Upload sigma_lambda_proto_v1/history.json with huggingface_hub
Browse files
sigma_lambda_proto_v1/history.json
CHANGED
|
@@ -4,20 +4,23 @@
|
|
| 4 |
2,
|
| 5 |
3,
|
| 6 |
4,
|
| 7 |
-
5
|
|
|
|
| 8 |
],
|
| 9 |
"gen_loss": [
|
| 10 |
14.378914150180238,
|
| 11 |
10.41937110927972,
|
| 12 |
9.95253911739046,
|
| 13 |
9.72740114264416,
|
| 14 |
-
9.55466846523863
|
|
|
|
| 15 |
],
|
| 16 |
"ret_loss": [
|
| 17 |
0.0,
|
| 18 |
0.0,
|
| 19 |
0.0,
|
| 20 |
0.0,
|
|
|
|
| 21 |
0.0
|
| 22 |
],
|
| 23 |
"ppl": [
|
|
@@ -25,34 +28,39 @@
|
|
| 25 |
36.01809758482572,
|
| 26 |
29.905926998953085,
|
| 27 |
27.759728442854154,
|
| 28 |
-
26.567988921309055
|
|
|
|
| 29 |
],
|
| 30 |
"tf_acc": [
|
| 31 |
0.44998506448388675,
|
| 32 |
0.4889349317509485,
|
| 33 |
0.5010759556689941,
|
| 34 |
0.5067917151141438,
|
| 35 |
-
0.5109845416047981
|
|
|
|
| 36 |
],
|
| 37 |
"fr_acc": [
|
| 38 |
0.15843538324420675,
|
| 39 |
0.21595909319438727,
|
| 40 |
0.22207020430549843,
|
| 41 |
0.17343382924265277,
|
| 42 |
-
0.21467075734722796
|
|
|
|
| 43 |
],
|
| 44 |
"mrr": [
|
| 45 |
0.01984136737883091,
|
| 46 |
0.02093728445470333,
|
| 47 |
0.021167319267988205,
|
| 48 |
0.02771509811282158,
|
| 49 |
-
0.025447646155953407
|
|
|
|
| 50 |
],
|
| 51 |
"r1": [
|
| 52 |
0.0031250000465661287,
|
| 53 |
0.0031250000465661287,
|
| 54 |
0.0031250000465661287,
|
| 55 |
0.00937500037252903,
|
|
|
|
| 56 |
0.0031250000465661287
|
| 57 |
],
|
| 58 |
"he_proxy": [
|
|
@@ -60,41 +68,47 @@
|
|
| 60 |
0.0,
|
| 61 |
0.0692528888652218,
|
| 62 |
0.0,
|
| 63 |
-
0.0
|
|
|
|
| 64 |
],
|
| 65 |
"grad_norm": [
|
| 66 |
69556.54960830766,
|
| 67 |
524881.9476341035,
|
| 68 |
129648.56937981857,
|
| 69 |
68050.81607155707,
|
| 70 |
-
164660.7477113645
|
|
|
|
| 71 |
],
|
| 72 |
"step_var": [
|
| 73 |
0.2371368706226349,
|
| 74 |
0.276157945394516,
|
| 75 |
0.3243665397167206,
|
| 76 |
0.37456902861595154,
|
| 77 |
-
0.4257376194000244
|
|
|
|
| 78 |
],
|
| 79 |
"pos_slope": [
|
| 80 |
0.4982355535030365,
|
| 81 |
0.4918489158153534,
|
| 82 |
0.48427340388298035,
|
| 83 |
0.4768323302268982,
|
| 84 |
-
0.4699003994464874
|
|
|
|
| 85 |
],
|
| 86 |
"beta_mean": [
|
| 87 |
0.09964697062969208,
|
| 88 |
0.0983697697520256,
|
| 89 |
0.09685240685939789,
|
| 90 |
0.0953640341758728,
|
| 91 |
-
0.09398408234119415
|
|
|
|
| 92 |
],
|
| 93 |
"w_ret": [
|
| 94 |
0.0,
|
| 95 |
0.0,
|
| 96 |
0.0,
|
| 97 |
0.0,
|
| 98 |
-
0.03
|
|
|
|
| 99 |
]
|
| 100 |
}
|
|
|
|
| 4 |
2,
|
| 5 |
3,
|
| 6 |
4,
|
| 7 |
+
5,
|
| 8 |
+
6
|
| 9 |
],
|
| 10 |
"gen_loss": [
|
| 11 |
14.378914150180238,
|
| 12 |
10.41937110927972,
|
| 13 |
9.95253911739046,
|
| 14 |
9.72740114264416,
|
| 15 |
+
9.55466846523863,
|
| 16 |
+
9.424415427897916
|
| 17 |
],
|
| 18 |
"ret_loss": [
|
| 19 |
0.0,
|
| 20 |
0.0,
|
| 21 |
0.0,
|
| 22 |
0.0,
|
| 23 |
+
0.0,
|
| 24 |
0.0
|
| 25 |
],
|
| 26 |
"ppl": [
|
|
|
|
| 28 |
36.01809758482572,
|
| 29 |
29.905926998953085,
|
| 30 |
27.759728442854154,
|
| 31 |
+
26.567988921309055,
|
| 32 |
+
25.650242317335714
|
| 33 |
],
|
| 34 |
"tf_acc": [
|
| 35 |
0.44998506448388675,
|
| 36 |
0.4889349317509485,
|
| 37 |
0.5010759556689941,
|
| 38 |
0.5067917151141438,
|
| 39 |
+
0.5109845416047981,
|
| 40 |
+
0.5145881089351385
|
| 41 |
],
|
| 42 |
"fr_acc": [
|
| 43 |
0.15843538324420675,
|
| 44 |
0.21595909319438727,
|
| 45 |
0.22207020430549843,
|
| 46 |
0.17343382924265277,
|
| 47 |
+
0.21467075734722796,
|
| 48 |
+
0.132664598473422
|
| 49 |
],
|
| 50 |
"mrr": [
|
| 51 |
0.01984136737883091,
|
| 52 |
0.02093728445470333,
|
| 53 |
0.021167319267988205,
|
| 54 |
0.02771509811282158,
|
| 55 |
+
0.025447646155953407,
|
| 56 |
+
0.028147408738732338
|
| 57 |
],
|
| 58 |
"r1": [
|
| 59 |
0.0031250000465661287,
|
| 60 |
0.0031250000465661287,
|
| 61 |
0.0031250000465661287,
|
| 62 |
0.00937500037252903,
|
| 63 |
+
0.0031250000465661287,
|
| 64 |
0.0031250000465661287
|
| 65 |
],
|
| 66 |
"he_proxy": [
|
|
|
|
| 68 |
0.0,
|
| 69 |
0.0692528888652218,
|
| 70 |
0.0,
|
| 71 |
+
0.0,
|
| 72 |
+
0.067495606794824
|
| 73 |
],
|
| 74 |
"grad_norm": [
|
| 75 |
69556.54960830766,
|
| 76 |
524881.9476341035,
|
| 77 |
129648.56937981857,
|
| 78 |
68050.81607155707,
|
| 79 |
+
164660.7477113645,
|
| 80 |
+
169132.88870452382
|
| 81 |
],
|
| 82 |
"step_var": [
|
| 83 |
0.2371368706226349,
|
| 84 |
0.276157945394516,
|
| 85 |
0.3243665397167206,
|
| 86 |
0.37456902861595154,
|
| 87 |
+
0.4257376194000244,
|
| 88 |
+
0.47667616605758667
|
| 89 |
],
|
| 90 |
"pos_slope": [
|
| 91 |
0.4982355535030365,
|
| 92 |
0.4918489158153534,
|
| 93 |
0.48427340388298035,
|
| 94 |
0.4768323302268982,
|
| 95 |
+
0.4699003994464874,
|
| 96 |
+
0.4636712968349457
|
| 97 |
],
|
| 98 |
"beta_mean": [
|
| 99 |
0.09964697062969208,
|
| 100 |
0.0983697697520256,
|
| 101 |
0.09685240685939789,
|
| 102 |
0.0953640341758728,
|
| 103 |
+
0.09398408234119415,
|
| 104 |
+
0.09273404628038406
|
| 105 |
],
|
| 106 |
"w_ret": [
|
| 107 |
0.0,
|
| 108 |
0.0,
|
| 109 |
0.0,
|
| 110 |
0.0,
|
| 111 |
+
0.03,
|
| 112 |
+
0.06
|
| 113 |
]
|
| 114 |
}
|