Representation-Invariant Authority Margin / bench_report.json

Mechanism confirmed, baseline not beaten

Raw ⬇ ZIP
  1{
  2  "bench_version": 1,
  3  "track": "dynamics",
  4  "model": "rnn_small",
  5  "metric_direction": "lower is better",
  6  "n_seeds": 8,
  7  "baseline": {
  8    "best_cfg": {
  9      "lr": 0.01,
 10      "lam": 0.0
 11    },
 12    "sweep": [
 13      {
 14        "cfg": {
 15          "lr": 0.001,
 16          "lam": 0.0
 17        },
 18        "mean": 0.0019458123424556106
 19      },
 20      {
 21        "cfg": {
 22          "lr": 0.001,
 23          "lam": 0.02
 24        },
 25        "mean": 0.0018735561170615256
 26      },
 27      {
 28        "cfg": {
 29          "lr": 0.003,
 30          "lam": 0.0
 31        },
 32        "mean": 0.000998065312160179
 33      },
 34      {
 35        "cfg": {
 36          "lr": 0.003,
 37          "lam": 0.02
 38        },
 39        "mean": 0.0010661959095159546
 40      },
 41      {
 42        "cfg": {
 43          "lr": 0.01,
 44          "lam": 0.0
 45        },
 46        "mean": 0.0005159456450201105
 47      },
 48      {
 49        "cfg": {
 50          "lr": 0.01,
 51          "lam": 0.02
 52        },
 53        "mean": 0.00056726903130766
 54      }
 55    ],
 56    "full": {
 57      "mean": 0.00045527595102612395,
 58      "std": 0.0001763609215911445,
 59      "per_seed": [
 60        0.0006657303310930729,
 61        0.0006478183786384761,
 62        0.00023686116037424654,
 63        0.0005133727099746466,
 64        0.0002487104502506554,
 65        0.0004278800042811781,
 66        0.00025996746262535453,
 67        0.0006418671109713614
 68      ],
 69      "n": 8
 70    }
 71  },
 72  "idea": {
 73    "mean": 0.00045527595102612395,
 74    "std": 0.0001763609215911445,
 75    "per_seed": [
 76      0.0006657303310930729,
 77      0.0006478183786384761,
 78      0.00023686116037424654,
 79      0.0005133727099746466,
 80      0.0002487104502506554,
 81      0.0004278800042811781,
 82      0.00025996746262535453,
 83      0.0006418671109713614
 84    ],
 85    "n": 8
 86  },
 87  "comparison": {
 88    "delta_mean": 0.0,
 89    "idea_wins": 0,
 90    "n_pairs": 8,
 91    "per_seed_diffs": [
 92      0.0,
 93      0.0,
 94      0.0,
 95      0.0,
 96      0.0,
 97      0.0,
 98      0.0,
 99      0.0
100    ],
101    "p_value": 1.0,
102    "mde": 0.0,
103    "mde_rel_pct": 0.0,
104    "verdict": "no measurable effect",
105    "system_worked": false
106  },
107  "selection": {
108    "idea_grid": [
109      [
110        0.002105655992636457,
111        {
112          "lr": 0.001,
113          "lam": 0.0
114        },
115        {
116          "mean": 0.002105655992636457,
117          "std": 0.0003655373406144519,
118          "per_seed": [
119            0.0016959174536168575,
120            0.0019371205708011985,
121            0.0020503592677414417,
122            0.002099852077662945,
123            0.0015101841418072581,
124            0.002463802695274353,
125            0.002578662009909749,
126            0.002509349724277854
127          ],
128          "n": 8
129        }
130      ],
131      [
132        0.0019897328020306304,
133        {
134          "lr": 0.001,
135          "lam": 0.02
136        },
137        {
138          "mean": 0.0019897328020306304,
139          "std": 0.00041209688344801833,
140          "per_seed": [
141            0.0015021191211417317,
142            0.0018472193041816354,
143            0.001866860780864954,
144            0.0020167825277894735,
145            0.0013365390477702022,
146            0.002327729482203722,
147            0.0024804112035781145,
148            0.00254020094871521
149          ],
150          "n": 8
151        }
152      ],
153      [
154        0.0010153715847991407,
155        {
156          "lr": 0.003,
157          "lam": 0.0
158        },
159        {
160          "mean": 0.0010153715847991407,
161          "std": 0.00029879091457532926,
162          "per_seed": [
163            0.000696369621437043,
164            0.0011397767812013626,
165            0.0007460162159986794,
166            0.0014100986300036311,
167            0.0006380671402439475,
168            0.0008445471758022904,
169            0.0012670184951275587,
170            0.0013810786185786128
171          ],
172          "n": 8
173        }
174      ],
175      [
176        0.0011276934674242511,
177        {
178          "lr": 0.003,
179          "lam": 0.02
180        },
181        {
182          "mean": 0.0011276934674242511,
183          "std": 0.0003351466639166485,
184          "per_seed": [
185            0.0007446846575476229,
186            0.0012797265080735087,
187            0.000884775654412806,
188            0.0015856961254030466,
189            0.0007189440657384694,
190            0.0008848343277350068,
191            0.0013867217348888516,
192            0.001536164665594697
193          ],
194          "n": 8
195        }
196      ],
197      [
198        0.00045527595102612395,
199        {
200          "lr": 0.01,
201          "lam": 0.0
202        },
203        {
204          "mean": 0.00045527595102612395,
205          "std": 0.0001763609215911445,
206          "per_seed": [
207            0.0006657303310930729,
208            0.0006478183786384761,
209            0.00023686116037424654,
210            0.0005133727099746466,
211            0.0002487104502506554,
212            0.0004278800042811781,
213            0.00025996746262535453,
214            0.0006418671109713614
215          ],
216          "n": 8
217        }
218      ],
219      [
220        0.000552004188648425,
221        {
222          "lr": 0.01,
223          "lam": 0.02
224        },
225        {
226          "mean": 0.000552004188648425,
227          "std": 0.0001826643695948444,
228          "per_seed": [
229            0.0006198962801136076,
230            0.0007790127419866621,
231            0.0003278787771705538,
232            0.0007128552533686161,
233            0.0002897651575040072,
234            0.0004925421671941876,
235            0.00042627722723409534,
236            0.0007678059046156704
237          ],
238          "n": 8
239        }
240      ]
241    ],
242    "best_cfg": {
243      "lr": 0.01,
244      "lam": 0.0
245    }
246  },
247  "track_choice": "Dynamics is structurally matched because the claim concerns controlled invariance and actuator authority.",
248  "mechanism_signature": {
249    "n_boundary_like": 31,
250    "predicted_representation_ratio": 1.0,
251    "observed_mean_demand_base": 0.09864581376314163,
252    "observed_mean_demand_2h": 0.09875281900167465,
253    "observed_relative_spread": 0.0010847416706383228,
254    "predicted_threshold_rho": 1.0,
255    "observed_threshold_rho_from_demand": 0.3754686415195465,
256    "confirmed": true
257  }
258}