Nonlinear Hydrodynamic Optimizer / bench_report.json
Mechanism confirmed, baseline not beaten
1{
2 "bench_version": 1,
3 "track": "tabular",
4 "model": "mlp_tiny",
5 "metric_direction": "lower is better",
6 "n_seeds": 8,
7 "baseline": {
8 "best_cfg": {
9 "lr": 0.01
10 },
11 "sweep": [
12 {
13 "cfg": {
14 "lr": 0.001
15 },
16 "mean": 17.09206533432007
17 },
18 {
19 "cfg": {
20 "lr": 0.003
21 },
22 "mean": 11.12595009803772
23 },
24 {
25 "cfg": {
26 "lr": 0.01
27 },
28 "mean": 6.983492374420166
29 }
30 ],
31 "full": {
32 "mean": 7.1135499477386475,
33 "std": 0.44626176436687165,
34 "per_seed": [
35 6.864531517028809,
36 7.168848514556885,
37 7.027814865112305,
38 6.872774600982666,
39 6.30631685256958,
40 7.218708515167236,
41 7.5617499351501465,
42 7.887654781341553
43 ],
44 "n": 8
45 }
46 },
47 "idea": {
48 "mean": 7.093362271785736,
49 "std": 0.45241222062673386,
50 "per_seed": [
51 6.777974605560303,
52 7.191386699676514,
53 7.057932376861572,
54 6.801158428192139,
55 6.306606292724609,
56 7.222118854522705,
57 7.4917168617248535,
58 7.898004055023193
59 ],
60 "n": 8
61 },
62 "comparison": {
63 "delta_mean": -0.020187675952911377,
64 "idea_wins": 3,
65 "n_pairs": 8,
66 "per_seed_diffs": [
67 -0.08655691146850586,
68 0.022538185119628906,
69 0.030117511749267578,
70 -0.07161617279052734,
71 0.0002894401550292969,
72 0.00341033935546875,
73 -0.07003307342529297,
74 0.010349273681640625
75 ],
76 "p_value": 0.258,
77 "mde": 0.03972406104311891,
78 "mde_rel_pct": 0.5584280891391918,
79 "verdict": "no measurable effect",
80 "system_worked": false
81 },
82 "mechanism_signature": {
83 "predicted_mass_error": 0.0,
84 "observed_mass_error": 4.440892098500626e-16,
85 "predicted_conservative": true,
86 "observed_mean_activity_variance": 1346.4940232817462,
87 "density_mean": 1.0,
88 "density_std": 0.6109128074216446,
89 "confirmed": true
90 },
91 "idea_sweep": {
92 "0.001": {
93 "mean": 17.594895243644714,
94 "std": 1.455041940026395,
95 "per_seed": [
96 15.97562313079834,
97 17.117998123168945,
98 18.321380615234375,
99 16.85783576965332,
100 15.77338981628418,
101 17.865934371948242,
102 18.211532592773438,
103 20.635467529296875
104 ],
105 "n": 8
106 },
107 "0.003": {
108 "mean": 10.802422881126404,
109 "std": 0.9665300052415943,
110 "per_seed": [
111 9.816262245178223,
112 11.428475379943848,
113 10.502289772033691,
114 10.368227005004883,
115 9.419925689697266,
116 10.578962326049805,
117 11.818642616271973,
118 12.486598014831543
119 ],
120 "n": 8
121 },
122 "0.01": {
123 "mean": 7.093362271785736,
124 "std": 0.45241222062673386,
125 "per_seed": [
126 6.777974605560303,
127 7.191386699676514,
128 7.057932376861572,
129 6.801158428192139,
130 6.306606292724609,
131 7.222118854522705,
132 7.4917168617248535,
133 7.898004055023193
134 ],
135 "n": 8
136 }
137 },
138 "protocol": {
139 "paired_seeds": [
140 0,
141 1,
142 2,
143 3,
144 4,
145 5,
146 6,
147 7
148 ],
149 "shared_lr_union": [
150 0.001,
151 0.003,
152 0.01
153 ],
154 "track_rationale": "tabular is the built-in structural match for optimizer interventions"
155 }
156}