Upload sigma_lambda_proto_v1/history.json with huggingface_hub
Browse files
sigma_lambda_proto_v1/history.json
CHANGED
|
@@ -7,7 +7,8 @@
|
|
| 7 |
5,
|
| 8 |
6,
|
| 9 |
7,
|
| 10 |
-
8
|
|
|
|
| 11 |
],
|
| 12 |
"gen_loss": [
|
| 13 |
14.378914150180238,
|
|
@@ -17,7 +18,8 @@
|
|
| 17 |
9.55466846523863,
|
| 18 |
9.424415427897916,
|
| 19 |
9.337661864071181,
|
| 20 |
-
9.261172692017121
|
|
|
|
| 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 |
26.567988921309055,
|
| 38 |
25.650242317335714,
|
| 39 |
25.075453392796494,
|
| 40 |
-
24.59513520962833
|
|
|
|
| 41 |
],
|
| 42 |
"tf_acc": [
|
| 43 |
0.44998506448388675,
|
|
@@ -47,7 +51,8 @@
|
|
| 47 |
0.5109845416047981,
|
| 48 |
0.5145881089351385,
|
| 49 |
0.5173037067829007,
|
| 50 |
-
0.5190997774091205
|
|
|
|
| 51 |
],
|
| 52 |
"fr_acc": [
|
| 53 |
0.15843538324420675,
|
|
@@ -57,7 +62,8 @@
|
|
| 57 |
0.21467075734722796,
|
| 58 |
0.132664598473422,
|
| 59 |
0.2427502285296403,
|
| 60 |
-
0.14725610174139586
|
|
|
|
| 61 |
],
|
| 62 |
"mrr": [
|
| 63 |
0.01984136737883091,
|
|
@@ -67,7 +73,8 @@
|
|
| 67 |
0.025447646155953407,
|
| 68 |
0.028147408738732338,
|
| 69 |
0.031810443848371506,
|
| 70 |
-
0.036788471043109894
|
|
|
|
| 71 |
],
|
| 72 |
"r1": [
|
| 73 |
0.0031250000465661287,
|
|
@@ -77,7 +84,8 @@
|
|
| 77 |
0.0031250000465661287,
|
| 78 |
0.0031250000465661287,
|
| 79 |
0.0031250000465661287,
|
| 80 |
-
0.0062500000931322575
|
|
|
|
| 81 |
],
|
| 82 |
"he_proxy": [
|
| 83 |
0.0,
|
|
@@ -87,7 +95,8 @@
|
|
| 87 |
0.0,
|
| 88 |
0.067495606794824,
|
| 89 |
0.0,
|
| 90 |
-
0.0
|
|
|
|
| 91 |
],
|
| 92 |
"grad_norm": [
|
| 93 |
69556.54960830766,
|
|
@@ -97,7 +106,8 @@
|
|
| 97 |
164660.7477113645,
|
| 98 |
169132.88870452382,
|
| 99 |
347530.32776330994,
|
| 100 |
-
319075.012837759
|
|
|
|
| 101 |
],
|
| 102 |
"step_var": [
|
| 103 |
0.2371368706226349,
|
|
@@ -107,7 +117,8 @@
|
|
| 107 |
0.4257376194000244,
|
| 108 |
0.47667616605758667,
|
| 109 |
0.5235044360160828,
|
| 110 |
-
0.5645045638084412
|
|
|
|
| 111 |
],
|
| 112 |
"pos_slope": [
|
| 113 |
0.4982355535030365,
|
|
@@ -117,7 +128,8 @@
|
|
| 117 |
0.4699003994464874,
|
| 118 |
0.4636712968349457,
|
| 119 |
0.4582301080226898,
|
| 120 |
-
0.4536069333553314
|
|
|
|
| 121 |
],
|
| 122 |
"beta_mean": [
|
| 123 |
0.09964697062969208,
|
|
@@ -127,7 +139,8 @@
|
|
| 127 |
0.09398408234119415,
|
| 128 |
0.09273404628038406,
|
| 129 |
0.09163296222686768,
|
| 130 |
-
0.09070558100938797
|
|
|
|
| 131 |
],
|
| 132 |
"w_ret": [
|
| 133 |
0.0,
|
|
@@ -137,6 +150,7 @@
|
|
| 137 |
0.03,
|
| 138 |
0.06,
|
| 139 |
0.09,
|
| 140 |
-
0.12
|
|
|
|
| 141 |
]
|
| 142 |
}
|
|
|
|
| 7 |
5,
|
| 8 |
6,
|
| 9 |
7,
|
| 10 |
+
8,
|
| 11 |
+
9
|
| 12 |
],
|
| 13 |
"gen_loss": [
|
| 14 |
14.378914150180238,
|
|
|
|
| 18 |
9.55466846523863,
|
| 19 |
9.424415427897916,
|
| 20 |
9.337661864071181,
|
| 21 |
+
9.261172692017121,
|
| 22 |
+
9.224323922919504
|
| 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 |
26.567988921309055,
|
| 41 |
25.650242317335714,
|
| 42 |
25.075453392796494,
|
| 43 |
+
24.59513520962833,
|
| 44 |
+
24.370497339906883
|
| 45 |
],
|
| 46 |
"tf_acc": [
|
| 47 |
0.44998506448388675,
|
|
|
|
| 51 |
0.5109845416047981,
|
| 52 |
0.5145881089351385,
|
| 53 |
0.5173037067829007,
|
| 54 |
+
0.5190997774091205,
|
| 55 |
+
0.5202924423023874
|
| 56 |
],
|
| 57 |
"fr_acc": [
|
| 58 |
0.15843538324420675,
|
|
|
|
| 62 |
0.21467075734722796,
|
| 63 |
0.132664598473422,
|
| 64 |
0.2427502285296403,
|
| 65 |
+
0.14725610174139586,
|
| 66 |
+
0.16435602632661456
|
| 67 |
],
|
| 68 |
"mrr": [
|
| 69 |
0.01984136737883091,
|
|
|
|
| 73 |
0.025447646155953407,
|
| 74 |
0.028147408738732338,
|
| 75 |
0.031810443848371506,
|
| 76 |
+
0.036788471043109894,
|
| 77 |
+
0.03900999575853348
|
| 78 |
],
|
| 79 |
"r1": [
|
| 80 |
0.0031250000465661287,
|
|
|
|
| 84 |
0.0031250000465661287,
|
| 85 |
0.0031250000465661287,
|
| 86 |
0.0031250000465661287,
|
| 87 |
+
0.0062500000931322575,
|
| 88 |
+
0.012500000186264515
|
| 89 |
],
|
| 90 |
"he_proxy": [
|
| 91 |
0.0,
|
|
|
|
| 95 |
0.0,
|
| 96 |
0.067495606794824,
|
| 97 |
0.0,
|
| 98 |
+
0.0,
|
| 99 |
+
0.03818094680227914
|
| 100 |
],
|
| 101 |
"grad_norm": [
|
| 102 |
69556.54960830766,
|
|
|
|
| 106 |
164660.7477113645,
|
| 107 |
169132.88870452382,
|
| 108 |
347530.32776330994,
|
| 109 |
+
319075.012837759,
|
| 110 |
+
164171.89088809182
|
| 111 |
],
|
| 112 |
"step_var": [
|
| 113 |
0.2371368706226349,
|
|
|
|
| 117 |
0.4257376194000244,
|
| 118 |
0.47667616605758667,
|
| 119 |
0.5235044360160828,
|
| 120 |
+
0.5645045638084412,
|
| 121 |
+
0.5979146957397461
|
| 122 |
],
|
| 123 |
"pos_slope": [
|
| 124 |
0.4982355535030365,
|
|
|
|
| 128 |
0.4699003994464874,
|
| 129 |
0.4636712968349457,
|
| 130 |
0.4582301080226898,
|
| 131 |
+
0.4536069333553314,
|
| 132 |
+
0.44983890652656555
|
| 133 |
],
|
| 134 |
"beta_mean": [
|
| 135 |
0.09964697062969208,
|
|
|
|
| 139 |
0.09398408234119415,
|
| 140 |
0.09273404628038406,
|
| 141 |
0.09163296222686768,
|
| 142 |
+
0.09070558100938797,
|
| 143 |
+
0.08995532989501953
|
| 144 |
],
|
| 145 |
"w_ret": [
|
| 146 |
0.0,
|
|
|
|
| 150 |
0.03,
|
| 151 |
0.06,
|
| 152 |
0.09,
|
| 153 |
+
0.12,
|
| 154 |
+
0.15
|
| 155 |
]
|
| 156 |
}
|