Flow-Efficiency Drift Scheduler / 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      "epochs": 10
 11    },
 12    "sweep": [
 13      {
 14        "cfg": {
 15          "lr": 0.0015,
 16          "epochs": 10
 17        },
 18        "mean": 0.012571729021146894
 19      },
 20      {
 21        "cfg": {
 22          "lr": 0.003,
 23          "epochs": 10
 24        },
 25        "mean": 0.00390045705717057
 26      },
 27      {
 28        "cfg": {
 29          "lr": 0.006,
 30          "epochs": 10
 31        },
 32        "mean": 0.0027824681892525405
 33      }
 34    ],
 35    "full": {
 36      "mean": 0.003422115682042204,
 37      "std": 0.0013572789915321942,
 38      "per_seed": [
 39        0.003793726908043027,
 40        0.0012769249733537436,
 41        0.004308650735765696,
 42        0.0017505701398476958,
 43        0.005116218701004982,
 44        0.00515680992975831,
 45        0.0032697871793061495,
 46        0.002704236889258027
 47      ],
 48      "n": 8
 49    }
 50  },
 51  "idea": {
 52    "mean": 0.007973545085405931,
 53    "std": 0.002709377013131552,
 54    "per_seed": [
 55      0.0038756646681576967,
 56      0.01088617742061615,
 57      0.004916933830827475,
 58      0.007167609874159098,
 59      0.00999990664422512,
 60      0.011676865629851818,
 61      0.009276396594941616,
 62      0.0059888060204684734
 63    ],
 64    "n": 8
 65  },
 66  "comparison": {
 67    "delta_mean": 0.004551429403363727,
 68    "idea_wins": 0,
 69    "n_pairs": 8,
 70    "per_seed_diffs": [
 71      8.19377601146698e-05,
 72      0.009609252447262406,
 73      0.000608283095061779,
 74      0.005417039734311402,
 75      0.0048836879432201385,
 76      0.006520055700093508,
 77      0.006006609415635467,
 78      0.0032845691312104464
 79    ],
 80    "p_value": 0.0081,
 81    "mde": 0.0026359025513121824,
 82    "mde_rel_pct": 77.02552444805619,
 83    "verdict": "idea worse (significant)",
 84    "system_worked": false
 85  },
 86  "math_check": {
 87    "ess_in_range": true,
 88    "ess_shift_error": 1.1102230246251565e-16
 89  },
 90  "idea_sweep": [
 91    {
 92      "cfg": {
 93        "lr": 0.0015,
 94        "epochs": 10
 95      },
 96      "mean": 0.010918576503172517
 97    },
 98    {
 99      "cfg": {
100        "lr": 0.003,
101        "epochs": 10
102      },
103      "mean": 0.006711596448440105
104    },
105    {
106      "cfg": {
107        "lr": 0.006,
108        "epochs": 10
109      },
110      "mean": 0.006873207807075232
111    }
112  ],
113  "runtime_sec": 28.547458171844482,
114  "mechanism_signature": {
115    "mean_eta": 0.7777270545781338,
116    "min_eta": 0.6528541115925,
117    "mean_drift": 0.32532993668262633,
118    "low_eta_levels": 0,
119    "initial_K": 4,
120    "final_K": 1,
121    "total_steps": 40,
122    "predicted_low_eta_under_drift": true,
123    "observed_low_eta_fraction": 0.0,
124    "confirmed": false
125  }
126}