Non-Abelian Event-Order Memory / bench_report.json

Failed on benchmark

Raw ⬇ ZIP
  1{
  2  "bench_version": 1,
  3  "track": "sequence",
  4  "model": "transformer_tiny",
  5  "metric_direction": "lower is better",
  6  "n_seeds": 8,
  7  "baseline": {
  8    "best_cfg": {
  9      "lr": 0.001
 10    },
 11    "sweep": [
 12      {
 13        "cfg": {
 14          "lr": 0.001
 15        },
 16        "mean": 0.22030843049287796
 17      },
 18      {
 19        "cfg": {
 20          "lr": 0.003
 21        },
 22        "mean": 0.2290698178112507
 23      },
 24      {
 25        "cfg": {
 26          "lr": 0.01
 27        },
 28        "mean": 1.0897138714790344
 29      }
 30    ],
 31    "full": {
 32      "mean": 0.22894670069217682,
 33      "std": 0.02468231743126849,
 34      "per_seed": [
 35        0.2348448932170868,
 36        0.18358114361763,
 37        0.2119988203048706,
 38        0.25080886483192444,
 39        0.21615827083587646,
 40        0.258085697889328,
 41        0.258510559797287,
 42        0.21758535504341125
 43      ],
 44      "n": 8
 45    }
 46  },
 47  "idea": {
 48    "mean": 0.23161649331450462,
 49    "std": 0.0217385249680007,
 50    "per_seed": [
 51      0.23855143785476685,
 52      0.20657582581043243,
 53      0.2038293331861496,
 54      0.27797847986221313,
 55      0.22990234196186066,
 56      0.2412124127149582,
 57      0.23263387382030487,
 58      0.22224824130535126
 59    ],
 60    "n": 8
 61  },
 62  "comparison": {
 63    "delta_mean": 0.0026697926223278046,
 64    "idea_wins": 3,
 65    "n_pairs": 8,
 66    "per_seed_diffs": [
 67      0.0037065446376800537,
 68      0.02299468219280243,
 69      -0.008169487118721008,
 70      0.027169615030288696,
 71      0.013744071125984192,
 72      -0.016873285174369812,
 73      -0.025876685976982117,
 74      0.0046628862619400024
 75    ],
 76    "p_value": 0.6818,
 77    "mde": 0.015660083437405745,
 78    "mde_rel_pct": 6.8400563930646125,
 79    "verdict": "no significant win",
 80    "system_worked": false
 81  },
 82  "mechanism_signature": {
 83    "idea_sweep": [
 84      {
 85        "cfg": {
 86          "lr": 0.001
 87        },
 88        "mean": 0.2317337691783905
 89      },
 90      {
 91        "cfg": {
 92          "lr": 0.003
 93        },
 94        "mean": 0.26143352687358856
 95      },
 96      {
 97        "cfg": {
 98          "lr": 0.01
 99        },
100        "mean": 1.1830730140209198
101      }
102    ],
103    "selected_idea_cfg": {
104      "lr": 0.001
105    },
106    "behavior": {
107      "prediction": "ordered noncommuting updates preserve sphere norm and create order sensitivity",
108      "predicted_norm_error": 0.0,
109      "observed_max_norm_error": 1.3113021850585938e-06,
110      "predicted_order_delta_nonzero": true,
111      "observed_mean_reversal_spin_delta": 1.0346672534942627,
112      "observed_axis_cross_norm": 0.9981767535209656,
113      "confirmed": true,
114      "note": "All values are measured from the trained benchmark spin model."
115    }
116  }
117}