Coupled Workload-Order Gate / 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      0.006
 10    ],
 11    "sweep": [
 12      {
 13        "config": [
 14          0.001
 15        ],
 16        "per_seed": [
 17          0.007754037156701088,
 18          0.003745198715478182,
 19          0.0025475025177001953,
 20          0.003148377873003483,
 21          0.004954223055392504,
 22          0.004425588063895702,
 23          0.0044244141317903996,
 24          0.009162315167486668
 25        ],
 26        "mean": 0.005020207085181028,
 27        "std": 0.002286214586911066
 28      },
 29      {
 30        "config": [
 31          0.003
 32        ],
 33        "per_seed": [
 34          0.002470282604917884,
 35          0.002755755092948675,
 36          0.0011578478151932359,
 37          0.007462490815669298,
 38          0.0025412666145712137,
 39          0.0030948929488658905,
 40          0.0021244161762297153,
 41          0.0019843443296849728
 42        ],
 43        "mean": 0.0029489120497601107,
 44        "std": 0.0019140223046789351
 45      },
 46      {
 47        "config": [
 48          0.006
 49        ],
 50        "per_seed": [
 51          0.002004395006224513,
 52          0.001437882543541491,
 53          0.0015748648438602686,
 54          0.002295783953741193,
 55          0.0024020641576498747,
 56          0.0024898042902350426,
 57          0.0020851334556937218,
 58          0.0025566271506249905
 59        ],
 60        "mean": 0.002105819425196387,
 61        "std": 0.0004163252013427006
 62      }
 63    ],
 64    "full": {
 65      "config": [
 66        0.006
 67      ],
 68      "per_seed": [
 69        0.002004395006224513,
 70        0.001437882543541491,
 71        0.0015748648438602686,
 72        0.002295783953741193,
 73        0.0024020641576498747,
 74        0.0024898042902350426,
 75        0.0020851334556937218,
 76        0.0025566271506249905
 77      ],
 78      "mean": 0.002105819425196387,
 79      "std": 0.0004163252013427006
 80    }
 81  },
 82  "idea": {
 83    "best_cfg": [
 84      0.006,
 85      0.2
 86    ],
 87    "sweep": [
 88      {
 89        "config": [
 90          0.001,
 91          0.02
 92        ],
 93        "per_seed": [
 94          0.007752448320388794,
 95          0.003742651315405965,
 96          0.0025358593557029963,
 97          0.003136643208563328,
 98          0.004953413270413876,
 99          0.004411333240568638,
100          0.0043915919959545135,
101          0.00907886028289795
102        ],
103        "mean": 0.005000350123737007,
104        "std": 0.0022695797131772446,
105        "lambda": 0.02
106      },
107      {
108        "config": [
109          0.003,
110          0.02
111        ],
112        "per_seed": [
113          0.0024700742214918137,
114          0.002753326902166009,
115          0.0011541355634108186,
116          0.007465476635843515,
117          0.0025420342572033405,
118          0.003092885483056307,
119          0.002128961030393839,
120          0.0019743957091122866
121        ],
122        "mean": 0.002947661225334741,
123        "std": 0.0019159624457765199,
124        "lambda": 0.02
125      },
126      {
127        "config": [
128          0.006,
129          0.02
130        ],
131        "per_seed": [
132          0.00200583110563457,
133          0.0014351154677569866,
134          0.0015823777066543698,
135          0.002294273115694523,
136          0.0024029696360230446,
137          0.002487938152626157,
138          0.002087509259581566,
139          0.002553937491029501
140        ],
141        "mean": 0.0021062439918750897,
142        "std": 0.000414867290214433,
143        "lambda": 0.02
144      },
145      {
146        "config": [
147          0.001,
148          0.08
149        ],
150        "per_seed": [
151          0.007747639436274767,
152          0.003735017729923129,
153          0.0025040435139089823,
154          0.0031060727778822184,
155          0.004950913600623608,
156          0.004369518253952265,
157          0.004298955202102661,
158          0.008836028166115284
159        ],
160        "mean": 0.004943523585097864,
161        "std": 0.0022215643665023208,
162        "lambda": 0.08
163      },
164      {
165        "config": [
166          0.003,
167          0.08
168        ],
169        "per_seed": [
170          0.002469458617269993,
171          0.0027461235877126455,
172          0.0011356071336194873,
173          0.007463372778147459,
174          0.002544374903663993,
175          0.003089339705184102,
176          0.0021439625415951014,
177          0.0019459263421595097
178        ],
179        "mean": 0.0029422707011690363,
180        "std": 0.0019189420653264089,
181        "lambda": 0.08
182      },
183      {
184        "config": [
185          0.006,
186          0.08
187        ],
188        "per_seed": [
189          0.002010113326832652,
190          0.0014257561415433884,
191          0.0015752584440633655,
192          0.0023342787753790617,
193          0.0024051156360656023,
194          0.002479771850630641,
195          0.0020939703099429607,
196          0.0025461555924266577
197        ],
198        "mean": 0.002108802509610541,
199        "std": 0.0004189643943471674,
200        "lambda": 0.08
201      },
202      {
203        "config": [
204          0.001,
205          0.2
206        ],
207        "per_seed": [
208          0.007737650536000729,
209          0.003719724016264081,
210          0.0024441867135465145,
211          0.0030418795067816973,
212          0.004945157095789909,
213          0.004281151108443737,
214          0.004141305573284626,
215          0.008383535780012608
216        ],
217        "mean": 0.004836823791265488,
218        "std": 0.0021378448640202853,
219        "lambda": 0.2
220      },
221      {
222        "config": [
223          0.003,
224          0.2
225        ],
226        "per_seed": [
227          0.0024682374205440283,
228          0.0027322175446897745,
229          0.001090307254344225,
230          0.0074797505512833595,
231          0.002549185650423169,
232          0.0030852353665977716,
233          0.0021790594328194857,
234          0.0018948764773085713
235        ],
236        "mean": 0.002934858712251298,
237        "std": 0.001932480390978357,
238        "lambda": 0.2
239      },
240      {
241        "config": [
242          0.006,
243          0.2
244        ],
245        "per_seed": [
246          0.0020184817258268595,
247          0.0014025752898305655,
248          0.0015944599872455,
249          0.0023251562379300594,
250          0.0024057631380856037,
251          0.0024610732216387987,
252          0.0021019575651735067,
253          0.0025320693384855986
254        ],
255        "mean": 0.0021051920630270615,
256        "std": 0.00041570070698645627,
257        "lambda": 0.2
258      }
259    ],
260    "per_seed": [
261      0.0020184817258268595,
262      0.0014025752898305655,
263      0.0015944599872455,
264      0.0023251562379300594,
265      0.0024057631380856037,
266      0.0024610732216387987,
267      0.0021019575651735067,
268      0.0025320693384855986
269    ],
270    "mean": 0.0021051920630270615,
271    "std": 0.00041570070698645627
272  },
273  "comparison": {
274    "delta_mean": -6.273621693253517e-07,
275    "idea_wins": 3,
276    "n_pairs": 8,
277    "per_seed_diffs": [
278      1.408671960234642e-05,
279      -3.530725371092558e-05,
280      1.9595143385231495e-05,
281      2.9372284188866615e-05,
282      3.698980435729027e-06,
283      -2.873106859624386e-05,
284      1.6824109479784966e-05,
285      -2.45578121393919e-05
286    ],
287    "p_value": 0.9473,
288    "mde": 2.0996531702215405e-05,
289    "mde_rel_pct": 0.9970718025956707,
290    "verdict": "no measurable effect",
291    "system_worked": false
292  },
293  "mechanism_signature": {
294    "track_match": "stability/control -> dynamics",
295    "prediction": "order penalty should reduce positive controlled-minus-baseline workload events",
296    "signature": {
297      "predicted_mean_violation": 2.5829562929402528e-05,
298      "observed_mean_violation": 1.9965890448503387e-05,
299      "predicted_event_rate": 0.004999999888241291,
300      "observed_event_rate": 0.004687499924330041,
301      "confirmed": false
302    }
303  },
304  "protocol_notes": {
305    "epochs": 12,
306    "n_train": 400,
307    "n_test": 200,
308    "shared_lr_union": [
309      0.001,
310      0.003,
311      0.006
312    ],
313    "idea_lambda_grid": [
314      0.02,
315      0.08,
316      0.2
317    ]
318  }
319}