Tau-leaped parallel discrete Hamiltonian sampler / bench_report.json

Mechanism confirmed, baseline not beaten

Raw ⬇ ZIP
  1{
  2  "bench_version": 1,
  3  "track": "categorical_energy",
  4  "model": "mlp_tiny",
  5  "metric_direction": "lower is better",
  6  "n_seeds": 8,
  7  "baseline": {
  8    "best_cfg": {
  9      "lr": 0.01
 10    },
 11    "sweep": [
 12      {
 13        "cfg": {
 14          "lr": 0.001
 15        },
 16        "mean": 0.29375000298023224
 17      },
 18      {
 19        "cfg": {
 20          "lr": 0.003
 21        },
 22        "mean": 0.26999999582767487
 23      },
 24      {
 25        "cfg": {
 26          "lr": 0.01
 27        },
 28        "mean": 0.20749999955296516
 29      }
 30    ],
 31    "full": {
 32      "mean": 0.2024999987334013,
 33      "std": 0.03614207525864678,
 34      "per_seed": [
 35        0.2199999988079071,
 36        0.20000000298023224,
 37        0.19499999284744263,
 38        0.2150000035762787,
 39        0.26499998569488525,
 40        0.15000000596046448,
 41        0.15000000596046448,
 42        0.22499999403953552
 43      ],
 44      "n": 8
 45    },
 46    "sweep_space": [
 47      {
 48        "lr": 0.001
 49      },
 50      {
 51        "lr": 0.003
 52      },
 53      {
 54        "lr": 0.01
 55      }
 56    ]
 57  },
 58  "idea": {
 59    "mean": 0.21249999850988388,
 60    "std": 0.0482182497334198,
 61    "per_seed": [
 62      0.2549999952316284,
 63      0.1850000023841858,
 64      0.18000000715255737,
 65      0.23000000417232513,
 66      0.29499998688697815,
 67      0.20499999821186066,
 68      0.125,
 69      0.22499999403953552
 70    ],
 71    "n": 8
 72  },
 73  "comparison": {
 74    "delta_mean": 0.009999999776482582,
 75    "idea_wins": 3,
 76    "n_pairs": 8,
 77    "per_seed_diffs": [
 78      0.034999996423721313,
 79      -0.015000000596046448,
 80      -0.014999985694885254,
 81      0.015000000596046448,
 82      0.030000001192092896,
 83      0.05499999225139618,
 84      -0.025000005960464478,
 85      0.0
 86    ],
 87    "p_value": 0.3679,
 88    "mde": 0.023755343439915824,
 89    "mde_rel_pct": 11.731033870864666,
 90    "verdict": "no significant win",
 91    "system_worked": false
 92  },
 93  "mechanism_signature": {
 94    "custom_track": {
 95      "name": "categorical_energy",
 96      "file": "categorical_track.py",
 97      "domain": "discrete_energy_sampling"
 98    },
 99    "idea_sweep": [
100      {
101        "h": 0.02,
102        "result": {
103          "mean": 0.21249999850988388,
104          "std": 0.0482182497334198,
105          "per_seed": [
106            0.2549999952316284,
107            0.1850000023841858,
108            0.18000000715255737,
109            0.23000000417232513,
110            0.29499998688697815,
111            0.20499999821186066,
112            0.125,
113            0.22499999403953552
114          ],
115          "n": 8
116        }
117      },
118      {
119        "h": 0.05,
120        "result": {
121          "mean": 0.22187499795109034,
122          "std": 0.045616437498093354,
123          "per_seed": [
124            0.2800000011920929,
125            0.20499999821186066,
126            0.2199999988079071,
127            0.23000000417232513,
128            0.26499998569488525,
129            0.20999999344348907,
130            0.11999999731779099,
131            0.24500000476837158
132          ],
133          "n": 8
134        }
135      },
136      {
137        "h": 0.1,
138        "result": {
139          "mean": 0.24874999932944775,
140          "std": 0.03927069276915124,
141          "per_seed": [
142            0.26499998569488525,
143            0.24500000476837158,
144            0.24500000476837158,
145            0.29499998688697815,
146            0.30000001192092896,
147            0.23499999940395355,
148            0.16500000655651093,
149            0.23999999463558197
150          ],
151          "n": 8
152        }
153      }
154    ],
155    "mechanism_signature": {
156      "quantity": "per-leap active proposal count versus resolved coordinate updates",
157      "predicted": 26.59152908809483,
158      "observed": 26.3515625,
159      "relative_error": 0.009024174100701296,
160      "confirmed": true,
161      "per_seed": [
162        {
163          "seed": 0,
164          "predicted_mean_counts": 26.762684151530266,
165          "observed_updates": 25.84375
166        },
167        {
168          "seed": 1,
169          "predicted_mean_counts": 27.059062629938126,
170          "observed_updates": 26.375
171        },
172        {
173          "seed": 2,
174          "predicted_mean_counts": 26.548943385481834,
175          "observed_updates": 26.21875
176        },
177        {
178          "seed": 3,
179          "predicted_mean_counts": 26.508921772241592,
180          "observed_updates": 26.78125
181        },
182        {
183          "seed": 4,
184          "predicted_mean_counts": 26.205237537622452,
185          "observed_updates": 25.40625
186        },
187        {
188          "seed": 5,
189          "predicted_mean_counts": 26.47282662987709,
190          "observed_updates": 25.3125
191        },
192        {
193          "seed": 6,
194          "predicted_mean_counts": 26.590720653533936,
195          "observed_updates": 28.15625
196        },
197        {
198          "seed": 7,
199          "predicted_mean_counts": 26.583835944533348,
200          "observed_updates": 26.71875
201        }
202      ]
203    }
204  }
205}