nur-dev's picture
Publish reproducibility artifacts (part 2)
fe31b61 verified
Raw History Blame Contribute Delete
12 kB
{
"status": "CONTROLLED_CONFIRMATION_COMPLETE",
"compact_representation_advantage": true,
"strongest_full_information_noninferiority": true,
"complete_cost_saving_gate": false,
"controlled_quality_cost_gate": false,
"practical_algorithm_promotion": false,
"promotion_requires_independent_external_confirmation": true,
"external_generalization_established": false,
"learned_control_advantage_tested": false,
"independent_external_authorship": false,
"worlds": 192,
"confirmation_families": [
"context",
"independent",
"orientation_alias",
"parity",
"redundant",
"uninformative"
],
"held_mechanisms": [
"redundant",
"orientation_alias"
],
"compact_comparisons": [
{
"control": "posterior_clusters",
"mean_gain": 0.14697129968150166,
"paired_interval": [
0.13471196139390745,
0.15968306092574677
],
"simultaneous_compact_familywise_95": true
},
{
"control": "posterior_bins",
"mean_gain": 0.16138521643219014,
"paired_interval": [
0.147432804001906,
0.17584175844992272
],
"simultaneous_compact_familywise_95": true
},
{
"control": "future_marginal",
"mean_gain": 0.32264581338832143,
"paired_interval": [
0.2985787360521876,
0.3464689306627599
],
"simultaneous_compact_familywise_95": true
},
{
"control": "raw_information",
"mean_gain": 0.10134525577106183,
"paired_interval": [
0.08829703798292718,
0.11530101552134753
],
"simultaneous_compact_familywise_95": true
},
{
"control": "raw_entropy",
"mean_gain": 0.13230635118148604,
"paired_interval": [
0.10888420366853153,
0.15790720468260205
],
"simultaneous_compact_familywise_95": true
},
{
"control": "hash_code",
"mean_gain": 0.442815633391068,
"paired_interval": [
0.41421657602254525,
0.4709814074719597
],
"simultaneous_compact_familywise_95": true
},
{
"control": "identity",
"mean_gain": 0.009383933645959215,
"paired_interval": [
0.007968504722197413,
0.010786756165418913
],
"simultaneous_compact_familywise_95": false
}
],
"full_model_comparisons": [
{
"model": "joint_jeffreys",
"mean_gain": -0.004071171127615503,
"paired_interval": [
-0.00771759383814986,
-0.0006007768451799707
]
},
{
"model": "naive_bayes",
"mean_gain": 0.34755162582326143,
"paired_interval": [
0.3297362785418623,
0.3653320771060183
]
},
{
"model": "support_mixture",
"mean_gain": 0.009383933645959215,
"paired_interval": [
0.007659803108744751,
0.011158058055535202
]
}
],
"complete_cost_comparisons": [
{
"requests": 1,
"reference_model": "joint_jeffreys",
"mean_saving_fraction": -0.013921024857006444,
"simultaneous_interval": [
-0.013969240432909374,
-0.013877061937501574
]
},
{
"requests": 1,
"reference_model": "naive_bayes",
"mean_saving_fraction": -0.013845724131802567,
"simultaneous_interval": [
-0.013880954679142564,
-0.013811733225877244
]
},
{
"requests": 1,
"reference_model": "support_mixture",
"mean_saving_fraction": -0.013842032727919758,
"simultaneous_interval": [
-0.013884064612882033,
-0.013796852723081092
]
},
{
"requests": 32,
"reference_model": "joint_jeffreys",
"mean_saving_fraction": -0.013920455094430356,
"simultaneous_interval": [
-0.013968832559706549,
-0.013876478486611422
]
},
{
"requests": 32,
"reference_model": "naive_bayes",
"mean_saving_fraction": -0.01384542449313253,
"simultaneous_interval": [
-0.013880269697593876,
-0.013811323288608854
]
},
{
"requests": 32,
"reference_model": "support_mixture",
"mean_saving_fraction": -0.013844039121007935,
"simultaneous_interval": [
-0.013884546586254001,
-0.01380148168833231
]
},
{
"requests": 1024,
"reference_model": "joint_jeffreys",
"mean_saving_fraction": -0.013920371340398464,
"simultaneous_interval": [
-0.013968424220477059,
-0.013876687819006523
]
},
{
"requests": 1024,
"reference_model": "naive_bayes",
"mean_saving_fraction": -0.013845093031752369,
"simultaneous_interval": [
-0.013879887384242996,
-0.013811179045041536
]
},
{
"requests": 1024,
"reference_model": "support_mixture",
"mean_saving_fraction": -0.013843992910080549,
"simultaneous_interval": [
-0.013884262398485883,
-0.013801808369034135
]
},
{
"requests": 65536,
"reference_model": "joint_jeffreys",
"mean_saving_fraction": -0.013920669189388219,
"simultaneous_interval": [
-0.013967500838619938,
-0.013877967585586374
]
},
{
"requests": 65536,
"reference_model": "naive_bayes",
"mean_saving_fraction": -0.013844868091559069,
"simultaneous_interval": [
-0.01387957315760266,
-0.013811546996287807
]
},
{
"requests": 65536,
"reference_model": "support_mixture",
"mean_saving_fraction": -0.013844312846015646,
"simultaneous_interval": [
-0.013883840436769558,
-0.013802728440505964
]
}
],
"threshold_changes_after_opening": false,
"primary_sample_budget": 1024,
"explicit_information_loss": 0.1991192480780211,
"primary_metrics": [
{
"method": "future_marginal",
"utility": 0.32892238335837554,
"queries": 0.5544834201523068,
"abstention": 0.46806334164019353,
"full_oracle": 0.713985016995153,
"available_oracle": 0.713985016995153,
"encoded_oracle": 0.3613625440057875,
"information_loss": 0.0,
"representation_loss": 0.35262247298936567,
"model_loss": 0.03244016064741188,
"total_regret": 0.3850626336367775,
"posterior_brier": 0.18667180330285535,
"posterior_log_loss": 0.5439733224103968,
"posterior_absolute_error": 0.04995626393083208,
"joint_forecast_log_loss": 1.5008272180173519,
"estimated_conditional_dispersion": 0.16299993023324447,
"true_conditional_dispersion": 0.20488680472036472
},
{
"method": "hash_code",
"utility": 0.20875256335562895,
"queries": 0.7309687722558658,
"abstention": 0.5406940283450717,
"full_oracle": 0.713985016995153,
"available_oracle": 0.713985016995153,
"encoded_oracle": 0.23524076669617813,
"information_loss": 0.0,
"representation_loss": 0.47874425029897494,
"model_loss": 0.026488203340549143,
"total_regret": 0.5052324536395242,
"posterior_brier": 0.22622160322616844,
"posterior_log_loss": 0.6436761288360193,
"posterior_absolute_error": 0.03708260319522386,
"joint_forecast_log_loss": 1.8019787478500033,
"estimated_conditional_dispersion": 0.30544615080378384,
"true_conditional_dispersion": 0.455228645801743
},
{
"method": "identity",
"utility": 0.6421842631007378,
"queries": 0.8457815187513584,
"abstention": 0.2138083247552466,
"full_oracle": 0.713985016995153,
"available_oracle": 0.713985016995153,
"encoded_oracle": 0.713985016995153,
"information_loss": 0.0,
"representation_loss": 0.0,
"model_loss": 0.07180075389441531,
"total_regret": 0.07180075389441531,
"posterior_brier": 0.14333381976139412,
"posterior_log_loss": 0.4452827495877168,
"posterior_absolute_error": 0.08959697227294662,
"joint_forecast_log_loss": 1.3050584378398493,
"estimated_conditional_dispersion": 0.0,
"true_conditional_dispersion": 0.0
},
{
"method": "posterior_bins",
"utility": 0.49018298031450686,
"queries": 0.5895838966415413,
"abstention": 0.31021806802597485,
"full_oracle": 0.713985016995153,
"available_oracle": 0.713985016995153,
"encoded_oracle": 0.545985131135212,
"information_loss": 0.0,
"representation_loss": 0.1679998858599411,
"model_loss": 0.05580215082070506,
"total_regret": 0.22380203668064624,
"posterior_brier": 0.14291202653073196,
"posterior_log_loss": 0.4449479999569366,
"posterior_absolute_error": 0.08309387780050025,
"joint_forecast_log_loss": 1.6503074140379088,
"estimated_conditional_dispersion": 0.2281804892849899,
"true_conditional_dispersion": 0.3687416760339481
},
{
"method": "posterior_clusters",
"utility": 0.5045968970651954,
"queries": 0.6181947218653467,
"abstention": 0.3066951890071943,
"full_oracle": 0.713985016995153,
"available_oracle": 0.713985016995153,
"encoded_oracle": 0.5646410630749022,
"information_loss": 0.0,
"representation_loss": 0.14934395392025088,
"model_loss": 0.06004416600970683,
"total_regret": 0.20938811992995776,
"posterior_brier": 0.1434919887557921,
"posterior_log_loss": 0.44586164115898436,
"posterior_absolute_error": 0.0854958494033341,
"joint_forecast_log_loss": 1.6175041666159624,
"estimated_conditional_dispersion": 0.21230823773731464,
"true_conditional_dispersion": 0.34122591050593654
},
{
"method": "query_joint",
"utility": 0.651568196746697,
"queries": 0.8261666952832235,
"abstention": 0.22481242601684215,
"full_oracle": 0.713985016995153,
"available_oracle": 0.713985016995153,
"encoded_oracle": 0.7039206512469436,
"information_loss": 0.0,
"representation_loss": 0.010064365748209458,
"model_loss": 0.05235245450024661,
"total_regret": 0.062416820248456094,
"posterior_brier": 0.14251900862053932,
"posterior_log_loss": 0.4446969468578407,
"posterior_absolute_error": 0.08044187730691645,
"joint_forecast_log_loss": 1.2992309455933082,
"estimated_conditional_dispersion": 0.0820432908814774,
"true_conditional_dispersion": 0.02417009051344016
},
{
"method": "raw_entropy",
"utility": 0.5192618455652109,
"queries": 0.9577800957873994,
"abstention": 0.29482544574330866,
"full_oracle": 0.713985016995153,
"available_oracle": 0.713985016995153,
"encoded_oracle": 0.5845117900153592,
"information_loss": 0.0,
"representation_loss": 0.12947322697979383,
"model_loss": 0.06524994445014827,
"total_regret": 0.19472317142994214,
"posterior_brier": 0.1951405121255698,
"posterior_log_loss": 0.5718216062912428,
"posterior_absolute_error": 0.043561287480735796,
"joint_forecast_log_loss": 1.5883154087248315,
"estimated_conditional_dispersion": 0.22589109948736963,
"true_conditional_dispersion": 0.30928958591753636
},
{
"method": "raw_information",
"utility": 0.5502229409756352,
"queries": 0.7165589001818898,
"abstention": 0.28175128030473223,
"full_oracle": 0.713985016995153,
"available_oracle": 0.713985016995153,
"encoded_oracle": 0.6070011596289279,
"information_loss": 0.0,
"representation_loss": 0.10698385736622527,
"model_loss": 0.05677821865329261,
"total_regret": 0.16376207601951792,
"posterior_brier": 0.15589534525171694,
"posterior_log_loss": 0.4736992751366403,
"posterior_absolute_error": 0.06634683165683118,
"joint_forecast_log_loss": 1.572122398926246,
"estimated_conditional_dispersion": 0.20662404968920522,
"true_conditional_dispersion": 0.31060705387670834
}
]
}