
989 |
+
{
|
990 |
+
"loss": 3.349375,
|
991 |
+
"learning_rate": 1.7769465427077282e-05,
|
992 |
+
"epoch": 0.6446106914584544,
|
993 |
+
"total_flos": 217531844166214656,
|
994 |
+
"step": 71000
|
995 |
+
},
|
996 |
+
{
|
997 |
+
"loss": 3.3093125,
|
998 |
+
"learning_rate": 1.7542489831493318e-05,
|
999 |
+
"epoch": 0.6491502033701336,
|
1000 |
+
"total_flos": 219061257606844416,
|
1001 |
+
"step": 71500
|
1002 |
+
},
|
1003 |
+
{
|
1004 |
+
"loss": 3.314375,
|
1005 |
+
"learning_rate": 1.7315514235909354e-05,
|
1006 |
+
"epoch": 0.6536897152818129,
|
1007 |
+
"total_flos": 220591969708151808,
|
1008 |
+
"step": 72000
|
1009 |
+
},
|
1010 |
+
{
|
1011 |
+
"loss": 3.3166875,
|
1012 |
+
"learning_rate": 1.7088538640325393e-05,
|
1013 |
+
"epoch": 0.6582292271934922,
|
1014 |
+
"total_flos": 222126144904599552,
|
1015 |
+
"step": 72500
|
1016 |
+
},
|
1017 |
+
{
|
1018 |
+
"loss": 3.32175,
|
1019 |
+
"learning_rate": 1.686156304474143e-05,
|
1020 |
+
"epoch": 0.6627687391051714,
|
1021 |
+
"total_flos": 223658997391097856,
|
1022 |
+
"step": 73000
|
1023 |
+
},
|
1024 |
+
{
|
1025 |
+
"loss": 3.2955625,
|
1026 |
+
"learning_rate": 1.6634587449157468e-05,
|
1027 |
+
"epoch": 0.6673082510168507,
|
1028 |
+
"total_flos": 225189493048958976,
|
1029 |
+
"step": 73500
|
1030 |
+
},
|
1031 |
+
{
|
1032 |
+
"loss": 3.35425,
|
1033 |
+
"learning_rate": 1.6407611853573504e-05,
|
1034 |
+
"epoch": 0.6718477629285299,
|
1035 |
+
"total_flos": 226722032894923776,
|
1036 |
+
"step": 74000
|
1037 |
+
},
|
1038 |
+
{
|
1039 |
+
"loss": 3.2941875,
|
1040 |
+
"learning_rate": 1.6180636257989543e-05,
|
1041 |
+
"epoch": 0.6763872748402092,
|
1042 |
+
"total_flos": 228254404395985920,
|
1043 |
+
"step": 74500
|
1044 |
+
},
|
1045 |
+
{
|
1046 |
+
"loss": 3.2930625,
|
1047 |
+
"learning_rate": 1.595366066240558e-05,
|
1048 |
+
"epoch": 0.6809267867518884,
|
1049 |
+
"total_flos": 229787834065007616,
|
1050 |
+
"step": 75000
|
1051 |
+
},
|
1052 |
+
{
|
1053 |
+
"loss": 3.3018125,
|
1054 |
+
"learning_rate": 1.5726685066821618e-05,
|
1055 |
+
"epoch": 0.6854662986635677,
|
1056 |
+
"total_flos": 231321937113639936,
|
1057 |
+
"step": 75500
|
1058 |
+
},
|
1059 |
+
{
|
1060 |
+
"loss": 3.2865,
|
1061 |
+
"learning_rate": 1.5499709471237654e-05,
|
1062 |
+
"epoch": 0.6900058105752469,
|
1063 |
+
"total_flos": 232855607275379712,
|
1064 |
+
"step": 76000
|
1065 |
+
},
|
1066 |
+
{
|
1067 |
+
"loss": 3.2755,
|
1068 |
+
"learning_rate": 1.5272733875653693e-05,
|
1069 |
+
"epoch": 0.6945453224869262,
|
1070 |
+
"total_flos": 234390022964545536,
|
1071 |
+
"step": 76500
|
1072 |
+
},
|
1073 |
+
{
|
1074 |
+
"loss": 3.2875625,
|
1075 |
+
"learning_rate": 1.5045758280069727e-05,
|
1076 |
+
"epoch": 0.6990848343986055,
|
1077 |
+
"total_flos": 235921119854201856,
|
1078 |
+
"step": 77000
|
1079 |
+
},
|
1080 |
+
{
|
1081 |
+
"loss": 3.254,
|
1082 |
+
"learning_rate": 1.4818782684485766e-05,
|
1083 |
+
"epoch": 0.7036243463102847,
|
1084 |
+
"total_flos": 237454693818854400,
|
1085 |
+
"step": 77500
|
1086 |
+
},
|
1087 |
+
{
|
1088 |
+
"loss": 3.25675,
|
1089 |
+
"learning_rate": 1.4591807088901802e-05,
|
1090 |
+
"epoch": 0.7081638582219639,
|
1091 |
+
"total_flos": 238986728630111232,
|
1092 |
+
"step": 78000
|
1093 |
+
},
|
1094 |
+
{
|
1095 |
+
"loss": 3.2600625,
|
1096 |
+
"learning_rate": 1.436483149331784e-05,
|
1097 |
+
"epoch": 0.7127033701336433,
|
1098 |
+
"total_flos": 240520519038210048,
|
1099 |
+
"step": 78500
|
1100 |
+
},
|
1101 |
+
{
|
1102 |
+
"loss": 3.26825,
|
1103 |
+
"learning_rate": 1.4137855897733875e-05,
|
1104 |
+
"epoch": 0.7172428820453225,
|
1105 |
+
"total_flos": 242053515820339200,
|
1106 |
+
"step": 79000
|
1107 |
+
},
|
1108 |
+
{
|
1109 |
+
"loss": 3.2588125,
|
1110 |
+
"learning_rate": 1.3910880302149915e-05,
|
1111 |
+
"epoch": 0.7217823939570017,
|
1112 |
+
"total_flos": 243586416405381120,
|
1113 |
+
"step": 79500
|
1114 |
+
},
|
1115 |
+
{
|
1116 |
+
"loss": 3.249,
|
1117 |
+
"learning_rate": 1.368390470656595e-05,
|
1118 |
+
"epoch": 0.726321905868681,
|
1119 |
+
"total_flos": 245117368999406592,
|
1120 |
+
"step": 80000
|
1121 |
+
},
|
1122 |
+
{
|
1123 |
+
"loss": 3.24175,
|
1124 |
+
"learning_rate": 1.345692911098199e-05,
|
1125 |
+
"epoch": 0.7308614177803603,
|
1126 |
+
"total_flos": 246648778529596416,
|
1127 |
+
"step": 80500
|
1128 |
+
},
|
1129 |
+
{
|
1130 |
+
"loss": 3.2466875,
|
1131 |
+
"learning_rate": 1.3229953515398025e-05,
|
1132 |
+
"epoch": 0.7354009296920395,
|
1133 |
+
"total_flos": 248182496789879808,
|
1134 |
+
"step": 81000
|
1135 |
+
},
|
1136 |
+
{
|
1137 |
+
"loss": 3.2416875,
|
1138 |
+
"learning_rate": 1.3002977919814063e-05,
|
1139 |
+
"epoch": 0.7399404416037187,
|
1140 |
+
"total_flos": 249716022655988736,
|
1141 |
+
"step": 81500
|
1142 |
+
},
|
1143 |
+
{
|
1144 |
+
"loss": 3.2241875,
|
1145 |
+
"learning_rate": 1.2776002324230099e-05,
|
1146 |
+
"epoch": 0.744479953515398,
|
1147 |
+
"total_flos": 251249211832292352,
|
1148 |
+
"step": 82000
|
1149 |
+
},
|
1150 |
+
{
|
1151 |
+
"loss": 3.2393125,
|
1152 |
+
"learning_rate": 1.2549026728646138e-05,
|
1153 |
+
"epoch": 0.7490194654270773,
|
1154 |
+
"total_flos": 252781751678257152,
|
1155 |
+
"step": 82500
|
1156 |
+
},
|
1157 |
+
{
|
1158 |
+
"loss": 3.2358125,
|
1159 |
+
"learning_rate": 1.2322051133062174e-05,
|
1160 |
+
"epoch": 0.7535589773387565,
|
1161 |
+
"total_flos": 254314387721309184,
|
1162 |
+
"step": 83000
|
1163 |
+
},
|
1164 |
+
{
|
1165 |
+
"loss": 3.205125,
|
1166 |
+
"learning_rate": 1.2095075537478211e-05,
|
1167 |
+
"epoch": 0.7580984892504358,
|
1168 |
+
"total_flos": 255847456651253760,
|
1169 |
+
"step": 83500
|
1170 |
+
},
|
1171 |
+
{
|
1172 |
+
"loss": 3.2190625,
|
1173 |
+
"learning_rate": 1.1868099941894249e-05,
|
1174 |
+
"epoch": 0.762638001162115,
|
1175 |
+
"total_flos": 257379034526346240,
|
1176 |
+
"step": 84000
|
1177 |
+
},
|
1178 |
+
{
|
1179 |
+
"loss": 3.2224375,
|
1180 |
+
"learning_rate": 1.1641124346310285e-05,
|
1181 |
+
"epoch": 0.7671775130737943,
|
1182 |
+
"total_flos": 258912247751921664,
|
1183 |
+
"step": 84500
|
1184 |
+
},
|
1185 |
+
{
|
1186 |
+
"loss": 3.225,
|
1187 |
+
"learning_rate": 1.1414148750726322e-05,
|
1188 |
+
"epoch": 0.7717170249854736,
|
1189 |
+
"total_flos": 260445460977497088,
|
1190 |
+
"step": 85000
|
1191 |
+
},
|
1192 |
+
{
|
1193 |
+
"loss": 3.2075625,
|
1194 |
+
"learning_rate": 1.118717315514236e-05,
|
1195 |
+
"epoch": 0.7762565368971528,
|
1196 |
+
"total_flos": 261975668044096512,
|
1197 |
+
"step": 85500
|
1198 |
+
},
|
1199 |
+
{
|
1200 |
+
"loss": 3.19675,
|
1201 |
+
"learning_rate": 1.0960197559558397e-05,
|
1202 |
+
"epoch": 0.780796048808832,
|
1203 |
+
"total_flos": 263509434402923520,
|
1204 |
+
"step": 86000
|
1205 |
+
},
|
1206 |
+
{
|
1207 |
+
"loss": 3.2095,
|
1208 |
+
"learning_rate": 1.0733221963974434e-05,
|
1209 |
+
"epoch": 0.7853355607205114,
|
1210 |
+
"total_flos": 265042815973401600,
|
1211 |
+
"step": 86500
|
1212 |
+
},
|
1213 |
+
{
|
1214 |
+
"loss": 3.2146875,
|
1215 |
+
"learning_rate": 1.050624636839047e-05,
|
1216 |
+
"epoch": 0.7898750726321906,
|
1217 |
+
"total_flos": 266577448106013696,
|
1218 |
+
"step": 87000
|
1219 |
+
},
|
1220 |
+
{
|
1221 |
+
"loss": 3.1964375,
|
1222 |
+
"learning_rate": 1.0279270772806508e-05,
|
1223 |
+
"epoch": 0.7944145845438698,
|
1224 |
+
"total_flos": 268109699360716800,
|
1225 |
+
"step": 87500
|
1226 |
+
},
|
1227 |
+
{
|
1228 |
+
"loss": 3.1685625,
|
1229 |
+
"learning_rate": 1.0052295177222545e-05,
|
1230 |
+
"epoch": 0.798954096455549,
|
1231 |
+
"total_flos": 269643032832651264,
|
1232 |
+
"step": 88000
|
1233 |
+
},
|
1234 |
+
{
|
1235 |
+
"loss": 3.1924375,
|
1236 |
+
"learning_rate": 9.825319581638583e-06,
|
1237 |
+
"epoch": 0.8034936083672284,
|
1238 |
+
"total_flos": 271174490461384704,
|
1239 |
+
"step": 88500
|
1240 |
+
},
|
1241 |
+
{
|
1242 |
+
"loss": 3.198875,
|
1243 |
+
"learning_rate": 9.59834398605462e-06,
|
1244 |
+
"epoch": 0.8080331202789076,
|
1245 |
+
"total_flos": 272707270800067584,
|
1246 |
+
"step": 89000
|
1247 |
+
},
|
1248 |
+
{
|
1249 |
+
"loss": 3.2030625,
|
1250 |
+
"learning_rate": 9.371368390470658e-06,
|
1251 |
+
"epoch": 0.8125726321905868,
|
1252 |
+
"total_flos": 274237501915938816,
|
1253 |
+
"step": 89500
|
1254 |
+
},
|
1255 |
+
{
|
1256 |
+
"loss": 3.183125,
|
1257 |
+
"learning_rate": 9.144392794886694e-06,
|
1258 |
+
"epoch": 0.8171121441022661,
|
1259 |
+
"total_flos": 275768502608507904,
|
1260 |
+
"step": 90000
|
1261 |
+
},
|
1262 |
+
{
|
1263 |
+
"loss": 3.1576875,
|
1264 |
+
"learning_rate": 8.917417199302731e-06,
|
1265 |
+
"epoch": 0.8216516560139454,
|
1266 |
+
"total_flos": 277300705764667392,
|
1267 |
+
"step": 90500
|
1268 |
+
},
|
1269 |
+
{
|
1270 |
+
"loss": 3.1826875,
|
1271 |
+
"learning_rate": 8.690441603718769e-06,
|
1272 |
+
"epoch": 0.8261911679256246,
|
1273 |
+
"total_flos": 278832163393400832,
|
1274 |
+
"step": 91000
|
1275 |
+
},
|
1276 |
+
{
|
1277 |
+
"loss": 3.1574375,
|
1278 |
+
"learning_rate": 8.463466008134806e-06,
|
1279 |
+
"epoch": 0.8307306798373039,
|
1280 |
+
"total_flos": 280364631091550208,
|
1281 |
+
"step": 91500
|
1282 |
+
},
|
1283 |
+
{
|
1284 |
+
"loss": 3.160125,
|
1285 |
+
"learning_rate": 8.236490412550844e-06,
|
1286 |
+
"epoch": 0.8352701917489832,
|
1287 |
+
"total_flos": 281895944424652800,
|
1288 |
+
"step": 92000
|
1289 |
+
},
|
1290 |
+
{
|
1291 |
+
"loss": 3.163875,
|
1292 |
+
"learning_rate": 8.009514816966881e-06,
|
1293 |
+
"epoch": 0.8398097036606624,
|
1294 |
+
"total_flos": 283429542438577152,
|
1295 |
+
"step": 92500
|
1296 |
+
},
|
1297 |
+
{
|
1298 |
+
"loss": 3.177625,
|
1299 |
+
"learning_rate": 7.782539221382917e-06,
|
1300 |
+
"epoch": 0.8443492155723417,
|
1301 |
+
"total_flos": 284961793693280256,
|
1302 |
+
"step": 93000
|
1303 |
+
},
|
1304 |
+
{
|
1305 |
+
"loss": 3.161875,
|
1306 |
+
"learning_rate": 7.555563625798954e-06,
|
1307 |
+
"epoch": 0.8488887274840209,
|
1308 |
+
"total_flos": 286495920791184384,
|
1309 |
+
"step": 93500
|
1310 |
+
},
|
1311 |
+
{
|
1312 |
+
"loss": 3.16925,
|
1313 |
+
"learning_rate": 7.328588030214992e-06,
|
1314 |
+
"epoch": 0.8534282393957002,
|
1315 |
+
"total_flos": 288029470706565120,
|
1316 |
+
"step": 94000
|
1317 |
+
},
|
1318 |
+
{
|
1319 |
+
"loss": 3.1579375,
|
1320 |
+
"learning_rate": 7.101612434631029e-06,
|
1321 |
+
"epoch": 0.8579677513073795,
|
1322 |
+
"total_flos": 289561505517821952,
|
1323 |
+
"step": 94500
|
1324 |
+
},
|
1325 |
+
{
|
1326 |
+
"loss": 3.1499375,
|
1327 |
+
"learning_rate": 6.874636839047066e-06,
|
1328 |
+
"epoch": 0.8625072632190587,
|
1329 |
+
"total_flos": 291092097372770304,
|
1330 |
+
"step": 95000
|
1331 |
+
},
|
1332 |
+
{
|
1333 |
+
"loss": 3.1406875,
|
1334 |
+
"learning_rate": 6.6476612434631034e-06,
|
1335 |
+
"epoch": 0.8670467751307379,
|
1336 |
+
"total_flos": 292625791583781888,
|
1337 |
+
"step": 95500
|
1338 |
+
},
|
1339 |
+
{
|
1340 |
+
"loss": 3.1599375,
|
1341 |
+
"learning_rate": 6.420685647879141e-06,
|
1342 |
+
"epoch": 0.8715862870424173,
|
1343 |
+
"total_flos": 294158114986300416,
|
1344 |
+
"step": 96000
|
1345 |
+
},
|
1346 |
+
{
|
1347 |
+
"loss": 3.1481875,
|
1348 |
+
"learning_rate": 6.1937100522951776e-06,
|
1349 |
+
"epoch": 0.8761257989540965,
|
1350 |
+
"total_flos": 295691833246583808,
|
1351 |
+
"step": 96500
|
1352 |
+
},
|
1353 |
+
{
|
1354 |
+
"loss": 3.13575,
|
1355 |
+
"learning_rate": 5.966734456711215e-06,
|
1356 |
+
"epoch": 0.8806653108657757,
|
1357 |
+
"total_flos": 297225791999585280,
|
1358 |
+
"step": 97000
|
1359 |
+
},
|
1360 |
+
{
|
1361 |
+
"loss": 3.1423125,
|
1362 |
+
"learning_rate": 5.739758861127252e-06,
|
1363 |
+
"epoch": 0.8852048227774549,
|
1364 |
+
"total_flos": 298758620436811776,
|
1365 |
+
"step": 97500
|
1366 |
+
},
|
1367 |
+
{
|
1368 |
+
"loss": 3.157625,
|
1369 |
+
"learning_rate": 5.512783265543289e-06,
|
1370 |
+
"epoch": 0.8897443346891343,
|
1371 |
+
"total_flos": 300291689366756352,
|
1372 |
+
"step": 98000
|
1373 |
+
},
|
1374 |
+
{
|
1375 |
+
"loss": 3.126125,
|
1376 |
+
"learning_rate": 5.285807669959327e-06,
|
1377 |
+
"epoch": 0.8942838466008135,
|
1378 |
+
"total_flos": 301824734247429120,
|
1379 |
+
"step": 98500
|
1380 |
+
},
|
1381 |
+
{
|
1382 |
+
"loss": 3.1455625,
|
1383 |
+
"learning_rate": 5.058832074375363e-06,
|
1384 |
+
"epoch": 0.8988233585124927,
|
1385 |
+
"total_flos": 303358115817907200,
|
1386 |
+
"step": 99000
|
1387 |
+
},
|
1388 |
+
{
|
1389 |
+
"loss": 3.1365625,
|
1390 |
+
"learning_rate": 4.831856478791401e-06,
|
1391 |
+
"epoch": 0.903362870424172,
|
1392 |
+
"total_flos": 304890920205861888,
|
1393 |
+
"step": 99500
|
1394 |
+
},
|
1395 |
+
{
|
1396 |
+
"loss": 3.131625,
|
1397 |
+
"learning_rate": 4.604880883207438e-06,
|
1398 |
+
"epoch": 0.9079023823358513,
|
1399 |
+
"total_flos": 306422858820031488,
|
1400 |
+
"step": 100000
|
1401 |
+
},
|
1402 |
+
{
|
1403 |
+
"loss": 3.15225,
|
1404 |
+
"learning_rate": 4.377905287623475e-06,
|
1405 |
+
"epoch": 0.9124418942475305,
|
1406 |
+
"total_flos": 307953498773523456,
|
1407 |
+
"step": 100500
|
1408 |
+
},
|
1409 |
+
{
|
1410 |
+
"loss": 3.145,
|
1411 |
+
"learning_rate": 4.1509296920395125e-06,
|
1412 |
+
"epoch": 0.9169814061592098,
|
1413 |
+
"total_flos": 309485365239877632,
|
1414 |
+
"step": 101000
|
1415 |
+
},
|
1416 |
+
{
|
1417 |
+
"loss": 3.1263125,
|
1418 |
+
"learning_rate": 3.92395409645555e-06,
|
1419 |
+
"epoch": 0.921520918070889,
|
1420 |
+
"total_flos": 311017039312057344,
|
1421 |
+
"step": 101500
|
1422 |
+
},
|
1423 |
+
{
|
1424 |
+
"loss": 3.12225,
|
1425 |
+
"learning_rate": 3.696978500871586e-06,
|
1426 |
+
"epoch": 0.9260604299825683,
|
1427 |
+
"total_flos": 312548977926226944,
|
1428 |
+
"step": 102000
|
1429 |
+
},
|
1430 |
+
{
|
1431 |
+
"loss": 3.1128125,
|
1432 |
+
"learning_rate": 3.4700029052876232e-06,
|
1433 |
+
"epoch": 0.9305999418942476,
|
1434 |
+
"total_flos": 314081493722919936,
|
1435 |
+
"step": 102500
|
1436 |
+
},
|
1437 |
+
{
|
1438 |
+
"loss": 3.1099375,
|
1439 |
+
"learning_rate": 3.2430273097036603e-06,
|
1440 |
+
"epoch": 0.9351394538059268,
|
1441 |
+
"total_flos": 315613744977623040,
|
1442 |
+
"step": 103000
|
1443 |
+
},
|
1444 |
+
{
|
1445 |
+
"loss": 3.114,
|
1446 |
+
"learning_rate": 3.0160517141196982e-06,
|
1447 |
+
"epoch": 0.939678965717606,
|
1448 |
+
"total_flos": 317144096339853312,
|
1449 |
+
"step": 103500
|
1450 |
+
},
|
1451 |
+
{
|
1452 |
+
"loss": 3.1200625,
|
1453 |
+
"learning_rate": 2.7890761185357353e-06,
|
1454 |
+
"epoch": 0.9442184776292853,
|
1455 |
+
"total_flos": 318675169180237824,
|
1456 |
+
"step": 104000
|
1457 |
+
},
|
1458 |
+
{
|
1459 |
+
"loss": 3.104625,
|
1460 |
+
"learning_rate": 2.5621005229517724e-06,
|
1461 |
+
"epoch": 0.9487579895409646,
|
1462 |
+
"total_flos": 320208959588336640,
|
1463 |
+
"step": 104500
|
1464 |
+
},
|
1465 |
+
{
|
1466 |
+
"loss": 3.1181875,
|
1467 |
+
"learning_rate": 2.3351249273678094e-06,
|
1468 |
+
"epoch": 0.9532975014526438,
|
1469 |
+
"total_flos": 321739960280905728,
|
1470 |
+
"step": 105000
|
1471 |
+
},
|
1472 |
+
{
|
1473 |
+
"loss": 3.1073125,
|
1474 |
+
"learning_rate": 2.108149331783847e-06,
|
1475 |
+
"epoch": 0.957837013364323,
|
1476 |
+
"total_flos": 323273894984635392,
|
1477 |
+
"step": 105500
|
1478 |
+
},
|
1479 |
+
{
|
1480 |
+
"loss": 3.141375,
|
1481 |
+
"learning_rate": 1.881173736199884e-06,
|
1482 |
+
"epoch": 0.9623765252760024,
|
1483 |
+
"total_flos": 324804607085942784,
|
1484 |
+
"step": 106000
|
1485 |
+
},
|
1486 |
+
{
|
1487 |
+
"loss": 3.1269375,
|
1488 |
+
"learning_rate": 1.6541981406159213e-06,
|
1489 |
+
"epoch": 0.9669160371876816,
|
1490 |
+
"total_flos": 326337627917343744,
|
1491 |
+
"step": 106500
|
1492 |
+
},
|
1493 |
+
{
|
1494 |
+
"loss": 3.118,
|
1495 |
+
"learning_rate": 1.4272225450319581e-06,
|
1496 |
+
"epoch": 0.9714555490993608,
|
1497 |
+
"total_flos": 327871298079083520,
|
1498 |
+
"step": 107000
|
1499 |
+
},
|
1500 |
+
{
|
1501 |
+
"loss": 3.1175,
|
1502 |
+
"learning_rate": 1.2002469494479954e-06,
|
1503 |
+
"epoch": 0.97599506101104,
|
1504 |
+
"total_flos": 329401745638401024,
|
1505 |
+
"step": 107500
|
1506 |
+
},
|
1507 |
+
{
|
1508 |
+
"loss": 3.120375,
|
1509 |
+
"learning_rate": 9.732713538640327e-07,
|
1510 |
+
"epoch": 0.9805345729227194,
|
1511 |
+
"total_flos": 330934525977083904,
|
1512 |
+
"step": 108000
|
1513 |
+
},
|
1514 |
+
{
|
1515 |
+
"loss": 3.0920625,
|
1516 |
+
"learning_rate": 7.462957582800698e-07,
|
1517 |
+
"epoch": 0.9850740848343986,
|
1518 |
+
"total_flos": 332464155861159936,
|
1519 |
+
"step": 108500
|
1520 |
+
},
|
1521 |
+
{
|
1522 |
+
"loss": 3.0940625,
|
1523 |
+
"learning_rate": 5.19320162696107e-07,
|
1524 |
+
"epoch": 0.9896135967460779,
|
1525 |
+
"total_flos": 333995685637708800,
|
1526 |
+
"step": 109000
|
1527 |
+
},
|
1528 |
+
{
|
1529 |
+
"loss": 3.10975,
|
1530 |
+
"learning_rate": 2.923445671121441e-07,
|
1531 |
+
"epoch": 0.9941531086577571,
|
1532 |
+
"total_flos": 335529836784884736,
|
1533 |
+
"step": 109500
|
1534 |
+
},
|
1535 |
+
{
|
1536 |
+
"loss": 3.115375,
|
1537 |
+
"learning_rate": 6.536897152818129e-08,
|
1538 |
+
"epoch": 0.9986926205694364,
|
1539 |
+
"total_flos": 337063434798809088,
|
1540 |
+
"step": 110000
|
1541 |
+
}
|
1542 |
+
]
|