Joint Modeling for Stochastic Interventions / custom_bench_results.json

Mechanism confirmed, baseline not beaten

Raw ⬇ ZIP
  1{
  2  "track": {
  3    "name": "stochastic_intervention_scm",
  4    "domain": "causal-world-model",
  5    "description": "Random intervention X causes mediator M and outcome Y; tests joint modeling under intervention shift and mediator selection."
  6  },
  7  "official_bench_available": false,
  8  "infrastructure_note": "Specified /home/maxwelhelp/all/math2nn/bench and README.md were absent; this is the required-contract local fallback, not bench.make_report output.",
  9  "math_check": {
 10    "selected_m_equals_1": {
 11      "sigma1": [
 12        1.483637689390105,
 13        1.5,
 14        2617
 15      ],
 16      "sigma2": [
 17        1.7395092655751987,
 18        1.8,
 19        1897
 20      ]
 21    },
 22    "shift_observed": true
 23  },
 24  "budget": {
 25    "seeds": [
 26      0,
 27      1,
 28      2,
 29      3,
 30      4,
 31      5,
 32      6,
 33      7
 34    ],
 35    "epochs": 90,
 36    "batch": 128,
 37    "lr_union": [
 38      0.003,
 39      0.01,
 40      0.03
 41    ]
 42  },
 43  "baseline_sweep": {
 44    "0.003": 2.854752391576767,
 45    "0.01": 2.711322247982025,
 46    "0.03": 2.849326878786087
 47  },
 48  "idea_sweep": {
 49    "0.003": 1.5709003955125809,
 50    "0.01": 1.6843366771936417,
 51    "0.03": 1.821063444018364
 52  },
 53  "best_baseline_lr": 0.01,
 54  "best_idea_lr": 0.003,
 55  "baseline_per_seed": [
 56    {
 57      "seed": 0,
 58      "lr": 0.01,
 59      "mse": 2.560922861099243,
 60      "selected_pred": 1.608640432357788,
 61      "selected_obs": 1.8415178060531616,
 62      "n_selected": 145
 63    },
 64    {
 65      "seed": 1,
 66      "lr": 0.01,
 67      "mse": 2.782212495803833,
 68      "selected_pred": 1.5625216960906982,
 69      "selected_obs": 2.0218238830566406,
 70      "n_selected": 117
 71    },
 72    {
 73      "seed": 2,
 74      "lr": 0.01,
 75      "mse": 2.7840850353240967,
 76      "selected_pred": 1.3886590003967285,
 77      "selected_obs": 1.9218453168869019,
 78      "n_selected": 122
 79    },
 80    {
 81      "seed": 3,
 82      "lr": 0.01,
 83      "mse": 2.6477458477020264,
 84      "selected_pred": 1.7123332023620605,
 85      "selected_obs": 1.9120752811431885,
 86      "n_selected": 150
 87    },
 88    {
 89      "seed": 4,
 90      "lr": 0.01,
 91      "mse": 2.5227646827697754,
 92      "selected_pred": 1.793519377708435,
 93      "selected_obs": 1.784800410270691,
 94      "n_selected": 126
 95    },
 96    {
 97      "seed": 5,
 98      "lr": 0.01,
 99      "mse": 2.7045862674713135,
100      "selected_pred": 1.6540354490280151,
101      "selected_obs": 1.926216721534729,
102      "n_selected": 132
103    },
104    {
105      "seed": 6,
106      "lr": 0.01,
107      "mse": 2.809804677963257,
108      "selected_pred": 1.7043554782867432,
109      "selected_obs": 1.8070753812789917,
110      "n_selected": 138
111    },
112    {
113      "seed": 7,
114      "lr": 0.01,
115      "mse": 2.8784561157226562,
116      "selected_pred": 1.50861656665802,
117      "selected_obs": 1.8454731702804565,
118      "n_selected": 132
119    }
120  ],
121  "idea_per_seed": [
122    {
123      "seed": 0,
124      "lr": 0.003,
125      "mse": 1.385545253753662,
126      "selected_pred": 1.8287450075149536,
127      "selected_obs": 1.8415178060531616,
128      "n_selected": 145
129    },
130    {
131      "seed": 1,
132      "lr": 0.003,
133      "mse": 1.5307773351669312,
134      "selected_pred": 1.9122552871704102,
135      "selected_obs": 2.0218238830566406,
136      "n_selected": 117
137    },
138    {
139      "seed": 2,
140      "lr": 0.003,
141      "mse": 1.6703295707702637,
142      "selected_pred": 1.9163973331451416,
143      "selected_obs": 1.9218453168869019,
144      "n_selected": 122
145    },
146    {
147      "seed": 3,
148      "lr": 0.003,
149      "mse": 1.5728068351745605,
150      "selected_pred": 1.8979436159133911,
151      "selected_obs": 1.9120752811431885,
152      "n_selected": 150
153    },
154    {
155      "seed": 4,
156      "lr": 0.003,
157      "mse": 1.3386306762695312,
158      "selected_pred": 1.7606240510940552,
159      "selected_obs": 1.784800410270691,
160      "n_selected": 126
161    },
162    {
163      "seed": 5,
164      "lr": 0.003,
165      "mse": 1.760671854019165,
166      "selected_pred": 1.8243013620376587,
167      "selected_obs": 1.926216721534729,
168      "n_selected": 132
169    },
170    {
171      "seed": 6,
172      "lr": 0.003,
173      "mse": 1.8169323205947876,
174      "selected_pred": 1.844454288482666,
175      "selected_obs": 1.8070753812789917,
176      "n_selected": 138
177    },
178    {
179      "seed": 7,
180      "lr": 0.003,
181      "mse": 1.4915093183517456,
182      "selected_pred": 1.7713148593902588,
183      "selected_obs": 1.8454731702804565,
184      "n_selected": 132
185    }
186  ],
187  "paired_delta_mean": -1.1404218524694443,
188  "paired_deltas": [
189    -1.175377607345581,
190    -1.2514351606369019,
191    -1.113755464553833,
192    -1.0749390125274658,
193    -1.1841340065002441,
194    -0.9439144134521484,
195    -0.9928723573684692,
196    -1.3869467973709106
197  ],
198  "permutation_p": 0.007782101167315175,
199  "mechanism_signature": {
200    "selected_m": 1.0,
201    "baseline_predicted_mean": 1.616585150361061,
202    "idea_predicted_mean": 1.844504475593567,
203    "observed_mean": 1.882603496313095,
204    "baseline_abs_error": 0.266018345952034,
205    "idea_abs_error": 0.0380990207195282,
206    "confirmed": true
207  },
208  "bench_report": {
209    "custom_track": {
210      "name": "stochastic_intervention_scm",
211      "file": "custom_stochastic_intervention_track.py",
212      "domain": "causal-world-model"
213    },
214    "verdict": "idea better (significant)",
215    "permutation_p": 0.007782101167315175,
216    "paired_delta_mean": -1.1404218524694443
217  },
218  "permutation_p_raw": 0.00390625
219}