Mean-field rainbow relation router / bench_report.json
Mechanism confirmed, baseline not beaten
1{
2 "bench_version": 1,
3 "track": "relational_graph_classification",
4 "model": "local_relational_gnn",
5 "metric_direction": "lower is better",
6 "n_seeds": 8,
7 "baseline": {
8 "best_cfg": {
9 "lr": 0.006,
10 "tau": 0.5
11 },
12 "sweep": [
13 {
14 "cfg": {
15 "lr": 0.0015,
16 "tau": 0.5
17 },
18 "mean": 0.012500000186264515
19 },
20 {
21 "cfg": {
22 "lr": 0.0015,
23 "tau": 1.0
24 },
25 "mean": 0.012500000186264515
26 },
27 {
28 "cfg": {
29 "lr": 0.0015,
30 "tau": 2.0
31 },
32 "mean": 0.012500000186264515
33 },
34 {
35 "cfg": {
36 "lr": 0.003,
37 "tau": 0.5
38 },
39 "mean": 0.009375000256113708
40 },
41 {
42 "cfg": {
43 "lr": 0.003,
44 "tau": 1.0
45 },
46 "mean": 0.009375000256113708
47 },
48 {
49 "cfg": {
50 "lr": 0.003,
51 "tau": 2.0
52 },
53 "mean": 0.009375000256113708
54 },
55 {
56 "cfg": {
57 "lr": 0.006,
58 "tau": 0.5
59 },
60 "mean": 0.004687500069849193
61 },
62 {
63 "cfg": {
64 "lr": 0.006,
65 "tau": 1.0
66 },
67 "mean": 0.004687500069849193
68 },
69 {
70 "cfg": {
71 "lr": 0.006,
72 "tau": 2.0
73 },
74 "mean": 0.0062500000931322575
75 }
76 ],
77 "full": {
78 "mean": 0.003906250058207661,
79 "std": 0.004349815973278263,
80 "per_seed": [
81 0.0062500000931322575,
82 0.0,
83 0.0062500000931322575,
84 0.0062500000931322575,
85 0.0,
86 0.012500000186264515,
87 0.0,
88 0.0
89 ],
90 "n": 8
91 }
92 },
93 "idea": {
94 "mean": 0.0031250000465661287,
95 "std": 0.004419417448270373,
96 "per_seed": [
97 0.0062500000931322575,
98 0.0,
99 0.0062500000931322575,
100 0.0,
101 0.0,
102 0.012500000186264515,
103 0.0,
104 0.0
105 ],
106 "n": 8
107 },
108 "comparison": {
109 "delta_mean": -0.0007812500116415322,
110 "idea_wins": 1,
111 "n_pairs": 8,
112 "per_seed_diffs": [
113 0.0,
114 0.0,
115 0.0,
116 -0.0062500000931322575,
117 0.0,
118 0.0,
119 0.0,
120 0.0
121 ],
122 "p_value": 1.0,
123 "mde": 0.0018476562775322238,
124 "mde_rel_pct": 47.300000000000004,
125 "verdict": "no significant win",
126 "system_worked": false
127 },
128 "idea_sweep": [
129 {
130 "cfg": {
131 "lr": 0.006,
132 "tau": 0.5,
133 "beta4": 0.75
134 },
135 "result": {
136 "mean": 0.003906250058207661,
137 "std": 0.004349815973278263,
138 "per_seed": [
139 0.0062500000931322575,
140 0.0,
141 0.0062500000931322575,
142 0.0062500000931322575,
143 0.0,
144 0.012500000186264515,
145 0.0,
146 0.0
147 ],
148 "n": 8
149 }
150 },
151 {
152 "cfg": {
153 "lr": 0.006,
154 "tau": 0.5,
155 "beta4": 1.5
156 },
157 "result": {
158 "mean": 0.003906250058207661,
159 "std": 0.004349815973278263,
160 "per_seed": [
161 0.0062500000931322575,
162 0.0,
163 0.0062500000931322575,
164 0.0062500000931322575,
165 0.0,
166 0.012500000186264515,
167 0.0,
168 0.0
169 ],
170 "n": 8
171 }
172 },
173 {
174 "cfg": {
175 "lr": 0.006,
176 "tau": 0.5,
177 "beta4": 3.0
178 },
179 "result": {
180 "mean": 0.0031250000465661287,
181 "std": 0.004419417448270373,
182 "per_seed": [
183 0.0062500000931322575,
184 0.0,
185 0.0062500000931322575,
186 0.0,
187 0.0,
188 0.012500000186264515,
189 0.0,
190 0.0
191 ],
192 "n": 8
193 }
194 }
195 ],
196 "protocol_note": "Built-in relational custom track used; no structurally matching built-in track exists.",
197 "mechanism_signature": {
198 "prediction": "trained positive beta4 increases rainbow density while beta4=0 is unary",
199 "observed_mean_density_change": 0.15861905947260094,
200 "per_seed": [
201 {
202 "seed": 0,
203 "density0": 0.0045055123046040535,
204 "density_beta4": 0.08666115999221802,
205 "delta_density": 0.08215564768761396,
206 "entropy_beta4": 0.8868252635002136
207 },
208 {
209 "seed": 1,
210 "density0": 0.0016223047859966755,
211 "density_beta4": 0.22191891074180603,
212 "delta_density": 0.22029660595580935,
213 "entropy_beta4": 1.0981340408325195
214 },
215 {
216 "seed": 2,
217 "density0": 0.0005412318860180676,
218 "density_beta4": 0.2031792253255844,
219 "delta_density": 0.20263799343956634,
220 "entropy_beta4": 1.0696096420288086
221 },
222 {
223 "seed": 3,
224 "density0": 0.006103047169744968,
225 "density_beta4": 0.18493475019931793,
226 "delta_density": 0.17883170302957296,
227 "entropy_beta4": 1.0392907857894897
228 },
229 {
230 "seed": 4,
231 "density0": 2.5453034124822693e-10,
232 "density_beta4": 0.22103427350521088,
233 "delta_density": 0.22103427325068054,
234 "entropy_beta4": 1.096804141998291
235 },
236 {
237 "seed": 5,
238 "density0": 0.06863658130168915,
239 "density_beta4": 0.19528405368328094,
240 "delta_density": 0.1266474723815918,
241 "entropy_beta4": 1.0535038709640503
242 },
243 {
244 "seed": 6,
245 "density0": 0.05335409194231033,
246 "density_beta4": 0.1256251484155655,
247 "delta_density": 0.07227105647325516,
248 "entropy_beta4": 0.914698600769043
249 },
250 {
251 "seed": 7,
252 "density0": 0.04267265647649765,
253 "density_beta4": 0.2077503800392151,
254 "delta_density": 0.16507772356271744,
255 "entropy_beta4": 1.0758920907974243
256 }
257 ],
258 "beta4_zero_max_change": 0.0,
259 "confirmed": true
260 },
261 "wall_clock_note": "canonical train_model; equal 18 epochs, batch 64; timing not primary metric"
262}