Pole-radius tuning for gradient tracking / bench_report.json

Failed on benchmark

Raw ⬇ ZIP
  1{
  2  "bench_version": 1,
  3  "track": "tabular",
  4  "model": "mlp_tiny",
  5  "metric_direction": "lower is better",
  6  "n_seeds": 8,
  7  "baseline": {
  8    "best_cfg": {
  9      "lr": 0.003
 10    },
 11    "sweep": [
 12      {
 13        "cfg": {
 14          "lr": 0.001
 15        },
 16        "mean": 210.5964012145996
 17      },
 18      {
 19        "cfg": {
 20          "lr": 0.003
 21        },
 22        "mean": 19.421141624450684
 23      },
 24      {
 25        "cfg": {
 26          "lr": 0.01
 27        },
 28        "mean": 89.58010625839233
 29      }
 30    ],
 31    "full": {
 32      "mean": 22.994831800460815,
 33      "std": 11.385712752026535,
 34      "per_seed": [
 35        27.76300621032715,
 36        13.660076141357422,
 37        19.283185958862305,
 38        16.97829818725586,
 39        17.96718406677246,
 40        18.604217529296875,
 41        18.24533462524414,
 42        51.45735168457031
 43      ],
 44      "n": 8
 45    }
 46  },
 47  "idea": {
 48    "mean": 209.8788948059082,
 49    "std": 12.027703233030866,
 50    "per_seed": [
 51      225.73562622070312,
 52      209.03338623046875,
 53      215.49949645996094,
 54      192.11709594726562,
 55      217.17715454101562,
 56      194.1483612060547,
 57      201.66188049316406,
 58      223.6581573486328
 59    ],
 60    "n": 8
 61  },
 62  "comparison": {
 63    "delta_mean": 186.8840630054474,
 64    "idea_wins": 0,
 65    "n_pairs": 8,
 66    "per_seed_diffs": [
 67      197.97262001037598,
 68      195.37331008911133,
 69      196.21631050109863,
 70      175.13879776000977,
 71      199.20997047424316,
 72      175.5441436767578,
 73      183.41654586791992,
 74      172.2008056640625
 75    ],
 76    "p_value": 0.0081,
 77    "mde": 9.628530681945449,
 78    "mde_rel_pct": 41.872585829275316,
 79    "verdict": "idea worse (significant)",
 80    "system_worked": false
 81  },
 82  "mechanism_signature": {
 83    "track_structure": "optimizer/decentralized regression",
 84    "alpha": 0.001,
 85    "mu_est": 0.249891459941864,
 86    "L_est": 0.249891459941864,
 87    "predicted_pole_radius": 0.5110536651589729,
 88    "observed_tracker_disagreement_ratio": 0.9515644814489328,
 89    "nondesignated_training_model": "mlp_tiny_worker_ensemble",
 90    "confirmed": false,
 91    "pole_spectrum": [
 92      -2.1467203015212988e-16,
 93      0.5,
 94      0.5000000000000001
 95    ],
 96    "idea_grid": [
 97      {
 98        "lr": 0.001
 99      },
100      {
101        "lr": 0.003
102      },
103      {
104        "lr": 0.01
105      }
106    ],
107    "selected_idea_cfg": {
108      "lr": 0.001
109    },
110    "candidate_full_means": {
111      "0.001": 209.8788948059082,
112      "0.003": 209.8788948059082,
113      "0.01": 209.8788948059082
114    }
115  }
116}