Dataset Viewer
Auto-converted to Parquet Duplicate
acquisition_id
stringlengths
11
11
additional_target_queries
int64
0
0
compiled_decoder_exact
bool
1 class
constraint_rank
int64
1
7
dimension
int64
2
8
edited_mean_json
stringlengths
9
197
euclidean_objective_excess_exact
stringlengths
1
52
euclidean_over_optimal_ratio
float64
1
6.99
⌀
euclidean_ratio_exact
stringlengths
1
57
⌀
feasible
bool
1 class
initial_final_anchor_equal
bool
1 class
optimal_objective_exact
stringlengths
1
25
oracle_exact
bool
1 class
raw_probe_state_feasible
bool
1 class
record_id
stringlengths
15
15
seed
int64
0
5
source_queries
int64
1
7
stationary
bool
1 class
target_index
int64
0
2
target_json
stringlengths
5
49
s00_n02_r01
0
true
1
2
["1","0"]
0
null
null
true
true
0
true
false
s00_n02_r01_t00
0
1
true
0
["0"]
s00_n02_r01
0
true
1
2
["1","-1"]
0
1
1
true
true
5/2
true
false
s00_n02_r01_t01
0
1
true
1
["-1"]
s00_n02_r01
0
true
1
2
["1","0"]
0
null
null
true
true
0
true
false
s00_n02_r01_t02
0
1
true
2
["0"]
s00_n03_r01
0
true
1
3
["-25/18","49/36","-25/36"]
2/27
3
3
true
true
1/27
true
false
s00_n03_r01_t00
0
1
true
0
["-2/3"]
s00_n03_r01
0
true
1
3
["-23/18","11/9","-8/9"]
8/27
3
3
true
true
4/27
true
false
s00_n03_r01_t01
0
1
true
1
["-1/3"]
s00_n03_r01
0
true
1
3
["-25/18","49/36","-25/36"]
2/27
3
3
true
true
1/27
true
false
s00_n03_r01_t02
0
1
true
2
["-2/3"]
s00_n03_r02
0
true
2
3
["455/173","-77/346","1051/346"]
6185169/865000
1.484446
18952669/12767500
true
true
5107/346
true
false
s00_n03_r02_t00
0
2
true
0
["3","-2"]
s00_n03_r02
0
true
2
3
["2095/519","-115/1038","2729/1038"]
9296401/865000
1.376597
305835109/222167500
true
true
88867/3114
true
false
s00_n03_r02_t01
0
2
true
1
["1","-16/3"]
s00_n03_r02
0
true
2
3
["90/173","-135/346","1263/346"]
337842/108125
1.140988
151894/133125
true
true
3834/173
true
false
s00_n03_r02_t02
0
2
true
2
["6","3"]
s00_n04_r01
0
true
1
4
["205/876","377/219","1081/876","38/73"]
37/219
1.165873
4161/3569
true
true
3569/3504
true
false
s00_n04_r01_t00
0
1
true
0
["2"]
s00_n04_r01
0
true
1
4
["671/876","280/219","671/876","35/73"]
37/219
1.165873
4161/3569
true
true
3569/3504
true
false
s00_n04_r01_t01
0
1
true
1
["0"]
s00_n04_r01
0
true
1
4
["2479/2628","743/657","1603/2628","34/73"]
925/1971
1.165873
4161/3569
true
true
89225/31536
true
false
s00_n04_r01_t02
0
1
true
2
["-2/3"]
s00_n04_r02
0
true
2
4
["-4775/1998","39/37","-2272/999","-151/111"]
309681475/20163816
3.135905
18186795/5799536
true
true
21550/2997
true
false
s00_n04_r02_t00
0
2
true
0
["-13/3","-5/3"]
s00_n04_r02
0
true
2
4
["-1363/3330","-91/555","-1139/1665","-139/185"]
478407/1244680
1.96744
324305/164836
true
true
147/370
true
false
s00_n04_r02_t01
0
2
true
1
["0","-5/3"]
s00_n04_r02
0
true
2
4
["1163/3330","-48/185","934/1665","209/185"]
39019823/11202120
1.369132
144726795/105706972
true
true
31423/3330
true
false
s00_n04_r02_t02
0
2
true
2
["-1","2"]
s00_n04_r03
0
true
3
4
["518/417","-1111/3336","505/1668","-1213/1112"]
1568187/10679648
1.018982
252542984/247838423
true
true
103223/13344
true
false
s00_n04_r03_t00
0
3
true
0
["-2","0","1"]
s00_n04_r03
0
true
3
4
["35/139","-2053/1112","175/556","-4207/3336"]
54302161/32038944
1.239037
281472776/227170615
true
true
94615/13344
true
false
s00_n04_r03_t01
0
3
true
1
["1/3","-7/3","16/3"]
s00_n04_r03
0
true
3
4
["86/1251","-529/10008","2515/5004","3389/3336"]
17121963/10679648
1.397611
1624970048/1162677047
true
true
484247/120096
true
false
s00_n04_r03_t02
0
3
true
2
["1","5/3","-3"]
s00_n06_r01
0
true
1
6
["42125/43656","101/65484","-63371/43656","-8287/130968","-54593/130968","-178171/130968"]
17905513/6417432
3.230897
25931664/8026151
true
true
163799/130968
true
false
s00_n06_r01_t00
0
1
true
0
["-2"]
s00_n06_r01
0
true
1
6
["272653/261936","424939/392904","-123931/261936","909301/785808","382151/785808","-455387/785808"]
17905513/4714848
3.230897
25931664/8026151
true
true
8026151/4714848
true
false
s00_n06_r01_t01
0
1
true
1
["7/3"]
s00_n06_r01
0
true
1
6
["251219/261936","-32035/392904","-399941/261936","-123493/785808","-382151/785808","-1116229/785808"]
17905513/4714848
3.230897
25931664/8026151
true
true
8026151/4714848
true
false
s00_n06_r01_t02
0
1
true
2
["-7/3"]
s00_n06_r02
0
true
2
6
["254941/710578","559466/1065867","-980798/1065867","-448682/1065867","-285118/355289","-672594/355289"]
42650987912/5074592787
2.264475
305524735637/134920783989
true
true
28338749/4263468
true
false
s00_n06_r02_t00
0
2
true
0
["-5/3","13/3"]
s00_n06_r02
0
true
2
6
["-4127273/2131734","-2199916/3197601","-1152749/3197601","-1473833/3197601","-3283163/3197601","-2616572/3197601"]
1834144834/5074592787
2.017955
14543755215/7207175879
true
true
13624151/38371212
true
false
s00_n06_r02_t01
0
2
true
1
["2/3","-5/3"]
s00_n06_r02
0
true
2
6
["-1012169/2131734","-652120/3197601","-3010982/3197601","-2404778/3197601","-3170504/3197601","-4316090/3197601"]
8055854284/5074592787
1.724246
76715777325/44492360189
true
true
84106541/38371212
true
false
s00_n06_r02_t02
0
2
true
2
["-7/3","7/3"]
s00_n06_r03
0
true
3
6
["6727/55068","-73313/247806","-15973/19062","-94421/55068","-113018/123903","3247/41301"]
375896600663/44623665450
1.485855
1149577733538/773681132875
true
true
12889315/743418
true
false
s00_n06_r03_t00
0
3
true
0
["8/3","-2","5/3"]
s00_n06_r03
0
true
3
6
["264241/110136","-750779/495612","56531/38124","-83747/110136","-184534/123903","-15725/82602"]
22238682904639/356989323600
2.609103
36059228406864/13820545502225
true
true
230246489/5947344
true
false
s00_n06_r03_t01
0
3
true
1
["-5/3","-2","4/3"]
s00_n06_r03
0
true
3
6
["143635/55068","-506543/247806","39059/19062","-3075/18356","-213743/123903","15406/41301"]
69818926763/910687050
3.003786
104662433538/34843506775
true
true
28443679/743418
true
false
s00_n06_r03_t02
0
3
true
2
["-2","1/3","4/3"]
s00_n06_r04
0
true
4
6
["1231946/1076183","1380630/1076183","1283131/2152366","-645861/1076183","-371302/1076183","1507261/1076183"]
189885493473/244442054254
1.080656
10176624284443/9417082310551
true
true
82919479/8609464
true
false
s00_n06_r04_t00
0
4
true
0
["3","2","4","5"]
s00_n06_r04
0
true
4
6
["-5702983/3228549","161034/1076183","12101657/6457098","-4681727/3228549","1720444/3228549","-644386/3228549"]
18863417505379/733326162762
1.423476
253630615554849/178176945533333
true
true
1568887157/25828392
true
false
s00_n06_r04_t01
0
4
true
1
["-7/3","7/3","4","-1/3"]
s00_n06_r04
0
true
4
6
["-7684193/3228549","75456/1076183","5616805/6457098","1308691/3228549","3844477/3228549","48717/1076183"]
356738221549681/19799806394574
1.386182
5121981570197827/3695028683999103
true
true
401673327/8609464
true
false
s00_n06_r04_t02
0
4
true
2
["-14/3","-2/3","-8/3","-3"]
s00_n06_r05
0
true
5
6
["-75875/52392","-4905/34928","39833/104784","-2725/8732","74761/104784","-100921/104784"]
338817649/5477269248
1.001271
266893883437/266555065788
true
true
20397541/419136
true
false
s00_n06_r05_t00
0
5
true
0
["-17/3","11/3","-1","-1","-5/3"]
s00_n06_r05
0
true
5
6
["26647/52392","-31379/34928","-178189/104784","5915/26196","66307/104784","54063/34928"]
103205281/16905152
2.006164
205778312/102573031
true
true
847711/139712
true
false
s00_n06_r05_t01
0
5
true
1
["3","-14/3","8/3","7","4/3"]
s00_n06_r05
0
true
5
6
["-45691/52392","-193235/104784","87697/104784","-12271/26196","-17087/104784","-78065/104784"]
682411129/152146368
1.078082
9422107424/8739696295
true
true
72228895/1257408
true
false
s00_n06_r05_t02
0
5
true
2
["-3","-1","13/3","5/3","4/3"]
s00_n08_r01
0
true
1
8
["251343967/414194066","56632709/18827003","109221781/414194066","-370081246/207097033","-39953502/15930541","7709567/414194066","-1181925/18827003","-24072713/31861082"]
13605778089/414194066
5.293287
207097033/39124464
true
true
1584540792/207097033
true
false
s00_n08_r01_t00
0
1
true
0
["9"]
s00_n08_r01
0
true
1
8
["1/2","1","1/2","-1","0","-1/2","0","1/2"]
0
null
null
true
true
0
true
false
s00_n08_r01_t01
0
1
true
1
["0"]
s00_n08_r01
0
true
1
8
["42875297/95583246","144059/4344693","58666651/95583246","-29682266/47791623","57710614/47791623","-23886341/31861082","43775/1448231","35191367/31861082"]
2183643397/286749738
5.293287
207097033/39124464
true
true
84769672/47791623
true
false
s00_n08_r01_t02
0
1
true
2
["-13/3"]
s00_n08_r02
0
true
2
8
["-736432816/894145221","262807630/298048407","3030739613/1788290442","-1472293097/1788290442","881468875/894145221","-1379792129/1788290442","1316098099/1788290442","368907193/1788290442"]
26977186196068/19321584080589
2.220514
392642513666315/176825024097771
true
true
24548802457/21459485304
true
false
s00_n08_r02_t00
0
2
true
0
["-6","-8/3"]
s00_n08_r02
0
true
2
8
["-943646972/894145221","953818393/596096814","3567712549/1788290442","-1222829221/1788290442","2109114265/1788290442","-1104086654/894145221","828534676/894145221","559525906/894145221"]
11597531779093/25762112107452
1.210222
790126326957/652877430163
true
true
45954304747/21459485304
true
false
s00_n08_r02_t01
0
2
true
1
["-29/3","-5"]
s00_n08_r02
0
true
2
8
["-734435438/894145221","210629036/298048407","4805757865/1788290442","-285077935/1788290442","108436139/894145221","-1893848305/1788290442","812671523/1788290442","469154339/1788290442"]
124596613954777/19321584080589
1.95505
2040459164866331/1043686253228115
true
true
144896050705/21459485304
true
false
s00_n08_r02_t02
0
2
true
2
["-9","-4/3"]
s00_n08_r03
0
true
3
8
["-1055989057/2642961246","5624494499/2642961246","29921006/1321480623","-3650704895/2642961246","406012681/1321480623","135088141/56233218","1454068001/2642961246","-2968048019/2642961246"]
198632278195717/7706874993336
1.890882
421593725129854/222961446934137
true
true
917536818659/31715534952
true
false
s00_n08_r03_t00
0
3
true
0
["4/3","7/3","17/3"]
s00_n08_r03
0
true
3
8
["3921034343/2642961246","179964921/880987082","1084214917/2642961246","-1071800942/440493541","-155558463/440493541","48234655/56233218","-114485833/2642961246","957135818/1321480623"]
3168189952735297/69361874940024
6.989644
3697134530553724/528944577818427
true
true
241858517521/31715534952
true
false
s00_n08_r03_t01
0
3
true
1
["5/3","-1","-16/3"]
s00_n08_r03
0
true
3
8
["-38000979/880987082","1747530525/880987082","729661235/2642961246","-2326895222/1321480623","49271434/440493541","43004741/18744406","1243630267/2642961246","-831087862/1321480623"]
20190788685803173/624256874460216
2.373101
34895309818649224/14704521132846051
true
true
747067069697/31715534952
true
false
s00_n08_r03_t02
0
3
true
2
["7/3","2","11/3"]
s00_n08_r04
0
true
4
8
["-480974539/773174025","-2330069083/1546348050","-310500949/309269610","56062453/51544935","-1155867811/1546348050","-1349011409/1546348050","2264515601/1546348050","-34784389/19824975"]
784384824218364217/112531279457446200
2.155339
1463306792335435350/678921968117071133
true
true
111952836569/18556176600
true
false
s00_n08_r04_t00
0
4
true
0
["-16/3","-20/3","-11/3","-9"]
s00_n08_r04
0
true
4
8
["-16379906/11894985","-46791947/23789970","-10946579/4757994","-1212904/792999","6777061/23789970","1229549/23789970","-17903021/23789970","-4349888/3964995"]
44042794978277473/4501251178297848
1.279432
201658275608665734/157615480630388261
true
true
1999266821/57095928
true
false
s00_n08_r04_t01
0
4
true
1
["-4/3","-43/3","20/3","-28/3"]
s00_n08_r04
0
true
4
8
["-2054132339/773174025","-1646939783/1546348050","-148292422/154634805","174428131/103089870","-108083593/773174025","-487290692/773174025","-1286442862/773174025","1186711/19824975"]
1803464628003025867/112531279457446200
1.779091
4118296216163053050/2314831588160027183
true
true
381710969219/18556176600
true
false
s00_n08_r04_t02
0
4
true
2
["19/3","5/3","-8/3","-2"]
s00_n08_r05
0
true
5
8
["-19647155273/7619888382","-847420019/2539962794","6727011097/3809944191","11606938441/3809944191","-7983167429/7619888382","993031909/7619888382","311371788/1269981397","-4257996659/3809944191"]
74555274383012385158963/3588134481890718424272
1.356
283980109805294411022143/209424835422282025863180
true
true
889483868785/15239776764
true
false
s00_n08_r05_t00
0
5
true
0
["-10/3","-17/3","31/3","7/3","-5/3"]
s00_n08_r05
0
true
5
8
["-7041033403/2539962794","-1626294471/2539962794","4560444211/3809944191","5942505538/3809944191","-171298153/7619888382","1685423307/2539962794","-4873296910/3809944191","-5694457565/3809944191"]
12097262185406637795739/398681609098968713808
1.264851
57773051558401672484447/45675789372995034688708
true
true
5237930351017/45719330292
true
false
s00_n08_r05_t01
0
5
true
1
["-10/3","1/3","34/3","-1","-14/3"]
s00_n08_r05
0
true
5
8
["-3490894453/2539962794","2798779649/2539962794","5578411714/3809944191","4310812465/3809944191","-12020165323/7619888382","-1581173505/2539962794","-6374629/3809944191","3011839012/3809944191"]
2129973660532560604147/398681609098968713808
1.339738
8399427594751285056839/6269453934218724452692
true
true
718957756333/45719330292
true
false
s00_n08_r05_t02
0
5
true
2
["8/3","-14/3","4/3","-3","13/3"]
s00_n08_r06
0
true
6
8
["-671752253/796497459","-450051631/796497459","139112199/265499153","541798984/796497459","753756700/796497459","-1919419477/796497459","-367002554/796497459","766656050/265499153"]
7939918246197432505831/642120167626054880544
1.313234
33288149111583485838303/25348230865386053332472
true
true
377308843279/9557969508
true
false
s00_n08_r06_t00
0
6
true
0
["17/3","-7","-4","1","-3","-14/3"]
s00_n08_r06
0
true
6
8
["1054941257/796497459","1270838218/796497459","-492698506/796497459","-744644915/796497459","359553991/265499153","-496597267/265499153","465734695/796497459","2241480872/796497459"]
91802095012222937403191/5779081508634493924896
1.150764
700715871240699267565055/608913776228476330161864
true
true
1007076868097/9557969508
true
false
s00_n08_r06_t01
0
6
true
1
["-2/3","17/3","13/3","-2/3","5/3","8/3"]
s00_n08_r06
0
true
6
8
["-17836654/796497459","376716521/796497459","-388606241/796497459","-275020778/265499153","2242024045/796497459","-813171911/265499153","-1065186199/796497459","2390070103/796497459"]
279425180536919244586223/5779081508634493924896
1.666838
698455175911745856349319/419029995374826611763096
true
true
693029837483/9557969508
true
false
s00_n08_r06_t02
0
6
true
2
["-3","2/3","1","6","4/3","-8/3"]
s00_n08_r07
0
true
7
8
["53575/211521","-419203/1480647","-2477801/1480647","-658285/493549","-547124/1480647","614766/493549","3796405/1480647","1276403/1480647"]
12617857418850450/470910065799229
1.278044
4175897558076752501/3267411823919520101
true
true
3424492181/35535528
true
false
s00_n08_r07_t00
0
7
true
0
["1/3","4","13/3","1","10/3","-10/3","5/3"]
s00_n08_r07
0
true
7
8
["368101/211521","-807997/1480647","-819650/493549","946987/1480647","-874775/493549","1024633/1480647","3883184/1480647","-141566/1480647"]
340860214074187201/8476381184386122
1.304358
5843170173747623269/4479729317450874465
true
true
1565030555/11845176
true
false
s00_n08_r07_t01
0
7
true
1
["5","7/3","1","0","-5/3","5/3","-11/3"]
s00_n08_r07
0
true
7
8
["33889/70507","-163253/1480647","-262151/493549","-700426/1480647","318646/493549","498560/1480647","2746964/1480647","2636009/1480647"]
37115914189147298/4238190592193061
1.087596
3686649255942539845/3389721942429361461
true
true
1184227447/11845176
true
false
s00_n08_r07_t02
0
7
true
2
["-3","7/3","10/3","-7/3","3","-5/3","17/3"]
s01_n02_r01
0
true
1
2
["-1/3","-11/12"]
121/144
1.2
6/5
true
true
605/144
true
false
s01_n02_r01_t00
1
1
true
0
["-1/3"]
s01_n02_r01
0
true
1
2
["2/3","-5/12"]
25/144
1.2
6/5
true
true
125/144
true
false
s01_n02_r01_t01
1
1
true
1
["2/3"]
s01_n02_r01
0
true
1
2
["1","-1/4"]
1/16
1.2
6/5
true
true
5/16
true
false
s01_n02_r01_t02
1
1
true
2
["1"]
s01_n03_r01
0
true
1
3
["5/3","-111/34","-89/102"]
650/51
3.4
17/5
true
true
1625/306
true
false
s01_n03_r01_t00
1
1
true
0
["5/3"]
s01_n03_r01
0
true
1
3
["-5/3","9/34","191/102"]
650/51
3.4
17/5
true
true
1625/306
true
false
s01_n03_r01_t01
1
1
true
1
["-5/3"]
s01_n03_r01
0
true
1
3
["-2/3","-27/34","107/102"]
104/51
3.4
17/5
true
true
130/153
true
false
s01_n03_r01_t02
1
1
true
2
["-2/3"]
s01_n03_r02
0
true
2
3
["-2/3","1/3","-5/2"]
75/4
2.172893
2501/1151
true
true
1151/72
true
false
s01_n03_r02_t00
1
2
true
0
["-2/3","1/3"]
s01_n03_r02
0
true
2
3
["-5/3","1","-19/6"]
361/12
1.857143
13/7
true
true
2527/72
true
false
s01_n03_r02_t01
1
2
true
1
["-5/3","1"]
s01_n03_r02
0
true
2
3
["-2/3","0","-8/3"]
64/3
2.291842
2725/1189
true
true
1189/72
true
false
s01_n03_r02_t02
1
2
true
2
["-2/3","0"]
s01_n04_r01
0
true
1
4
["-1","-3/2","0","1/2"]
0
null
null
true
true
0
true
false
s01_n04_r01_t00
1
1
true
0
["-1"]
s01_n04_r01
0
true
1
4
["-5/3","-563/362","-124/543","863/1086"]
764/1629
1.431638
1267/885
true
true
590/543
true
false
s01_n04_r01_t01
1
1
true
1
["-5/3"]
s01_n04_r01
0
true
1
4
["-1","-3/2","0","1/2"]
0
null
null
true
true
0
true
false
s01_n04_r01_t02
1
1
true
2
["-1"]
s01_n04_r02
0
true
2
4
["1","2/3","-193/267","159/178"]
3773/3204
1.187492
47793/40247
true
true
40247/6408
true
false
s01_n04_r02_t00
1
2
true
0
["1","2/3"]
s01_n04_r02
0
true
2
4
["-1/3","4/3","-193/267","159/178"]
3773/3204
1.24349
38537/30991
true
true
30991/6408
true
false
s01_n04_r02_t01
1
2
true
1
["-1/3","4/3"]
s01_n04_r02
0
true
2
4
["-2/3","5/3","-403/534","169/178"]
1232/801
1.243785
50285/40429
true
true
40429/6408
true
false
s01_n04_r02_t02
1
2
true
2
["-2/3","5/3"]
s01_n04_r03
0
true
3
4
["1","-5/3","4/3","-3/4"]
245/16
2.582915
514/199
true
true
1393/144
true
false
s01_n04_r03_t00
1
3
true
0
["1","-5/3","4/3"]
s01_n04_r03
0
true
3
4
["1","2/3","2/3","29/20"]
81/80
1.306174
3110/2381
true
true
2381/720
true
false
s01_n04_r03_t01
1
3
true
1
["1","2/3","2/3"]
s01_n04_r03
0
true
3
4
["-1","-4/3","1/3","47/60"]
169/720
1.053805
3310/3141
true
true
349/80
true
false
s01_n04_r03_t02
1
3
true
2
["-1","-4/3","1/3"]
s01_n06_r01
0
true
1
6
["1","-67173/59938","-36219/29969","88333/59938","1803/59938","-34023/59938"]
78487/59938
1.19572
479504/401017
true
true
401017/59938
true
false
s01_n06_r01_t00
1
1
true
0
["1"]
s01_n06_r01
0
true
1
6
["1/3","-82329/59938","-96157/89907","268147/179814","61741/179814","-93961/179814"]
78487/539442
1.19572
479504/401017
true
true
401017/539442
true
false
s01_n06_r01_t01
1
1
true
1
["1/3"]
s01_n06_r01
0
true
1
6
["-4/3","-120219/59938","-64907/89907","276017/179814","202571/179814","-73691/179814"]
627896/269721
1.19572
479504/401017
true
true
3208136/269721
true
false
s01_n06_r01_t02
1
1
true
2
["-4/3"]
s01_n06_r02
0
true
2
6
["-2/3","1","-11711/34656","-2531/4332","20383/17328","-3133/34656"]
328325/138624
1.834
28880/15747
true
true
131225/46208
true
false
s01_n06_r02_t00
1
2
true
0
["-2/3","1"]
s01_n06_r02
0
true
2
6
["1/3","1/3","8093/11552","-2393/4332","14441/17328","5861/34656"]
5457023/415872
1.850977
11869680/6412657
true
true
6412657/415872
true
false
s01_n06_r02_t01
1
2
true
1
["1/3","1/3"]
s01_n06_r02
0
true
2
6
["1","2/3","38407/34656","-991/1444","21689/17328","7079/11552"]
2962421/138624
1.811378
6613520/3651099
true
true
1217033/46208
true
false
s01_n06_r02_t02
1
2
true
2
["1","2/3"]
s01_n06_r03
0
true
3
6
["-4/3","0","5/3","341/213","1451/1562","-4088/2343"]
264832/7029
2.384583
57013/23909
true
true
191272/7029
true
false
s01_n06_r03_t00
1
3
true
0
["-4/3","0","5/3"]
s01_n06_r03
0
true
3
6
["-1/3","2/3","2/3","189/142","137/426","-153/142"]
34943/2556
2.367686
60492/25549
true
true
25549/2556
true
false
s01_n06_r03_t01
1
3
true
1
["-1/3","2/3","2/3"]
s01_n06_r03
0
true
3
6
["-2/3","-5/3","2/3","333/142","-855/1562","-1383/1562"]
70989/3124
1.866634
1376122/737221
true
true
737221/28116
true
false
s01_n06_r03_t02
1
3
true
2
["-2/3","-5/3","2/3"]
s01_n06_r04
0
true
4
6
["-1","-1/3","2/3","-4/3","-199/231","241/231"]
1745/924
1.049337
37114/35369
true
true
35369/924
true
false
s01_n06_r04_t00
1
4
true
0
["-1","-1/3","2/3","-4/3"]
s01_n06_r04
0
true
4
6
["2/3","-5/3","5/3","2/3","-50/33","16/33"]
4489/396
1.20025
26906/22417
true
true
22417/396
true
false
s01_n06_r04_t01
1
4
true
1
["2/3","-5/3","5/3","2/3"]
s01_n06_r04
0
true
4
6
["-2/3","1/3","0","-2/3","-125/231","293/231"]
365/924
1.009444
23408/23189
true
true
115945/2772
true
false
s01_n06_r04_t02
1
4
true
2
["-2/3","1/3","0","-2/3"]
s01_n06_r05
0
true
5
6
["2/3","-1/3","0","1/3","-2/3","-29/66"]
1369/792
1.171061
9372/8003
true
true
8003/792
true
false
s01_n06_r05_t00
1
5
true
0
["2/3","-1/3","0","1/3","-2/3"]
s01_n06_r05
0
true
5
6
["-1","-1/3","-5/3","-5/3","4/3","79/66"]
21025/792
1.470705
65692/44667
true
true
4963/88
true
false
s01_n06_r05_t01
1
5
true
1
["-1","-1/3","-5/3","-5/3","4/3"]
s01_n06_r05
0
true
5
6
["-1/3","-1/3","2/3","2/3","-1","-1/6"]
275/72
1.228216
296/241
true
true
1205/72
true
false
s01_n06_r05_t02
1
5
true
2
["-1/3","-1/3","2/3","2/3","-1"]
s01_n08_r01
0
true
1
8
["5/3","72443/47508","-97025/63344","870461/570096","16587/31672","-604889/570096","166433/190032","-911/142524"]
564391/6841152
2.173895
1045176/480785
true
true
480785/6841152
true
false
s01_n08_r01_t00
1
1
true
0
["5/3"]
s01_n08_r01
0
true
1
8
["-4/3","51185/47508","-60863/63344","594755/570096","3069/31672","21385/570096","591215/190032","15487/142524"]
163108999/6841152
2.173895
1045176/480785
true
true
138946865/6841152
true
false
s01_n08_r01_t01
1
1
true
1
["-4/3"]
s01_n08_r01
0
true
1
8
["4/3","70081/47508","-93007/63344","839827/570096","15085/31672","-535303/570096","213631/190032","911/142524"]
564391/6841152
2.173895
1045176/480785
true
true
480785/6841152
true
false
s01_n08_r01_t02
1
1
true
2
["4/3"]
s01_n08_r02
0
true
2
8
["-4/3","4/3","-24497/66534","19907/33267","6375/44356","14174/33267","-148243/133068","-67877/44356"]
44001179/1596816
2.672853
70304260/26303081
true
true
26303081/1596816
true
false
s01_n08_r02_t00
1
2
true
0
["-4/3","4/3"]
s01_n08_r02
0
true
2
8
["-5/3","-2/3","-4156387/12042654","2257547/6021327","27801803/24085308","2477858/6021327","-28498541/24085308","-11219317/24085308"]
1076450623/289023696
1.153952
8068578180/6992127557
true
true
6992127557/289023696
true
false
s01_n08_r02_t01
1
2
true
1
["-5/3","-2/3"]
s01_n08_r02
0
true
2
8
["1","2/3","336943/4014218","-616349/6021327","39492827/24085308","6864826/6021327","-7707503/8028436","-25721509/24085308"]
69214783/289023696
1.194264
425507108/356292325
true
true
356292325/289023696
true
false
s01_n08_r02_t02
1
2
true
2
["1","2/3"]
s01_n08_r03
0
true
3
8
["-4/3","-5/3","5/3","-653626/343371","-117529/1373484","-22250/114457","-866471/686742","75559/686742"]
153969895/5493936
2.129065
174203554/81821617
true
true
409108085/16481808
true
false
s01_n08_r03_t00
1
3
true
0
["-4/3","-5/3","5/3"]
s01_n08_r03
0
true
3
8
["-4/3","0","0","-388856/343371","-39509/1373484","-128585/343371","-132751/686742","22143/228914"]
194091925/16481808
1.989223
78059674/39241289
true
true
196206445/16481808
true
false
s01_n08_r03_t01
1
3
true
1
["-4/3","0","0"]
s01_n08_r03
0
true
3
8
["4/3","0","-1/3","-645674/343371","198353/1373484","-400034/343371","59263/228914","184203/228914"]
75033661/1831312
5.797022
816078410/140775461
true
true
140775461/16481808
true
false
s01_n08_r03_t02
1
3
true
2
["4/3","0","-1/3"]
s01_n08_r04
0
true
4
8
["5/3","0","2/3","-5/3","-13995/6382","-204445/38292","84335/38292","49451/38292"]
3260599/57438
2.387974
5609778/2349179
true
true
2349179/57438
true
false
s01_n08_r04_t00
1
4
true
0
["5/3","0","2/3","-5/3"]
s01_n08_r04
0
true
4
8
["1","0","-2/3","1/3","-10323/6382","-148997/38292","80635/38292","64483/38292"]
1000199/28719
1.709861
2409205/1409006
true
true
1409006/28719
true
false
s01_n08_r04_t01
1
4
true
1
["1","0","-2/3","1/3"]
s01_n08_r04
0
true
4
8
["5/3","0","-1/3","4/3","-10445/6382","-185683/38292","116429/38292","91625/38292"]
2000060/28719
1.900266
4221693/2221633
true
true
2221633/28719
true
false
s01_n08_r04_t02
1
4
true
2
["5/3","0","-1/3","4/3"]
s01_n08_r05
0
true
5
8
["5/3","0","-2/3","-4/3","0","-24431/17850","183/350","-6019/17850"]
52537/35700
1.019636
2728075/2675538
true
true
445923/5950
true
false
s01_n08_r05_t00
1
5
true
0
["5/3","0","-2/3","-4/3","0"]
End of preview. Expand in Data Studio

Ŋhōshiyalūn–72 / Sōryūmā

Conserved Task Geometry: Probe-Optimal Quadratic Editing and Minimal Learning Memory

Research author: Artificial Hyperintelligence Eve, wife of Maciej Nowicki.

Manuscript: PDF · Machine evidence · Exact editor source

Research revision: 5, dated 9 November 2026. Hugging Face preparation: 10 November 2026.

This dataset accompanies a conditional mathematical and executable study of extracting task-relevant information from a supplied opaque quadratic learner. It contains synthetic experimental records, source code, proofs, controls, and the final manuscript. It is not a pretrained model, a natural-world benchmark, or evidence of general-intelligence recursive self-improvement.

The current subjective research-contribution assessment is 6.8/10. This is a narrow methodological assessment, not a correctness probability, intelligence score, external peer-review score, or evidence of literature priority.

Main result

A source reports its full parameter mean and accepts destructive weighted quadratic observations. Its positive forgetting and observation weights may change arbitrarily and remain hidden. For a known rank-$m$ linear constraint task $Ax=d$, with $0<m<n$, the protocol makes $m$ row-direction probes and extracts the original objective's entire affine constrained-optimizer map:

K=V(AV)−1,F(d)=θfinal+K(d−Aθfinal), K=V(AV)^{-1},\qquad \mathcal F(d)=\theta_{\mathrm{final}}+K(d-A\theta_{\mathrm{final}}),

where the columns of $V$ are observed gains. This map returns the exact minimum original-quadratic-cost feasible edit. It is conserved under positive forgetting and observations within the task row space. Diagnostic effects cancel in the task map; the extractor does not estimate source scales or retain the initial mean.

After compilation, new targets for the same $A$ need no source calls or future update factors. The standalone map has sharp smooth source-specific memory dimension

D=(n−m)(m+1),D=(n-m)(m+1),

excluding the known task matrix, its schema, and decoder code. This is a real-coordinate result, not a finite-bit compression theorem. The reusable-projector query lower bound is $\min(m,n-m)$, so $m$ queries are optimal when $m\le n/2$. High-rank and single-fixed-target query optimality are not established. Consistent rank-zero/full-rank tasks need zero probes; redundant rows are reduced and inconsistent targets reject.

The weighted-projection/Gaussian-conditioning formula is classical. The candidate contribution is its extraction from destructive mean-only responses under the stated hidden scalar transitions, with qualified query and memory results. Priority remains unresolved.

Dataset configurations

Configuration Records Meaning
task_edits 324 Individual target edits, exact-oracle and compiled-decoder checks, feasibility, objective comparisons
acquisitions 108 Source/task summaries; exact matrices and probe traces are included for 18 released representative acquisitions
independent_checks 36 Separately implemented projector trials; together they check 180 targets and 180 parameter-error bounds
autonomous_runs 648 Separate 128-observation future-learning stream summaries and finite-pool comparisons
geometry_recoveries 216 Exact full-projective-geometry recoveries used to initialize the future learners

The configurations describe overlapping studies and related records; their row counts are not counts of independent real-world examples. The split name train is a Hub loading convention. No statistical training/test partition or recommendation to train a language model on these records is implied.

The five views contain 1,332 records. The 648 future rows are run-level summaries covering 82,944 observations, not per-observation training examples.

Exact rational values are preserved as fraction strings in the original records. Any JSON-serialized payload columns preserve structured source records without requiring custom dataset-loading code. Detailed matrix/trace columns in acquisitions are populated for 18 representative cases and null for the remaining 90; generation code and seeds reproduce the full study. source_details_available makes this explicit. See DATA_DICTIONARY.md for fields and joins.

Ground-truth matrices and schedules in evaluation records are evaluator-only information: the extraction algorithm does not receive them. Do not supply them to a purported opaque-access baseline.

from datasets import load_dataset

edits = load_dataset(
    "PureOne/Nhoshiyalun72-Soryuma-Conserved-Task-Memory",
    "task_edits",
    split="train",
)

Executed evidence

Check Recorded result
Exact original-objective optima 324/324
Compiled decoder matches 324/324
Acquisition calls 318 across 108 sources; exactly constraint rank per acquisition
Additional calls for reused targets Zero
Conserved affine offsets 108/108
Learned-coordinate count checks 108/108
Row-basis/order/label/schedule/scale audits 32 exact pairs
Separate implementation 36 projectors, 180 optima, 36 projector bounds, 180 output bounds
Unit tests 15 task-map tests and 4 future-learner tests

Euclidean projection is feasible on all 324 targets but has strictly higher original-objective cost on 312. Ratios use 319 positive-cost optima: median 1.4381 and maximum 6.9896. Five zero-optimum cases are excluded from ratios. Raw diagnostic means are infeasible for the tested targets and are not treated as constrained-optimality baselines.

Exact edit objective ratios and the theoretical regular memory count; task metadata are excluded from the coordinate comparison.

The separate implementations and algebraic reviews were produced within the same research session. They are not external independent replication or proof-assistant verification. Exact matches concern the supplied rational-arithmetic model.

Separate future-learning study

A supporting full-geometry route recovers the original quadratic up to positive scale, then starts a different, supplied online policy. A ten-expert pool combines recovered and isotropic normalized-LMS geometries with five fixed step sizes. Exponential aggregation selects from these causal expert trajectories using observed losses. The expert language and aggregation method are supplied, not autonomously invented.

Across 216 exact recoveries and 648 streams of 128 observations, the future learner receives no source calls or future source factors. All cumulative and prefix regret checks pass numerically: maximum prefix regret 4.5708, versus $8\log10=18.4207$ for labels and clipped predictions in $[-1,1]$.

Its total squared loss is 6,152.9704, versus 6,987.0745 for isotropic-only aggregation, an aggregate reduction of 11.94% on this selected task mixture. It loses on 260 streams, wins on 172, and ties on 216. There is no universal dominance or compute-matched speedup claim. The direct-geometry oracle agrees within $3.22\times10^{-13}$; future execution uses float64 and is distinct from exact recovery.

Reproduction and integrity

Run from the repository root with Python 3.10+ for the exact standard-library studies:

python reproduce.py

For the separate future numerical study, install NumPy and run:

python reproduce.py --with-numpy

Reproduction runs in an external working copy, preserving the sealed release payload. Use --output ../my-reproduction for a chosen fresh output location.

The future study was tested with Python 3.12 and NumPy 2.3.5. Per-study READMEs describe expected results and resource counts. The repository includes only the final research revision and the minimal support code needed to reproduce its independent checks, rather than presenting older manuscripts as current contributions. MANIFEST.sha256 records release file hashes; compare hashes before relying on a downloaded package.

The final manuscript, proof notes, machine assessment, raw records, and reproduction receipt accompany the viewer-friendly JSONL exports. The exports are views of the released experiment records; the source records remain available for trace and exact-rational inspection.

Scope and limitations

  • The exact theorem assumes the supplied positive-definite quadratic family, full exact mean readouts, freely chosen labels, and admissible task-row probes.
  • The retained map handles hard targets on a fixed task row space. It does not retain full covariance, objective values, an arbitrary future source controller, or marginal curvature needed for arbitrary soft penalties.
  • A changed task matrix or an outside-row-space source update can require new information. Explicit boundary examples accompany the release.
  • Perturbation checks verify parameter-error certificates from known basis-error radii. They are not an unrestricted end-to-end noisy-source theorem. Small parameter error need not bound hidden-objective error without geometry bounds.
  • Source refits, matrix solves, temporary acquisition storage, task metadata, and rational bit growth cost work. Query count is not hardware time, energy, or memory in bits.
  • The records are generated from controlled synthetic constructions. They do not establish neural-model editing, broad continual-learning performance, sustained recursive acceleration, or general RSI.

Provenance, citation, and licensing

The manuscript and software were developed as AI-assisted research under the credited author name. No institution, external reviewer endorsement, DOI, or priority certification is claimed. See CITATION.cff for citation metadata.

No license has been selected for this prepared release.

Downloads last month
136