Pre-Training Depth Feasibility Certificates / bench_report.json
Failed on benchmark
1{
2 "bench_version": 1,
3 "track": "tabular",
4 "model": "variable_mlp",
5 "metric_direction": "lower is better",
6 "n_seeds": 8,
7 "baseline": {
8 "best_cfg": {
9 "depth": 4,
10 "lr": 0.006
11 },
12 "sweep": [
13 {
14 "cfg": {
15 "depth": 1,
16 "lr": 0.0015
17 },
18 "mean": 179.61644744873047
19 },
20 {
21 "cfg": {
22 "depth": 1,
23 "lr": 0.003
24 },
25 "mean": 61.134450912475586
26 },
27 {
28 "cfg": {
29 "depth": 1,
30 "lr": 0.006
31 },
32 "mean": 19.637452602386475
33 },
34 {
35 "cfg": {
36 "depth": 2,
37 "lr": 0.0015
38 },
39 "mean": 48.220858573913574
40 },
41 {
42 "cfg": {
43 "depth": 2,
44 "lr": 0.003
45 },
46 "mean": 18.314094066619873
47 },
48 {
49 "cfg": {
50 "depth": 2,
51 "lr": 0.006
52 },
53 "mean": 15.357292652130127
54 },
55 {
56 "cfg": {
57 "depth": 3,
58 "lr": 0.0015
59 },
60 "mean": 24.715059757232666
61 },
62 {
63 "cfg": {
64 "depth": 3,
65 "lr": 0.003
66 },
67 "mean": 16.38268733024597
68 },
69 {
70 "cfg": {
71 "depth": 3,
72 "lr": 0.006
73 },
74 "mean": 12.54684829711914
75 },
76 {
77 "cfg": {
78 "depth": 4,
79 "lr": 0.0015
80 },
81 "mean": 22.451377391815186
82 },
83 {
84 "cfg": {
85 "depth": 4,
86 "lr": 0.003
87 },
88 "mean": 14.139441013336182
89 },
90 {
91 "cfg": {
92 "depth": 4,
93 "lr": 0.006
94 },
95 "mean": 11.652069807052612
96 }
97 ],
98 "full": {
99 "per_seed": [
100 10.326997756958008,
101 11.77970027923584,
102 10.673856735229492,
103 11.692980766296387,
104 9.679404258728027,
105 11.413150787353516,
106 11.966154098510742,
107 13.212504386901855
108 ],
109 "mean": 11.343093633651733,
110 "std": 1.098638538871677
111 }
112 },
113 "idea": {
114 "per_seed": [
115 10.326997756958008,
116 11.77970027923584,
117 10.673856735229492,
118 11.692980766296387,
119 9.679404258728027,
120 11.413150787353516,
121 11.966154098510742,
122 13.212504386901855
123 ],
124 "mean": 11.343093633651733,
125 "std": 1.098638538871677
126 },
127 "comparison": {
128 "delta_mean": 0.0,
129 "idea_wins": 0,
130 "n_pairs": 8,
131 "per_seed_diffs": [
132 0.0,
133 0.0,
134 0.0,
135 0.0,
136 0.0,
137 0.0,
138 0.0,
139 0.0
140 ],
141 "p_value": 1.0,
142 "mde": 0.0,
143 "mde_rel_pct": 0.0,
144 "verdict": "no measurable effect",
145 "system_worked": false
146 },
147 "mechanism_signature": {
148 "predicted_radius_depth_exponent": -1.0,
149 "observed_test_error_depth_slope": -0.38086177111636127,
150 "predicted_vs_observed_tolerance": 0.35,
151 "confirmed": false,
152 "observed_depth_errors": [
153 [
154 1,
155 18.856087684631348
156 ],
157 [
158 2,
159 13.66954231262207
160 ],
161 [
162 3,
163 12.318396091461182
164 ],
165 [
166 4,
167 11.053349018096924
168 ]
169 ],
170 "note": "Measured on independently trained benchmark models."
171 },
172 "bench_report": {
173 "track": "tabular",
174 "model": "variable_mlp",
175 "calibration": {
176 "floor_L": -19.786727674992903,
177 "floor_U": 5.600807448724559,
178 "C_syn": 70.65927382249097,
179 "rows": [
180 [
181 1,
182 66.96246337890625
183 ],
184 [
185 2,
186 17.991180419921875
187 ],
188 [
189 3,
190 16.63127326965332
191 ],
192 [
193 4,
194 17.250062942504883
195 ]
196 ],
197 "residual_std": 6.346883780929366
198 },
199 "chosen_depth": 4
200 }
201}