Multiplicity-balanced symmetric interaction layer / bench_report.json

Mechanism confirmed, baseline not beaten

Raw ⬇ ZIP
  1{
  2  "bench_version": 1,
  3  "track": "custom_symmetric_polynomial_regression",
  4  "model": "interaction_layer",
  5  "metric_direction": "lower is better",
  6  "n_seeds": 8,
  7  "baseline": {
  8    "best_cfg": {
  9      "lr": 0.01,
 10      "epochs": 20
 11    },
 12    "sweep": [
 13      {
 14        "cfg": {
 15          "lr": 0.001,
 16          "epochs": 20
 17        },
 18        "mean": 288.5231819152832
 19      },
 20      {
 21        "cfg": {
 22          "lr": 0.003,
 23          "epochs": 20
 24        },
 25        "mean": 186.22671699523926
 26      },
 27      {
 28        "cfg": {
 29          "lr": 0.01,
 30          "epochs": 20
 31        },
 32        "mean": 147.9050407409668
 33      }
 34    ],
 35    "full": {
 36      "mean": 167.82896614074707,
 37      "std": 48.00322837292885,
 38      "per_seed": [
 39        141.9176483154297,
 40        198.45465087890625,
 41        160.6289825439453,
 42        90.61888122558594,
 43        214.09580993652344,
 44        127.07957458496094,
 45        158.7684783935547,
 46        251.0677032470703
 47      ],
 48      "n": 8
 49    }
 50  },
 51  "idea": {
 52    "mean": 182.48091316223145,
 53    "std": 44.266280792836994,
 54    "per_seed": [
 55      149.2514190673828,
 56      157.55279541015625,
 57      199.7967529296875,
 58      138.394287109375,
 59      229.0579833984375,
 60      128.88735961914062,
 61      193.53683471679688,
 62      263.369873046875
 63    ],
 64    "n": 8
 65  },
 66  "comparison": {
 67    "delta_mean": 14.651947021484375,
 68    "idea_wins": 1,
 69    "n_pairs": 8,
 70    "per_seed_diffs": [
 71      7.333770751953125,
 72      -40.90185546875,
 73      39.16777038574219,
 74      47.77540588378906,
 75      14.962173461914062,
 76      1.8077850341796875,
 77      34.76835632324219,
 78      12.302169799804688
 79    ],
 80    "p_value": 0.1557,
 81    "mde": 23.25227057624662,
 82    "mde_rel_pct": 13.854742188393438,
 83    "verdict": "no significant win",
 84    "system_worked": false
 85  },
 86  "mechanism_signature": {
 87    "prediction": "weighted feature variance ratio equals multinomial multiplicity N_alpha",
 88    "predicted_mean_ratio": 12.412121212121212,
 89    "observed_mean_ratio": 12.412121438416662,
 90    "median_relative_error": 3.4938962389929884e-08,
 91    "max_relative_error": 1.6764265427700783e-07,
 92    "n_feature_supports": 330,
 93    "confirmed": true
 94  },
 95  "idea_sweep": [
 96    {
 97      "cfg": {
 98        "lr": 0.001,
 99        "epochs": 20
100      },
101      "result": {
102        "mean": 408.1041793823242,
103        "std": 92.88364047045734,
104        "per_seed": [
105          297.24658203125,
106          362.6485595703125,
107          517.3526611328125,
108          342.9815979003906,
109          553.795166015625,
110          291.0316467285156,
111          444.25634765625,
112          455.5208740234375
113        ],
114        "n": 8
115      }
116    },
117    {
118      "cfg": {
119        "lr": 0.003,
120        "epochs": 20
121      },
122      "result": {
123        "mean": 300.2734432220459,
124        "std": 83.23694316909196,
125        "per_seed": [
126          215.8817596435547,
127          243.940185546875,
128          370.4352111816406,
129          231.63311767578125,
130          437.728271484375,
131          198.990478515625,
132          329.0107421875,
133          374.5677795410156
134        ],
135        "n": 8
136      }
137    },
138    {
139      "cfg": {
140        "lr": 0.01,
141        "epochs": 20
142      },
143      "result": {
144        "mean": 182.48091316223145,
145        "std": 44.266280792836994,
146        "per_seed": [
147          149.2514190673828,
148          157.55279541015625,
149          199.7967529296875,
150          138.394287109375,
151          229.0579833984375,
152          128.88735961914062,
153          193.53683471679688,
154          263.369873046875
155        ],
156        "n": 8
157      }
158    }
159  ],
160  "custom_track": {
161    "name": "symmetric_polynomial_regression",
162    "file": "custom_symmetric_poly.py",
163    "domain": "symmetric_interactions"
164  },
165  "runtime_sec": 434.1459777355194
166}