Recursive variation-norm regularization / results.json
Mechanism failed
1{
2 "device": "cuda",
3 "checks": {
4 "normalized_formula_max_error": 0.0,
5 "recursive_q_max_error": 0.0,
6 "silu_shape_delta_s025_vs_s4": 0.7544106841087341
7 },
8 "results": [
9 {
10 "kind": "l2",
11 "lambda": 0.0001,
12 "train_mse": 0.4700239896774292,
13 "val_mse": 0.46993404626846313,
14 "V": 54.94351577758789,
15 "param_l2": 157.51809692382812,
16 "max_prediction": 0.4512426257133484,
17 "seconds": 3.7238762378692627,
18 "steps": 700
19 },
20 {
21 "kind": "var",
22 "lambda": 0.0001,
23 "train_mse": 0.47400546073913574,
24 "val_mse": 0.4738515317440033,
25 "V": 40.160892486572266,
26 "param_l2": 102.81482696533203,
27 "max_prediction": 0.40470099449157715,
28 "seconds": 4.986510992050171,
29 "steps": 700
30 },
31 {
32 "kind": "l2",
33 "lambda": 0.001,
34 "train_mse": 0.4502234160900116,
35 "val_mse": 0.4381687343120575,
36 "V": 107.67620086669922,
37 "param_l2": 165.28672790527344,
38 "max_prediction": 0.801859974861145,
39 "seconds": 3.3919320106506348,
40 "steps": 700
41 },
42 {
43 "kind": "var",
44 "lambda": 0.001,
45 "train_mse": 0.4767749607563019,
46 "val_mse": 0.47259318828582764,
47 "V": 6.340097427368164,
48 "param_l2": 112.36546325683594,
49 "max_prediction": 0.33097904920578003,
50 "seconds": 4.659990310668945,
51 "steps": 700
52 },
53 {
54 "kind": "l2",
55 "lambda": 0.01,
56 "train_mse": 0.27451881766319275,
57 "val_mse": 0.2583797574043274,
58 "V": 183.54513549804688,
59 "param_l2": 358.7202453613281,
60 "max_prediction": 0.7563812732696533,
61 "seconds": 3.564720630645752,
62 "steps": 700
63 },
64 {
65 "kind": "var",
66 "lambda": 0.01,
67 "train_mse": 0.488555371761322,
68 "val_mse": 0.4729349613189697,
69 "V": 0.25097382068634033,
70 "param_l2": 119.76573944091797,
71 "max_prediction": 0.19171901047229767,
72 "seconds": 5.0653510093688965,
73 "steps": 700
74 }
75 ],
76 "best_l2": {
77 "kind": "l2",
78 "lambda": 0.01,
79 "train_mse": 0.27451881766319275,
80 "val_mse": 0.2583797574043274,
81 "V": 183.54513549804688,
82 "param_l2": 358.7202453613281,
83 "max_prediction": 0.7563812732696533,
84 "seconds": 3.564720630645752,
85 "steps": 700
86 },
87 "best_var": {
88 "kind": "var",
89 "lambda": 0.001,
90 "train_mse": 0.4767749607563019,
91 "val_mse": 0.47259318828582764,
92 "V": 6.340097427368164,
93 "param_l2": 112.36546325683594,
94 "max_prediction": 0.33097904920578003,
95 "seconds": 4.659990310668945,
96 "steps": 700
97 }
98}