1{
 2  "math_check": {
 3    "weights": [
 4      0.1506291547769241,
 5      0.2744642148059653,
 6      0.5001064058323684,
 7      0.0748002245847421
 8    ],
 9    "rates": [
10      0.048687351573742065,
11      0.31336168751822285,
12      0.7982388693815918,
13      1.620517409918451
14    ],
15    "checks": {
16      "0": {
17        "min_signed_exact": 0.09040247034962548,
18        "max_abs_identity_error": 0.0
19      },
20      "1": {
21        "min_signed_exact": 0.006118228130790459,
22        "max_abs_identity_error": 0.0
23      },
24      "2": {
25        "min_signed_exact": 0.0008484244473373037,
26        "max_abs_identity_error": 0.0
27      },
28      "3": {
29        "min_signed_exact": 0.0002238643079255148,
30        "max_abs_identity_error": 0.0
31      },
32      "4": {
33        "min_signed_exact": 7.611858109805141e-05,
34        "max_abs_identity_error": 0.0
35      },
36      "finite_difference_first": {
37        "max_dK": -0.0002452044972530987
38      },
39      "K_min": 0.09040247034962548
40    }
41  },
42  "toy_attention": {
43    "train_context": 32,
44    "test_context": 64,
45    "train_cross_entropy_mixture": 3.301440289954469,
46    "train_cross_entropy_arbitrary": 2.454133274258428,
47    "train_KL_mixture": 6.585829063255329e-07,
48    "test_KL_arbitrary_table": 0.012451793425620486,
49    "learned_weights": [
50      0.2783703540171039,
51      0.29271218705933405,
52      0.3237646493854326,
53      0.10515280953812949
54    ],
55    "learned_rates": [
56      0.01783418011609201,
57      0.17182854144462814,
58      0.17991811733428217,
59      0.20221956117274106
60    ],
61    "target_train_mass_lags_0_7": 0.4844556612421408,
62    "target_test_mass_lags_32_plus": 0.2566478401066379
63  }
64}