Removable-Pole Negative-Shifted Optimizer / 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": 0.01,
 10      "weight_decay": 0.0001
 11    },
 12    "sweep": [
 13      {
 14        "cfg": {
 15          "lr": 0.001,
 16          "weight_decay": 0.0
 17        },
 18        "mean": 25.60164165496826
 19      },
 20      {
 21        "cfg": {
 22          "lr": 0.001,
 23          "weight_decay": 0.0001
 24        },
 25        "mean": 25.508389472961426
 26      },
 27      {
 28        "cfg": {
 29          "lr": 0.003,
 30          "weight_decay": 0.0
 31        },
 32        "mean": 13.860755920410156
 33      },
 34      {
 35        "cfg": {
 36          "lr": 0.003,
 37          "weight_decay": 0.0001
 38        },
 39        "mean": 13.839483976364136
 40      },
 41      {
 42        "cfg": {
 43          "lr": 0.01,
 44          "weight_decay": 0.0
 45        },
 46        "mean": 8.352817177772522
 47      },
 48      {
 49        "cfg": {
 50          "lr": 0.01,
 51          "weight_decay": 0.0001
 52        },
 53        "mean": 8.268931150436401
 54      }
 55    ],
 56    "full": {
 57      "mean": 8.206986248493195,
 58      "std": 0.7130848431694788,
 59      "per_seed": [
 60        7.954751968383789,
 61        9.628533363342285,
 62        7.705244541168213,
 63        7.787194728851318,
 64        7.160106182098389,
 65        8.077467918395996,
 66        8.747560501098633,
 67        8.595030784606934
 68      ],
 69      "n": 8
 70    }
 71  },
 72  "idea": {
 73    "mean": 7.733814001083374,
 74    "std": 0.6299701216517714,
 75    "per_seed": [
 76      6.852144241333008,
 77      7.3872199058532715,
 78      7.651295185089111,
 79      8.405105590820312,
 80      7.134853839874268,
 81      7.516578197479248,
 82      8.040628433227539,
 83      8.882686614990234
 84    ],
 85    "n": 8
 86  },
 87  "comparison": {
 88    "delta_mean": -0.47317224740982056,
 89    "idea_wins": 6,
 90    "n_pairs": 8,
 91    "per_seed_diffs": [
 92      -1.1026077270507812,
 93      -2.2413134574890137,
 94      -0.05394935607910156,
 95      0.6179108619689941,
 96      -0.025252342224121094,
 97      -0.560889720916748,
 98      -0.7069320678710938,
 99      0.2876558303833008
100    ],
101    "p_value": 0.1835,
102    "mde": 0.756546047206517,
103    "mde_rel_pct": 9.218317471232746,
104    "verdict": "no significant win",
105    "system_worked": false
106  },
107  "mechanism_signature": {
108    "predicted": {
109      "one_step_displacement_factor": 1.089914695739746,
110      "nu": 8.991469573974609,
111      "curvature_scale": 59.94313049316406
112    },
113    "observed": {
114      "trained_positive_shift_disp": 50.0323600769043,
115      "trained_zero_shift_disp": 7.432165145874023,
116      "whole_run_ratio": 6.731868721281016
117    },
118    "confirmed": false
119  },
120  "idea_sweep": {
121    "grid": [
122      {
123        "cfg": {
124          "lr": 0.001,
125          "weight_decay": 0.0001,
126          "nu_mult": 0.0
127        },
128        "mean": 48.51860332489014
129      },
130      {
131        "cfg": {
132          "lr": 0.001,
133          "weight_decay": 0.0001,
134          "nu_mult": 0.05
135        },
136        "mean": 53.31177806854248
137      },
138      {
139        "cfg": {
140          "lr": 0.001,
141          "weight_decay": 0.0001,
142          "nu_mult": 0.15
143        },
144        "mean": 59.071059226989746
145      },
146      {
147        "cfg": {
148          "lr": 0.003,
149          "weight_decay": 0.0001,
150          "nu_mult": 0.0
151        },
152        "mean": 9.312313795089722
153      },
154      {
155        "cfg": {
156          "lr": 0.003,
157          "weight_decay": 0.0001,
158          "nu_mult": 0.05
159        },
160        "mean": 12.079964399337769
161      },
162      {
163        "cfg": {
164          "lr": 0.003,
165          "weight_decay": 0.0001,
166          "nu_mult": 0.15
167        },
168        "mean": 15.533822059631348
169      },
170      {
171        "cfg": {
172          "lr": 0.01,
173          "weight_decay": 0.0001,
174          "nu_mult": 0.0
175        },
176        "mean": 7.573941230773926
177      },
178      {
179        "cfg": {
180          "lr": 0.01,
181          "weight_decay": 0.0001,
182          "nu_mult": 0.05
183        },
184        "mean": 9.034136652946472
185      },
186      {
187        "cfg": {
188          "lr": 0.01,
189          "weight_decay": 0.0001,
190          "nu_mult": 0.15
191        },
192        "mean": 10.087280511856079
193      }
194    ],
195    "best_cfg": {
196      "lr": 0.01,
197      "weight_decay": 0.0001,
198      "nu_mult": 0.0
199    },
200    "shared_lr_union": [
201      0.001,
202      0.003,
203      0.01
204    ]
205  }
206}