Upload sigma_lambda_proto_v3/history.json with huggingface_hub
Browse files
sigma_lambda_proto_v3/history.json
CHANGED
|
@@ -7,7 +7,8 @@
|
|
| 7 |
5,
|
| 8 |
6,
|
| 9 |
7,
|
| 10 |
-
8
|
|
|
|
| 11 |
],
|
| 12 |
"gen_loss": [
|
| 13 |
18.370433433850607,
|
|
@@ -17,7 +18,8 @@
|
|
| 17 |
11.67701819376512,
|
| 18 |
11.95506536512664,
|
| 19 |
11.651441534504746,
|
| 20 |
-
11.484277419465961
|
|
|
|
| 21 |
],
|
| 22 |
"ret_loss": [
|
| 23 |
0.0,
|
|
@@ -27,6 +29,7 @@
|
|
| 27 |
0.0,
|
| 28 |
0.0,
|
| 29 |
0.0,
|
|
|
|
| 30 |
0.0
|
| 31 |
],
|
| 32 |
"ent_loss": [
|
|
@@ -37,7 +40,8 @@
|
|
| 37 |
0.6095350569667238,
|
| 38 |
0.6105213370467677,
|
| 39 |
0.6074923021143133,
|
| 40 |
-
0.606140976638505
|
|
|
|
| 41 |
],
|
| 42 |
"ppl": [
|
| 43 |
40729.60576844635,
|
|
@@ -47,7 +51,8 @@
|
|
| 47 |
81.4022446898097,
|
| 48 |
73.53574793147658,
|
| 49 |
66.62547611610893,
|
| 50 |
-
61.79489383359564
|
|
|
|
| 51 |
],
|
| 52 |
"tf_acc": [
|
| 53 |
9.249682110189831e-06,
|
|
@@ -57,7 +62,8 @@
|
|
| 57 |
0.4514476024551921,
|
| 58 |
0.4529803291907453,
|
| 59 |
0.4559930050639289,
|
| 60 |
-
0.45953291281739334
|
|
|
|
| 61 |
],
|
| 62 |
"fr_acc": [
|
| 63 |
0.0,
|
|
@@ -67,7 +73,8 @@
|
|
| 67 |
0.1671656922337565,
|
| 68 |
0.16827387097913035,
|
| 69 |
0.1727579081837558,
|
| 70 |
-
0.18450658056400712
|
|
|
|
| 71 |
],
|
| 72 |
"def_rate": [
|
| 73 |
0.0,
|
|
@@ -77,6 +84,7 @@
|
|
| 77 |
1.0,
|
| 78 |
1.0,
|
| 79 |
1.0,
|
|
|
|
| 80 |
1.0
|
| 81 |
],
|
| 82 |
"mrr": [
|
|
@@ -87,7 +95,8 @@
|
|
| 87 |
0.004973841365426779,
|
| 88 |
0.004973897710442543,
|
| 89 |
0.004973262082785368,
|
| 90 |
-
0.004975350573658943
|
|
|
|
| 91 |
],
|
| 92 |
"r1": [
|
| 93 |
0.0006249999860301614,
|
|
@@ -97,6 +106,7 @@
|
|
| 97 |
0.0006249999860301614,
|
| 98 |
0.0006249999860301614,
|
| 99 |
0.0006249999860301614,
|
|
|
|
| 100 |
0.0006249999860301614
|
| 101 |
],
|
| 102 |
"ret_ok_pct": [
|
|
@@ -107,6 +117,7 @@
|
|
| 107 |
0.0,
|
| 108 |
0.0,
|
| 109 |
0.0,
|
|
|
|
| 110 |
0.0
|
| 111 |
],
|
| 112 |
"grad_norm": [
|
|
@@ -117,7 +128,8 @@
|
|
| 117 |
1152.3023670419057,
|
| 118 |
2138.4472111783607,
|
| 119 |
1115.3582513533938,
|
| 120 |
-
1271.3913563283284
|
|
|
|
| 121 |
],
|
| 122 |
"step_var": [
|
| 123 |
0.22542811930179596,
|
|
@@ -127,7 +139,8 @@
|
|
| 127 |
0.23098886013031006,
|
| 128 |
0.23129060864448547,
|
| 129 |
0.23215171694755554,
|
| 130 |
-
0.23333927989006042
|
|
|
|
| 131 |
],
|
| 132 |
"pos_slope": [
|
| 133 |
0.49990350008010864,
|
|
@@ -137,7 +150,8 @@
|
|
| 137 |
0.4983888268470764,
|
| 138 |
0.4982922673225403,
|
| 139 |
0.49802443385124207,
|
| 140 |
-
0.49764952063560486
|
|
|
|
| 141 |
],
|
| 142 |
"beta_mean": [
|
| 143 |
0.09998118132352829,
|
|
@@ -147,7 +161,8 @@
|
|
| 147 |
0.09967698156833649,
|
| 148 |
0.0996585488319397,
|
| 149 |
0.09960417449474335,
|
| 150 |
-
0.09952875226736069
|
|
|
|
| 151 |
],
|
| 152 |
"w_ret": [
|
| 153 |
0.0,
|
|
@@ -157,6 +172,7 @@
|
|
| 157 |
0.03,
|
| 158 |
0.06,
|
| 159 |
0.09,
|
| 160 |
-
0.12
|
|
|
|
| 161 |
]
|
| 162 |
}
|
|
|
|
| 7 |
5,
|
| 8 |
6,
|
| 9 |
7,
|
| 10 |
+
8,
|
| 11 |
+
9
|
| 12 |
],
|
| 13 |
"gen_loss": [
|
| 14 |
18.370433433850607,
|
|
|
|
| 18 |
11.67701819376512,
|
| 19 |
11.95506536512664,
|
| 20 |
11.651441534504746,
|
| 21 |
+
11.484277419465961,
|
| 22 |
+
11.311316739862615
|
| 23 |
],
|
| 24 |
"ret_loss": [
|
| 25 |
0.0,
|
|
|
|
| 29 |
0.0,
|
| 30 |
0.0,
|
| 31 |
0.0,
|
| 32 |
+
0.0,
|
| 33 |
0.0
|
| 34 |
],
|
| 35 |
"ent_loss": [
|
|
|
|
| 40 |
0.6095350569667238,
|
| 41 |
0.6105213370467677,
|
| 42 |
0.6074923021143133,
|
| 43 |
+
0.606140976638505,
|
| 44 |
+
0.6052777969114708
|
| 45 |
],
|
| 46 |
"ppl": [
|
| 47 |
40729.60576844635,
|
|
|
|
| 51 |
81.4022446898097,
|
| 52 |
73.53574793147658,
|
| 53 |
66.62547611610893,
|
| 54 |
+
61.79489383359564,
|
| 55 |
+
57.91288486607438
|
| 56 |
],
|
| 57 |
"tf_acc": [
|
| 58 |
9.249682110189831e-06,
|
|
|
|
| 62 |
0.4514476024551921,
|
| 63 |
0.4529803291907453,
|
| 64 |
0.4559930050639289,
|
| 65 |
+
0.45953291281739334,
|
| 66 |
+
0.46299827901502855
|
| 67 |
],
|
| 68 |
"fr_acc": [
|
| 69 |
0.0,
|
|
|
|
| 73 |
0.1671656922337565,
|
| 74 |
0.16827387097913035,
|
| 75 |
0.1727579081837558,
|
| 76 |
+
0.18450658056400712,
|
| 77 |
+
0.19845598231894296
|
| 78 |
],
|
| 79 |
"def_rate": [
|
| 80 |
0.0,
|
|
|
|
| 84 |
1.0,
|
| 85 |
1.0,
|
| 86 |
1.0,
|
| 87 |
+
1.0,
|
| 88 |
1.0
|
| 89 |
],
|
| 90 |
"mrr": [
|
|
|
|
| 95 |
0.004973841365426779,
|
| 96 |
0.004973897710442543,
|
| 97 |
0.004973262082785368,
|
| 98 |
+
0.004975350573658943,
|
| 99 |
+
0.004973086062818766
|
| 100 |
],
|
| 101 |
"r1": [
|
| 102 |
0.0006249999860301614,
|
|
|
|
| 106 |
0.0006249999860301614,
|
| 107 |
0.0006249999860301614,
|
| 108 |
0.0006249999860301614,
|
| 109 |
+
0.0006249999860301614,
|
| 110 |
0.0006249999860301614
|
| 111 |
],
|
| 112 |
"ret_ok_pct": [
|
|
|
|
| 117 |
0.0,
|
| 118 |
0.0,
|
| 119 |
0.0,
|
| 120 |
+
0.0,
|
| 121 |
0.0
|
| 122 |
],
|
| 123 |
"grad_norm": [
|
|
|
|
| 128 |
1152.3023670419057,
|
| 129 |
2138.4472111783607,
|
| 130 |
1115.3582513533938,
|
| 131 |
+
1271.3913563283284,
|
| 132 |
+
954.6965318550485
|
| 133 |
],
|
| 134 |
"step_var": [
|
| 135 |
0.22542811930179596,
|
|
|
|
| 139 |
0.23098886013031006,
|
| 140 |
0.23129060864448547,
|
| 141 |
0.23215171694755554,
|
| 142 |
+
0.23333927989006042,
|
| 143 |
+
0.23468747735023499
|
| 144 |
],
|
| 145 |
"pos_slope": [
|
| 146 |
0.49990350008010864,
|
|
|
|
| 150 |
0.4983888268470764,
|
| 151 |
0.4982922673225403,
|
| 152 |
0.49802443385124207,
|
| 153 |
+
0.49764952063560486,
|
| 154 |
+
0.49722111225128174
|
| 155 |
],
|
| 156 |
"beta_mean": [
|
| 157 |
0.09998118132352829,
|
|
|
|
| 161 |
0.09967698156833649,
|
| 162 |
0.0996585488319397,
|
| 163 |
0.09960417449474335,
|
| 164 |
+
0.09952875226736069,
|
| 165 |
+
0.0994425043463707
|
| 166 |
],
|
| 167 |
"w_ret": [
|
| 168 |
0.0,
|
|
|
|
| 172 |
0.03,
|
| 173 |
0.06,
|
| 174 |
0.09,
|
| 175 |
+
0.12,
|
| 176 |
+
0.15
|
| 177 |
]
|
| 178 |
}
|