Floquet Monodromy Optimizer / 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    "sweep": {
  9      "best_cfg": {
 10        "lr": 0.006
 11      },
 12      "sweep": [
 13        {
 14          "cfg": {
 15            "lr": 0.001
 16          },
 17          "mean": 0.5709687098860741
 18        },
 19        {
 20          "cfg": {
 21            "lr": 0.003
 22          },
 23          "mean": 0.28142324835062027
 24        },
 25        {
 26          "cfg": {
 27            "lr": 0.006
 28          },
 29          "mean": 0.052330534905195236
 30        }
 31      ],
 32      "full": {
 33        "mean": 0.04564603161998093,
 34        "std": 0.019217470172042173,
 35        "per_seed": [
 36          0.0358407162129879,
 37          0.04454701393842697,
 38          0.03819940611720085,
 39          0.09073500335216522,
 40          0.030078278854489326,
 41          0.030246704816818237,
 42          0.03592348471283913,
 43          0.059597644954919815
 44        ],
 45        "n": 8
 46      }
 47    },
 48    "best_config": {
 49      "lr": 0.006
 50    },
 51    "full": {
 52      "lr": 0.006,
 53      "periodic": false,
 54      "per_seed": [
 55        0.0358407162129879,
 56        0.04454701393842697,
 57        0.03819940611720085,
 58        0.09073500335216522,
 59        0.030078278854489326,
 60        0.030246704816818237,
 61        0.03592348471283913,
 62        0.059597644954919815
 63      ],
 64      "mean": 0.04564603161998093,
 65      "std": 0.020544339778379852
 66    }
 67  },
 68  "idea": {
 69    "lr": 0.006,
 70    "periodic": true,
 71    "per_seed": [
 72      0.03649410977959633,
 73      0.04531322419643402,
 74      0.03744680434465408,
 75      0.08633020520210266,
 76      0.030837101861834526,
 77      0.02951119653880596,
 78      0.03519521281123161,
 79      0.062146708369255066
 80    ],
 81    "mean": 0.04540932038798928,
 82    "std": 0.01951914347909807
 83  },
 84  "comparison": {
 85    "delta_mean": -0.00023671123199164867,
 86    "idea_wins": 4,
 87    "n_pairs": 8,
 88    "per_seed_diffs": [
 89      0.000653393566608429,
 90      0.0007662102580070496,
 91      -0.0007526017725467682,
 92      -0.004404798150062561,
 93      0.0007588230073451996,
 94      -0.0007355082780122757,
 95      -0.0007282719016075134,
 96      0.002549063414335251
 97    ],
 98    "p_value": 0.7555,
 99    "mde": 0.0016924713304113021,
100    "mde_rel_pct": 3.707817022302648,
101    "verdict": "no measurable effect",
102    "system_worked": false
103  },
104  "mechanism_signature": {
105    "track_match": "dynamics is the built-in stability/control track",
106    "schedule": {
107      "phase_steps": [
108        1,
109        1
110      ],
111      "lr_multipliers": [
112        1.6,
113        0.3999999999999999
114      ]
115    },
116    "predicted_vs_observed": {
117      "seed": 0,
118      "test_mse": 0.03649410977959633,
119      "predicted_rho_product": 0.9999960247634755,
120      "observed_two_phase_gain": 1.000000023644702,
121      "relative_error": 3.998881131933376e-06,
122      "confirmed": true
123    }
124  },
125  "idea_sweep": [
126    {
127      "lr": 0.006,
128      "periodic": true,
129      "per_seed": [
130        0.03649410977959633,
131        0.04531322419643402,
132        0.03744680434465408,
133        0.08633020520210266,
134        0.030837101861834526,
135        0.02951119653880596,
136        0.03519521281123161,
137        0.062146708369255066
138      ],
139      "mean": 0.04540932038798928,
140      "std": 0.01951914347909807
141    },
142    {
143      "lr": 0.001,
144      "periodic": true,
145      "per_seed": [
146        0.4697585105895996,
147        0.6145305633544922,
148        0.5424293875694275,
149        0.6560655236244202,
150        0.49593931436538696,
151        0.5435400009155273,
152        0.5226424336433411,
153        0.6273190379142761
154      ],
155      "mean": 0.5590280964970589,
156      "std": 0.0664927567198825
157    },
158    {
159      "lr": 0.003,
160      "periodic": true,
161      "per_seed": [
162        0.22396183013916016,
163        0.29183024168014526,
164        0.2440265566110611,
165        0.3627980649471283,
166        0.21796903014183044,
167        0.23902076482772827,
168        0.23642562329769135,
169        0.3251190781593323
170      ],
171      "mean": 0.26764389872550964,
172      "std": 0.05300950027246798
173    }
174  ],
175  "runtime_sec": 87.40311336517334
176}