Upload sigma_lambda_v6/history.json with huggingface_hub
Browse files- sigma_lambda_v6/history.json +22 -10
sigma_lambda_v6/history.json
CHANGED
|
@@ -9,7 +9,8 @@
|
|
| 9 |
7,
|
| 10 |
8,
|
| 11 |
5,
|
| 12 |
-
6
|
|
|
|
| 13 |
],
|
| 14 |
"gen_loss": [
|
| 15 |
16.730865429819772,
|
|
@@ -21,7 +22,8 @@
|
|
| 21 |
18.599128044941526,
|
| 22 |
20.198475095520426,
|
| 23 |
11.820103142087035,
|
| 24 |
-
13.183105084420808
|
|
|
|
| 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": [
|
|
@@ -45,7 +48,8 @@
|
|
| 45 |
61.441551617998655,
|
| 46 |
74.18379990559752,
|
| 47 |
32.02608510289743,
|
| 48 |
-
35.48630770546718
|
|
|
|
| 49 |
],
|
| 50 |
"tf_acc": [
|
| 51 |
0.4522865752576864,
|
|
@@ -57,7 +61,8 @@
|
|
| 57 |
0.49550384227161876,
|
| 58 |
0.49492731733553386,
|
| 59 |
0.49543937462700866,
|
| 60 |
-
0.49583907402359145
|
|
|
|
| 61 |
],
|
| 62 |
"fr_acc": [
|
| 63 |
0.12302286153389094,
|
|
@@ -69,7 +74,8 @@
|
|
| 69 |
0.25665842531651356,
|
| 70 |
0.2002858762969057,
|
| 71 |
0.21672786976095798,
|
| 72 |
-
0.22766301304904246
|
|
|
|
| 73 |
],
|
| 74 |
"mrr": [
|
| 75 |
0.006079697050154209,
|
|
@@ -81,7 +87,8 @@
|
|
| 81 |
0.006045354530215263,
|
| 82 |
0.006043205503374338,
|
| 83 |
0.006071251351386309,
|
| 84 |
-
0.006076182704418898
|
|
|
|
| 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 |
0.19729147851467133,
|
| 106 |
0.20193856954574585,
|
| 107 |
0.17992281913757324,
|
| 108 |
-
0.1854308843612671
|
|
|
|
| 109 |
],
|
| 110 |
"pos_slope": [
|
| 111 |
0.49914172291755676,
|
|
@@ -117,7 +126,8 @@
|
|
| 117 |
0.4844416379928589,
|
| 118 |
0.48276612162590027,
|
| 119 |
0.49052613973617554,
|
| 120 |
-
0.48830142617225647
|
|
|
|
| 121 |
],
|
| 122 |
"grad_norm": [
|
| 123 |
NaN,
|
|
@@ -129,7 +139,8 @@
|
|
| 129 |
Infinity,
|
| 130 |
NaN,
|
| 131 |
NaN,
|
| 132 |
-
NaN
|
|
|
|
| 133 |
],
|
| 134 |
"w_ret": [
|
| 135 |
0.0,
|
|
@@ -141,6 +152,7 @@
|
|
| 141 |
0.04,
|
| 142 |
0.05,
|
| 143 |
0.0,
|
| 144 |
-
0.005
|
|
|
|
| 145 |
]
|
| 146 |
}
|
|
|
|
| 9 |
7,
|
| 10 |
8,
|
| 11 |
5,
|
| 12 |
+
6,
|
| 13 |
+
7
|
| 14 |
],
|
| 15 |
"gen_loss": [
|
| 16 |
16.730865429819772,
|
|
|
|
| 22 |
18.599128044941526,
|
| 23 |
20.198475095520426,
|
| 24 |
11.820103142087035,
|
| 25 |
+
13.183105084420808,
|
| 26 |
+
14.745438242800887
|
| 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 |
61.441551617998655,
|
| 49 |
74.18379990559752,
|
| 50 |
32.02608510289743,
|
| 51 |
+
35.48630770546718,
|
| 52 |
+
39.99732084159513
|
| 53 |
],
|
| 54 |
"tf_acc": [
|
| 55 |
0.4522865752576864,
|
|
|
|
| 61 |
0.49550384227161876,
|
| 62 |
0.49492731733553386,
|
| 63 |
0.49543937462700866,
|
| 64 |
+
0.49583907402359145,
|
| 65 |
+
0.49596432544740543
|
| 66 |
],
|
| 67 |
"fr_acc": [
|
| 68 |
0.12302286153389094,
|
|
|
|
| 74 |
0.25665842531651356,
|
| 75 |
0.2002858762969057,
|
| 76 |
0.21672786976095798,
|
| 77 |
+
0.22766301304904246,
|
| 78 |
+
0.24383791249600073
|
| 79 |
],
|
| 80 |
"mrr": [
|
| 81 |
0.006079697050154209,
|
|
|
|
| 87 |
0.006045354530215263,
|
| 88 |
0.006043205503374338,
|
| 89 |
0.006071251351386309,
|
| 90 |
+
0.006076182704418898,
|
| 91 |
+
0.005990722682327032
|
| 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 |
0.19729147851467133,
|
| 114 |
0.20193856954574585,
|
| 115 |
0.17992281913757324,
|
| 116 |
+
0.1854308843612671,
|
| 117 |
+
0.19223669171333313
|
| 118 |
],
|
| 119 |
"pos_slope": [
|
| 120 |
0.49914172291755676,
|
|
|
|
| 126 |
0.4844416379928589,
|
| 127 |
0.48276612162590027,
|
| 128 |
0.49052613973617554,
|
| 129 |
+
0.48830142617225647,
|
| 130 |
+
0.4855189025402069
|
| 131 |
],
|
| 132 |
"grad_norm": [
|
| 133 |
NaN,
|
|
|
|
| 139 |
Infinity,
|
| 140 |
NaN,
|
| 141 |
NaN,
|
| 142 |
+
NaN,
|
| 143 |
+
25002.312624738646
|
| 144 |
],
|
| 145 |
"w_ret": [
|
| 146 |
0.0,
|
|
|
|
| 152 |
0.04,
|
| 153 |
0.05,
|
| 154 |
0.0,
|
| 155 |
+
0.005,
|
| 156 |
+
0.01
|
| 157 |
]
|
| 158 |
}
|