Upload sigma_lambda_v3/history.json with huggingface_hub
Browse files- sigma_lambda_v3/history.json +23 -11
sigma_lambda_v3/history.json
CHANGED
|
@@ -6,7 +6,8 @@
|
|
| 6 |
4,
|
| 7 |
5,
|
| 8 |
6,
|
| 9 |
-
4
|
|
|
|
| 10 |
],
|
| 11 |
"gen_loss": [
|
| 12 |
30.399562595346868,
|
|
@@ -15,7 +16,8 @@
|
|
| 15 |
13.480450591629728,
|
| 16 |
13.051396704105676,
|
| 17 |
12.889946385535268,
|
| 18 |
-
12.94567153010407
|
|
|
|
| 19 |
],
|
| 20 |
"ret_loss": [
|
| 21 |
796116674.0579515,
|
|
@@ -24,6 +26,7 @@
|
|
| 24 |
108628.19653386119,
|
| 25 |
58865.573026343496,
|
| 26 |
34838.68130633002,
|
|
|
|
| 27 |
0.0
|
| 28 |
],
|
| 29 |
"ppl": [
|
|
@@ -33,7 +36,8 @@
|
|
| 33 |
264.47601817816616,
|
| 34 |
127.8872290404574,
|
| 35 |
96.4789019458568,
|
| 36 |
-
220.29541970512514
|
|
|
|
| 37 |
],
|
| 38 |
"tf_acc": [
|
| 39 |
9.734349599431513e-06,
|
|
@@ -42,7 +46,8 @@
|
|
| 42 |
0.44354888428129674,
|
| 43 |
0.4487827195825911,
|
| 44 |
0.4525466680943713,
|
| 45 |
-
0.4534487178239186
|
|
|
|
| 46 |
],
|
| 47 |
"fr_acc": [
|
| 48 |
0.0,
|
|
@@ -51,7 +56,8 @@
|
|
| 51 |
0.23937194341606105,
|
| 52 |
0.23937194341606105,
|
| 53 |
0.21731311988664928,
|
| 54 |
-
0.24257707162118927
|
|
|
|
| 55 |
],
|
| 56 |
"mrr": [
|
| 57 |
0.007931923493742943,
|
|
@@ -60,7 +66,8 @@
|
|
| 60 |
0.009372583590447903,
|
| 61 |
0.010170594789087772,
|
| 62 |
0.011823299340903759,
|
| 63 |
-
0.008585811592638493
|
|
|
|
| 64 |
],
|
| 65 |
"r1": [
|
| 66 |
0.0010416667209938169,
|
|
@@ -69,7 +76,8 @@
|
|
| 69 |
0.0010416667209938169,
|
| 70 |
0.0010416667209938169,
|
| 71 |
0.0020833334419876337,
|
| 72 |
-
0.0010416667209938169
|
|
|
|
| 73 |
],
|
| 74 |
"step_var": [
|
| 75 |
0.15913955867290497,
|
|
@@ -78,7 +86,8 @@
|
|
| 78 |
0.17514024674892426,
|
| 79 |
0.18433114886283875,
|
| 80 |
0.19140200316905975,
|
| 81 |
-
0.16667772829532623
|
|
|
|
| 82 |
],
|
| 83 |
"pos_slope": [
|
| 84 |
0.4995734989643097,
|
|
@@ -87,7 +96,8 @@
|
|
| 87 |
0.49392592906951904,
|
| 88 |
0.4921555817127228,
|
| 89 |
0.49082034826278687,
|
| 90 |
-
0.4955560266971588
|
|
|
|
| 91 |
],
|
| 92 |
"grad_norm": [
|
| 93 |
2887832256709451.0,
|
|
@@ -96,7 +106,8 @@
|
|
| 96 |
NaN,
|
| 97 |
NaN,
|
| 98 |
NaN,
|
| 99 |
-
NaN
|
|
|
|
| 100 |
],
|
| 101 |
"w_ret": [
|
| 102 |
0.05,
|
|
@@ -105,6 +116,7 @@
|
|
| 105 |
0.083,
|
| 106 |
0.116,
|
| 107 |
0.14900000000000002,
|
| 108 |
-
0.05
|
|
|
|
| 109 |
]
|
| 110 |
}
|
|
|
|
| 6 |
4,
|
| 7 |
5,
|
| 8 |
6,
|
| 9 |
+
4,
|
| 10 |
+
5
|
| 11 |
],
|
| 12 |
"gen_loss": [
|
| 13 |
30.399562595346868,
|
|
|
|
| 16 |
13.480450591629728,
|
| 17 |
13.051396704105676,
|
| 18 |
12.889946385535268,
|
| 19 |
+
12.94567153010407,
|
| 20 |
+
10.66514055876719
|
| 21 |
],
|
| 22 |
"ret_loss": [
|
| 23 |
796116674.0579515,
|
|
|
|
| 26 |
108628.19653386119,
|
| 27 |
58865.573026343496,
|
| 28 |
34838.68130633002,
|
| 29 |
+
0.0,
|
| 30 |
0.0
|
| 31 |
],
|
| 32 |
"ppl": [
|
|
|
|
| 36 |
264.47601817816616,
|
| 37 |
127.8872290404574,
|
| 38 |
96.4789019458568,
|
| 39 |
+
220.29541970512514,
|
| 40 |
+
52.08889899500112
|
| 41 |
],
|
| 42 |
"tf_acc": [
|
| 43 |
9.734349599431513e-06,
|
|
|
|
| 46 |
0.44354888428129674,
|
| 47 |
0.4487827195825911,
|
| 48 |
0.4525466680943713,
|
| 49 |
+
0.4534487178239186,
|
| 50 |
+
0.47825833016966973
|
| 51 |
],
|
| 52 |
"fr_acc": [
|
| 53 |
0.0,
|
|
|
|
| 56 |
0.23937194341606105,
|
| 57 |
0.23937194341606105,
|
| 58 |
0.21731311988664928,
|
| 59 |
+
0.24257707162118927,
|
| 60 |
+
0.11165922517025459
|
| 61 |
],
|
| 62 |
"mrr": [
|
| 63 |
0.007931923493742943,
|
|
|
|
| 66 |
0.009372583590447903,
|
| 67 |
0.010170594789087772,
|
| 68 |
0.011823299340903759,
|
| 69 |
+
0.008585811592638493,
|
| 70 |
+
0.008481163531541824
|
| 71 |
],
|
| 72 |
"r1": [
|
| 73 |
0.0010416667209938169,
|
|
|
|
| 76 |
0.0010416667209938169,
|
| 77 |
0.0010416667209938169,
|
| 78 |
0.0020833334419876337,
|
| 79 |
+
0.0010416667209938169,
|
| 80 |
+
0.0
|
| 81 |
],
|
| 82 |
"step_var": [
|
| 83 |
0.15913955867290497,
|
|
|
|
| 86 |
0.17514024674892426,
|
| 87 |
0.18433114886283875,
|
| 88 |
0.19140200316905975,
|
| 89 |
+
0.16667772829532623,
|
| 90 |
+
0.17024925351142883
|
| 91 |
],
|
| 92 |
"pos_slope": [
|
| 93 |
0.4995734989643097,
|
|
|
|
| 96 |
0.49392592906951904,
|
| 97 |
0.4921555817127228,
|
| 98 |
0.49082034826278687,
|
| 99 |
+
0.4955560266971588,
|
| 100 |
+
0.49401724338531494
|
| 101 |
],
|
| 102 |
"grad_norm": [
|
| 103 |
2887832256709451.0,
|
|
|
|
| 106 |
NaN,
|
| 107 |
NaN,
|
| 108 |
NaN,
|
| 109 |
+
NaN,
|
| 110 |
+
Infinity
|
| 111 |
],
|
| 112 |
"w_ret": [
|
| 113 |
0.05,
|
|
|
|
| 116 |
0.083,
|
| 117 |
0.116,
|
| 118 |
0.14900000000000002,
|
| 119 |
+
0.05,
|
| 120 |
+
0.08
|
| 121 |
]
|
| 122 |
}
|