Recycled-curvature proximal optimizer / bench_report.json
Failed on benchmark
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": 0.03
10 },
11 "sweep": [
12 {
13 "cfg": {
14 "lr": 0.001
15 },
16 "mean": 24.142436981201172
17 },
18 {
19 "cfg": {
20 "lr": 0.003
21 },
22 "mean": 13.75401242574056
23 },
24 {
25 "cfg": {
26 "lr": 0.01
27 },
28 "mean": 8.325917879740397
29 },
30 {
31 "cfg": {
32 "lr": 0.03
33 },
34 "mean": 6.865411599477132
35 }
36 ],
37 "full": {
38 "mean": 7.418325066566467,
39 "std": 0.9955428790176792,
40 "per_seed": [
41 6.716945648193359,
42 5.94451379776001,
43 7.934775352478027,
44 9.200451850891113,
45 7.416287899017334,
46 6.373697280883789,
47 8.270235061645508,
48 7.489693641662598
49 ],
50 "n": 8
51 }
52 },
53 "idea": {
54 "mean": 124.02495861053467,
55 "std": 19.888619557248344,
56 "per_seed": [
57 133.7542266845703,
58 95.59589385986328,
59 161.67819213867188,
60 123.55552673339844,
61 96.50812530517578,
62 130.4013214111328,
63 120.75763702392578,
64 129.94874572753906
65 ],
66 "n": 8,
67 "chosen_cfg": {
68 "lr": 0.03
69 },
70 "trials": [
71 {
72 "cfg": {
73 "lr": 0.003
74 },
75 "mean": 235.4395955403646
76 },
77 {
78 "cfg": {
79 "lr": 0.01
80 },
81 "mean": 223.33053588867188
82 },
83 {
84 "cfg": {
85 "lr": 0.03
86 },
87 "mean": 130.3427708943685
88 }
89 ]
90 },
91 "comparison": {
92 "delta_mean": 116.6066335439682,
93 "idea_wins": 0,
94 "n_pairs": 8,
95 "per_seed_diffs": [
96 127.03728103637695,
97 89.65138006210327,
98 153.74341678619385,
99 114.35507488250732,
100 89.09183740615845,
101 124.02762413024902,
102 112.48740196228027,
103 122.45905208587646
104 ],
105 "p_value": 0.0081,
106 "mde": 17.547701974609687,
107 "mde_rel_pct": 236.54533627402162,
108 "verdict": "idea worse (significant)",
109 "system_worked": false
110 },
111 "mechanism_signature": {
112 "predicted": {
113 "transport_error": 0.0,
114 "call_reduction_fraction": 0.5
115 },
116 "observed": {
117 "mean_transport_error": 1.59325485071804e-06,
118 "mean_idea_grad_calls": 73.0,
119 "mean_baseline_grad_calls": 72.0,
120 "call_reduction_fraction": -0.013888888888888888
121 },
122 "confirmed": false
123 }
124}