Barrier-Controlled Basin Switching / bench_report.json

Failed on benchmark

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": 0.006,
 10      "weight_decay": 0.0
 11    },
 12    "sweep": [
 13      {
 14        "cfg": {
 15          "lr": 0.001,
 16          "weight_decay": 0.0
 17        },
 18        "mean": 0.0015310924500226974
 19      },
 20      {
 21        "cfg": {
 22          "lr": 0.001,
 23          "weight_decay": 0.0001
 24        },
 25        "mean": 0.0015161814808379859
 26      },
 27      {
 28        "cfg": {
 29          "lr": 0.003,
 30          "weight_decay": 0.0
 31        },
 32        "mean": 0.0008559521666029468
 33      },
 34      {
 35        "cfg": {
 36          "lr": 0.003,
 37          "weight_decay": 0.0001
 38        },
 39        "mean": 0.0008401070081163198
 40      },
 41      {
 42        "cfg": {
 43          "lr": 0.006,
 44          "weight_decay": 0.0
 45        },
 46        "mean": 0.0005687399534508586
 47      },
 48      {
 49        "cfg": {
 50          "lr": 0.006,
 51          "weight_decay": 0.0001
 52        },
 53        "mean": 0.000572658478631638
 54      }
 55    ],
 56    "full": {
 57      "mean": 0.0005194859513721894,
 58      "std": 0.0001693891077925739,
 59      "per_seed": [
 60        0.0005914251087233424,
 61        0.0007362987380474806,
 62        0.00035295303678140044,
 63        0.0005942829302512109,
 64        0.0002958264376502484,
 65        0.0005069610779173672,
 66        0.00032333534909412265,
 67        0.0007548049325123429
 68      ],
 69      "n": 8
 70    }
 71  },
 72  "idea": {
 73    "per_seed": [
 74      0.0005676778964698315,
 75      0.0006774243083782494,
 76      0.00039852733607403934,
 77      0.0005248879315331578,
 78      0.0003072890976909548,
 79      0.000477546826004982,
 80      0.00033985383925028145,
 81      0.000700155389495194
 82    ],
 83    "mean": 0.0004991703281120863
 84  },
 85  "comparison": {
 86    "delta_mean": -2.0315623260103166e-05,
 87    "idea_wins": 5,
 88    "n_pairs": 8,
 89    "per_seed_diffs": [
 90      -2.3747212253510952e-05,
 91      -5.8874429669231176e-05,
 92      4.55742992926389e-05,
 93      -6.93949987180531e-05,
 94      1.1462660040706396e-05,
 95      -2.9414251912385225e-05,
 96      1.6518490156158805e-05,
 97      -5.464954301714897e-05
 98    ],
 99    "p_value": 0.20575,
100    "mde": 3.4442226319100816e-05,
101    "mde_rel_pct": 6.630059239932061,
102    "verdict": "no significant win",
103    "system_worked": false
104  },
105  "mechanism_signature": {
106    "barrier_noise_calibration": {
107      "predicted_slope": NaN,
108      "observed_slope": NaN,
109      "n_checkpoints": 0,
110      "confirmed": false
111    },
112    "q_safe": 8.0,
113    "custom_track": null
114  }
115}