1{
 2  "math_checks": {
 3    "device": "cuda",
 4    "epsilon_values": [
 5      0.001,
 6      0.0022758459260747888,
 7      0.005179474679231213,
 8      0.011787686347935873,
 9      0.02682695795279726,
10      0.0610540229658533,
11      0.13894954943731375,
12      0.31622776601683794
13    ],
14    "remainder_values": [
15      2.8881004254799336e-09,
16      3.4055119613185525e-08,
17      4.0165468817576766e-07,
18      4.741479642689228e-06,
19      5.60772605240345e-05,
20      0.0006660250946879387,
21      0.007985111325979233,
22      0.09772586822509766
23    ],
24    "remainder_loglog_slope": 3.0096020091838245,
25    "predicted_remainder_slope": 3.0,
26    "permutation_error": 4.76837158203125e-07,
27    "quadratic_orthogonal_error": 4.76837158203125e-07,
28    "original_orthogonal_difference_slope": 2.9919965411889566,
29    "predicted_orthogonal_difference_slope": 3.0,
30    "widths": [
31      4,
32      8,
33      16,
34      32,
35      64
36    ],
37    "width_mean_abs_remainder": [
38      0.0001285379049174177,
39      0.00027992582957570754,
40      0.00039409029413945974,
41      0.0005606541293673217,
42      0.0009599066339433193
43    ],
44    "width_slope_random": 0.6803466874167912,
45    "predicted_width_slope_random": 0.5,
46    "width_slope_aligned": 1.0000041097811878,
47    "predicted_width_slope_aligned": 1.0,
48    "prediction_pass": true
49  },
50  "mini_experiment": {
51    "dataset": "synthetic cubic exchangeable branch teacher",
52    "k": 6,
53    "width": 16,
54    "steps": [
55      0,
56      99,
57      249,
58      499
59    ],
60    "baseline_branch_mse": [
61      0.08169986307621002,
62      9.217748447554186e-05,
63      2.138323543476872e-05,
64      1.1578095836739521e-05
65    ],
66    "idea_quadratic_mse": [
67      0.0817537009716034,
68      0.01447005569934845,
69      0.014469806104898453,
70      0.014469806104898453
71    ],
72    "parameter_count_each": 112
73  }
74}