Upload sigma_lambda_v5/history.json with huggingface_hub
Browse files- sigma_lambda_v5/history.json +21 -9
sigma_lambda_v5/history.json
CHANGED
|
@@ -7,7 +7,8 @@
|
|
| 7 |
5,
|
| 8 |
6,
|
| 9 |
7,
|
| 10 |
-
4
|
|
|
|
| 11 |
],
|
| 12 |
"gen_loss": [
|
| 13 |
16.586591267879488,
|
|
@@ -17,7 +18,8 @@
|
|
| 17 |
205.98175486637822,
|
| 18 |
333.0307641626024,
|
| 19 |
465.32868966012995,
|
| 20 |
-
48.11255316601957
|
|
|
|
| 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 |
"ppl": [
|
|
@@ -37,7 +40,8 @@
|
|
| 37 |
485165195.4097903,
|
| 38 |
485165195.4097903,
|
| 39 |
485165195.4097903,
|
| 40 |
-
15048.086253037915
|
|
|
|
| 41 |
],
|
| 42 |
"tf_acc": [
|
| 43 |
0.46026214386231185,
|
|
@@ -47,7 +51,8 @@
|
|
| 47 |
0.4302312730702071,
|
| 48 |
0.4411152534131013,
|
| 49 |
0.4291537424388662,
|
| 50 |
-
0.4800979171424993
|
|
|
|
| 51 |
],
|
| 52 |
"fr_acc": [
|
| 53 |
0.12302286153389094,
|
|
@@ -57,7 +62,8 @@
|
|
| 57 |
0.13397800116550115,
|
| 58 |
0.1528195558526441,
|
| 59 |
0.04862503713606654,
|
| 60 |
-
0.2002858762969057
|
|
|
|
| 61 |
],
|
| 62 |
"mrr": [
|
| 63 |
0.006059600971639156,
|
|
@@ -67,7 +73,8 @@
|
|
| 67 |
0.006043311208486557,
|
| 68 |
0.0060408106073737144,
|
| 69 |
0.006041784770786762,
|
| 70 |
-
0.0060499003157019615
|
|
|
|
| 71 |
],
|
| 72 |
"r1": [
|
| 73 |
0.0007812500116415322,
|
|
@@ -77,6 +84,7 @@
|
|
| 77 |
0.0007812500116415322,
|
| 78 |
0.0007812500116415322,
|
| 79 |
0.0007812500116415322,
|
|
|
|
| 80 |
0.0007812500116415322
|
| 81 |
],
|
| 82 |
"step_var": [
|
|
@@ -87,7 +95,8 @@
|
|
| 87 |
0.2775871157646179,
|
| 88 |
0.312656432390213,
|
| 89 |
0.34328973293304443,
|
| 90 |
-
0.20889317989349365
|
|
|
|
| 91 |
],
|
| 92 |
"pos_slope": [
|
| 93 |
0.4985616207122803,
|
|
@@ -97,7 +106,8 @@
|
|
| 97 |
0.47680673003196716,
|
| 98 |
0.47214341163635254,
|
| 99 |
0.46813833713531494,
|
| 100 |
-
0.48697665333747864
|
|
|
|
| 101 |
],
|
| 102 |
"grad_norm": [
|
| 103 |
NaN,
|
|
@@ -107,6 +117,7 @@
|
|
| 107 |
NaN,
|
| 108 |
NaN,
|
| 109 |
NaN,
|
|
|
|
| 110 |
NaN
|
| 111 |
],
|
| 112 |
"w_ret": [
|
|
@@ -117,6 +128,7 @@
|
|
| 117 |
0.02,
|
| 118 |
0.03,
|
| 119 |
0.04,
|
| 120 |
-
0.01
|
|
|
|
| 121 |
]
|
| 122 |
}
|
|
|
|
| 7 |
5,
|
| 8 |
6,
|
| 9 |
7,
|
| 10 |
+
4,
|
| 11 |
+
5
|
| 12 |
],
|
| 13 |
"gen_loss": [
|
| 14 |
16.586591267879488,
|
|
|
|
| 18 |
205.98175486637822,
|
| 19 |
333.0307641626024,
|
| 20 |
465.32868966012995,
|
| 21 |
+
48.11255316601957,
|
| 22 |
+
74.47382958961273
|
| 23 |
],
|
| 24 |
"ret_loss": [
|
| 25 |
0.0,
|
|
|
|
| 29 |
0.0,
|
| 30 |
0.0,
|
| 31 |
0.0,
|
| 32 |
+
0.0,
|
| 33 |
0.0
|
| 34 |
],
|
| 35 |
"ppl": [
|
|
|
|
| 40 |
485165195.4097903,
|
| 41 |
485165195.4097903,
|
| 42 |
485165195.4097903,
|
| 43 |
+
15048.086253037915,
|
| 44 |
+
1624266.5983328498
|
| 45 |
],
|
| 46 |
"tf_acc": [
|
| 47 |
0.46026214386231185,
|
|
|
|
| 51 |
0.4302312730702071,
|
| 52 |
0.4411152534131013,
|
| 53 |
0.4291537424388662,
|
| 54 |
+
0.4800979171424993,
|
| 55 |
+
0.4732017211019178
|
| 56 |
],
|
| 57 |
"fr_acc": [
|
| 58 |
0.12302286153389094,
|
|
|
|
| 62 |
0.13397800116550115,
|
| 63 |
0.1528195558526441,
|
| 64 |
0.04862503713606654,
|
| 65 |
+
0.2002858762969057,
|
| 66 |
+
0.24578219982631747
|
| 67 |
],
|
| 68 |
"mrr": [
|
| 69 |
0.006059600971639156,
|
|
|
|
| 73 |
0.006043311208486557,
|
| 74 |
0.0060408106073737144,
|
| 75 |
0.006041784770786762,
|
| 76 |
+
0.0060499003157019615,
|
| 77 |
+
0.006042419467121363
|
| 78 |
],
|
| 79 |
"r1": [
|
| 80 |
0.0007812500116415322,
|
|
|
|
| 84 |
0.0007812500116415322,
|
| 85 |
0.0007812500116415322,
|
| 86 |
0.0007812500116415322,
|
| 87 |
+
0.0007812500116415322,
|
| 88 |
0.0007812500116415322
|
| 89 |
],
|
| 90 |
"step_var": [
|
|
|
|
| 95 |
0.2775871157646179,
|
| 96 |
0.312656432390213,
|
| 97 |
0.34328973293304443,
|
| 98 |
+
0.20889317989349365,
|
| 99 |
+
0.22339501976966858
|
| 100 |
],
|
| 101 |
"pos_slope": [
|
| 102 |
0.4985616207122803,
|
|
|
|
| 106 |
0.47680673003196716,
|
| 107 |
0.47214341163635254,
|
| 108 |
0.46813833713531494,
|
| 109 |
+
0.48697665333747864,
|
| 110 |
+
0.48436233401298523
|
| 111 |
],
|
| 112 |
"grad_norm": [
|
| 113 |
NaN,
|
|
|
|
| 117 |
NaN,
|
| 118 |
NaN,
|
| 119 |
NaN,
|
| 120 |
+
NaN,
|
| 121 |
NaN
|
| 122 |
],
|
| 123 |
"w_ret": [
|
|
|
|
| 128 |
0.02,
|
| 129 |
0.03,
|
| 130 |
0.04,
|
| 131 |
+
0.01,
|
| 132 |
+
0.02
|
| 133 |
]
|
| 134 |
}
|