Upload sigma_lambda_v5/history.json with huggingface_hub
Browse files- sigma_lambda_v5/history.json +19 -7
sigma_lambda_v5/history.json
CHANGED
|
@@ -12,7 +12,8 @@
|
|
| 12 |
6,
|
| 13 |
7,
|
| 14 |
8,
|
| 15 |
-
9
|
|
|
|
| 16 |
],
|
| 17 |
"gen_loss": [
|
| 18 |
16.586591267879488,
|
|
@@ -27,7 +28,8 @@
|
|
| 27 |
116.53298210471324,
|
| 28 |
170.46740706801785,
|
| 29 |
214.5222631372081,
|
| 30 |
-
274.12421057906226
|
|
|
|
| 31 |
],
|
| 32 |
"ret_loss": [
|
| 33 |
0.0,
|
|
@@ -42,6 +44,7 @@
|
|
| 42 |
0.0,
|
| 43 |
0.0,
|
| 44 |
0.0,
|
|
|
|
| 45 |
0.0
|
| 46 |
],
|
| 47 |
"ppl": [
|
|
@@ -57,6 +60,7 @@
|
|
| 57 |
485165195.4097903,
|
| 58 |
485165195.4097903,
|
| 59 |
485165195.4097903,
|
|
|
|
| 60 |
485165195.4097903
|
| 61 |
],
|
| 62 |
"tf_acc": [
|
|
@@ -72,7 +76,8 @@
|
|
| 72 |
0.46131388743580864,
|
| 73 |
0.45428323030789747,
|
| 74 |
0.44858981632247086,
|
| 75 |
-
0.44045584150537476
|
|
|
|
| 76 |
],
|
| 77 |
"fr_acc": [
|
| 78 |
0.12302286153389094,
|
|
@@ -87,7 +92,8 @@
|
|
| 87 |
0.27369709538827186,
|
| 88 |
0.12652472119383884,
|
| 89 |
0.14813148452854336,
|
| 90 |
-
0.1719167780748663
|
|
|
|
| 91 |
],
|
| 92 |
"mrr": [
|
| 93 |
0.006059600971639156,
|
|
@@ -102,7 +108,8 @@
|
|
| 102 |
0.006052091252058744,
|
| 103 |
0.006054983474314213,
|
| 104 |
0.006042384542524815,
|
| 105 |
-
0.006041090935468674
|
|
|
|
| 106 |
],
|
| 107 |
"r1": [
|
| 108 |
0.0007812500116415322,
|
|
@@ -117,6 +124,7 @@
|
|
| 117 |
0.0007812500116415322,
|
| 118 |
0.0007812500116415322,
|
| 119 |
0.0007812500116415322,
|
|
|
|
| 120 |
0.0007812500116415322
|
| 121 |
],
|
| 122 |
"step_var": [
|
|
@@ -132,7 +140,8 @@
|
|
| 132 |
0.24468941986560822,
|
| 133 |
0.26731032133102417,
|
| 134 |
0.28763458132743835,
|
| 135 |
-
0.3055805563926697
|
|
|
|
| 136 |
],
|
| 137 |
"pos_slope": [
|
| 138 |
0.4985616207122803,
|
|
@@ -147,7 +156,8 @@
|
|
| 147 |
0.4812909960746765,
|
| 148 |
0.4783017337322235,
|
| 149 |
0.47553473711013794,
|
| 150 |
-
0.47306424379348755
|
|
|
|
| 151 |
],
|
| 152 |
"grad_norm": [
|
| 153 |
NaN,
|
|
@@ -162,6 +172,7 @@
|
|
| 162 |
NaN,
|
| 163 |
NaN,
|
| 164 |
Infinity,
|
|
|
|
| 165 |
Infinity
|
| 166 |
],
|
| 167 |
"w_ret": [
|
|
@@ -177,6 +188,7 @@
|
|
| 177 |
0.03,
|
| 178 |
0.04,
|
| 179 |
0.05,
|
|
|
|
| 180 |
0.08
|
| 181 |
]
|
| 182 |
}
|
|
|
|
| 12 |
6,
|
| 13 |
7,
|
| 14 |
8,
|
| 15 |
+
9,
|
| 16 |
+
10
|
| 17 |
],
|
| 18 |
"gen_loss": [
|
| 19 |
16.586591267879488,
|
|
|
|
| 28 |
116.53298210471324,
|
| 29 |
170.46740706801785,
|
| 30 |
214.5222631372081,
|
| 31 |
+
274.12421057906226,
|
| 32 |
+
323.5490314494401
|
| 33 |
],
|
| 34 |
"ret_loss": [
|
| 35 |
0.0,
|
|
|
|
| 44 |
0.0,
|
| 45 |
0.0,
|
| 46 |
0.0,
|
| 47 |
+
0.0,
|
| 48 |
0.0
|
| 49 |
],
|
| 50 |
"ppl": [
|
|
|
|
| 60 |
485165195.4097903,
|
| 61 |
485165195.4097903,
|
| 62 |
485165195.4097903,
|
| 63 |
+
485165195.4097903,
|
| 64 |
485165195.4097903
|
| 65 |
],
|
| 66 |
"tf_acc": [
|
|
|
|
| 76 |
0.46131388743580864,
|
| 77 |
0.45428323030789747,
|
| 78 |
0.44858981632247086,
|
| 79 |
+
0.44045584150537476,
|
| 80 |
+
0.41316392464284923
|
| 81 |
],
|
| 82 |
"fr_acc": [
|
| 83 |
0.12302286153389094,
|
|
|
|
| 92 |
0.27369709538827186,
|
| 93 |
0.12652472119383884,
|
| 94 |
0.14813148452854336,
|
| 95 |
+
0.1719167780748663,
|
| 96 |
+
0.1009754644864939
|
| 97 |
],
|
| 98 |
"mrr": [
|
| 99 |
0.006059600971639156,
|
|
|
|
| 108 |
0.006052091252058744,
|
| 109 |
0.006054983474314213,
|
| 110 |
0.006042384542524815,
|
| 111 |
+
0.006041090935468674,
|
| 112 |
+
0.006040983833372593
|
| 113 |
],
|
| 114 |
"r1": [
|
| 115 |
0.0007812500116415322,
|
|
|
|
| 124 |
0.0007812500116415322,
|
| 125 |
0.0007812500116415322,
|
| 126 |
0.0007812500116415322,
|
| 127 |
+
0.0007812500116415322,
|
| 128 |
0.0007812500116415322
|
| 129 |
],
|
| 130 |
"step_var": [
|
|
|
|
| 140 |
0.24468941986560822,
|
| 141 |
0.26731032133102417,
|
| 142 |
0.28763458132743835,
|
| 143 |
+
0.3055805563926697,
|
| 144 |
+
0.32141008973121643
|
| 145 |
],
|
| 146 |
"pos_slope": [
|
| 147 |
0.4985616207122803,
|
|
|
|
| 156 |
0.4812909960746765,
|
| 157 |
0.4783017337322235,
|
| 158 |
0.47553473711013794,
|
| 159 |
+
0.47306424379348755,
|
| 160 |
+
0.47092458605766296
|
| 161 |
],
|
| 162 |
"grad_norm": [
|
| 163 |
NaN,
|
|
|
|
| 172 |
NaN,
|
| 173 |
NaN,
|
| 174 |
Infinity,
|
| 175 |
+
Infinity,
|
| 176 |
Infinity
|
| 177 |
],
|
| 178 |
"w_ret": [
|
|
|
|
| 188 |
0.03,
|
| 189 |
0.04,
|
| 190 |
0.05,
|
| 191 |
+
0.08,
|
| 192 |
0.08
|
| 193 |
]
|
| 194 |
}
|