Dual Information-Demand Curiosity / 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    "best_cfg": {
  9      "lr": 0.01,
 10      "pressure": 0.2
 11    },
 12    "sweep": [
 13      {
 14        "cfg": {
 15          "lr": 0.001,
 16          "pressure": 0.0
 17        },
 18        "mean": 0.0035706528869923204
 19      },
 20      {
 21        "cfg": {
 22          "lr": 0.001,
 23          "pressure": 0.05
 24        },
 25        "mean": 0.003419904154725373
 26      },
 27      {
 28        "cfg": {
 29          "lr": 0.001,
 30          "pressure": 0.2
 31        },
 32        "mean": 0.0036505938041955233
 33      },
 34      {
 35        "cfg": {
 36          "lr": 0.003,
 37          "pressure": 0.0
 38        },
 39        "mean": 0.005002945079468191
 40      },
 41      {
 42        "cfg": {
 43          "lr": 0.003,
 44          "pressure": 0.05
 45        },
 46        "mean": 0.005080707313027233
 47      },
 48      {
 49        "cfg": {
 50          "lr": 0.003,
 51          "pressure": 0.2
 52        },
 53        "mean": 0.005320765834767371
 54      },
 55      {
 56        "cfg": {
 57          "lr": 0.01,
 58          "pressure": 0.0
 59        },
 60        "mean": 0.0013175007770769298
 61      },
 62      {
 63        "cfg": {
 64          "lr": 0.01,
 65          "pressure": 0.05
 66        },
 67        "mean": 0.0012840094132116064
 68      },
 69      {
 70        "cfg": {
 71          "lr": 0.01,
 72          "pressure": 0.2
 73        },
 74        "mean": 0.0011667622748063877
 75      }
 76    ],
 77    "full": {
 78      "mean": 0.0011755235827877186,
 79      "std": 0.0003522616089917298,
 80      "per_seed": [
 81        0.0008392606396228075,
 82        0.0014225966297090054,
 83        0.0006899926229380071,
 84        0.001715199206955731,
 85        0.0013351935194805264,
 86        0.0015198795590549707,
 87        0.0010780569864436984,
 88        0.0008040094980970025
 89      ],
 90      "n": 8
 91    }
 92  },
 93  "idea": {
 94    "mean": 0.0010581257738522254,
 95    "std": 0.0002584312332048161,
 96    "per_seed": [
 97      0.0007879937184043229,
 98      0.0010015293955802917,
 99      0.0013035451993346214,
100      0.001240157289430499,
101      0.0013518166961148381,
102      0.0013151094317436218,
103      0.0007202015840448439,
104      0.0007446528761647642
105    ],
106    "n": 8,
107    "details": [
108      {
109        "I_train": 0.6254459023475647,
110        "I_test": 0.8024823665618896,
111        "lambda": 0.5281313902530512,
112        "H_goal": 0.55,
113        "pressure": 0.2
114      },
115      {
116        "I_train": 0.8220813274383545,
117        "I_test": 1.0823215246200562,
118        "lambda": 0.0,
119        "H_goal": 0.55,
120        "pressure": 0.2
121      },
122      {
123        "I_train": 0.6989840865135193,
124        "I_test": 0.7982175350189209,
125        "lambda": 0.7593800543184935,
126        "H_goal": 0.55,
127        "pressure": 0.2
128      },
129      {
130        "I_train": 0.5356451272964478,
131        "I_test": 0.6818308234214783,
132        "lambda": 0.8704148933857491,
133        "H_goal": 0.55,
134        "pressure": 0.2
135      },
136      {
137        "I_train": 0.9433308243751526,
138        "I_test": 0.857042670249939,
139        "lambda": 0.5193827239810556,
140        "H_goal": 0.55,
141        "pressure": 0.2
142      },
143      {
144        "I_train": 0.7470368146896362,
145        "I_test": 0.62962806224823,
146        "lambda": 0.8655181587285236,
147        "H_goal": 0.55,
148        "pressure": 0.2
149      },
150      {
151        "I_train": 0.7524755001068115,
152        "I_test": 1.1005165576934814,
153        "lambda": 0.0,
154        "H_goal": 0.55,
155        "pressure": 0.2
156      },
157      {
158        "I_train": 0.5790018439292908,
159        "I_test": 0.9611145853996277,
160        "lambda": 0.40624716999978266,
161        "H_goal": 0.55,
162        "pressure": 0.2
163      }
164    ],
165    "best_cfg": {
166      "lr": 0.01,
167      "pressure": 0.2,
168      "eta_lambda": 0.08
169    },
170    "candidates": [
171      {
172        "cfg": {
173          "lr": 0.01,
174          "pressure": 0.2,
175          "eta_lambda": 0.04
176        },
177        "mean": 0.001126833783928305
178      },
179      {
180        "cfg": {
181          "lr": 0.01,
182          "pressure": 0.2,
183          "eta_lambda": 0.08
184        },
185        "mean": 0.0010581257738522254
186      },
187      {
188        "cfg": {
189          "lr": 0.01,
190          "pressure": 0.2,
191          "eta_lambda": 0.16
192        },
193        "mean": 0.0011072655142925214
194      }
195    ]
196  },
197  "comparison": {
198    "delta_mean": -0.00011739780893549323,
199    "idea_wins": 6,
200    "n_pairs": 8,
201    "per_seed_diffs": [
202      -5.126692121848464e-05,
203      -0.0004210672341287136,
204      0.0006135525763966143,
205      -0.00047504191752523184,
206      1.6623176634311676e-05,
207      -0.00020477012731134892,
208      -0.0003578554023988545,
209      -5.935662193223834e-05
210    ],
211    "p_value": 0.365,
212    "mde": 0.00029061097353965813,
213    "mde_rel_pct": 24.721832704579434,
214    "verdict": "no significant win",
215    "system_worked": false
216  },
217  "mechanism_signature": {
218    "bench_report": {
219      "track_match": "controlled dynamics / latent predictive information",
220      "custom_track": null
221    },
222    "signature": {
223      "H_goal": 0.55,
224      "baseline_fixed_pressure": 0.2,
225      "baseline_mean_I_test": 0.867881640791893,
226      "idea_mean_I_test": 0.8641442656517029,
227      "idea_mean_lambda": 0.493634298833332,
228      "predicted": "dual pressure increases when I_test/EMA is below H_goal and should reduce deficit",
229      "observed_deficit_baseline": -0.31788164079189296,
230      "observed_deficit_idea": -0.31414426565170284,
231      "confirmed": false
232    }
233  }
234}