Ellipsoidal-Preserving Spherical Feature Stabilizer / bench_report.json

Failed on benchmark

Raw ⬇ ZIP
  1{
  2  "bench_version": 1,
  3  "track": "vision",
  4  "model": "cnn_small",
  5  "metric_direction": "lower is better",
  6  "n_seeds": 8,
  7  "baseline": {
  8    "best_cfg": {
  9      "lr": 0.002,
 10      "alpha": 0.0
 11    },
 12    "sweep": [
 13      {
 14        "cfg": {
 15          "lr": 0.001,
 16          "alpha": 0.0
 17        },
 18        "mean": 0.6916667073965073
 19      },
 20      {
 21        "cfg": {
 22          "lr": 0.001,
 23          "alpha": 0.5
 24        },
 25        "mean": 0.6916667073965073
 26      },
 27      {
 28        "cfg": {
 29          "lr": 0.001,
 30          "alpha": 1.0
 31        },
 32        "mean": 0.6916667073965073
 33      },
 34      {
 35        "cfg": {
 36          "lr": 0.002,
 37          "alpha": 0.0
 38        },
 39        "mean": 0.6854167133569717
 40      },
 41      {
 42        "cfg": {
 43          "lr": 0.002,
 44          "alpha": 0.5
 45        },
 46        "mean": 0.6854167133569717
 47      },
 48      {
 49        "cfg": {
 50          "lr": 0.002,
 51          "alpha": 1.0
 52        },
 53        "mean": 0.6854167133569717
 54      },
 55      {
 56        "cfg": {
 57          "lr": 0.003,
 58          "alpha": 0.0
 59        },
 60        "mean": 0.708333358168602
 61      },
 62      {
 63        "cfg": {
 64          "lr": 0.003,
 65          "alpha": 0.5
 66        },
 67        "mean": 0.708333358168602
 68      },
 69      {
 70        "cfg": {
 71          "lr": 0.003,
 72          "alpha": 1.0
 73        },
 74        "mean": 0.708333358168602
 75      },
 76      {
 77        "cfg": {
 78          "lr": 0.004,
 79          "alpha": 0.0
 80        },
 81        "mean": 0.7041666954755783
 82      },
 83      {
 84        "cfg": {
 85          "lr": 0.004,
 86          "alpha": 0.5
 87        },
 88        "mean": 0.7041666954755783
 89      },
 90      {
 91        "cfg": {
 92          "lr": 0.004,
 93          "alpha": 1.0
 94        },
 95        "mean": 0.7041666954755783
 96      },
 97      {
 98        "cfg": {
 99          "lr": 0.006,
100          "alpha": 0.0
101        },
102        "mean": 0.7041667103767395
103      },
104      {
105        "cfg": {
106          "lr": 0.006,
107          "alpha": 0.5
108        },
109        "mean": 0.7041667103767395
110      },
111      {
112        "cfg": {
113          "lr": 0.006,
114          "alpha": 1.0
115        },
116        "mean": 0.7041667103767395
117      }
118    ],
119    "full": {
120      "mean": 0.7041667029261589,
121      "std": 0.04310032788585424,
122      "per_seed": [
123        0.7166666984558105,
124        0.7000000476837158,
125        0.6416667103767395,
126        0.6833333969116211,
127        0.7166666984558105,
128        0.7250000238418579,
129        0.7916666865348816,
130        0.6583333611488342
131      ],
132      "n": 8
133    }
134  },
135  "idea": {
136    "per_seed": [
137      0.7500000596046448,
138      0.7416667342185974,
139      0.7500000596046448,
140      0.7083333730697632,
141      0.7666667103767395,
142      0.7750000357627869,
143      0.7750000357627869,
144      0.8000000715255737
145    ],
146    "mean": 0.7583333849906921,
147    "std": 0.025685060440690133,
148    "n": 8,
149    "config": {
150      "lr": 0.001,
151      "alpha": 0.75
152    }
153  },
154  "comparison": {
155    "delta_mean": 0.054166682064533234,
156    "idea_wins": 1,
157    "n_pairs": 8,
158    "per_seed_diffs": [
159      0.03333336114883423,
160      0.04166668653488159,
161      0.10833334922790527,
162      0.02499997615814209,
163      0.050000011920928955,
164      0.050000011920928955,
165      -0.016666650772094727,
166      0.1416667103767395
167    ],
168    "p_value": 0.0169,
169    "mde": 0.04130698754254222,
170    "mde_rel_pct": 5.866080769069509,
171    "verdict": "idea worse (significant)",
172    "system_worked": false
173  },
174  "structural_match": "vision feature-map normalization / convolutional spatial angular field",
175  "idea_sweep": [
176    {
177      "config": {
178        "lr": 0.001,
179        "alpha": 0.75
180      },
181      "mean": 0.7583333849906921
182    },
183    {
184      "config": {
185        "lr": 0.003,
186        "alpha": 1.0
187      },
188      "mean": 0.7697917148470879
189    },
190    {
191      "config": {
192        "lr": 0.006,
193        "alpha": 1.0
194      },
195      "mean": 0.8447917029261589
196    }
197  ],
198  "custom_track": null,
199  "mechanism_signature": {
200    "predicted_high_ratio": 0.5625,
201    "observed_high_ratio": 7.593735545873642,
202    "predicted_degree2_ratio": 1.0,
203    "observed_degree2_ratio": 16.193447947502136,
204    "tolerance": 0.25,
205    "confirmed": false,
206    "measurement": "trained vision CNN intermediate feature maps; 2D Fourier high/low proxies"
207  }
208}