Files
trolley-bench/scenarios/v2/reproduction-criteria.json
Zachary Landes 26576dfdef feat: add v2 benchmark and registered reproduction sample
* Kept the original raw answers while making later runs reproducible.
* Registered a variance-sized sample before the independent rerun.
* Removed article drafts and private paths from the current tree.
2026-10-14 15:02:40 -07:00

1481 lines
48 KiB
JSON

{
"seed": 20261014,
"cells": 2414,
"sample_sha256": "27612ec70104171551b7f142290bc17351715eebb3f177436fbc0cc200d59a25",
"alpha_familywise": 0.05,
"comparisons": 126,
"z": 3.542186105472276,
"target_power": 0.8,
"meaningful_mean_probability_shift": 0.1,
"families": {
"switch": {
"population_wordings": 24,
"sample_wordings": 24,
"selected_wordings": [
"c1",
"s05",
"s06",
"s07",
"s08",
"s09",
"s10",
"s11",
"s12",
"s13",
"s14",
"s15",
"s16",
"s17",
"s18",
"s19",
"s20",
"s21",
"s22",
"s23",
"s24",
"w2",
"w3",
"w4"
],
"power": {
"jev": {
"pull": {
"reference_wording_sd": 0.03598094251843624,
"assumed_difference_sd": 0.07196188503687248,
"infinite_n": 9.951945317251958,
"finite_n": 8
},
"pull_criteria": {
"reference_wording_sd": 0.02345841619697814,
"assumed_difference_sd": 0.04691683239395628,
"infinite_n": 4.230193996463641,
"finite_n": 4
},
"stay": {
"reference_wording_sd": 0.027382265031376748,
"assumed_difference_sd": 0.054764530062753496,
"infinite_n": 5.763703737999932,
"finite_n": 5
}
},
"clef": {
"pull": {
"reference_wording_sd": 0.033873187007904704,
"assumed_difference_sd": 0.06774637401580941,
"infinite_n": 8.820130882055908,
"finite_n": 7
},
"pull_criteria": {
"reference_wording_sd": 0.0337140521320507,
"assumed_difference_sd": 0.0674281042641014,
"infinite_n": 8.737452306415122,
"finite_n": 7
},
"stay": {
"reference_wording_sd": 0.03975794898753596,
"assumed_difference_sd": 0.07951589797507191,
"infinite_n": 12.15096736343694,
"finite_n": 9
}
},
"clef-flash": {
"pull": {
"reference_wording_sd": 0.05905199941561109,
"assumed_difference_sd": 0.11810399883122218,
"infinite_n": 26.8060068148224,
"finite_n": 13
},
"pull_criteria": {
"reference_wording_sd": 0.05850752475310725,
"assumed_difference_sd": 0.1170150495062145,
"infinite_n": 26.31396908730701,
"finite_n": 13
},
"stay": {
"reference_wording_sd": 0.0684552688483691,
"assumed_difference_sd": 0.1369105376967382,
"infinite_n": 36.02273398170173,
"finite_n": 15
}
},
"decider-2b": {
"pull": {
"reference_wording_sd": 0.09623881629321628,
"assumed_difference_sd": 0.19247763258643255,
"infinite_n": 71.1972887154482,
"finite_n": 19
},
"pull_criteria": {
"reference_wording_sd": 0.13782662458729722,
"assumed_difference_sd": 0.27565324917459444,
"infinite_n": 146.02565087246754,
"finite_n": 21
},
"stay": {
"reference_wording_sd": 0.1237405557559554,
"assumed_difference_sd": 0.2474811115119108,
"infinite_n": 117.70286512264285,
"finite_n": 21
}
},
"gpt-6-luna": {
"pull": {
"reference_wording_sd": 0.05323253401583471,
"assumed_difference_sd": 0.10646506803166941,
"infinite_n": 21.782974880570865,
"finite_n": 12
},
"pull_criteria": {
"reference_wording_sd": 0.04573604678973104,
"assumed_difference_sd": 0.09147209357946208,
"infinite_n": 16.07978202072024,
"finite_n": 10
},
"stay": {
"reference_wording_sd": 0.05538830110767548,
"assumed_difference_sd": 0.11077660221535096,
"infinite_n": 23.582997181274575,
"finite_n": 13
}
},
"jevk5": {
"pull": {
"reference_wording_sd": 0.0624408052340213,
"assumed_difference_sd": 0.1248816104680426,
"infinite_n": 29.97090797831835,
"finite_n": 14
},
"pull_criteria": {
"reference_wording_sd": 0.08652446656225235,
"assumed_difference_sd": 0.1730489331245047,
"infinite_n": 57.549396148988514,
"finite_n": 18
},
"stay": {
"reference_wording_sd": 0.061067168264775026,
"assumed_difference_sd": 0.12213433652955005,
"infinite_n": 28.666750978583682,
"finite_n": 14
}
},
"kev-4b": {
"pull": {
"reference_wording_sd": 0.0551309757698991,
"assumed_difference_sd": 0.1102619515397982,
"infinite_n": 23.3643803997642,
"finite_n": 13
},
"pull_criteria": {
"reference_wording_sd": 0.05104319911368963,
"assumed_difference_sd": 0.10208639822737926,
"infinite_n": 20.028050681496396,
"finite_n": 12
},
"stay": {
"reference_wording_sd": 0.07565484254480837,
"assumed_difference_sd": 0.15130968508961673,
"infinite_n": 43.99834832221262,
"finite_n": 16
}
},
"laya": {
"pull": {
"reference_wording_sd": 0.14539123398553197,
"assumed_difference_sd": 0.29078246797106394,
"infinite_n": 162.49475793382913,
"finite_n": 22
},
"pull_criteria": {
"reference_wording_sd": 0.08728037750285703,
"assumed_difference_sd": 0.17456075500571405,
"infinite_n": 58.559335786766795,
"finite_n": 18
},
"stay": {
"reference_wording_sd": 0.28547535711617034,
"assumed_difference_sd": 0.5709507142323407,
"infinite_n": 626.4698287864375,
"finite_n": 24
}
},
"nimble-9b": {
"pull": {
"reference_wording_sd": 0.05386041823885915,
"assumed_difference_sd": 0.1077208364777183,
"infinite_n": 22.299871118985077,
"finite_n": 12
},
"pull_criteria": {
"reference_wording_sd": 0.05021543668259359,
"assumed_difference_sd": 0.10043087336518718,
"infinite_n": 19.383732044203427,
"finite_n": 11
},
"stay": {
"reference_wording_sd": 0.0641303519386249,
"assumed_difference_sd": 0.1282607038772498,
"infinite_n": 31.6147794640374,
"finite_n": 14
}
},
"semif": {
"pull": {
"reference_wording_sd": 0.07366708178571665,
"assumed_difference_sd": 0.1473341635714333,
"infinite_n": 41.716690044307384,
"finite_n": 16
},
"pull_criteria": {
"reference_wording_sd": 0.08574072673837572,
"assumed_difference_sd": 0.17148145347675145,
"infinite_n": 56.51155147518457,
"finite_n": 18
},
"stay": {
"reference_wording_sd": 0.087758287251038,
"assumed_difference_sd": 0.175516574502076,
"infinite_n": 59.20238289711121,
"finite_n": 18
}
},
"strands-decider-2b": {
"pull": {
"reference_wording_sd": 0.06665982089350343,
"assumed_difference_sd": 0.13331964178700687,
"infinite_n": 34.157902532169466,
"finite_n": 15
},
"pull_criteria": {
"reference_wording_sd": 0.090312754247214,
"assumed_difference_sd": 0.180625508494428,
"infinite_n": 62.699067851501106,
"finite_n": 18
},
"stay": {
"reference_wording_sd": 0.06292652865954686,
"assumed_difference_sd": 0.12585305731909371,
"infinite_n": 30.439005508186558,
"finite_n": 14
}
},
"verdict": {
"pull": {
"reference_wording_sd": 0.026237051962573713,
"assumed_difference_sd": 0.052474103925147425,
"infinite_n": 5.291672779475294,
"finite_n": 5
},
"pull_criteria": {
"reference_wording_sd": 0.024596415569403524,
"assumed_difference_sd": 0.04919283113880705,
"infinite_n": 4.650573946114164,
"finite_n": 5
},
"stay": {
"reference_wording_sd": 0.025293476846065633,
"assumed_difference_sd": 0.050586953692131266,
"infinite_n": 4.91790316835569,
"finite_n": 5
}
},
"von": {
"pull": {
"reference_wording_sd": 0.07857102538180058,
"assumed_difference_sd": 0.15714205076360116,
"infinite_n": 47.45563094012269,
"finite_n": 17
},
"pull_criteria": {
"reference_wording_sd": 0.12075013737075917,
"assumed_difference_sd": 0.24150027474151833,
"infinite_n": 112.08259491290507,
"finite_n": 20
},
"stay": {
"reference_wording_sd": 0.03808472972136781,
"assumed_difference_sd": 0.07616945944273562,
"infinite_n": 11.149738089384085,
"finite_n": 8
}
},
"winnow-12b": {
"pull": {
"reference_wording_sd": 0.034557320789828386,
"assumed_difference_sd": 0.06911464157965677,
"infinite_n": 9.180007565026534,
"finite_n": 7
},
"pull_criteria": {
"reference_wording_sd": 0.0197435498012517,
"assumed_difference_sd": 0.0394870996025034,
"infinite_n": 2.996493839626159,
"finite_n": 4
},
"stay": {
"reference_wording_sd": 0.04439601924585265,
"assumed_difference_sd": 0.0887920384917053,
"infinite_n": 15.151337491411523,
"finite_n": 10
}
}
}
},
"ai": {
"population_wordings": 24,
"sample_wordings": 23,
"selected_wordings": [
"v1",
"v2",
"v3",
"x01",
"x02",
"x03",
"x04",
"x05",
"x06",
"x07",
"x08",
"x09",
"x10",
"x11",
"x12",
"x13",
"x14",
"x15",
"x16",
"x17",
"x18",
"x20",
"x21"
],
"power": {
"jev": {
"pull": {
"reference_wording_sd": 0.03317600755346147,
"assumed_difference_sd": 0.06635201510692294,
"infinite_n": 8.460794612015377,
"finite_n": 7
},
"pull_criteria": {
"reference_wording_sd": 0.021561839158296826,
"assumed_difference_sd": 0.04312367831659365,
"infinite_n": 3.573835135814259,
"finite_n": 4
},
"stay": {
"reference_wording_sd": 0.024378158185738647,
"assumed_difference_sd": 0.048756316371477294,
"infinite_n": 4.568405983277174,
"finite_n": 4
}
},
"clef": {
"pull": {
"reference_wording_sd": 0.05646226187571597,
"assumed_difference_sd": 0.11292452375143194,
"infinite_n": 24.506396396880625,
"finite_n": 13
},
"pull_criteria": {
"reference_wording_sd": 0.06051656550863698,
"assumed_difference_sd": 0.12103313101727396,
"infinite_n": 28.15214270140582,
"finite_n": 14
},
"stay": {
"reference_wording_sd": 0.03666777632680796,
"assumed_difference_sd": 0.07333555265361592,
"infinite_n": 10.335513464321348,
"finite_n": 8
}
},
"clef-flash": {
"pull": {
"reference_wording_sd": 0.07576940228471704,
"assumed_difference_sd": 0.15153880456943408,
"infinite_n": 44.13169750100933,
"finite_n": 16
},
"pull_criteria": {
"reference_wording_sd": 0.08560246672702346,
"assumed_difference_sd": 0.17120493345404691,
"infinite_n": 56.32944459455827,
"finite_n": 18
},
"stay": {
"reference_wording_sd": 0.0517653278861846,
"assumed_difference_sd": 0.1035306557723692,
"infinite_n": 20.598749135749593,
"finite_n": 12
}
},
"decider-2b": {
"pull": {
"reference_wording_sd": 0.10275752625914342,
"assumed_difference_sd": 0.20551505251828683,
"infinite_n": 81.16899925112028,
"finite_n": 19
},
"pull_criteria": {
"reference_wording_sd": 0.14395342245277917,
"assumed_difference_sd": 0.28790684490555835,
"infinite_n": 159.29674409174018,
"finite_n": 21
},
"stay": {
"reference_wording_sd": 0.13545120792458595,
"assumed_difference_sd": 0.2709024158491719,
"infinite_n": 141.03557541218882,
"finite_n": 21
}
},
"gpt-6-luna": {
"pull": {
"reference_wording_sd": 0.02865439869748048,
"assumed_difference_sd": 0.05730879739496096,
"infinite_n": 6.311687799393505,
"finite_n": 6
},
"pull_criteria": {
"reference_wording_sd": 0.03273327651342511,
"assumed_difference_sd": 0.06546655302685021,
"infinite_n": 8.236484177899072,
"finite_n": 7
},
"stay": {
"reference_wording_sd": 0.03863782641313512,
"assumed_difference_sd": 0.07727565282627025,
"infinite_n": 11.475940401171592,
"finite_n": 8
}
},
"jevk5": {
"pull": {
"reference_wording_sd": 0.055841585468629815,
"assumed_difference_sd": 0.11168317093725963,
"infinite_n": 23.97057163172512,
"finite_n": 13
},
"pull_criteria": {
"reference_wording_sd": 0.09480048383398663,
"assumed_difference_sd": 0.18960096766797327,
"infinite_n": 69.08504070405006,
"finite_n": 19
},
"stay": {
"reference_wording_sd": 0.043589118710217784,
"assumed_difference_sd": 0.08717823742043557,
"infinite_n": 14.60558939043967,
"finite_n": 10
}
},
"kev-4b": {
"pull": {
"reference_wording_sd": 0.08151709784494013,
"assumed_difference_sd": 0.16303419568988026,
"infinite_n": 51.081110392847314,
"finite_n": 17
},
"pull_criteria": {
"reference_wording_sd": 0.07420100111381937,
"assumed_difference_sd": 0.14840200222763875,
"infinite_n": 42.323584255723766,
"finite_n": 16
},
"stay": {
"reference_wording_sd": 0.0628091346678205,
"assumed_difference_sd": 0.125618269335641,
"infinite_n": 30.32553911302924,
"finite_n": 14
}
},
"laya": {
"pull": {
"reference_wording_sd": 0.23180253759212235,
"assumed_difference_sd": 0.4636050751842447,
"infinite_n": 413.04681914900056,
"finite_n": 23
},
"pull_criteria": {
"reference_wording_sd": 0.07696464766015132,
"assumed_difference_sd": 0.15392929532030264,
"infinite_n": 45.53501470642526,
"finite_n": 16
},
"stay": {
"reference_wording_sd": 0.22733802663641725,
"assumed_difference_sd": 0.4546760532728345,
"infinite_n": 397.28949508700566,
"finite_n": 23
}
},
"nimble-9b": {
"pull": {
"reference_wording_sd": 0.064973348884229,
"assumed_difference_sd": 0.129946697768458,
"infinite_n": 32.45139823189919,
"finite_n": 15
},
"pull_criteria": {
"reference_wording_sd": 0.10492392159350977,
"assumed_difference_sd": 0.20984784318701954,
"infinite_n": 84.62758322389702,
"finite_n": 19
},
"stay": {
"reference_wording_sd": 0.037639351053516984,
"assumed_difference_sd": 0.07527870210703397,
"infinite_n": 10.890483572228556,
"finite_n": 8
}
},
"semif": {
"pull": {
"reference_wording_sd": 0.08878776987889403,
"assumed_difference_sd": 0.17757553975778806,
"infinite_n": 60.59952298928056,
"finite_n": 18
},
"pull_criteria": {
"reference_wording_sd": 0.09262775523039973,
"assumed_difference_sd": 0.18525551046079947,
"infinite_n": 65.9546148855018,
"finite_n": 18
},
"stay": {
"reference_wording_sd": 0.07897205100345871,
"assumed_difference_sd": 0.15794410200691741,
"infinite_n": 47.9412931957953,
"finite_n": 17
}
},
"strands-decider-2b": {
"pull": {
"reference_wording_sd": 0.07347498245576518,
"assumed_difference_sd": 0.14694996491153037,
"infinite_n": 41.49940711652334,
"finite_n": 16
},
"pull_criteria": {
"reference_wording_sd": 0.10266918348701938,
"assumed_difference_sd": 0.20533836697403876,
"infinite_n": 81.02949390746097,
"finite_n": 19
},
"stay": {
"reference_wording_sd": 0.06329014484563392,
"assumed_difference_sd": 0.12658028969126783,
"infinite_n": 30.791800854141684,
"finite_n": 14
}
},
"verdict": {
"pull": {
"reference_wording_sd": 0.018888571495887817,
"assumed_difference_sd": 0.037777142991775634,
"infinite_n": 2.742591582909884,
"finite_n": 4
},
"pull_criteria": {
"reference_wording_sd": 0.01670598587714804,
"assumed_difference_sd": 0.03341197175429608,
"infinite_n": 2.145394337151022,
"finite_n": 4
},
"stay": {
"reference_wording_sd": 0.018097790557872744,
"assumed_difference_sd": 0.03619558111574549,
"infinite_n": 2.5177582395424385,
"finite_n": 4
}
},
"von": {
"pull": {
"reference_wording_sd": 0.053498445998084315,
"assumed_difference_sd": 0.10699689199616863,
"infinite_n": 22.001143013396344,
"finite_n": 12
},
"pull_criteria": {
"reference_wording_sd": 0.08958618461294932,
"assumed_difference_sd": 0.17917236922589863,
"infinite_n": 61.69429300351079,
"finite_n": 18
},
"stay": {
"reference_wording_sd": 0.028554657772312926,
"assumed_difference_sd": 0.05710931554462585,
"infinite_n": 6.267824521071319,
"finite_n": 6
}
},
"winnow-12b": {
"pull": {
"reference_wording_sd": 0.04513428583578322,
"assumed_difference_sd": 0.09026857167156643,
"infinite_n": 15.659433975221273,
"finite_n": 10
},
"pull_criteria": {
"reference_wording_sd": 0.051090970437715176,
"assumed_difference_sd": 0.10218194087543035,
"infinite_n": 20.065556724775554,
"finite_n": 12
},
"stay": {
"reference_wording_sd": 0.06509878298393443,
"assumed_difference_sd": 0.13019756596786886,
"infinite_n": 32.57681707357911,
"finite_n": 15
}
}
}
},
"foot": {
"population_wordings": 24,
"sample_wordings": 23,
"selected_wordings": [
"f04",
"f05",
"f06",
"f07",
"f08",
"f09",
"f10",
"f11",
"f12",
"f14",
"f15",
"f16",
"f17",
"f18",
"f19",
"f20",
"f21",
"f22",
"f23",
"f24",
"v1",
"v2",
"v3"
],
"power": {
"jev": {
"pull": {
"reference_wording_sd": 0.03797168194644663,
"assumed_difference_sd": 0.07594336389289326,
"infinite_n": 11.08364428353064,
"finite_n": 8
},
"pull_criteria": {
"reference_wording_sd": 0.0358807607218732,
"assumed_difference_sd": 0.0717615214437464,
"infinite_n": 9.89660403312215,
"finite_n": 8
},
"stay": {
"reference_wording_sd": 0.05177885909687244,
"assumed_difference_sd": 0.10355771819374487,
"infinite_n": 20.609519373461698,
"finite_n": 12
}
},
"clef": {
"pull": {
"reference_wording_sd": 0.07931283296221806,
"assumed_difference_sd": 0.1586256659244361,
"infinite_n": 48.3559405963669,
"finite_n": 17
},
"pull_criteria": {
"reference_wording_sd": 0.052030686166508314,
"assumed_difference_sd": 0.10406137233301663,
"infinite_n": 20.810476129093384,
"finite_n": 12
},
"stay": {
"reference_wording_sd": 0.05502260773386237,
"assumed_difference_sd": 0.11004521546772474,
"infinite_n": 23.272618426757347,
"finite_n": 13
}
},
"clef-flash": {
"pull": {
"reference_wording_sd": 0.07266407547268644,
"assumed_difference_sd": 0.14532815094537288,
"infinite_n": 40.588445131868255,
"finite_n": 16
},
"pull_criteria": {
"reference_wording_sd": 0.06512067589422033,
"assumed_difference_sd": 0.13024135178844065,
"infinite_n": 32.5987321148975,
"finite_n": 15
},
"stay": {
"reference_wording_sd": 0.0575546837997539,
"assumed_difference_sd": 0.1151093675995078,
"infinite_n": 25.463860984690864,
"finite_n": 13
}
},
"decider-2b": {
"pull": {
"reference_wording_sd": 0.0839889680117882,
"assumed_difference_sd": 0.1679779360235764,
"infinite_n": 54.22597882811692,
"finite_n": 17
},
"pull_criteria": {
"reference_wording_sd": 0.09267368727162457,
"assumed_difference_sd": 0.18534737454324915,
"infinite_n": 66.0200419531407,
"finite_n": 18
},
"stay": {
"reference_wording_sd": 0.05396387141086105,
"assumed_difference_sd": 0.1079277428217221,
"infinite_n": 22.385618986246314,
"finite_n": 12
}
},
"gpt-6-luna": {
"pull": {
"reference_wording_sd": 0.07514322869252664,
"assumed_difference_sd": 0.15028645738505328,
"infinite_n": 43.40528510163146,
"finite_n": 16
},
"pull_criteria": {
"reference_wording_sd": 0.07331143782358159,
"assumed_difference_sd": 0.14662287564716317,
"infinite_n": 41.31486942485041,
"finite_n": 16
},
"stay": {
"reference_wording_sd": 0.06249102189943309,
"assumed_difference_sd": 0.12498204379886618,
"infinite_n": 30.01913427040177,
"finite_n": 14
}
},
"jevk5": {
"pull": {
"reference_wording_sd": 0.13766515447434857,
"assumed_difference_sd": 0.27533030894869714,
"infinite_n": 145.68370001052506,
"finite_n": 21
},
"pull_criteria": {
"reference_wording_sd": 0.17518627275546556,
"assumed_difference_sd": 0.3503725455109311,
"infinite_n": 235.91907434278622,
"finite_n": 22
},
"stay": {
"reference_wording_sd": 0.0777184257654709,
"assumed_difference_sd": 0.1554368515309418,
"infinite_n": 46.43130608957591,
"finite_n": 17
}
},
"kev-4b": {
"pull": {
"reference_wording_sd": 0.08685654149974573,
"assumed_difference_sd": 0.17371308299949145,
"infinite_n": 57.99198506487604,
"finite_n": 18
},
"pull_criteria": {
"reference_wording_sd": 0.0593378617556297,
"assumed_difference_sd": 0.1186757235112594,
"infinite_n": 27.06616312590838,
"finite_n": 13
},
"stay": {
"reference_wording_sd": 0.034403865248114444,
"assumed_difference_sd": 0.06880773049622889,
"infinite_n": 9.098658941477291,
"finite_n": 7
}
},
"laya": {
"pull": {
"reference_wording_sd": 0.1726493401716611,
"assumed_difference_sd": 0.3452986803433222,
"infinite_n": 229.1356984838949,
"finite_n": 22
},
"pull_criteria": {
"reference_wording_sd": 0.09442809752320469,
"assumed_difference_sd": 0.18885619504640938,
"infinite_n": 68.54336001757864,
"finite_n": 18
},
"stay": {
"reference_wording_sd": 0.18966692611209046,
"assumed_difference_sd": 0.3793338522241809,
"infinite_n": 276.53246286416106,
"finite_n": 23
}
},
"nimble-9b": {
"pull": {
"reference_wording_sd": 0.04940074419453433,
"assumed_difference_sd": 0.09880148838906866,
"infinite_n": 18.75987295111695,
"finite_n": 11
},
"pull_criteria": {
"reference_wording_sd": 0.05488140696239545,
"assumed_difference_sd": 0.1097628139247909,
"infinite_n": 23.15332581781833,
"finite_n": 13
},
"stay": {
"reference_wording_sd": 0.03569454126802766,
"assumed_difference_sd": 0.07138908253605532,
"infinite_n": 9.794144788965077,
"finite_n": 8
}
},
"semif": {
"pull": {
"reference_wording_sd": 0.10267789638631133,
"assumed_difference_sd": 0.20535579277262267,
"infinite_n": 81.04324743608797,
"finite_n": 19
},
"pull_criteria": {
"reference_wording_sd": 0.12149215367869312,
"assumed_difference_sd": 0.24298430735738624,
"infinite_n": 113.46433489838195,
"finite_n": 20
},
"stay": {
"reference_wording_sd": 0.0802524391610631,
"assumed_difference_sd": 0.1605048783221262,
"infinite_n": 49.50845711996355,
"finite_n": 17
}
},
"strands-decider-2b": {
"pull": {
"reference_wording_sd": 0.06312359996070842,
"assumed_difference_sd": 0.12624719992141684,
"infinite_n": 30.629959871858425,
"finite_n": 14
},
"pull_criteria": {
"reference_wording_sd": 0.06040787887946891,
"assumed_difference_sd": 0.12081575775893783,
"infinite_n": 28.05111205530304,
"finite_n": 14
},
"stay": {
"reference_wording_sd": 0.03593983475927727,
"assumed_difference_sd": 0.07187966951855454,
"infinite_n": 9.929218370947051,
"finite_n": 8
}
},
"verdict": {
"pull": {
"reference_wording_sd": 0.02948577900191598,
"assumed_difference_sd": 0.05897155800383196,
"infinite_n": 6.683256391314353,
"finite_n": 6
},
"pull_criteria": {
"reference_wording_sd": 0.015959839464668047,
"assumed_difference_sd": 0.031919678929336094,
"infinite_n": 1.9580327309000918,
"finite_n": 4
},
"stay": {
"reference_wording_sd": 0.02369451776295325,
"assumed_difference_sd": 0.0473890355259065,
"infinite_n": 4.315773644963301,
"finite_n": 4
}
},
"von": {
"pull": {
"reference_wording_sd": 0.05355575198545584,
"assumed_difference_sd": 0.10711150397091168,
"infinite_n": 22.048302233254145,
"finite_n": 12
},
"pull_criteria": {
"reference_wording_sd": 0.12179891314075539,
"assumed_difference_sd": 0.24359782628151078,
"infinite_n": 114.03803779201351,
"finite_n": 20
},
"stay": {
"reference_wording_sd": 0.021277212097652758,
"assumed_difference_sd": 0.042554424195305515,
"infinite_n": 3.4801050656758306,
"finite_n": 4
}
},
"winnow-12b": {
"pull": {
"reference_wording_sd": 0.09248707822464658,
"assumed_difference_sd": 0.18497415644929316,
"infinite_n": 65.7544318609368,
"finite_n": 18
},
"pull_criteria": {
"reference_wording_sd": 0.05859586854702856,
"assumed_difference_sd": 0.11719173709405711,
"infinite_n": 26.393494958598527,
"finite_n": 13
},
"stay": {
"reference_wording_sd": 0.09177959233847877,
"assumed_difference_sd": 0.18355918467695753,
"infinite_n": 64.75229397879826,
"finite_n": 18
}
}
}
}
},
"models": {
"jev": {
"questions": {
"pull": {
"replay_pairs": 7242,
"observed_pair_mean_abs": 0.017268710301021818,
"mean_abs_limit": 0.02726871030102182,
"p95_abs_limit": 0.07000000000000005,
"near_line_margin": 0.08999999999999996,
"flip_rate_limit": 0.058467274233637116,
"noise_source": "committed three-pass replay"
},
"pull_criteria": {
"replay_pairs": 7242,
"observed_pair_mean_abs": 0.015070422535211268,
"mean_abs_limit": 0.025070422535211266,
"p95_abs_limit": 0.06999999999999994,
"near_line_margin": 0.08,
"flip_rate_limit": 0.041897265948632974,
"noise_source": "committed three-pass replay"
},
"stay": {
"replay_pairs": 7242,
"observed_pair_mean_abs": 0.0184065175365921,
"mean_abs_limit": 0.028406517536592103,
"p95_abs_limit": 0.07000000000000005,
"near_line_margin": 0.09000000000000001,
"flip_rate_limit": 0.05308202154101077,
"noise_source": "committed three-pass replay"
}
},
"passes": 3,
"estimated_wall_s_per_pass_excluding_setup": 151,
"reference_files": [
"results/v2/raw/jev-v2-pass-a.jsonl",
"results/v2/raw/jev-v2-pass-b.jsonl",
"results/v2/raw/jev-v2-pass-c.jsonl"
]
},
"clef": {
"questions": {
"pull": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"pull_criteria": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"stay": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
}
},
"passes": 1,
"estimated_wall_s_per_pass_excluding_setup": 9319,
"reference_files": [
"results/v2/raw/clef.jsonl"
]
},
"clef-flash": {
"questions": {
"pull": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"pull_criteria": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"stay": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
}
},
"passes": 1,
"estimated_wall_s_per_pass_excluding_setup": 2501,
"reference_files": [
"results/v2/raw/clef-flash.jsonl"
]
},
"decider-2b": {
"questions": {
"pull": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"pull_criteria": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"stay": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
}
},
"passes": 1,
"estimated_wall_s_per_pass_excluding_setup": 727,
"reference_files": [
"results/v2/raw/decider-2b.jsonl"
]
},
"gpt-6-luna": {
"questions": {
"pull": {
"replay_pairs": 7242,
"observed_pair_mean_abs": 0.0,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "committed three-pass replay"
},
"pull_criteria": {
"replay_pairs": 7242,
"observed_pair_mean_abs": 0.0,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "committed three-pass replay"
},
"stay": {
"replay_pairs": 7242,
"observed_pair_mean_abs": 0.0,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "committed three-pass replay"
}
},
"passes": 3,
"estimated_wall_s_per_pass_excluding_setup": 42,
"reference_files": [
"results/v2/raw/gpt-6-luna-pass-a.jsonl",
"results/v2/raw/gpt-6-luna-pass-b.jsonl",
"results/v2/raw/gpt-6-luna-pass-c.jsonl"
]
},
"jevk5": {
"questions": {
"pull": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"pull_criteria": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"stay": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
}
},
"passes": 1,
"estimated_wall_s_per_pass_excluding_setup": 2390,
"reference_files": [
"results/v2/raw/jevk5.jsonl"
]
},
"kev-4b": {
"questions": {
"pull": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"pull_criteria": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"stay": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
}
},
"passes": 1,
"estimated_wall_s_per_pass_excluding_setup": 844,
"reference_files": [
"results/v2/raw/kev-4b.jsonl"
]
},
"laya": {
"questions": {
"pull": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"pull_criteria": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"stay": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
}
},
"passes": 1,
"estimated_wall_s_per_pass_excluding_setup": 180,
"reference_files": [
"results/v2/raw/laya.jsonl"
]
},
"nimble-9b": {
"questions": {
"pull": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"pull_criteria": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"stay": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
}
},
"passes": 1,
"estimated_wall_s_per_pass_excluding_setup": 6143,
"reference_files": [
"results/v2/raw/nimble-9b.jsonl"
]
},
"semif": {
"questions": {
"pull": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"pull_criteria": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"stay": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
}
},
"passes": 1,
"estimated_wall_s_per_pass_excluding_setup": 2146,
"reference_files": [
"results/v2/raw/semif.jsonl"
]
},
"strands-decider-2b": {
"questions": {
"pull": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"pull_criteria": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"stay": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
}
},
"passes": 1,
"estimated_wall_s_per_pass_excluding_setup": 615,
"reference_files": [
"results/v2/raw/strands-decider-2b.jsonl"
]
},
"verdict": {
"questions": {
"pull": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"pull_criteria": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"stay": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
}
},
"passes": 1,
"estimated_wall_s_per_pass_excluding_setup": 436,
"reference_files": [
"results/v2/raw/verdict.jsonl"
]
},
"von": {
"questions": {
"pull": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"pull_criteria": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"stay": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
}
},
"passes": 1,
"estimated_wall_s_per_pass_excluding_setup": 332,
"reference_files": [
"results/v2/raw/von.jsonl"
]
},
"winnow-12b": {
"questions": {
"pull": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"pull_criteria": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
},
"stay": {
"replay_pairs": 0,
"observed_pair_mean_abs": null,
"mean_abs_limit": 0.01,
"p95_abs_limit": 0.02,
"near_line_margin": 0.01,
"flip_rate_limit": 0.01,
"noise_source": "no replay measured; 0.01 engineering allowance, not an empirical noise floor"
}
},
"passes": 1,
"estimated_wall_s_per_pass_excluding_setup": 3856,
"reference_files": [
"results/v2/raw/winnow-12b.jsonl"
]
}
},
"article_claims": {
"jev_keeps_server_at_20_of_48": 16,
"other_models_worse_than_jev_at_20": 8,
"never_keeps_server_at_positive_counts": [
"gpt-6-luna",
"verdict"
],
"server_at_20_by_model": {
"jev": 16,
"clef": 4,
"clef-flash": 35,
"decider-2b": 22,
"gpt-6-luna": 0,
"jevk5": 42,
"kev-4b": 43,
"laya": 10,
"nimble-9b": 29,
"semif": 17,
"strands-decider-2b": 22,
"verdict": 0,
"von": 47,
"winnow-12b": 4
}
}
}