Upload sigma_lambda_v5/history.json with huggingface_hub
Browse files- sigma_lambda_v5/history.json +20 -8
sigma_lambda_v5/history.json
CHANGED
|
@@ -9,7 +9,8 @@
|
|
| 9 |
7,
|
| 10 |
4,
|
| 11 |
5,
|
| 12 |
-
6
|
|
|
|
| 13 |
],
|
| 14 |
"gen_loss": [
|
| 15 |
16.586591267879488,
|
|
@@ -21,7 +22,8 @@
|
|
| 21 |
465.32868966012995,
|
| 22 |
48.11255316601957,
|
| 23 |
74.47382958961273,
|
| 24 |
-
116.53298210471324
|
|
|
|
| 25 |
],
|
| 26 |
"ret_loss": [
|
| 27 |
0.0,
|
|
@@ -33,6 +35,7 @@
|
|
| 33 |
0.0,
|
| 34 |
0.0,
|
| 35 |
0.0,
|
|
|
|
| 36 |
0.0
|
| 37 |
],
|
| 38 |
"ppl": [
|
|
@@ -45,6 +48,7 @@
|
|
| 45 |
485165195.4097903,
|
| 46 |
15048.086253037915,
|
| 47 |
1624266.5983328498,
|
|
|
|
| 48 |
485165195.4097903
|
| 49 |
],
|
| 50 |
"tf_acc": [
|
|
@@ -57,7 +61,8 @@
|
|
| 57 |
0.4291537424388662,
|
| 58 |
0.4800979171424993,
|
| 59 |
0.4732017211019178,
|
| 60 |
-
0.46131388743580864
|
|
|
|
| 61 |
],
|
| 62 |
"fr_acc": [
|
| 63 |
0.12302286153389094,
|
|
@@ -69,7 +74,8 @@
|
|
| 69 |
0.04862503713606654,
|
| 70 |
0.2002858762969057,
|
| 71 |
0.24578219982631747,
|
| 72 |
-
0.27369709538827186
|
|
|
|
| 73 |
],
|
| 74 |
"mrr": [
|
| 75 |
0.006059600971639156,
|
|
@@ -81,7 +87,8 @@
|
|
| 81 |
0.006041784770786762,
|
| 82 |
0.0060499003157019615,
|
| 83 |
0.006042419467121363,
|
| 84 |
-
0.006052091252058744
|
|
|
|
| 85 |
],
|
| 86 |
"r1": [
|
| 87 |
0.0007812500116415322,
|
|
@@ -93,6 +100,7 @@
|
|
| 93 |
0.0007812500116415322,
|
| 94 |
0.0007812500116415322,
|
| 95 |
0.0007812500116415322,
|
|
|
|
| 96 |
0.0007812500116415322
|
| 97 |
],
|
| 98 |
"step_var": [
|
|
@@ -105,7 +113,8 @@
|
|
| 105 |
0.34328973293304443,
|
| 106 |
0.20889317989349365,
|
| 107 |
0.22339501976966858,
|
| 108 |
-
0.24468941986560822
|
|
|
|
| 109 |
],
|
| 110 |
"pos_slope": [
|
| 111 |
0.4985616207122803,
|
|
@@ -117,7 +126,8 @@
|
|
| 117 |
0.46813833713531494,
|
| 118 |
0.48697665333747864,
|
| 119 |
0.48436233401298523,
|
| 120 |
-
0.4812909960746765
|
|
|
|
| 121 |
],
|
| 122 |
"grad_norm": [
|
| 123 |
NaN,
|
|
@@ -129,6 +139,7 @@
|
|
| 129 |
NaN,
|
| 130 |
NaN,
|
| 131 |
NaN,
|
|
|
|
| 132 |
NaN
|
| 133 |
],
|
| 134 |
"w_ret": [
|
|
@@ -141,6 +152,7 @@
|
|
| 141 |
0.04,
|
| 142 |
0.01,
|
| 143 |
0.02,
|
| 144 |
-
0.03
|
|
|
|
| 145 |
]
|
| 146 |
}
|
|
|
|
| 9 |
7,
|
| 10 |
4,
|
| 11 |
5,
|
| 12 |
+
6,
|
| 13 |
+
7
|
| 14 |
],
|
| 15 |
"gen_loss": [
|
| 16 |
16.586591267879488,
|
|
|
|
| 22 |
465.32868966012995,
|
| 23 |
48.11255316601957,
|
| 24 |
74.47382958961273,
|
| 25 |
+
116.53298210471324,
|
| 26 |
+
170.46740706801785
|
| 27 |
],
|
| 28 |
"ret_loss": [
|
| 29 |
0.0,
|
|
|
|
| 35 |
0.0,
|
| 36 |
0.0,
|
| 37 |
0.0,
|
| 38 |
+
0.0,
|
| 39 |
0.0
|
| 40 |
],
|
| 41 |
"ppl": [
|
|
|
|
| 48 |
485165195.4097903,
|
| 49 |
15048.086253037915,
|
| 50 |
1624266.5983328498,
|
| 51 |
+
485165195.4097903,
|
| 52 |
485165195.4097903
|
| 53 |
],
|
| 54 |
"tf_acc": [
|
|
|
|
| 61 |
0.4291537424388662,
|
| 62 |
0.4800979171424993,
|
| 63 |
0.4732017211019178,
|
| 64 |
+
0.46131388743580864,
|
| 65 |
+
0.45428323030789747
|
| 66 |
],
|
| 67 |
"fr_acc": [
|
| 68 |
0.12302286153389094,
|
|
|
|
| 74 |
0.04862503713606654,
|
| 75 |
0.2002858762969057,
|
| 76 |
0.24578219982631747,
|
| 77 |
+
0.27369709538827186,
|
| 78 |
+
0.12652472119383884
|
| 79 |
],
|
| 80 |
"mrr": [
|
| 81 |
0.006059600971639156,
|
|
|
|
| 87 |
0.006041784770786762,
|
| 88 |
0.0060499003157019615,
|
| 89 |
0.006042419467121363,
|
| 90 |
+
0.006052091252058744,
|
| 91 |
+
0.006054983474314213
|
| 92 |
],
|
| 93 |
"r1": [
|
| 94 |
0.0007812500116415322,
|
|
|
|
| 100 |
0.0007812500116415322,
|
| 101 |
0.0007812500116415322,
|
| 102 |
0.0007812500116415322,
|
| 103 |
+
0.0007812500116415322,
|
| 104 |
0.0007812500116415322
|
| 105 |
],
|
| 106 |
"step_var": [
|
|
|
|
| 113 |
0.34328973293304443,
|
| 114 |
0.20889317989349365,
|
| 115 |
0.22339501976966858,
|
| 116 |
+
0.24468941986560822,
|
| 117 |
+
0.26731032133102417
|
| 118 |
],
|
| 119 |
"pos_slope": [
|
| 120 |
0.4985616207122803,
|
|
|
|
| 126 |
0.46813833713531494,
|
| 127 |
0.48697665333747864,
|
| 128 |
0.48436233401298523,
|
| 129 |
+
0.4812909960746765,
|
| 130 |
+
0.4783017337322235
|
| 131 |
],
|
| 132 |
"grad_norm": [
|
| 133 |
NaN,
|
|
|
|
| 139 |
NaN,
|
| 140 |
NaN,
|
| 141 |
NaN,
|
| 142 |
+
NaN,
|
| 143 |
NaN
|
| 144 |
],
|
| 145 |
"w_ret": [
|
|
|
|
| 152 |
0.04,
|
| 153 |
0.01,
|
| 154 |
0.02,
|
| 155 |
+
0.03,
|
| 156 |
+
0.04
|
| 157 |
]
|
| 158 |
}
|