Fundamental-Cycle Compatibility Basis / bench_report.json
Mechanism confirmed, baseline not beaten
1{
2 "bench_version": 1,
3 "track": "sparse_cycle_compatibility",
4 "model": "pair_mlp",
5 "metric_direction": "lower is better",
6 "n_seeds": 8,
7 "baseline": {
8 "best_cfg": {
9 "lr": 0.01,
10 "lambda": 0.0
11 },
12 "sweep": [
13 {
14 "cfg": {
15 "lr": 0.001,
16 "lambda": 0.0
17 },
18 "mean": 0.572812482714653
19 },
20 {
21 "cfg": {
22 "lr": 0.003,
23 "lambda": 0.0
24 },
25 "mean": 0.4990624710917473
26 },
27 {
28 "cfg": {
29 "lr": 0.01,
30 "lambda": 0.0
31 },
32 "mean": 0.4959374815225601
33 }
34 ],
35 "full": {
36 "mean": 0.4884374812245369,
37 "std": 0.07308691848966371,
38 "per_seed": [
39 0.5362499952316284,
40 0.5862499475479126,
41 0.5299999713897705,
42 0.33125001192092896,
43 0.44374996423721313,
44 0.45749998092651367,
45 0.4962499737739563,
46 0.5262500047683716
47 ],
48 "n": 8
49 }
50 },
51 "idea": {
52 "mean": 0.4884374812245369,
53 "std": 0.07308691848966371,
54 "per_seed": [
55 0.5362499952316284,
56 0.5862499475479126,
57 0.5299999713897705,
58 0.33125001192092896,
59 0.44374996423721313,
60 0.45749998092651367,
61 0.4962499737739563,
62 0.5262500047683716
63 ],
64 "n": 8
65 },
66 "comparison": {
67 "delta_mean": 0.0,
68 "idea_wins": 0,
69 "n_pairs": 8,
70 "per_seed_diffs": [
71 0.0,
72 0.0,
73 0.0,
74 0.0,
75 0.0,
76 0.0,
77 0.0,
78 0.0
79 ],
80 "p_value": 1.0,
81 "mde": 0.0,
82 "mde_rel_pct": 0.0,
83 "verdict": "no measurable effect",
84 "system_worked": false
85 },
86 "custom_track": {
87 "name": "sparse_cycle_compatibility",
88 "file": "custom_cycle_track.py",
89 "domain": "masked_categorical_compatibility"
90 },
91 "idea_cfg": {
92 "lr": 0.01,
93 "lambda": 0.01
94 },
95 "math_check": {
96 "basis_rank": 5,
97 "expected_rank": 5,
98 "compatible_max_residual": 2.220446049250313e-16
99 },
100 "mechanism_signature": {
101 "prediction": "basis constraints span all cycle constraints; basis count equals cycle rank",
102 "predicted_cycle_rank": 5,
103 "observed_exhaustive_cycles": 28,
104 "observed_basis_mean_abs": 0.055877745151519775,
105 "observed_unseen_cycle_mean_abs": 0.1098224903856005,
106 "observed_unseen_to_basis_ratio": 1.9654066227231812,
107 "confirmed": true
108 }
109}