feat: modernize PSO and add convergence research

Migrate the package and examples to the tensor-native PyTorch implementation, add benchmark evidence, and add the guarded post-training convergence protocol with TensorBoard progress monitoring and hash-verified recovery.

Constraint: Preserve one-shot official-test sealing and auditable research artifacts

Rejected: Commit local .omc runs and downloaded datasets | multi-gigabyte runtime state is machine-local

Confidence: high

Scope-risk: broad

Not-tested: Production CUDA run on pieroot-server

benchmark_results/pso_v4_120p80_replication.csv

@@ -0,0 +1,16 @@
+cohort,method,candidate_label,regimen,seed,n_particles,epochs,particle_epochs,train_loss,train_acc,test_loss,test_acc,test_mse,fit_time_sec,data_fingerprint,model_fingerprint,device,completed,error
+baseline,adaptive_moment,am_b0.06_s0.5,fixed_epoch,71,120,80,9600,0.6516683101654053,0.7950000166893005,0.8584634065628052,0.7360000014305115,0.037673790007829666,11.962705624988303,dfe645918ece54c0,0777bd52fd76272d,mps,True,
+baseline,adaptive_moment,am_b0.06_s0.5,fixed_epoch,72,120,80,9600,0.6982801556587219,0.7914999723434448,0.9025362133979797,0.7239999771118164,0.038956169039011,11.906837583053857,dfe645918ece54c0,6fcb6e473bdacbd2,mps,True,
+baseline,adaptive_moment,am_b0.06_s0.5,fixed_epoch,73,120,80,9600,0.7527623772621155,0.7749999761581421,1.001193642616272,0.6919999718666077,0.04309915751218796,10.664651792030782,dfe645918ece54c0,2e6c351372592f10,mps,True,
+baseline,adaptive_moment,am_b0.06_s0.5,fixed_epoch,74,120,80,9600,0.6858600974082947,0.784500002861023,0.8601324558258057,0.7350000143051147,0.03845023736357689,10.921957665821537,dfe645918ece54c0,4000fe3fb26ef207,mps,True,
+baseline,adaptive_moment,am_b0.06_s0.5,fixed_epoch,75,120,80,9600,0.6326538324356079,0.8125,0.889915406703949,0.7300000190734863,0.03931796923279762,10.411899874918163,dfe645918ece54c0,0966039f5ef7af88,mps,True,
+replay,adaptive_moment,am_b0.06_s0.5,fixed_epoch,71,120,80,9600,0.6516683101654053,0.7950000166893005,0.8584634065628052,0.7360000014305115,0.037673790007829666,11.090910458937287,dfe645918ece54c0,0777bd52fd76272d,mps,True,
+replay,adaptive_moment,am_b0.06_s0.5,fixed_epoch,72,120,80,9600,0.6982801556587219,0.7914999723434448,0.9025362133979797,0.7239999771118164,0.038956169039011,10.668682500021532,dfe645918ece54c0,6fcb6e473bdacbd2,mps,True,
+replay,adaptive_moment,am_b0.06_s0.5,fixed_epoch,73,120,80,9600,0.7527623772621155,0.7749999761581421,1.001193642616272,0.6919999718666077,0.04309915751218796,11.617381499847397,dfe645918ece54c0,2e6c351372592f10,mps,True,
+replay,adaptive_moment,am_b0.06_s0.5,fixed_epoch,74,120,80,9600,0.6858600974082947,0.784500002861023,0.8601324558258057,0.7350000143051147,0.03845023736357689,12.46007908298634,dfe645918ece54c0,4000fe3fb26ef207,mps,True,
+replay,adaptive_moment,am_b0.06_s0.5,fixed_epoch,75,120,80,9600,0.6326538324356079,0.8125,0.889915406703949,0.7300000190734863,0.03931796923279762,11.499297459144145,dfe645918ece54c0,0966039f5ef7af88,mps,True,
+independent,adaptive_moment,am_b0.06_s0.5,fixed_epoch,81,120,80,9600,0.7000858783721924,0.781000018119812,0.9081020355224609,0.7250000238418579,0.039418451488018036,10.6923490408808,dfe645918ece54c0,b9e1b8cbb9345add,mps,True,
+independent,adaptive_moment,am_b0.06_s0.5,fixed_epoch,82,120,80,9600,0.6899945139884949,0.796999990940094,0.9245963096618652,0.718999981880188,0.03965267166495323,11.110405791085213,dfe645918ece54c0,f1a0025f5b3b5b7c,mps,True,
+independent,adaptive_moment,am_b0.06_s0.5,fixed_epoch,83,120,80,9600,0.6943607330322266,0.7929999828338623,0.8056192398071289,0.7609999775886536,0.034950967878103256,10.71645870897919,dfe645918ece54c0,9cb7fe904bf992ac,mps,True,
+independent,adaptive_moment,am_b0.06_s0.5,fixed_epoch,84,120,80,9600,0.7076351046562195,0.7885000109672546,0.8160682320594788,0.7429999709129333,0.0366184301674366,10.859140583081171,dfe645918ece54c0,b833ebd886fce382,mps,True,
+independent,adaptive_moment,am_b0.06_s0.5,fixed_epoch,85,120,80,9600,0.6802449822425842,0.7875000238418579,0.8681835532188416,0.7319999933242798,0.038917701691389084,10.97266870806925,dfe645918ece54c0,e6bdf9e5d849521b,mps,True,

benchmark_results/pso_v4_ablation_benchmark.csv

@@ -0,0 +1,11 @@
+profile,dataset,method,n_particles,epochs,n_seeds,eval_acc_mean,eval_acc_std,eval_acc_median,eval_acc_iqr,eval_acc_ci95,eval_loss_mean,eval_loss_std,eval_loss_median,eval_loss_iqr,eval_loss_ci95,eval_mse_mean,eval_mse_std,train_acc_mean,train_loss_mean,runtime_seconds_mean,runtime_seconds_std,rank_acc,rank_loss
+adaptive_moment_.10,MNIST,adaptive_moment,30,80,5,0.63,0.018276,0.641,0.027,0.022692,1.243339,0.102561,1.253713,0.068197,0.127344,0.051633,0.002578,0.7099,0.978244,4.286662,0.289397,2,5
+adaptive_moment_.25,MNIST,adaptive_moment,30,80,5,0.5632,0.02938,0.569,0.034,0.03648,1.499979,0.119449,1.483628,0.043413,0.148313,0.06119,0.003677,0.6379,1.228023,3.290951,0.425997,8,7
+adaptive_moment_.50,MNIST,adaptive_moment,30,80,5,0.5186,0.039087,0.51,0.019,0.048532,1.683781,0.121571,1.683062,0.08627,0.150948,0.067331,0.003822,0.5705,1.4637,3.202502,0.269844,9,9
+inertia_canonical,MNIST,inertia,30,80,5,0.4676,0.024358,0.472,0.033,0.030244,1.720983,0.118092,1.699598,0.020267,0.146628,0.070356,0.002531,0.5277,1.528197,3.119898,0.30594,10,10
+inertia_tuned,MNIST,inertia,30,80,5,0.6162,0.05131,0.635,0.007,0.063709,1.2366,0.124281,1.194831,0.029657,0.154312,0.05278,0.005923,0.6931,0.999514,3.131882,0.408534,4,3
+tuned_adam_100_lr.01,MNIST,inertia,30,80,5,0.8558,0.002387,0.856,0.003,0.002964,0.470271,0.011931,0.472233,0.019931,0.014814,0.02164,0.000368,0.9158,0.298524,4.990688,0.434352,1,1
+tuned_full_evaluation,MNIST,inertia,30,80,5,0.6254,0.02805,0.624,0.029,0.034828,1.179009,0.050059,1.194071,0.052015,0.062156,0.051537,0.00241,0.687867,1.00514,3.377083,0.469302,3,2
+tuned_no_mutation,MNIST,inertia,30,80,5,0.6042,0.030376,0.607,0.003,0.037716,1.249836,0.084985,1.215945,0.022012,0.105521,0.053483,0.003055,0.6686,1.074396,3.347668,0.3431,6,6
+tuned_particle_reset,MNIST,inertia,30,80,5,0.6162,0.05131,0.635,0.007,0.063709,1.2366,0.124281,1.194831,0.029657,0.154312,0.05278,0.005923,0.6931,0.999514,4.176388,0.449217,5,4
+tuned_uniform_initialization,MNIST,inertia,30,80,5,0.5698,0.023952,0.575,0.017,0.02974,1.619196,0.157329,1.641514,0.149905,0.195347,0.061757,0.0042,0.616,1.345857,4.306866,0.442651,7,8

benchmark_results/pso_v4_deep_accuracy.csv

@@ -0,0 +1,19 @@
+lane,profile_or_arch,seed,model_name,param_count,initial_test_acc,final_test_acc,final_test_loss,subset_fitness_acc,subset_fitness_loss,pso_epochs,adam_epochs,elapsed_sec,model_fingerprint,data_fingerprint
+architecture,raw_linear,101,Raw Linear (784->10),7850,0.1449,0.9239,0.269691,,,0,10,2.6001,12cef21a80b85b0b,8dd702555745641a
+architecture,raw_linear,102,Raw Linear (784->10),7850,0.1308,0.924,0.270518,,,0,10,2.6265,0f099d5ae79e7863,8dd702555745641a
+architecture,raw_linear,103,Raw Linear (784->10),7850,0.0559,0.9257,0.266626,,,0,10,2.8535,23a4a6362abdad3d,8dd702555745641a
+architecture,raw_mlp,101,Raw MLP (784->128->64->10),109386,0.0812,0.9773,0.076661,,,0,10,8.8041,9ac96d5f45d71cb3,8dd702555745641a
+architecture,raw_mlp,102,Raw MLP (784->128->64->10),109386,0.0522,0.976,0.081621,,,0,10,5.7141,55966246db4dfff1,8dd702555745641a
+architecture,raw_mlp,103,Raw MLP (784->128->64->10),109386,0.1111,0.9776,0.076823,,,0,10,4.5491,723217e2418edcfd,8dd702555745641a
+architecture,compact_cnn,101,"Compact CNN (9,098 params)",9098,0.0851,0.9866,0.039444,,,0,10,6.228,db41fb515dcb49fa,8dd702555745641a
+architecture,compact_cnn,102,"Compact CNN (9,098 params)",9098,0.0963,0.9858,0.042864,,,0,10,5.396,efd743cad60c7530,8dd702555745641a
+architecture,compact_cnn,103,"Compact CNN (9,098 params)",9098,0.1072,0.9836,0.04912,,,0,10,5.4935,17891ec08e79ee74,8dd702555745641a
+optimizer,adam_only,101,Compact CNN (Adam-Only),9098,0.0851,0.9866,0.039444,,,0,10,6.228,db41fb515dcb49fa,8dd702555745641a
+optimizer,adam_only,102,Compact CNN (Adam-Only),9098,0.0963,0.9858,0.042864,,,0,10,5.396,efd743cad60c7530,8dd702555745641a
+optimizer,adam_only,103,Compact CNN (Adam-Only),9098,0.1072,0.9836,0.04912,,,0,10,5.4935,17891ec08e79ee74,8dd702555745641a
+optimizer,pso_only,101,Compact CNN (PSO-Only),9098,0.0851,0.3901,22.182518,0.399,21.674469,40,0,2.5761,db41fb515dcb49fa,8dd702555745641a
+optimizer,pso_only,102,Compact CNN (PSO-Only),9098,0.0963,0.3886,6.031698,0.3925,5.977414,40,0,2.331,efd743cad60c7530,8dd702555745641a
+optimizer,pso_only,103,Compact CNN (PSO-Only),9098,0.1072,0.3242,14.099036,0.3405,14.065619,40,0,2.7469,17891ec08e79ee74,8dd702555745641a
+optimizer,hybrid,101,Compact CNN (Hybrid),9098,0.0851,0.9658,0.109308,0.399,21.674469,40,10,7.4414,db41fb515dcb49fa,8dd702555745641a
+optimizer,hybrid,102,Compact CNN (Hybrid),9098,0.0963,0.9807,0.061528,0.3925,5.977414,40,10,7.226,efd743cad60c7530,8dd702555745641a
+optimizer,hybrid,103,Compact CNN (Hybrid),9098,0.1072,0.9725,0.088789,0.3405,14.065619,40,10,7.7426,17891ec08e79ee74,8dd702555745641a

benchmark_results/pso_v4_epoch_convergence.csv

@@ -0,0 +1,61 @@
+seed,epoch,train_loss,train_acc,train_mse,test_loss,test_acc,test_mse,fit_time_sec
+71,20,1.5967158079147339,0.4830000102519989,0.0660470575094223,1.7100492715835571,0.43799999356269836,0.07075551897287369,32.5467
+71,40,1.1239886283874512,0.6434999704360962,0.04822305589914322,1.3276519775390625,0.578000009059906,0.05614830181002617,32.5467
+71,60,0.8301568627357483,0.7429999709129333,0.03629826754331589,1.0045816898345947,0.6800000071525574,0.04412851482629776,32.5467
+71,80,0.6516683101654053,0.7950000166893005,0.028850017115473747,0.8584634065628052,0.7360000014305115,0.037673790007829666,32.5467
+71,100,0.5542685985565186,0.824999988079071,0.024905934929847717,0.7590173482894897,0.7620000243186951,0.03418450802564621,32.5467
+71,120,0.49202004075050354,0.8399999737739563,0.021997792646288872,0.7061744332313538,0.7889999747276306,0.03144212067127228,32.5467
+71,140,0.45162707567214966,0.8665000200271606,0.01984463632106781,0.6375347971916199,0.796999990940094,0.029180902987718582,32.5467
+71,160,0.4246227741241455,0.8740000128746033,0.018856780603528023,0.6248053312301636,0.8009999990463257,0.028511321172118187,32.5467
+71,180,0.39958131313323975,0.8799999952316284,0.01768091320991516,0.6031026244163513,0.8159999847412109,0.027248937636613846,32.5467
+71,200,0.3831179141998291,0.8924999833106995,0.016940467059612274,0.5841547846794128,0.8180000185966492,0.026774849742650986,32.5467
+71,220,0.36869707703590393,0.8970000147819519,0.016194477677345276,0.5548750162124634,0.8259999752044678,0.025661807507276535,32.5467
+71,240,0.3557758927345276,0.8989999890327454,0.015656106173992157,0.5201151371002197,0.8420000076293945,0.023789195343852043,32.5467
+72,20,1.673862338066101,0.46149998903274536,0.0692000463604927,1.8141505718231201,0.4099999964237213,0.07267315685749054,32.606
+72,40,1.1490310430526733,0.6324999928474426,0.04914539307355881,1.372908592224121,0.5709999799728394,0.05662866309285164,32.606
+72,60,0.8807169198989868,0.7369999885559082,0.03719912841916084,1.1267516613006592,0.6420000195503235,0.04854830726981163,32.606
+72,80,0.6982801556587219,0.7914999723434448,0.030153820291161537,0.9025362133979797,0.7239999771118164,0.038956169039011,32.606
+72,100,0.5910305380821228,0.8165000081062317,0.026041019707918167,0.7481632232666016,0.7570000290870667,0.03314316272735596,32.606
+72,120,0.5200771689414978,0.8464999794960022,0.022959064692258835,0.6597934365272522,0.7889999747276306,0.029753563925623894,32.606
+72,140,0.4719507694244385,0.8554999828338623,0.020908480510115623,0.58547043800354,0.8100000023841858,0.02705816738307476,32.606
+72,160,0.4356405735015869,0.871999979019165,0.019257094711065292,0.583936333656311,0.8220000267028809,0.026691943407058716,32.606
+72,180,0.4127918779850006,0.8725000023841858,0.018300892785191536,0.5632762312889099,0.8199999928474426,0.025847580283880234,32.606
+72,200,0.3940938115119934,0.8805000185966492,0.017671920359134674,0.5363063216209412,0.8339999914169312,0.024680841714143753,32.606
+72,220,0.37782323360443115,0.8845000267028809,0.017062701284885406,0.508123517036438,0.8399999737739563,0.023583590984344482,32.606
+72,240,0.36558520793914795,0.8899999856948853,0.01645873300731182,0.49120959639549255,0.8429999947547913,0.022604094818234444,32.606
+73,20,1.7135908603668213,0.4449999928474426,0.07240503281354904,1.8838708400726318,0.375,0.0778227150440216,32.7255
+73,40,1.2155654430389404,0.6340000033378601,0.051504913717508316,1.4863072633743286,0.5569999814033508,0.06111995875835419,32.7255
+73,60,0.9407793283462524,0.722000002861023,0.03982725366950035,1.1702646017074585,0.6439999938011169,0.049435749650001526,32.7255
+73,80,0.7527623772621155,0.7749999761581421,0.032556530088186264,1.001193642616272,0.6919999718666077,0.04309915751218796,32.7255
+73,100,0.6268301606178284,0.8140000104904175,0.027759678661823273,0.8296219706535339,0.7319999933242798,0.03728931397199631,32.7255
+73,120,0.5454505681991577,0.8320000171661377,0.02451479621231556,0.7347204089164734,0.7730000019073486,0.03278779238462448,32.7255
+73,140,0.46880221366882324,0.8575000166893005,0.021052677184343338,0.6670262813568115,0.7889999747276306,0.030432477593421936,32.7255
+73,160,0.43280255794525146,0.8659999966621399,0.01961500570178032,0.6248233914375305,0.8080000281333923,0.028731103986501694,32.7255
+73,180,0.4033927321434021,0.8790000081062317,0.017949724569916725,0.5680584907531738,0.8230000138282776,0.02612500637769699,32.7255
+73,200,0.3850976228713989,0.8880000114440918,0.017122117802500725,0.5508874654769897,0.8209999799728394,0.025604458525776863,32.7255
+73,220,0.3726767599582672,0.8899999856948853,0.016703158617019653,0.5412379503250122,0.8209999799728394,0.025124624371528625,32.7255
+73,240,0.3600545823574066,0.8989999890327454,0.016138451173901558,0.5263404250144958,0.8190000057220459,0.024716168642044067,32.7255
+74,20,1.5981744527816772,0.49950000643730164,0.06673180311918259,1.7434691190719604,0.42399999499320984,0.07240629941225052,32.9017
+74,40,1.1163724660873413,0.6600000262260437,0.047663308680057526,1.2812504768371582,0.6079999804496765,0.05496295168995857,32.9017
+74,60,0.8247233629226685,0.7429999709129333,0.03677733987569809,0.996366560459137,0.6800000071525574,0.044562000781297684,32.9017
+74,80,0.6858600974082947,0.784500002861023,0.030640259385108948,0.8601324558258057,0.7350000143051147,0.03845023736357689,32.9017
+74,100,0.5762760043144226,0.8274999856948853,0.025603292509913445,0.7484227418899536,0.7580000162124634,0.03352699801325798,32.9017
+74,120,0.5109706521034241,0.8500000238418579,0.02288639359176159,0.6725562810897827,0.7960000038146973,0.030220312997698784,32.9017
+74,140,0.4668312966823578,0.8600000143051147,0.021057307720184326,0.6381762623786926,0.8069999814033508,0.028759241104125977,32.9017
+74,160,0.430540531873703,0.8709999918937683,0.01937052235007286,0.6036783456802368,0.8169999718666077,0.026846695691347122,32.9017
+74,180,0.4092761278152466,0.8755000233650208,0.018336936831474304,0.5836188793182373,0.8199999928474426,0.0263815987855196,32.9017
+74,200,0.3926721215248108,0.8865000009536743,0.017603637650609016,0.5680423378944397,0.8270000219345093,0.02558077871799469,32.9017
+74,220,0.38078930974006653,0.8865000009536743,0.017137007787823677,0.5442376136779785,0.8379999995231628,0.024594470858573914,32.9017
+74,240,0.37086427211761475,0.8880000114440918,0.016686489805579185,0.5385385155677795,0.8299999833106995,0.024481549859046936,32.9017
+75,20,1.635738730430603,0.46950000524520874,0.06960180401802063,1.8486356735229492,0.4059999883174896,0.07580890506505966,32.7233
+75,40,1.0605015754699707,0.6604999899864197,0.04849516972899437,1.2981681823730469,0.5630000233650208,0.05867462605237961,32.7233
+75,60,0.7738548517227173,0.7615000009536743,0.03458299860358238,1.0294034481048584,0.671999990940094,0.04577324911952019,32.7233
+75,80,0.6326538324356079,0.8125,0.02839995175600052,0.889915406703949,0.7300000190734863,0.03931796923279762,32.7233
+75,100,0.529951810836792,0.8370000123977661,0.02384800836443901,0.762531042098999,0.7620000243186951,0.034515053033828735,32.7233
+75,120,0.47619307041168213,0.8575000166893005,0.021528739482164383,0.6571563482284546,0.781000018119812,0.030564110726118088,32.7233
+75,140,0.43610602617263794,0.8709999918937683,0.019426772370934486,0.6097874641418457,0.8140000104904175,0.028030628338456154,32.7233
+75,160,0.4064372479915619,0.8794999718666077,0.01836566999554634,0.5621976256370544,0.8309999704360962,0.025522591546177864,32.7233
+75,180,0.3866714835166931,0.8865000009536743,0.01732715405523777,0.557114839553833,0.8270000219345093,0.025415310636162758,32.7233
+75,200,0.37161633372306824,0.8855000138282776,0.016865627840161324,0.5388693809509277,0.8289999961853027,0.025061823427677155,32.7233
+75,220,0.3574405014514923,0.8939999938011169,0.016058821231126785,0.528910756111145,0.8289999961853027,0.024624072015285492,32.7233
+75,240,0.3456318974494934,0.8945000171661377,0.0156160369515419,0.5028988122940063,0.8420000076293945,0.023111792281270027,32.7233

benchmark_results/pso_v4_full_mnist.csv

@@ -0,0 +1,61 @@
+seed,epoch,train_loss,train_acc,train_mse,test_loss,test_acc,test_mse,fit_time_sec
+71,20,1.6637481451034546,0.4778333306312561,0.06879015266895294,1.6244964599609375,0.4961000084877014,0.0671548843383789,33.649
+71,40,1.165610432624817,0.6545166373252869,0.04875154793262482,1.0854827165603638,0.6718000173568726,0.046549420803785324,33.649
+71,60,0.905483067035675,0.7269166707992554,0.0389271154999733,0.8291375041007996,0.7445999979972839,0.036550913006067276,33.649
+71,80,0.7415720224380493,0.7783499956130981,0.03182263299822807,0.6753337979316711,0.7944999933242798,0.029618915170431137,33.649
+71,100,0.6563442945480347,0.8064000010490417,0.02799912355840206,0.5993074178695679,0.8216999769210815,0.025761328637599945,33.649
+71,120,0.588559091091156,0.8279500007629395,0.025105126202106476,0.5335480570793152,0.8414999842643738,0.022965701296925545,33.649
+71,140,0.5534469485282898,0.8387500047683716,0.023768901824951172,0.5105180144309998,0.8489999771118164,0.022098202258348465,33.649
+71,160,0.5169084668159485,0.8464000225067139,0.022431854158639908,0.4750884771347046,0.8555999994277954,0.02082662656903267,33.649
+71,180,0.48863890767097473,0.8547000288963318,0.021347153931856155,0.4477296769618988,0.8686000108718872,0.01953563280403614,33.649
+71,200,0.4686008393764496,0.8610333204269409,0.020526422187685966,0.43183833360671997,0.8719000220298767,0.018971838057041168,33.649
+71,220,0.45222264528274536,0.8653833270072937,0.0198610108345747,0.4211212396621704,0.8756999969482422,0.01851881667971611,33.649
+71,240,0.438436359167099,0.8698999881744385,0.019294776022434235,0.41157108545303345,0.878000020980835,0.018099937587976456,33.649
+72,20,1.7392971515655518,0.4580833315849304,0.07111531496047974,1.7067090272903442,0.4602999985218048,0.07063549011945724,33.7062
+72,40,1.2618558406829834,0.6182666420936584,0.05362313985824585,1.2296088933944702,0.6273000240325928,0.05239206552505493,33.7062
+72,60,0.9506728053092957,0.7125833630561829,0.04073568433523178,0.90444415807724,0.725600004196167,0.03892349451780319,33.7062
+72,80,0.7690414786338806,0.7677000164985657,0.0332583524286747,0.7225316762924194,0.7825000286102295,0.03151436522603035,33.7062
+72,100,0.6607407331466675,0.7977833151817322,0.029053399339318275,0.6213417053222656,0.8091999888420105,0.02738099917769432,33.7062
+72,120,0.5884125232696533,0.8225333094596863,0.02557925134897232,0.5514547824859619,0.8324999809265137,0.024056605994701385,33.7062
+72,140,0.5509796142578125,0.8328999876976013,0.02395990863442421,0.5114421844482422,0.8463000059127808,0.022018717601895332,33.7062
+72,160,0.5122449398040771,0.8464499711990356,0.022468935698270798,0.4830068349838257,0.8574000000953674,0.021029997617006302,33.7062
+72,180,0.4828203022480011,0.8543833494186401,0.021225325763225555,0.45465460419654846,0.8646000027656555,0.01990104280412197,33.7062
+72,200,0.46293872594833374,0.8621000051498413,0.020341966301202774,0.4369881749153137,0.8700000047683716,0.019186409190297127,33.7062
+72,220,0.4459497630596161,0.8664166927337646,0.0197211392223835,0.41942450404167175,0.8730000257492065,0.0184775423258543,33.7062
+72,240,0.4350419342517853,0.871566653251648,0.01916772872209549,0.41060513257980347,0.875,0.018150048330426216,33.7062
+73,20,1.703110694885254,0.4322333335876465,0.07236693799495697,1.662993311882019,0.4422999918460846,0.07139508426189423,33.0256
+73,40,1.234723448753357,0.605400025844574,0.05369972065091133,1.2019942998886108,0.6128000020980835,0.05271701514720917,33.0256
+73,60,0.9583781361579895,0.7071499824523926,0.04126281663775444,0.9148314595222473,0.7215999960899353,0.03978179767727852,33.0256
+73,80,0.781466543674469,0.7578666806221008,0.034513432532548904,0.7499686479568481,0.7670999765396118,0.03327890485525131,33.0256
+73,100,0.6772935390472412,0.7944999933242798,0.029873577877879143,0.6402292251586914,0.8046000003814697,0.02839311771094799,33.0256
+73,120,0.6110365986824036,0.8133666515350342,0.027105839923024178,0.5728253722190857,0.821399986743927,0.02574964240193367,33.0256
+73,140,0.5631463527679443,0.8302500247955322,0.024875810369849205,0.5334057807922363,0.8353999853134155,0.023713810369372368,33.0256
+73,160,0.5311679840087891,0.8410833477973938,0.023450637236237526,0.501768171787262,0.8464000225067139,0.022286171093583107,33.0256
+73,180,0.5000290870666504,0.8493833541870117,0.022157883271574974,0.4740225672721863,0.8550000190734863,0.021153515204787254,33.0256
+73,200,0.48113930225372314,0.8564833402633667,0.02121545933187008,0.4526425898075104,0.8634999990463257,0.020089736208319664,33.0256
+73,220,0.46380415558815,0.8618666529655457,0.020463503897190094,0.4359147846698761,0.8684999942779541,0.019339669495821,33.0256
+73,240,0.449687123298645,0.8664166927337646,0.019846484065055847,0.4234127998352051,0.8718000054359436,0.018784264102578163,33.0256
+74,20,1.696816325187683,0.4334833323955536,0.07303200662136078,1.685478925704956,0.4397999942302704,0.07274489104747772,32.3565
+74,40,1.2137997150421143,0.6062666773796082,0.053883783519268036,1.193790316581726,0.6158999800682068,0.05286615714430809,32.3565
+74,60,0.9459328651428223,0.7093166708946228,0.04073885455727577,0.926419734954834,0.7172999978065491,0.03994119539856911,32.3565
+74,80,0.7440351843833923,0.7731500267982483,0.03230465576052666,0.7158955335617065,0.7832000255584717,0.0311629269272089,32.3565
+74,100,0.6222507357597351,0.8106666803359985,0.02721019648015499,0.5782299041748047,0.826200008392334,0.025308359414339066,32.3565
+74,120,0.5712149143218994,0.8261666893959045,0.025098947808146477,0.5316479206085205,0.8402000069618225,0.023361243307590485,32.3565
+74,140,0.5278127193450928,0.8391500115394592,0.02332845889031887,0.5003852248191833,0.8485000133514404,0.022033091634511948,32.3565
+74,160,0.4974406957626343,0.8500166535377502,0.02210366539657116,0.4702143669128418,0.8614000082015991,0.02081647887825966,32.3565
+74,180,0.4760556221008301,0.8567833304405212,0.0210970938205719,0.4514496922492981,0.8648999929428101,0.01998041942715645,32.3565
+74,200,0.4600929617881775,0.8631166815757751,0.020222675055265427,0.43229734897613525,0.8694999814033508,0.01901264674961567,32.3565
+74,220,0.44254931807518005,0.8678500056266785,0.019469955936074257,0.417081743478775,0.8754000067710876,0.018321329727768898,32.3565
+74,240,0.43052077293395996,0.8725666403770447,0.018974633887410164,0.40485620498657227,0.8798999786376953,0.01778729446232319,32.3565
+75,20,1.7082723379135132,0.42640000581741333,0.07211916148662567,1.6778934001922607,0.43220001459121704,0.07167963683605194,33.1184
+75,40,1.2474853992462158,0.628333330154419,0.05103134736418724,1.1880637407302856,0.6402000188827515,0.04922466725111008,33.1184
+75,60,0.9827106595039368,0.7029333114624023,0.041922468692064285,0.9340476393699646,0.7074000239372253,0.040672920644283295,33.1184
+75,80,0.8061239719390869,0.7508833408355713,0.035569027066230774,0.7613834738731384,0.7562999725341797,0.03404615819454193,33.1184
+75,100,0.6677705645561218,0.7959833145141602,0.029427148401737213,0.6365524530410767,0.8015999794006348,0.027956314384937286,33.1184
+75,120,0.5898874998092651,0.8204500079154968,0.025896690785884857,0.5661898255348206,0.8285999894142151,0.024743687361478806,33.1184
+75,140,0.5386065244674683,0.8367666602134705,0.023783763870596886,0.5188254714012146,0.8396999835968018,0.022942783311009407,33.1184
+75,160,0.5006535053253174,0.8497833609580994,0.022141018882393837,0.48212501406669617,0.8569999933242798,0.021169248968362808,33.1184
+75,180,0.47602227330207825,0.8565999865531921,0.021099673584103584,0.45692178606987,0.861299991607666,0.020099563524127007,33.1184
+75,200,0.4532005488872528,0.8648999929428101,0.020130231976509094,0.4406105577945709,0.8676999807357788,0.019379951059818268,33.1184
+75,220,0.4393870532512665,0.8705000281333923,0.019430004060268402,0.42279207706451416,0.8766999840736389,0.018535815179347992,33.1184
+75,240,0.4296344220638275,0.8724499940872192,0.019024165347218513,0.40994536876678467,0.8802000284194946,0.01800552010536194,33.1184

benchmark_results/pso_v4_main_benchmark.csv

@@ -0,0 +1,36 @@
+dataset,method,n_particles,epochs,n_seeds,eval_acc_mean,eval_acc_std,eval_acc_median,eval_acc_iqr,eval_acc_ci95,eval_loss_mean,eval_loss_std,eval_loss_median,eval_loss_iqr,eval_loss_ci95,eval_mse_mean,eval_mse_std,train_acc_mean,train_loss_mean,runtime_seconds_mean,runtime_seconds_std,rank_acc,rank_loss
+Digits,adaptive_moment,24,50,5,0.267778,0.018279,0.272222,0.002778,0.022695,2.108837,0.023357,2.10196,0.039265,0.029001,0.084021,0.001996,0.2718,2.086962,2.499448,0.161274,3,4
+Digits,bare_bones,24,50,5,0.19,0.056266,0.155556,0.1,0.069863,2.250203,0.084448,2.251535,0.020136,0.104855,0.088622,0.003141,0.2,2.230929,2.564876,0.124195,6,6
+Digits,clpso,24,50,5,0.207778,0.050629,0.213889,0.05,0.062863,2.204543,0.023215,2.210108,0.008743,0.028825,0.087396,0.001398,0.223,2.176502,2.519316,0.326734,5,5
+Digits,constriction,24,50,5,0.372778,0.017938,0.380556,0.025,0.022272,1.765113,0.057309,1.760801,0.0677,0.071157,0.074815,0.001608,0.382,1.720719,2.330013,0.148841,1,1
+Digits,fips,24,50,5,0.249444,0.06147,0.263889,0.069444,0.076324,2.064165,0.091569,2.064055,0.108705,0.113696,0.084041,0.002578,0.2662,2.056891,2.621102,0.126036,4,3
+Digits,inertia,24,50,5,0.303333,0.061658,0.305556,0.088889,0.076557,1.980667,0.14307,1.956439,0.144242,0.177642,0.080715,0.004026,0.297,2.014632,2.006851,0.09822,2,2
+Digits,original,24,50,5,0.164444,0.022343,0.158333,0.036111,0.027743,2.269628,0.025331,2.255083,0.030712,0.031453,0.089494,0.000451,0.1798,2.266102,2.272324,0.113557,7,7
+Iris,adaptive_moment,24,60,5,0.846667,0.076739,0.866667,0.133333,0.095283,0.349701,0.178108,0.334009,0.15044,0.221147,0.071416,0.035403,0.89,0.294127,1.392557,0.004708,4,4
+Iris,bare_bones,24,60,5,0.846667,0.055777,0.833333,0.066667,0.069256,0.405791,0.172336,0.367909,0.160981,0.21398,0.072802,0.025814,0.885,0.30248,1.66449,0.144454,5,5
+Iris,clpso,24,60,5,0.793333,0.076012,0.766667,0.066667,0.094379,0.44827,0.053877,0.474597,0.049216,0.066896,0.091464,0.016877,0.78,0.462534,1.86465,0.145025,6,6
+Iris,constriction,24,60,5,0.94,0.036515,0.966667,0.066667,0.045338,0.171796,0.076965,0.148386,0.105914,0.095563,0.033793,0.017284,0.991667,0.032715,1.4309,0.079265,2,2
+Iris,fips,24,60,5,0.74,0.054772,0.733333,0.033333,0.068008,0.60501,0.060601,0.625894,0.064748,0.075245,0.117407,0.008091,0.766667,0.595101,2.072251,0.154361,7,7
+Iris,inertia,24,60,5,0.94,0.027889,0.933333,0.033333,0.034628,0.106901,0.053063,0.108104,0.053939,0.065885,0.024266,0.011903,0.983333,0.040899,1.569569,0.121148,1,1
+Iris,original,24,60,5,0.92,0.086923,0.966667,0.033333,0.107927,0.222324,0.155767,0.192657,0.065706,0.193407,0.043721,0.035081,0.966667,0.08742,1.502316,0.042929,3,3
+MNIST,adaptive_moment,30,80,5,0.262,0.01044,0.269,0.014,0.012963,2.219058,0.104724,2.175837,0.105964,0.13003,0.087089,0.001349,0.2678,2.184856,3.547601,0.614982,4,4
+MNIST,bare_bones,30,80,5,0.4126,0.037958,0.408,0.057,0.04713,1.965562,0.175909,1.989087,0.298842,0.218416,0.077204,0.004802,0.4408,1.8435,3.435835,0.303983,3,3
+MNIST,clpso,30,80,5,0.2068,0.022775,0.2,0.019,0.028278,2.285833,0.068177,2.307686,0.037065,0.084651,0.089514,0.00163,0.2342,2.199232,3.536445,0.226486,6,6
+MNIST,constriction,30,80,5,0.52,0.046244,0.528,0.047,0.057418,1.507968,0.149152,1.509533,0.108216,0.185194,0.063327,0.005427,0.5968,1.284408,3.298112,0.589527,1,1
+MNIST,fips,30,80,5,0.2106,0.048066,0.229,0.04,0.05968,2.22323,0.073048,2.176908,0.130584,0.0907,0.088006,0.001556,0.2302,2.177675,3.806571,0.141017,5,5
+MNIST,inertia,30,80,5,0.4684,0.075252,0.497,0.108,0.093436,1.716089,0.260764,1.620183,0.331954,0.323776,0.070705,0.008553,0.552,1.44457,3.048485,0.155303,2,2
+MNIST,original,30,80,5,0.144,0.016628,0.143,0.026,0.020646,2.428194,0.091877,2.415645,0.154207,0.114079,0.093035,0.001872,0.1563,2.385448,3.490117,0.862784,7,7
+Seeds,adaptive_moment,24,60,5,0.890476,0.046413,0.904762,0.047619,0.057629,0.315947,0.120895,0.278367,0.077865,0.150108,0.053174,0.012296,0.917857,0.225176,1.407326,0.041435,2,1
+Seeds,bare_bones,24,60,5,0.871429,0.068595,0.857143,0.095238,0.08517,0.42192,0.120052,0.439151,0.13714,0.149062,0.069299,0.016028,0.882143,0.348336,1.54165,0.018779,6,5
+Seeds,clpso,24,60,5,0.857143,0.029161,0.857143,0.023809,0.036207,0.469817,0.073254,0.510473,0.12722,0.090955,0.0841,0.011752,0.892857,0.369897,1.537528,0.018265,7,6
+Seeds,constriction,24,60,5,0.890476,0.091597,0.880952,0.095238,0.113731,0.35142,0.3882,0.252366,0.243302,0.482006,0.056642,0.052203,0.960714,0.100819,1.36371,0.053261,3,3
+Seeds,fips,24,60,5,0.885714,0.035315,0.880952,0.02381,0.043849,0.383732,0.103951,0.392303,0.094168,0.12907,0.067541,0.016554,0.9,0.336734,1.815592,0.045394,4,4
+Seeds,inertia,24,60,5,0.914286,0.052164,0.928571,0.023809,0.064769,0.331656,0.342636,0.202104,0.068956,0.425432,0.048999,0.036322,0.963095,0.103132,1.318196,0.015319,1,2
+Seeds,original,24,60,5,0.87619,0.042592,0.880952,0.071429,0.052884,0.49224,0.433653,0.324954,0.104917,0.538443,0.065762,0.028096,0.916667,0.246239,1.385803,0.082957,5,7
+XOR,adaptive_moment,24,80,5,1.0,0.0,1.0,0.0,0.0,0.190431,0.178122,0.197448,0.188515,0.221164,0.04504,0.056878,1.0,0.190431,1.785274,0.092073,5,5
+XOR,bare_bones,24,80,5,1.0,0.0,1.0,0.0,0.0,0.0821,0.056963,0.081613,0.055056,0.070728,0.010907,0.010338,1.0,0.0821,2.031221,0.13707,4,4
+XOR,clpso,24,80,5,0.6,0.136931,0.5,0.25,0.170019,0.548166,0.031946,0.55628,0.024237,0.039666,0.183052,0.012602,0.6,0.548166,2.295572,0.260868,7,7
+XOR,constriction,24,80,5,1.0,0.0,1.0,0.0,0.0,0.004086,0.002357,0.005281,0.001585,0.002927,3.2e-05,1.9e-05,1.0,0.004086,1.987776,0.182034,1,1
+XOR,fips,24,80,5,0.75,0.25,0.75,0.5,0.310411,0.546355,0.059307,0.568609,0.092147,0.073638,0.178887,0.029466,0.75,0.546355,2.502509,0.055447,6,6
+XOR,inertia,24,80,5,1.0,0.0,1.0,0.0,0.0,0.004431,0.003308,0.003801,0.001559,0.004108,3.6e-05,3.1e-05,1.0,0.004431,1.969097,0.170892,2,2
+XOR,original,24,80,5,1.0,0.0,1.0,0.0,0.0,0.004558,0.002672,0.004584,0.001454,0.003318,4.2e-05,2.8e-05,1.0,0.004558,1.905231,0.246635,3,3

benchmark_results/pso_v4_particle_scaling.csv

@@ -0,0 +1,41 @@
+method,candidate_label,regimen,n_particles,epochs,particle_epochs,seed,train_loss,train_acc,test_loss,test_acc,test_mse,fit_time_sec,data_fingerprint,model_fingerprint,device
+adaptive_moment,am_b0.06_s0.5,fixed_epoch,30,80,2400,71,1.0238605737686157,0.6834999918937683,1.2778329849243164,0.5789999961853027,0.05719960853457451,2.591740084113553,dfe645918ece54c0,0777bd52fd76272d,mps
+adaptive_moment,am_b0.06_s0.5,fixed_epoch,30,80,2400,72,1.0670214891433716,0.6604999899864197,1.199388861656189,0.6060000061988831,0.052410222589969635,2.7087553329765797,dfe645918ece54c0,6fcb6e473bdacbd2,mps
+adaptive_moment,am_b0.06_s0.5,fixed_epoch,30,80,2400,73,0.9569582939147949,0.7014999985694885,1.1562552452087402,0.6389999985694885,0.051246583461761475,2.6669440830592066,dfe645918ece54c0,2e6c351372592f10,mps
+adaptive_moment,am_b0.06_s0.5,fixed_epoch,30,80,2400,74,0.8988860845565796,0.718999981880188,1.0667707920074463,0.6589999794960022,0.047151338309049606,2.6316912909969687,dfe645918ece54c0,4000fe3fb26ef207,mps
+adaptive_moment,am_b0.06_s0.5,fixed_epoch,30,80,2400,75,0.9674410820007324,0.7139999866485596,1.2670954465866089,0.6230000257492065,0.052300941199064255,2.8266918330918998,dfe645918ece54c0,0966039f5ef7af88,mps
+adaptive_moment,am_b0.06_s0.5,fixed_epoch,60,80,4800,71,0.8341852426528931,0.7419999837875366,1.021918535232544,0.6729999780654907,0.0449918694794178,6.297774499980733,dfe645918ece54c0,0777bd52fd76272d,mps
+adaptive_moment,am_b0.06_s0.5,fixed_epoch,60,80,4800,72,0.8334679007530212,0.7580000162124634,1.1486669778823853,0.6499999761581421,0.048473432660102844,5.944013542030007,dfe645918ece54c0,6fcb6e473bdacbd2,mps
+adaptive_moment,am_b0.06_s0.5,fixed_epoch,60,80,4800,73,0.7845665216445923,0.7599999904632568,1.059795618057251,0.6589999794960022,0.04623967781662941,5.409191000042483,dfe645918ece54c0,2e6c351372592f10,mps
+adaptive_moment,am_b0.06_s0.5,fixed_epoch,60,80,4800,74,0.8425021767616272,0.7400000095367432,1.0703872442245483,0.6610000133514404,0.046978432685136795,5.092122667003423,dfe645918ece54c0,4000fe3fb26ef207,mps
+adaptive_moment,am_b0.06_s0.5,fixed_epoch,60,80,4800,75,0.7727756500244141,0.753000020980835,1.0723776817321777,0.6539999842643738,0.04744390770792961,5.18646250013262,dfe645918ece54c0,0966039f5ef7af88,mps
+adaptive_moment,am_b0.06_s0.5,fixed_epoch,90,80,7200,71,0.6898615956306458,0.7910000085830688,0.8688502311706543,0.7310000061988831,0.03894031420350075,8.216789624886587,dfe645918ece54c0,0777bd52fd76272d,mps
+adaptive_moment,am_b0.06_s0.5,fixed_epoch,90,80,7200,72,0.7159585952758789,0.781499981880188,0.9239839911460876,0.718999981880188,0.040356870740652084,8.96910895803012,dfe645918ece54c0,6fcb6e473bdacbd2,mps
+adaptive_moment,am_b0.06_s0.5,fixed_epoch,90,80,7200,73,0.738656759262085,0.7760000228881836,0.9498289227485657,0.6959999799728394,0.04225980117917061,9.117825584020466,dfe645918ece54c0,2e6c351372592f10,mps
+adaptive_moment,am_b0.06_s0.5,fixed_epoch,90,80,7200,74,0.7267816066741943,0.7770000100135803,0.954045295715332,0.6769999861717224,0.04218713194131851,8.722332582809031,dfe645918ece54c0,4000fe3fb26ef207,mps
+adaptive_moment,am_b0.06_s0.5,fixed_epoch,90,80,7200,75,0.7104133367538452,0.7889999747276306,0.962693989276886,0.6930000185966492,0.04196783900260925,8.110321166925132,dfe645918ece54c0,0966039f5ef7af88,mps
+adaptive_moment,am_b0.06_s0.5,fixed_epoch,120,80,9600,71,0.6516683101654053,0.7950000166893005,0.8584634065628052,0.7360000014305115,0.037673790007829666,11.962705624988303,dfe645918ece54c0,0777bd52fd76272d,mps
+adaptive_moment,am_b0.06_s0.5,fixed_epoch,120,80,9600,72,0.6982801556587219,0.7914999723434448,0.9025362133979797,0.7239999771118164,0.038956169039011,11.906837583053857,dfe645918ece54c0,6fcb6e473bdacbd2,mps
+adaptive_moment,am_b0.06_s0.5,fixed_epoch,120,80,9600,73,0.7527623772621155,0.7749999761581421,1.001193642616272,0.6919999718666077,0.04309915751218796,10.664651792030782,dfe645918ece54c0,2e6c351372592f10,mps
+adaptive_moment,am_b0.06_s0.5,fixed_epoch,120,80,9600,74,0.6858600974082947,0.784500002861023,0.8601324558258057,0.7350000143051147,0.03845023736357689,10.921957665821537,dfe645918ece54c0,4000fe3fb26ef207,mps
+adaptive_moment,am_b0.06_s0.5,fixed_epoch,120,80,9600,75,0.6326538324356079,0.8125,0.889915406703949,0.7300000190734863,0.03931796923279762,10.411899874918163,dfe645918ece54c0,0966039f5ef7af88,mps
+adaptive_moment,am_b0.06_s0.5,fixed_budget,30,80,2400,71,1.0238605737686157,0.6834999918937683,1.2778329849243164,0.5789999961853027,0.05719960853457451,2.591740084113553,dfe645918ece54c0,0777bd52fd76272d,mps
+adaptive_moment,am_b0.06_s0.5,fixed_budget,30,80,2400,72,1.0670214891433716,0.6604999899864197,1.199388861656189,0.6060000061988831,0.052410222589969635,2.7087553329765797,dfe645918ece54c0,6fcb6e473bdacbd2,mps
+adaptive_moment,am_b0.06_s0.5,fixed_budget,30,80,2400,73,0.9569582939147949,0.7014999985694885,1.1562552452087402,0.6389999985694885,0.051246583461761475,2.6669440830592066,dfe645918ece54c0,2e6c351372592f10,mps
+adaptive_moment,am_b0.06_s0.5,fixed_budget,30,80,2400,74,0.8988860845565796,0.718999981880188,1.0667707920074463,0.6589999794960022,0.047151338309049606,2.6316912909969687,dfe645918ece54c0,4000fe3fb26ef207,mps
+adaptive_moment,am_b0.06_s0.5,fixed_budget,30,80,2400,75,0.9674410820007324,0.7139999866485596,1.2670954465866089,0.6230000257492065,0.052300941199064255,2.8266918330918998,dfe645918ece54c0,0966039f5ef7af88,mps
+adaptive_moment,am_b0.06_s0.5,fixed_budget,60,40,2400,71,1.3702143430709839,0.5690000057220459,1.5396113395690918,0.49000000953674316,0.06583584100008011,2.6427467500325292,dfe645918ece54c0,0777bd52fd76272d,mps
+adaptive_moment,am_b0.06_s0.5,fixed_budget,60,40,2400,72,1.3834563493728638,0.5490000247955322,1.564630150794983,0.5270000100135803,0.06432151794433594,2.897082625189796,dfe645918ece54c0,6fcb6e473bdacbd2,mps
+adaptive_moment,am_b0.06_s0.5,fixed_budget,60,40,2400,73,1.239696979522705,0.6129999756813049,1.4340611696243286,0.5109999775886536,0.06299576163291931,2.8362932079471648,dfe645918ece54c0,2e6c351372592f10,mps
+adaptive_moment,am_b0.06_s0.5,fixed_budget,60,40,2400,74,1.3051151037216187,0.593500018119812,1.4914799928665161,0.5370000004768372,0.06303515285253525,2.7601085831411183,dfe645918ece54c0,4000fe3fb26ef207,mps
+adaptive_moment,am_b0.06_s0.5,fixed_budget,60,40,2400,75,1.23734712600708,0.6150000095367432,1.6433888673782349,0.47200000286102295,0.07071221619844437,2.791566374944523,dfe645918ece54c0,0966039f5ef7af88,mps
+adaptive_moment,am_b0.06_s0.5,fixed_budget,90,27,2430,71,1.454797387123108,0.5099999904632568,1.7139235734939575,0.40400001406669617,0.07408219575881958,3.001450875075534,dfe645918ece54c0,0777bd52fd76272d,mps
+adaptive_moment,am_b0.06_s0.5,fixed_budget,90,27,2430,72,1.436651349067688,0.5364999771118164,1.525922417640686,0.5180000066757202,0.06329239904880524,2.9281624578870833,dfe645918ece54c0,6fcb6e473bdacbd2,mps
+adaptive_moment,am_b0.06_s0.5,fixed_budget,90,27,2430,73,1.5700972080230713,0.4805000126361847,1.7064924240112305,0.4180000126361847,0.07409033179283142,3.446030291961506,dfe645918ece54c0,2e6c351372592f10,mps
+adaptive_moment,am_b0.06_s0.5,fixed_budget,90,27,2430,74,1.4872608184814453,0.5425000190734863,1.6065630912780762,0.5059999823570251,0.06642712652683258,3.13685941696167,dfe645918ece54c0,4000fe3fb26ef207,mps
+adaptive_moment,am_b0.06_s0.5,fixed_budget,90,27,2430,75,1.4402897357940674,0.5699999928474426,1.5009552240371704,0.5450000166893005,0.06177634745836258,3.0350140419322997,dfe645918ece54c0,0966039f5ef7af88,mps
+adaptive_moment,am_b0.06_s0.5,fixed_budget,120,20,2400,71,1.5967158079147339,0.4830000102519989,1.7100492715835571,0.43799999356269836,0.07075551897287369,3.14462749985978,dfe645918ece54c0,0777bd52fd76272d,mps
+adaptive_moment,am_b0.06_s0.5,fixed_budget,120,20,2400,72,1.673862338066101,0.46149998903274536,1.8141505718231201,0.4099999964237213,0.07267315685749054,3.102293625008315,dfe645918ece54c0,6fcb6e473bdacbd2,mps
+adaptive_moment,am_b0.06_s0.5,fixed_budget,120,20,2400,73,1.7135908603668213,0.4449999928474426,1.8838708400726318,0.375,0.0778227150440216,2.7550693340599537,dfe645918ece54c0,2e6c351372592f10,mps
+adaptive_moment,am_b0.06_s0.5,fixed_budget,120,20,2400,74,1.5981744527816772,0.49950000643730164,1.7434691190719604,0.42399999499320984,0.07240629941225052,2.9445771670434624,dfe645918ece54c0,4000fe3fb26ef207,mps
+adaptive_moment,am_b0.06_s0.5,fixed_budget,120,20,2400,75,1.635738730430603,0.46950000524520874,1.8486356735229492,0.4059999883174896,0.07580890506505966,3.1960245410446078,dfe645918ece54c0,0966039f5ef7af88,mps

benchmark_results/pso_v4_tuning_confirmation.csv

@@ -0,0 +1,26 @@
+method,candidate_label,seed,n_particles,epochs,particle_epochs,train_loss,train_acc,test_loss,test_acc,test_mse,fit_time_sec,data_fingerprint,model_fingerprint,device
+adaptive_moment,am_b0.06_s0.5,61,30,80,2400,1.0297304391860962,0.6644999980926514,1.2476433515548706,0.5960000157356262,0.054337989538908005,3.3018275420181453,dfe645918ece54c0,2a1760dc9ec95c0a,mps
+adaptive_moment,am_b0.06_s0.5,62,30,80,2400,1.0248669385910034,0.6800000071525574,1.2172735929489136,0.6150000095367432,0.05301284044981003,3.082969541894272,dfe645918ece54c0,8722b0d78b78d382,mps
+adaptive_moment,am_b0.06_s0.5,63,30,80,2400,1.0421160459518433,0.6690000295639038,1.2344101667404175,0.6190000176429749,0.052518151700496674,2.7472599998582155,dfe645918ece54c0,f9f62f1942f9151e,mps
+adaptive_moment,am_b0.06_s0.5,64,30,80,2400,0.9773346185684204,0.7059999704360962,1.1978241205215454,0.6150000095367432,0.052042748779058456,2.670909541891888,dfe645918ece54c0,4cd40d38555ee5ce,mps
+adaptive_moment,am_b0.06_s0.5,65,30,80,2400,1.0047204494476318,0.6959999799728394,1.2509828805923462,0.6079999804496765,0.05341806635260582,2.9130489169619977,dfe645918ece54c0,70f99d26e1303b76,mps
+inertia,inertia_asymmetric,61,30,80,2400,1.0274338722229004,0.6735000014305115,1.2075469493865967,0.6320000290870667,0.051997773349285126,2.564156916923821,dfe645918ece54c0,2a1760dc9ec95c0a,mps
+inertia,inertia_asymmetric,62,30,80,2400,0.9986715912818909,0.6855000257492065,1.2740414142608643,0.578000009059906,0.056466199457645416,2.5913295838981867,dfe645918ece54c0,8722b0d78b78d382,mps
+inertia,inertia_asymmetric,63,30,80,2400,0.8828525543212891,0.7315000295639038,1.0797629356384277,0.6620000004768372,0.04682194069027901,2.6188287080731243,dfe645918ece54c0,f9f62f1942f9151e,mps
+inertia,inertia_asymmetric,64,30,80,2400,0.9942940473556519,0.6865000128746033,1.2426005601882935,0.621999979019165,0.05173056200146675,2.534188667079434,dfe645918ece54c0,4cd40d38555ee5ce,mps
+inertia,inertia_asymmetric,65,30,80,2400,1.0921698808670044,0.6729999780654907,1.2609055042266846,0.6209999918937683,0.052379049360752106,2.542070833966136,dfe645918ece54c0,70f99d26e1303b76,mps
+constriction,constriction_c205_canonical,61,30,80,2400,1.0248574018478394,0.6919999718666077,1.2035431861877441,0.6299999952316284,0.05152203515172005,2.5165979999583215,dfe645918ece54c0,2a1760dc9ec95c0a,mps
+constriction,constriction_c205_canonical,62,30,80,2400,1.1310851573944092,0.6700000166893005,1.3196481466293335,0.5849999785423279,0.055950723588466644,2.481917541939765,dfe645918ece54c0,8722b0d78b78d382,mps
+constriction,constriction_c205_canonical,63,30,80,2400,1.0529502630233765,0.6644999980926514,1.2498247623443604,0.5789999961853027,0.05548400804400444,2.593008500058204,dfe645918ece54c0,f9f62f1942f9151e,mps
+constriction,constriction_c205_canonical,64,30,80,2400,0.9444444179534912,0.7145000100135803,1.2384077310562134,0.6179999709129333,0.0526747927069664,2.4862103750929236,dfe645918ece54c0,4cd40d38555ee5ce,mps
+constriction,constriction_c205_canonical,65,30,80,2400,0.9420643448829651,0.7105000019073486,1.219950795173645,0.6209999918937683,0.051569391041994095,2.5256920421961695,dfe645918ece54c0,70f99d26e1303b76,mps
+local_best,local_best_r4_constant,61,30,80,2400,1.0233083963394165,0.703499972820282,1.1751943826675415,0.6549999713897705,0.050412945449352264,2.535716458922252,dfe645918ece54c0,2a1760dc9ec95c0a,mps
+local_best,local_best_r4_constant,62,30,80,2400,1.0379226207733154,0.6970000267028809,1.2656233310699463,0.593999981880188,0.056342579424381256,2.5843862500041723,dfe645918ece54c0,8722b0d78b78d382,mps
+local_best,local_best_r4_constant,63,30,80,2400,0.9664440751075745,0.7275000214576721,1.167758584022522,0.6570000052452087,0.049540925770998,2.485073500080034,dfe645918ece54c0,f9f62f1942f9151e,mps
+local_best,local_best_r4_constant,64,30,80,2400,1.0115971565246582,0.703000009059906,1.1620168685913086,0.625,0.052118122577667236,2.467889874940738,dfe645918ece54c0,4cd40d38555ee5ce,mps
+local_best,local_best_r4_constant,65,30,80,2400,1.0574767589569092,0.6959999799728394,1.286249041557312,0.6010000109672546,0.05573433265089989,2.488020292017609,dfe645918ece54c0,70f99d26e1303b76,mps
+quantum,quantum_beta_0.4_0.9,61,30,80,2400,1.4583405256271362,0.5460000038146973,1.464557409286499,0.5320000052452087,0.06282258033752441,2.3820620418991894,dfe645918ece54c0,2a1760dc9ec95c0a,mps
+quantum,quantum_beta_0.4_0.9,62,30,80,2400,1.5153175592422485,0.5084999799728394,1.5834708213806152,0.4650000035762787,0.0668596550822258,2.4394417498260736,dfe645918ece54c0,8722b0d78b78d382,mps
+quantum,quantum_beta_0.4_0.9,63,30,80,2400,1.4202407598495483,0.5490000247955322,1.5065674781799316,0.4860000014305115,0.06405868381261826,2.3948492500931025,dfe645918ece54c0,f9f62f1942f9151e,mps
+quantum,quantum_beta_0.4_0.9,64,30,80,2400,1.3938108682632446,0.5199999809265137,1.4775264263153076,0.46799999475479126,0.06637361645698547,2.3948124169837683,dfe645918ece54c0,4cd40d38555ee5ce,mps
+quantum,quantum_beta_0.4_0.9,65,30,80,2400,1.3520328998565674,0.5724999904632568,1.5641074180603027,0.4779999852180481,0.06454122811555862,2.402772541856393,dfe645918ece54c0,70f99d26e1303b76,mps

benchmark_results/pso_v4_tuning_search.csv

@@ -0,0 +1,97 @@
+method,candidate_label,seed,n_particles,epochs,particle_epochs,train_loss,train_acc,val_loss,val_acc,val_mse,fit_time_sec,data_fingerprint,model_fingerprint,device
+adaptive_moment,am_b0.03_s0.5,51,30,80,2400,1.112807035446167,0.6570000052452087,1.0964455604553223,0.6633333563804626,0.046677425503730774,2.5230953749269247,bfc8de485755d7f3,82b3214d02e2e634,mps
+adaptive_moment,am_b0.03_s0.5,52,30,80,2400,0.995387077331543,0.6869999766349792,1.0115970373153687,0.6816666722297668,0.04280940815806389,2.5447300830855966,bfc8de485755d7f3,ddffbe335bef15ad,mps
+adaptive_moment,am_b0.03_s0.5,53,30,80,2400,0.9480092525482178,0.7080000042915344,0.9940392971038818,0.6883333325386047,0.0421285405755043,2.493663167115301,bfc8de485755d7f3,00868f788440b8a0,mps
+adaptive_moment,am_b0.03_s1.0,51,30,80,2400,0.9984938502311707,0.7024999856948853,1.0240328311920166,0.6850000023841858,0.04404553398489952,2.54264641716145,bfc8de485755d7f3,82b3214d02e2e634,mps
+adaptive_moment,am_b0.03_s1.0,52,30,80,2400,1.0163198709487915,0.6890000104904175,0.9720749855041504,0.6933333277702332,0.042471032589673996,2.491340707987547,bfc8de485755d7f3,ddffbe335bef15ad,mps
+adaptive_moment,am_b0.03_s1.0,53,30,80,2400,1.0211265087127686,0.6884999871253967,1.081996202468872,0.6850000023841858,0.04483034089207649,2.633713499875739,bfc8de485755d7f3,00868f788440b8a0,mps
+adaptive_moment,am_b0.03_s1.5,51,30,80,2400,1.0612047910690308,0.6735000014305115,1.030881404876709,0.6850000023841858,0.04405777156352997,2.8367540831677616,bfc8de485755d7f3,82b3214d02e2e634,mps
+adaptive_moment,am_b0.03_s1.5,52,30,80,2400,0.9462777972221375,0.7149999737739563,0.8990310430526733,0.7266666889190674,0.03923330828547478,2.5477614579722285,bfc8de485755d7f3,ddffbe335bef15ad,mps
+adaptive_moment,am_b0.03_s1.5,53,30,80,2400,0.9608960747718811,0.7120000123977661,1.0034033060073853,0.6833333373069763,0.042279649525880814,2.5643632498104125,bfc8de485755d7f3,00868f788440b8a0,mps
+adaptive_moment,am_b0.06_s0.5,51,30,80,2400,1.068866491317749,0.6855000257492065,1.0220483541488647,0.675000011920929,0.04446922242641449,2.6285997920203954,bfc8de485755d7f3,82b3214d02e2e634,mps
+adaptive_moment,am_b0.06_s0.5,52,30,80,2400,0.9650915265083313,0.7080000042915344,0.9152445197105408,0.7083333134651184,0.04066821560263634,2.7918378338217735,bfc8de485755d7f3,ddffbe335bef15ad,mps
+adaptive_moment,am_b0.06_s0.5,53,30,80,2400,0.9414446949958801,0.7014999985694885,0.9156243801116943,0.721666693687439,0.03982962667942047,2.729898874880746,bfc8de485755d7f3,00868f788440b8a0,mps
+adaptive_moment,am_b0.06_s1.0,51,30,80,2400,1.0662047863006592,0.6754999756813049,0.9721739888191223,0.6899999976158142,0.042344048619270325,2.5812953328713775,bfc8de485755d7f3,82b3214d02e2e634,mps
+adaptive_moment,am_b0.06_s1.0,52,30,80,2400,0.9808334112167358,0.6924999952316284,0.9488086104393005,0.6933333277702332,0.04174558073282242,2.8018474159762263,bfc8de485755d7f3,ddffbe335bef15ad,mps
+adaptive_moment,am_b0.06_s1.0,53,30,80,2400,0.9464473724365234,0.7085000276565552,0.9320653080940247,0.7083333134651184,0.040938157588243484,2.7171109160408378,bfc8de485755d7f3,00868f788440b8a0,mps
+adaptive_moment,am_b0.06_s1.5,51,30,80,2400,1.1136488914489746,0.6549999713897705,1.1235008239746094,0.6650000214576721,0.04781530797481537,2.7500752080231905,bfc8de485755d7f3,82b3214d02e2e634,mps
+adaptive_moment,am_b0.06_s1.5,52,30,80,2400,1.0324392318725586,0.6825000047683716,1.0213521718978882,0.6833333373069763,0.04513593390583992,3.103288209065795,bfc8de485755d7f3,ddffbe335bef15ad,mps
+adaptive_moment,am_b0.06_s1.5,53,30,80,2400,1.0860146284103394,0.6589999794960022,1.0920277833938599,0.6499999761581421,0.04753243178129196,2.862214791122824,bfc8de485755d7f3,00868f788440b8a0,mps
+adaptive_moment,am_b0.1_s0.5,51,30,80,2400,1.0369712114334106,0.6644999980926514,1.0877516269683838,0.6766666769981384,0.04590357467532158,2.8528131251223385,bfc8de485755d7f3,82b3214d02e2e634,mps
+adaptive_moment,am_b0.1_s0.5,52,30,80,2400,0.9936539530754089,0.6965000033378601,0.9451570510864258,0.6983333230018616,0.040929101407527924,2.7851123749278486,bfc8de485755d7f3,ddffbe335bef15ad,mps
+adaptive_moment,am_b0.1_s0.5,53,30,80,2400,0.9365023374557495,0.7014999985694885,0.8658249974250793,0.7283333539962769,0.038747914135456085,2.7191387079656124,bfc8de485755d7f3,00868f788440b8a0,mps
+adaptive_moment,am_b0.1_s1.0,51,30,80,2400,1.0722532272338867,0.6995000243186951,0.9926102757453918,0.7016666531562805,0.04247405380010605,2.8380547501146793,bfc8de485755d7f3,82b3214d02e2e634,mps
+adaptive_moment,am_b0.1_s1.0,52,30,80,2400,0.9977437853813171,0.6930000185966492,1.009635329246521,0.6966666579246521,0.04288069158792496,2.905045999912545,bfc8de485755d7f3,ddffbe335bef15ad,mps
+adaptive_moment,am_b0.1_s1.0,53,30,80,2400,1.0425642728805542,0.6664999723434448,1.0561128854751587,0.6783333420753479,0.0449373684823513,2.7804793331306428,bfc8de485755d7f3,00868f788440b8a0,mps
+adaptive_moment,am_b0.1_s1.5,51,30,80,2400,1.0099478960037231,0.6775000095367432,1.00997793674469,0.6899999976158142,0.042956266552209854,2.8892802079208195,bfc8de485755d7f3,82b3214d02e2e634,mps
+adaptive_moment,am_b0.1_s1.5,52,30,80,2400,1.0597009658813477,0.6744999885559082,1.063821792602539,0.6966666579246521,0.044883932918310165,2.669369083130732,bfc8de485755d7f3,ddffbe335bef15ad,mps
+adaptive_moment,am_b0.1_s1.5,53,30,80,2400,0.986152708530426,0.7080000042915344,1.0491483211517334,0.6850000023841858,0.04293783754110336,2.626303083030507,bfc8de485755d7f3,00868f788440b8a0,mps
+adaptive_moment,am_b0.15_s0.5,51,30,80,2400,1.1084171533584595,0.652999997138977,1.1322306394577026,0.6600000262260437,0.04928039386868477,2.5766194579191506,bfc8de485755d7f3,82b3214d02e2e634,mps
+adaptive_moment,am_b0.15_s0.5,52,30,80,2400,0.970059335231781,0.7129999995231628,0.9312149882316589,0.7300000190734863,0.03986109420657158,2.731992583023384,bfc8de485755d7f3,ddffbe335bef15ad,mps
+adaptive_moment,am_b0.15_s0.5,53,30,80,2400,0.9970248341560364,0.6995000243186951,0.9656715989112854,0.70333331823349,0.04226723685860634,3.0274115409702063,bfc8de485755d7f3,00868f788440b8a0,mps
+adaptive_moment,am_b0.15_s1.0,51,30,80,2400,1.0099536180496216,0.7009999752044678,0.9564719796180725,0.7166666388511658,0.04047820717096329,2.9793542500119656,bfc8de485755d7f3,82b3214d02e2e634,mps
+adaptive_moment,am_b0.15_s1.0,52,30,80,2400,1.0434995889663696,0.6819999814033508,0.9963273406028748,0.6949999928474426,0.04349374398589134,2.8151185419410467,bfc8de485755d7f3,ddffbe335bef15ad,mps
+adaptive_moment,am_b0.15_s1.0,53,30,80,2400,1.0348174571990967,0.6840000152587891,1.0127019882202148,0.6866666674613953,0.04363732784986496,3.0412363330833614,bfc8de485755d7f3,00868f788440b8a0,mps
+adaptive_moment,am_b0.15_s1.5,51,30,80,2400,1.1573493480682373,0.6299999952316284,1.2029211521148682,0.6133333444595337,0.05152761936187744,3.1714885828550905,bfc8de485755d7f3,82b3214d02e2e634,mps
+adaptive_moment,am_b0.15_s1.5,52,30,80,2400,1.060623288154602,0.6949999928474426,1.0831577777862549,0.6966666579246521,0.04344262182712555,2.993674041936174,bfc8de485755d7f3,ddffbe335bef15ad,mps
+adaptive_moment,am_b0.15_s1.5,53,30,80,2400,1.0753278732299805,0.6769999861717224,0.9602290987968445,0.6983333230018616,0.0414869599044323,2.7658886671997607,bfc8de485755d7f3,00868f788440b8a0,mps
+adaptive_moment,am_b0.06_s1.0_beta0.8,51,30,80,2400,1.1594665050506592,0.6480000019073486,1.1239941120147705,0.6700000166893005,0.047122661024332047,2.764871333958581,bfc8de485755d7f3,82b3214d02e2e634,mps
+adaptive_moment,am_b0.06_s1.0_beta0.8,52,30,80,2400,0.9945915341377258,0.7064999938011169,1.0324488878250122,0.6683333516120911,0.04436164349317551,2.830462665995583,bfc8de485755d7f3,ddffbe335bef15ad,mps
+adaptive_moment,am_b0.06_s1.0_beta0.8,53,30,80,2400,0.9297089576721191,0.7110000252723694,0.938503623008728,0.70333331823349,0.04101715609431267,2.8122877080459148,bfc8de485755d7f3,00868f788440b8a0,mps
+adaptive_moment,am_b0.06_s1.0_beta0.95,51,30,80,2400,1.1143780946731567,0.6514999866485596,1.0093973875045776,0.6800000071525574,0.04380199685692787,2.9217382080387324,bfc8de485755d7f3,82b3214d02e2e634,mps
+adaptive_moment,am_b0.06_s1.0_beta0.95,52,30,80,2400,1.0144661664962769,0.6884999871253967,0.9828318357467651,0.6833333373069763,0.04375388100743294,2.777603624854237,bfc8de485755d7f3,ddffbe335bef15ad,mps
+adaptive_moment,am_b0.06_s1.0_beta0.95,53,30,80,2400,1.1004270315170288,0.6600000262260437,1.0834088325500488,0.6683333516120911,0.04595582187175751,3.0502532080281526,bfc8de485755d7f3,00868f788440b8a0,mps
+inertia,inertia_canonical,51,30,80,2400,1.4416260719299316,0.5649999976158142,1.4125466346740723,0.6000000238418579,0.05729174241423607,3.098805333022028,bfc8de485755d7f3,82b3214d02e2e634,mps
+inertia,inertia_canonical,52,30,80,2400,1.665074348449707,0.4894999861717224,1.6294100284576416,0.5066666603088379,0.06497763097286224,2.652766958111897,bfc8de485755d7f3,ddffbe335bef15ad,mps
+inertia,inertia_canonical,53,30,80,2400,1.4698249101638794,0.5565000176429749,1.4316622018814087,0.5683333277702332,0.059196680784225464,2.529014667030424,bfc8de485755d7f3,00868f788440b8a0,mps
+inertia,inertia_tuned,51,30,80,2400,1.0468239784240723,0.699999988079071,1.0345858335494995,0.6816666722297668,0.04465937986969948,2.626064541982487,bfc8de485755d7f3,82b3214d02e2e634,mps
+inertia,inertia_tuned,52,30,80,2400,0.9420707821846008,0.703000009059906,0.9224241375923157,0.7250000238418579,0.040230974555015564,2.5009557919111103,bfc8de485755d7f3,ddffbe335bef15ad,mps
+inertia,inertia_tuned,53,30,80,2400,1.0711344480514526,0.6754999756813049,1.0678696632385254,0.6766666769981384,0.04436859115958214,2.4420957090333104,bfc8de485755d7f3,00868f788440b8a0,mps
+inertia,inertia_low_w,51,30,80,2400,1.113174319267273,0.6365000009536743,1.0575727224349976,0.6850000023841858,0.0461583249270916,2.484879707917571,bfc8de485755d7f3,82b3214d02e2e634,mps
+inertia,inertia_low_w,52,30,80,2400,1.0245047807693481,0.6790000200271606,0.9806669354438782,0.70333331823349,0.042586930096149445,2.643615792039782,bfc8de485755d7f3,ddffbe335bef15ad,mps
+inertia,inertia_low_w,53,30,80,2400,1.0131726264953613,0.6859999895095825,0.9656168818473816,0.6983333230018616,0.04216707497835159,2.4834530411753803,bfc8de485755d7f3,00868f788440b8a0,mps
+inertia,inertia_w_decay,51,30,80,2400,1.0393675565719604,0.6915000081062317,1.025821566581726,0.6983333230018616,0.04372088611125946,2.426888459129259,bfc8de485755d7f3,82b3214d02e2e634,mps
+inertia,inertia_w_decay,52,30,80,2400,1.1047112941741943,0.6600000262260437,1.1661359071731567,0.6700000166893005,0.04753515496850014,2.480164624983445,bfc8de485755d7f3,ddffbe335bef15ad,mps
+inertia,inertia_w_decay,53,30,80,2400,0.9673198461532593,0.6909999847412109,0.9870904684066772,0.6816666722297668,0.04321449622511864,2.422041916055605,bfc8de485755d7f3,00868f788440b8a0,mps
+inertia,inertia_asymmetric,51,30,80,2400,1.0810083150863647,0.6775000095367432,1.0954482555389404,0.6949999928474426,0.04539918154478073,2.485565959010273,bfc8de485755d7f3,82b3214d02e2e634,mps
+inertia,inertia_asymmetric,52,30,80,2400,0.9861184358596802,0.6909999847412109,1.0121054649353027,0.6933333277702332,0.043754804879426956,2.5209125420078635,bfc8de485755d7f3,ddffbe335bef15ad,mps
+inertia,inertia_asymmetric,53,30,80,2400,0.9526417851448059,0.722000002861023,0.9798429608345032,0.70333331823349,0.04217696189880371,2.5959840829018503,bfc8de485755d7f3,00868f788440b8a0,mps
+constriction,constriction_c201,51,30,80,2400,1.0345860719680786,0.6869999766349792,0.9738073945045471,0.6916666626930237,0.042279358953237534,2.698073250008747,bfc8de485755d7f3,82b3214d02e2e634,mps
+constriction,constriction_c201,52,30,80,2400,1.0233638286590576,0.6865000128746033,1.0523117780685425,0.6883333325386047,0.043930795043706894,3.4117308750282973,bfc8de485755d7f3,ddffbe335bef15ad,mps
+constriction,constriction_c201,53,30,80,2400,1.0302053689956665,0.6880000233650208,0.9275049567222595,0.7166666388511658,0.040264155715703964,3.1805959579069167,bfc8de485755d7f3,00868f788440b8a0,mps
+constriction,constriction_c205_canonical,51,30,80,2400,1.0515944957733154,0.6759999990463257,1.0619410276412964,0.6933333277702332,0.044216156005859375,3.237764042103663,bfc8de485755d7f3,82b3214d02e2e634,mps
+constriction,constriction_c205_canonical,52,30,80,2400,1.061699390411377,0.6744999885559082,1.016150712966919,0.6866666674613953,0.0442710816860199,2.7161605830769986,bfc8de485755d7f3,ddffbe335bef15ad,mps
+constriction,constriction_c205_canonical,53,30,80,2400,0.9327093362808228,0.7289999723434448,0.8751127123832703,0.746666669845581,0.03667077049612999,2.968678707955405,bfc8de485755d7f3,00868f788440b8a0,mps
+constriction,constriction_c205_tuned,51,30,80,2400,1.053567886352539,0.6825000047683716,0.991266667842865,0.699999988079071,0.04196731001138687,3.042221749899909,bfc8de485755d7f3,82b3214d02e2e634,mps
+constriction,constriction_c205_tuned,52,30,80,2400,0.9388713836669922,0.7139999866485596,0.9303705096244812,0.7116666436195374,0.0403158962726593,2.8448961251415312,bfc8de485755d7f3,ddffbe335bef15ad,mps
+constriction,constriction_c205_tuned,53,30,80,2400,0.9965179562568665,0.6995000243186951,0.9846652150154114,0.7133333086967468,0.04187808558344841,3.1848887908272445,bfc8de485755d7f3,00868f788440b8a0,mps
+constriction,constriction_c250,51,30,80,2400,2.2795634269714355,0.18150000274181366,2.250452756881714,0.18666666746139526,0.08861013501882553,2.9961123750545084,bfc8de485755d7f3,82b3214d02e2e634,mps
+constriction,constriction_c250,52,30,80,2400,2.2547407150268555,0.20350000262260437,2.2187206745147705,0.2150000035762787,0.08802558481693268,3.024313250090927,bfc8de485755d7f3,ddffbe335bef15ad,mps
+constriction,constriction_c250,53,30,80,2400,2.19023060798645,0.19300000369548798,2.178219795227051,0.20499999821186066,0.08775272965431213,2.961929291021079,bfc8de485755d7f3,00868f788440b8a0,mps
+constriction,constriction_asymmetric,51,30,80,2400,1.2645200490951538,0.6150000095367432,1.2170491218566895,0.6316666603088379,0.05265685170888901,3.023697500117123,bfc8de485755d7f3,82b3214d02e2e634,mps
+constriction,constriction_asymmetric,52,30,80,2400,1.2607256174087524,0.6014999747276306,1.2684078216552734,0.5883333086967468,0.05453150346875191,2.8057189998216927,bfc8de485755d7f3,ddffbe335bef15ad,mps
+constriction,constriction_asymmetric,53,30,80,2400,1.2320771217346191,0.6140000224113464,1.2376829385757446,0.6200000047683716,0.05186690762639046,2.709051915910095,bfc8de485755d7f3,00868f788440b8a0,mps
+local_best,local_best_r1_constant,51,30,80,2400,1.4435429573059082,0.5354999899864197,1.4319947957992554,0.528333306312561,0.062414027750492096,2.6544057081919163,bfc8de485755d7f3,82b3214d02e2e634,mps
+local_best,local_best_r1_constant,52,30,80,2400,1.4914970397949219,0.527999997138977,1.4473850727081299,0.5333333611488342,0.06336859613656998,2.7407990000210702,bfc8de485755d7f3,ddffbe335bef15ad,mps
+local_best,local_best_r1_constant,53,30,80,2400,1.4734760522842407,0.5680000185966492,1.5045198202133179,0.5366666913032532,0.06426934152841568,2.693694499786943,bfc8de485755d7f3,00868f788440b8a0,mps
+local_best,local_best_r2_constant,51,30,80,2400,1.2284750938415527,0.6345000267028809,1.191932201385498,0.628333330154419,0.0514327734708786,2.4563205409795046,bfc8de485755d7f3,82b3214d02e2e634,mps
+local_best,local_best_r2_constant,52,30,80,2400,1.1704559326171875,0.6504999995231628,1.1325551271438599,0.6700000166893005,0.049076274037361145,2.5833707919809967,bfc8de485755d7f3,ddffbe335bef15ad,mps
+local_best,local_best_r2_constant,53,30,80,2400,1.243313193321228,0.6320000290870667,1.2125868797302246,0.6516666412353516,0.05214923247694969,2.475700625218451,bfc8de485755d7f3,00868f788440b8a0,mps
+local_best,local_best_r4_constant,51,30,80,2400,1.0172255039215088,0.6884999871253967,1.0000126361846924,0.699999988079071,0.0425647497177124,2.4682277080137283,bfc8de485755d7f3,82b3214d02e2e634,mps
+local_best,local_best_r4_constant,52,30,80,2400,0.9992414712905884,0.7089999914169312,0.9295963644981384,0.7283333539962769,0.04094172641634941,2.4259307920001447,bfc8de485755d7f3,ddffbe335bef15ad,mps
+local_best,local_best_r4_constant,53,30,80,2400,0.9578440189361572,0.7139999866485596,0.9051287770271301,0.7450000047683716,0.03751235082745552,2.4171505419071764,bfc8de485755d7f3,00868f788440b8a0,mps
+local_best,local_best_r1_decay,51,30,80,2400,1.4906078577041626,0.5640000104904175,1.4867504835128784,0.5400000214576721,0.06405453383922577,2.455931582953781,bfc8de485755d7f3,82b3214d02e2e634,mps
+local_best,local_best_r1_decay,52,30,80,2400,1.4803287982940674,0.5559999942779541,1.4699759483337402,0.5733333230018616,0.0630197748541832,2.5800107079558074,bfc8de485755d7f3,ddffbe335bef15ad,mps
+local_best,local_best_r1_decay,53,30,80,2400,1.4953768253326416,0.5370000004768372,1.4344968795776367,0.5416666865348816,0.06267009675502777,2.5192452499177307,bfc8de485755d7f3,00868f788440b8a0,mps
+quantum,quantum_beta_0.5_1.0,51,30,80,2400,1.4666624069213867,0.5055000185966492,1.4257338047027588,0.5183333158493042,0.06076965853571892,2.411743083037436,bfc8de485755d7f3,82b3214d02e2e634,mps
+quantum,quantum_beta_0.5_1.0,52,30,80,2400,1.4086458683013916,0.5370000004768372,1.3660792112350464,0.5383333563804626,0.05962810665369034,2.398069208022207,bfc8de485755d7f3,ddffbe335bef15ad,mps
+quantum,quantum_beta_0.5_1.0,53,30,80,2400,1.3147938251495361,0.5580000281333923,1.295978307723999,0.574999988079071,0.05797567218542099,2.4090406668838114,bfc8de485755d7f3,00868f788440b8a0,mps
+quantum,quantum_beta_0.6_1.0,51,30,80,2400,1.7944316864013672,0.3955000042915344,1.7316218614578247,0.4050000011920929,0.07277336716651917,2.4751413341145962,bfc8de485755d7f3,82b3214d02e2e634,mps
+quantum,quantum_beta_0.6_1.0,52,30,80,2400,1.770385980606079,0.3785000145435333,1.7482523918151855,0.3866666555404663,0.07508829981088638,2.41624170797877,bfc8de485755d7f3,ddffbe335bef15ad,mps
+quantum,quantum_beta_0.6_1.0,53,30,80,2400,1.711459994316101,0.4320000112056732,1.684290885925293,0.4399999976158142,0.07037489116191864,2.4244089999701828,bfc8de485755d7f3,00868f788440b8a0,mps
+quantum,quantum_beta_0.5_1.2,51,30,80,2400,1.6640392541885376,0.4580000042915344,1.6313401460647583,0.43833333253860474,0.0688972994685173,2.379183917073533,bfc8de485755d7f3,82b3214d02e2e634,mps
+quantum,quantum_beta_0.5_1.2,52,30,80,2400,1.5431870222091675,0.49149999022483826,1.515285849571228,0.5133333206176758,0.06496633589267731,2.3730562501586974,bfc8de485755d7f3,ddffbe335bef15ad,mps
+quantum,quantum_beta_0.5_1.2,53,30,80,2400,1.7410157918930054,0.4059999883174896,1.7098183631896973,0.3916666805744171,0.07386184483766556,2.438973000040278,bfc8de485755d7f3,00868f788440b8a0,mps
+quantum,quantum_beta_0.4_0.9,51,30,80,2400,1.4330989122390747,0.512499988079071,1.369968295097351,0.5433333516120911,0.06111948564648628,2.5881015001796186,bfc8de485755d7f3,82b3214d02e2e634,mps
+quantum,quantum_beta_0.4_0.9,52,30,80,2400,1.4685417413711548,0.5249999761581421,1.3982360363006592,0.5716666579246521,0.06072157621383667,2.5672016669996083,bfc8de485755d7f3,ddffbe335bef15ad,mps
+quantum,quantum_beta_0.4_0.9,53,30,80,2400,1.3139362335205078,0.5789999961853027,1.244922161102295,0.5799999833106995,0.056043192744255066,2.4583064999897033,bfc8de485755d7f3,00868f788440b8a0,mps

benchmark_results/pso_v5_deep_methods.csv

@@ -0,0 +1,10 @@
+section,metric,value
+protocol,version,MNIST-PSO-RAW-V5 1.0.0
+pilot,selected_dimension,full
+single_model,test_accuracy,83.23
+single_model,test_nll,0.556332
+single_model,test_brier,0.253921
+single_model,test_ece,0.085242
+ensemble,test_accuracy,87.0
+ensemble,test_nll,0.535928
+ensemble,pairwise_disagreement,0.1585

benchmark_results/pso_v6_heavy_autoresearch.csv

@@ -0,0 +1,9 @@
+workload,split,seeds,accuracy_mean,accuracy_sd,nll_mean,nll_sd,accuracy_gain_vs_baseline_pp,relative_nll_reduction_vs_baseline_pct,state_ratio,ratio,projection_mode,projection_seed
+mnist_compact,development_seeds_101_103,101-103,49.053333,0.417652,1.525704,0.01494,-0.1,-0.501617,0.4918032786885246,0.5,fixed,1800044939
+mnist_compact,confirmation_seeds_111_113,111-113,48.466667,2.515002,1.550212,0.046101,-0.686666,-2.116016,0.4918032786885246,0.5,fixed,1800044939
+mnist_wide,development_seeds_101_103,101-103,48.346667,3.214052,1.677712,0.072138,4.486667,2.52995,0.5,0.5,explicit,592157828
+mnist_wide,confirmation_seeds_111_113,111-113,45.723333,1.05633,1.734905,0.034461,1.863333,-0.792792,0.5,0.5,explicit,592157828
+fashion_compact,development_seeds_101_103,101-103,50.796667,3.095228,1.385324,0.035646,3.793334,8.330571,0.4918032786885246,0.5,fixed,1363313651
+fashion_compact,confirmation_seeds_111_113,111-113,53.196667,2.025669,1.328168,0.078488,6.193334,12.112688,0.4918032786885246,0.5,fixed,1363313651
+fashion_wide,development_seeds_101_103,101-103,48.263333,3.71862,1.531421,0.069986,1.953333,-0.371883,0.5,0.5,fixed,189641451
+fashion_wide,confirmation_seeds_111_113,111-113,49.476667,4.250251,1.513876,0.067096,3.166667,0.778045,0.5,0.5,fixed,189641451

benchmark_results/pso_v6_heavy_tasks.csv

@@ -0,0 +1,164 @@
+section,workload,method,metric,value
+protocol,global,all,version,HEAVY-TASK-PSO-V6 1.0.0
+protocol,global,all,official_test_data_loaded,False
+protocol,global,all,official_test_evaluations,0
+baseline,mnist_compact,untrained,val_nll,2.325095
+baseline,mnist_compact,untrained,val_accuracy,6.65
+baseline,mnist_wide,untrained,val_nll,2.32458
+baseline,mnist_wide,untrained,val_accuracy,9.31
+baseline,fashion_compact,untrained,val_nll,2.320302
+baseline,fashion_compact,untrained,val_accuracy,8.08
+baseline,fashion_wide,untrained,val_nll,2.311222
+baseline,fashion_wide,untrained,val_accuracy,8.77
+screen,mnist_compact,G0,val_nll,2.019988
+screen,mnist_compact,G0,val_acc,33.32
+screen,mnist_compact,G0,gbest_loss,2.014459
+screen,mnist_compact,G0,gbest_acc,34.4
+screen,mnist_compact,G0,throughput_sps,1106883.43
+screen,mnist_compact,G5,val_nll,1.959743
+screen,mnist_compact,G5,val_acc,33.95
+screen,mnist_compact,G5,gbest_loss,1.971574
+screen,mnist_compact,G5,gbest_acc,32.35
+screen,mnist_compact,G5,throughput_sps,1139465.88
+screen,mnist_compact,G6,val_nll,1.957307
+screen,mnist_compact,G6,val_acc,35.11
+screen,mnist_compact,G6,gbest_loss,1.941859
+screen,mnist_compact,G6,gbest_acc,35.35
+screen,mnist_compact,G6,throughput_sps,1340969.41
+screen,mnist_compact,G8,val_nll,1.690442
+screen,mnist_compact,G8,val_acc,43.88
+screen,mnist_compact,G8,gbest_loss,1.675063
+screen,mnist_compact,G8,gbest_acc,43.6
+screen,mnist_compact,G8,throughput_sps,984615.38
+screen,mnist_wide,G0,val_nll,2.088156
+screen,mnist_wide,G0,val_acc,33.19
+screen,mnist_wide,G0,gbest_loss,2.087951
+screen,mnist_wide,G0,gbest_acc,32.45
+screen,mnist_wide,G0,throughput_sps,595459.62
+screen,mnist_wide,G5,val_nll,1.913566
+screen,mnist_wide,G5,val_acc,39.87
+screen,mnist_wide,G5,gbest_loss,1.924357
+screen,mnist_wide,G5,gbest_acc,39.5
+screen,mnist_wide,G5,throughput_sps,772511.47
+screen,mnist_wide,G6,val_nll,2.017511
+screen,mnist_wide,G6,val_acc,29.46
+screen,mnist_wide,G6,gbest_loss,2.014518
+screen,mnist_wide,G6,gbest_acc,27.8
+screen,mnist_wide,G6,throughput_sps,777013.35
+screen,mnist_wide,G8,val_nll,2.056341
+screen,mnist_wide,G8,val_acc,20.43
+screen,mnist_wide,G8,gbest_loss,2.062368
+screen,mnist_wide,G8,gbest_acc,20.3
+screen,mnist_wide,G8,throughput_sps,855005.34
+screen,fashion_compact,G0,val_nll,1.969548
+screen,fashion_compact,G0,val_acc,28.25
+screen,fashion_compact,G0,gbest_loss,1.96974
+screen,fashion_compact,G0,gbest_acc,28.15
+screen,fashion_compact,G0,throughput_sps,1219977.13
+screen,fashion_compact,G5,val_nll,1.892402
+screen,fashion_compact,G5,val_acc,34.31
+screen,fashion_compact,G5,gbest_loss,1.874629
+screen,fashion_compact,G5,gbest_acc,36.6
+screen,fashion_compact,G5,throughput_sps,1122281.97
+screen,fashion_compact,G6,val_nll,1.824273
+screen,fashion_compact,G6,val_acc,35.2
+screen,fashion_compact,G6,gbest_loss,1.844914
+screen,fashion_compact,G6,gbest_acc,33.5
+screen,fashion_compact,G6,throughput_sps,1152322.65
+screen,fashion_compact,G8,val_nll,1.88018
+screen,fashion_compact,G8,val_acc,30.85
+screen,fashion_compact,G8,gbest_loss,1.895269
+screen,fashion_compact,G8,gbest_acc,28.85
+screen,fashion_compact,G8,throughput_sps,1193139.45
+screen,fashion_wide,G0,val_nll,1.981316
+screen,fashion_wide,G0,val_acc,36.39
+screen,fashion_wide,G0,gbest_loss,1.98436
+screen,fashion_wide,G0,gbest_acc,36.05
+screen,fashion_wide,G0,throughput_sps,540327.57
+screen,fashion_wide,G5,val_nll,1.945862
+screen,fashion_wide,G5,val_acc,31.97
+screen,fashion_wide,G5,gbest_loss,1.938569
+screen,fashion_wide,G5,gbest_acc,33.3
+screen,fashion_wide,G5,throughput_sps,807401.18
+screen,fashion_wide,G6,val_nll,1.975825
+screen,fashion_wide,G6,val_acc,28.64
+screen,fashion_wide,G6,gbest_loss,1.960139
+screen,fashion_wide,G6,gbest_acc,28.1
+screen,fashion_wide,G6,throughput_sps,735350.44
+screen,fashion_wide,G8,val_nll,1.987923
+screen,fashion_wide,G8,val_acc,28.69
+screen,fashion_wide,G8,gbest_loss,1.976977
+screen,fashion_wide,G8,gbest_acc,28.45
+screen,fashion_wide,G8,throughput_sps,865800.87
+confirm,mnist_compact,G8,val_acc_mean,49.153333
+confirm,mnist_compact,G8,val_acc_std,1.320656
+confirm,mnist_compact,G8,val_nll_mean,1.518089
+confirm,mnist_compact,G8,val_nll_std,0.061523
+confirm,mnist_compact,G8,wall_time_sec_mean,4.530167
+confirm,mnist_compact,G6,val_acc_mean,45.966667
+confirm,mnist_compact,G6,val_acc_std,3.545438
+confirm,mnist_compact,G6,val_nll_mean,1.642081
+confirm,mnist_compact,G6,val_nll_std,0.091398
+confirm,mnist_compact,G6,wall_time_sec_mean,5.2706
+confirm,mnist_wide,G8,val_acc_mean,41.03
+confirm,mnist_wide,G8,val_acc_std,4.055083
+confirm,mnist_wide,G8,val_nll_mean,1.733351
+confirm,mnist_wide,G8,val_nll_std,0.084331
+confirm,mnist_wide,G8,wall_time_sec_mean,9.4569
+confirm,mnist_wide,G5,val_acc_mean,43.86
+confirm,mnist_wide,G5,val_acc_std,3.512222
+confirm,mnist_wide,G5,val_nll_mean,1.721259
+confirm,mnist_wide,G5,val_nll_std,0.098656
+confirm,mnist_wide,G5,wall_time_sec_mean,9.868467
+confirm,fashion_compact,G8,val_acc_mean,47.003333
+confirm,fashion_compact,G8,val_acc_std,5.2259
+confirm,fashion_compact,G8,val_nll_mean,1.511217
+confirm,fashion_compact,G8,val_nll_std,0.168628
+confirm,fashion_compact,G8,wall_time_sec_mean,5.0146
+confirm,fashion_compact,G6,val_acc_mean,45.616667
+confirm,fashion_compact,G6,val_acc_std,3.585392
+confirm,fashion_compact,G6,val_nll_mean,1.584443
+confirm,fashion_compact,G6,val_nll_std,0.053077
+confirm,fashion_compact,G6,wall_time_sec_mean,4.839267
+confirm,fashion_wide,G8,val_acc_mean,41.666667
+confirm,fashion_wide,G8,val_acc_std,2.269016
+confirm,fashion_wide,G8,val_nll_mean,1.615937
+confirm,fashion_wide,G8,val_nll_std,0.036692
+confirm,fashion_wide,G8,wall_time_sec_mean,9.499333
+confirm,fashion_wide,G5,val_acc_mean,46.31
+confirm,fashion_wide,G5,val_acc_std,7.568494
+confirm,fashion_wide,G5,val_nll_mean,1.525747
+confirm,fashion_wide,G5,val_nll_std,0.149706
+confirm,fashion_wide,G5,wall_time_sec_mean,10.038833
+feasibility,mnist_compact,G8,execution_feasible,True
+feasibility,mnist_compact,G8,optimization_feasible,True
+feasibility,mnist_compact,G8,mean_val_nll,1.518089
+feasibility,mnist_compact,G8,mean_val_acc,49.1533
+feasibility,mnist_compact,G6,execution_feasible,True
+feasibility,mnist_compact,G6,optimization_feasible,True
+feasibility,mnist_compact,G6,mean_val_nll,1.642081
+feasibility,mnist_compact,G6,mean_val_acc,45.9667
+feasibility,mnist_wide,G8,execution_feasible,True
+feasibility,mnist_wide,G8,optimization_feasible,True
+feasibility,mnist_wide,G8,mean_val_nll,1.733351
+feasibility,mnist_wide,G8,mean_val_acc,41.03
+feasibility,mnist_wide,G5,execution_feasible,True
+feasibility,mnist_wide,G5,optimization_feasible,True
+feasibility,mnist_wide,G5,mean_val_nll,1.721259
+feasibility,mnist_wide,G5,mean_val_acc,43.86
+feasibility,fashion_compact,G8,execution_feasible,True
+feasibility,fashion_compact,G8,optimization_feasible,True
+feasibility,fashion_compact,G8,mean_val_nll,1.511217
+feasibility,fashion_compact,G8,mean_val_acc,47.0033
+feasibility,fashion_compact,G6,execution_feasible,True
+feasibility,fashion_compact,G6,optimization_feasible,True
+feasibility,fashion_compact,G6,mean_val_nll,1.584443
+feasibility,fashion_compact,G6,mean_val_acc,45.6167
+feasibility,fashion_wide,G8,execution_feasible,True
+feasibility,fashion_wide,G8,optimization_feasible,True
+feasibility,fashion_wide,G8,mean_val_nll,1.615937
+feasibility,fashion_wide,G8,mean_val_acc,41.6667
+feasibility,fashion_wide,G5,execution_feasible,True
+feasibility,fashion_wide,G5,optimization_feasible,True
+feasibility,fashion_wide,G5,mean_val_nll,1.525747
+feasibility,fashion_wide,G5,mean_val_acc,46.31

benchmark_results/pso_v6_phase_b.csv

@@ -0,0 +1,21 @@
+protocol_version,MNIST-PSO-RAW-V6 1.0.0
+
+--- Phase B Screen Results ---
+config_id,description,val_nll,val_acc_%,queries,sample_evals,wall_time_sec
+G0,exact V5 control,1.183117,67.73,4800,9600000,8.4356
+G1,isolate anisotropic per-tensor scaling,1.437604,62.32,4800,9600000,6.6136
+G2,isolate nonzero launch velocity,1.296105,60.72,4800,9600000,6.6528
+G3,isolate mutation,1.130839,68.07,4800,9600000,6.4943
+G4,velocity x mutation interaction,1.11628,65.76,4800,9600000,6.4837
+G5,test sufficient bound expansion,0.90151,70.93,4800,9600000,6.76
+G6,test broader normalized initialization,0.96129,69.56,4800,9600000,6.7408
+G7,isolate antithetic position coupling against G2,1.291266,58.79,4800,9600000,6.4273
+G8,retained semantic control (public Optimizer),0.883557,70.99,4800,9600000,8.7632
+
+--- Phase B Confirmation Aggregates ---
+config_id,mean_val_acc_%,std_val_acc,mean_val_nll,std_val_nll,num_seeds
+G0,79.526667,0.576397,0.689687,0.008079,3
+G1,76.953333,2.490067,0.833759,0.123667,3
+G5,84.236667,0.903678,0.510297,0.024899,3
+G6,84.25,0.219317,0.503809,0.015789,3
+G8,84.916667,0.988804,0.48144,0.029447,3

benchmark_results/pso_v7_heavy_cross_split.csv

@@ -0,0 +1,73 @@
+variant_id,phase,split_seed,workload_id,baseline_method,baseline_acc,candidate_acc,acc_gain_pp,baseline_nll,candidate_nll,nll_reduction_fraction,state_ratio,score,pass,is_best_observed,candidate_path,evaluation_path
+iteration-0001-development,development,20260905,mnist_compact,G8,50.713333,50.723333,0.00999999999999801,1.534298,1.492805,0.0270436381980554,0.4918032786885246,-289.2921035434933,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0001-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0001-development.json
+iteration-0001-development,development,20260905,mnist_wide,G5,46.656667,46.04,-0.6166669999999996,1.677872,1.675644,0.0013278724479579603,0.5,-289.2921035434933,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0001-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0001-development.json
+iteration-0001-development,development,20260905,fashion_compact,G8,46.366667,49.576667,3.210000000000001,1.497844,1.400574,0.06494000710354347,0.4918032786885246,-289.2921035434933,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0001-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0001-development.json
+iteration-0001-development,development,20260905,fashion_wide,G5,48.516667,45.01,-3.506667,1.490304,1.584923,-0.06348973095422142,0.5,-289.2921035434933,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0001-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0001-development.json
+iteration-0001-development,development,20260906,mnist_compact,G8,46.81,51.663333,4.853332999999999,1.58986,1.436337,0.09656384838916639,0.4918032786885246,-289.2921035434933,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0001-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0001-development.json
+iteration-0001-development,development,20260906,mnist_wide,G5,44.936667,44.83,-0.10666700000000162,1.665007,1.749135,-0.050527114901018556,0.5,-289.2921035434933,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0001-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0001-development.json
+iteration-0001-development,development,20260906,fashion_compact,G8,44.193333,48.753333,4.559999999999995,1.553097,1.375751,0.11418861796784104,0.4918032786885246,-289.2921035434933,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0001-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0001-development.json
+iteration-0001-development,development,20260906,fashion_wide,G5,48.37,41.193333,-7.176666999999995,1.490657,1.707819,-0.14568207173078723,0.5,-289.2921035434933,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0001-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0001-development.json
+iteration-0002-development,development,20260905,mnist_compact,G8,50.713333,50.723333,0.00999999999999801,1.534298,1.492805,0.0270436381980554,0.4918032786885246,-186.3459040597986,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0002-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0002-development.json
+iteration-0002-development,development,20260905,mnist_wide,G5,46.656667,45.193333,-1.4633339999999961,1.677872,1.733563,-0.03319144726176963,0.5,-186.3459040597986,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0002-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0002-development.json
+iteration-0002-development,development,20260905,fashion_compact,G8,46.366667,49.576667,3.210000000000001,1.497844,1.400574,0.06494000710354347,0.4918032786885246,-186.3459040597986,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0002-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0002-development.json
+iteration-0002-development,development,20260905,fashion_wide,G5,48.516667,47.223333,-1.2933340000000015,1.490304,1.536216,-0.03080713733573818,0.5,-186.3459040597986,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0002-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0002-development.json
+iteration-0002-development,development,20260906,mnist_compact,G8,46.81,51.663333,4.853332999999999,1.58986,1.436337,0.09656384838916639,0.4918032786885246,-186.3459040597986,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0002-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0002-development.json
+iteration-0002-development,development,20260906,mnist_wide,G5,44.936667,45.09,0.1533330000000035,1.665007,1.734207,-0.04156138682900441,0.5,-186.3459040597986,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0002-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0002-development.json
+iteration-0002-development,development,20260906,fashion_compact,G8,44.193333,48.753333,4.559999999999995,1.553097,1.375751,0.11418861796784104,0.4918032786885246,-186.3459040597986,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0002-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0002-development.json
+iteration-0002-development,development,20260906,fashion_wide,G5,48.37,46.813333,-1.5566669999999974,1.490657,1.475127,0.010418224984016995,0.5,-186.3459040597986,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0002-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0002-development.json
+iteration-0003-replica1-development,development,20260905,mnist_compact,G8,50.713333,50.723333,0.00999999999999801,1.534298,1.492805,0.0270436381980554,0.4918032786885246,-287.7250973286179,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0003-replica1-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0003-replica1-development.json
+iteration-0003-replica1-development,development,20260905,mnist_wide,G5,46.656667,44.923333,-1.7333339999999993,1.677872,1.728906,-0.030415907768888223,0.5,-287.7250973286179,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0003-replica1-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0003-replica1-development.json
+iteration-0003-replica1-development,development,20260905,fashion_compact,G8,46.366667,49.576667,3.210000000000001,1.497844,1.400574,0.06494000710354347,0.4918032786885246,-287.7250973286179,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0003-replica1-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0003-replica1-development.json
+iteration-0003-replica1-development,development,20260905,fashion_wide,G5,48.516667,47.176667,-1.3399999999999963,1.490304,1.56267,-0.048557878124194744,0.5,-287.7250973286179,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0003-replica1-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0003-replica1-development.json
+iteration-0003-replica1-development,development,20260906,mnist_compact,G8,46.81,51.663333,4.853332999999999,1.58986,1.436337,0.09656384838916639,0.4918032786885246,-287.7250973286179,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0003-replica1-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0003-replica1-development.json
+iteration-0003-replica1-development,development,20260906,mnist_wide,G5,44.936667,42.113333,-2.8233340000000027,1.665007,1.713134,-0.028904983582651624,0.5,-287.7250973286179,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0003-replica1-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0003-replica1-development.json
+iteration-0003-replica1-development,development,20260906,fashion_compact,G8,44.193333,48.753333,4.559999999999995,1.553097,1.375751,0.11418861796784104,0.4918032786885246,-287.7250973286179,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0003-replica1-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0003-replica1-development.json
+iteration-0003-replica1-development,development,20260906,fashion_wide,G5,48.37,46.433333,-1.936667,1.490657,1.581386,-0.06086510847230454,0.5,-287.7250973286179,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0003-replica1-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0003-replica1-development.json
+iteration-0003-replica2-development,development,20260905,mnist_compact,G8,50.713333,50.723333,0.00999999999999801,1.534298,1.492805,0.0270436381980554,0.4918032786885246,-286.97652369317115,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0003-replica2-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0003-replica2-development.json
+iteration-0003-replica2-development,development,20260905,mnist_wide,G5,46.656667,46.01,-0.6466670000000008,1.677872,1.72888,-0.0304004119503752,0.5,-286.97652369317115,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0003-replica2-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0003-replica2-development.json
+iteration-0003-replica2-development,development,20260905,fashion_compact,G8,46.366667,49.576667,3.210000000000001,1.497844,1.400574,0.06494000710354347,0.4918032786885246,-286.97652369317115,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0003-replica2-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0003-replica2-development.json
+iteration-0003-replica2-development,development,20260905,fashion_wide,G5,48.516667,48.613333,0.09666599999999903,1.490304,1.48049,0.006585233616765431,0.5,-286.97652369317115,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0003-replica2-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0003-replica2-development.json
+iteration-0003-replica2-development,development,20260906,mnist_compact,G8,46.81,51.663333,4.853332999999999,1.58986,1.436337,0.09656384838916639,0.4918032786885246,-286.97652369317115,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0003-replica2-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0003-replica2-development.json
+iteration-0003-replica2-development,development,20260906,mnist_wide,G5,44.936667,41.556667,-3.3800000000000026,1.665007,1.801698,-0.08209635154687045,0.5,-286.97652369317115,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0003-replica2-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0003-replica2-development.json
+iteration-0003-replica2-development,development,20260906,fashion_compact,G8,44.193333,48.753333,4.559999999999995,1.553097,1.375751,0.11418861796784104,0.4918032786885246,-286.97652369317115,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0003-replica2-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0003-replica2-development.json
+iteration-0003-replica2-development,development,20260906,fashion_wide,G5,48.37,45.913333,-2.456666999999996,1.490657,1.516614,-0.017413127231817923,0.5,-286.97652369317115,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0003-replica2-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0003-replica2-development.json
+iteration-0004-development,development,20260905,mnist_compact,G8,50.713333,50.723333,0.00999999999999801,1.534298,1.492805,0.0270436381980554,0.4918032786885246,-390.14004501856266,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0004-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0004-development.json
+iteration-0004-development,development,20260905,mnist_wide,G5,46.656667,44.763333,-1.8933339999999959,1.677872,1.698999,-0.012591544527830428,0.5,-390.14004501856266,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0004-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0004-development.json
+iteration-0004-development,development,20260905,fashion_compact,G8,46.366667,49.576667,3.210000000000001,1.497844,1.400574,0.06494000710354347,0.4918032786885246,-390.14004501856266,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0004-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0004-development.json
+iteration-0004-development,development,20260905,fashion_wide,G5,48.516667,44.816667,-3.6999999999999957,1.490304,1.582588,-0.06192293652838617,0.5,-390.14004501856266,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0004-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0004-development.json
+iteration-0004-development,development,20260906,mnist_compact,G8,46.81,51.663333,4.853332999999999,1.58986,1.436337,0.09656384838916639,0.4918032786885246,-390.14004501856266,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0004-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0004-development.json
+iteration-0004-development,development,20260906,mnist_wide,G5,44.936667,38.946667,-5.990000000000002,1.665007,1.819662,-0.09288549537629572,0.5,-390.14004501856266,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0004-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0004-development.json
+iteration-0004-development,development,20260906,fashion_compact,G8,44.193333,48.753333,4.559999999999995,1.553097,1.375751,0.11418861796784104,0.4918032786885246,-390.14004501856266,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0004-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0004-development.json
+iteration-0004-development,development,20260906,fashion_wide,G5,48.37,43.9,-4.469999999999999,1.490657,1.658117,-0.11233972671110803,0.5,-390.14004501856266,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0004-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0004-development.json
+iteration-0005-development,development,20260905,mnist_compact,G8,50.713333,50.723333,0.00999999999999801,1.534298,1.492805,0.0270436381980554,0.4918032786885246,-185.61068572934795,False,True,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0005-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0005-development.json
+iteration-0005-development,development,20260905,mnist_wide,G5,46.656667,43.723333,-2.933334000000002,1.677872,1.675155,0.0016193130346057866,0.5,-185.61068572934795,False,True,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0005-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0005-development.json
+iteration-0005-development,development,20260905,fashion_compact,G8,46.366667,49.576667,3.210000000000001,1.497844,1.400574,0.06494000710354347,0.4918032786885246,-185.61068572934795,False,True,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0005-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0005-development.json
+iteration-0005-development,development,20260905,fashion_wide,G5,48.516667,45.45,-3.0666669999999954,1.490304,1.441667,0.03263562333591002,0.5,-185.61068572934795,False,True,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0005-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0005-development.json
+iteration-0005-development,development,20260906,mnist_compact,G8,46.81,51.663333,4.853332999999999,1.58986,1.436337,0.09656384838916639,0.4918032786885246,-185.61068572934795,False,True,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0005-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0005-development.json
+iteration-0005-development,development,20260906,mnist_wide,G5,44.936667,42.543333,-2.393334000000003,1.665007,1.740544,-0.04536737683385127,0.5,-185.61068572934795,False,True,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0005-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0005-development.json
+iteration-0005-development,development,20260906,fashion_compact,G8,44.193333,48.753333,4.559999999999995,1.553097,1.375751,0.11418861796784104,0.4918032786885246,-185.61068572934795,False,True,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0005-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0005-development.json
+iteration-0005-development,development,20260906,fashion_wide,G5,48.37,47.65,-0.7199999999999989,1.490657,1.454402,0.02432149045689245,0.5,-185.61068572934795,False,True,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0005-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0005-development.json
+iteration-0006-development,development,20260905,mnist_compact,G8,50.713333,50.723333,0.00999999999999801,1.534298,1.492805,0.0270436381980554,0.4918032786885246,-186.8633001166316,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0006-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0006-development.json
+iteration-0006-development,development,20260905,mnist_wide,G5,46.656667,42.006667,-4.649999999999999,1.677872,1.749895,-0.04292520525999596,0.5,-186.8633001166316,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0006-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0006-development.json
+iteration-0006-development,development,20260905,fashion_compact,G8,46.366667,49.576667,3.210000000000001,1.497844,1.400574,0.06494000710354347,0.4918032786885246,-186.8633001166316,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0006-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0006-development.json
+iteration-0006-development,development,20260905,fashion_wide,G5,48.516667,49.08,0.5633330000000001,1.490304,1.466855,0.015734373657991962,0.5,-186.8633001166316,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0006-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0006-development.json
+iteration-0006-development,development,20260906,mnist_compact,G8,46.81,51.663333,4.853332999999999,1.58986,1.436337,0.09656384838916639,0.4918032786885246,-186.8633001166316,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0006-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0006-development.json
+iteration-0006-development,development,20260906,mnist_wide,G5,44.936667,40.84,-4.0966669999999965,1.665007,1.73617,-0.04274036085133581,0.5,-186.8633001166316,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0006-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0006-development.json
+iteration-0006-development,development,20260906,fashion_compact,G8,44.193333,48.753333,4.559999999999995,1.553097,1.375751,0.11418861796784104,0.4918032786885246,-186.8633001166316,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0006-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0006-development.json
+iteration-0006-development,development,20260906,fashion_wide,G5,48.37,46.946667,-1.4233329999999995,1.490657,1.508747,-0.012135588535793386,0.5,-186.8633001166316,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0006-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0006-development.json
+iteration-0007-development,development,20260905,mnist_compact,G8,50.713333,50.723333,0.00999999999999801,1.534298,1.492805,0.0270436381980554,0.4918032786885246,-391.33742460463645,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0007-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0007-development.json
+iteration-0007-development,development,20260905,mnist_wide,G5,46.656667,41.05,-5.606667000000002,1.677872,1.794261,-0.06936703157332626,0.5,-391.33742460463645,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0007-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0007-development.json
+iteration-0007-development,development,20260905,fashion_compact,G8,46.366667,49.576667,3.210000000000001,1.497844,1.400574,0.06494000710354347,0.4918032786885246,-391.33742460463645,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0007-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0007-development.json
+iteration-0007-development,development,20260905,fashion_wide,G5,48.516667,42.513333,-6.003333999999995,1.490304,1.615206,-0.08380974619943303,0.5,-391.33742460463645,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0007-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0007-development.json
+iteration-0007-development,development,20260906,mnist_compact,G8,46.81,51.663333,4.853332999999999,1.58986,1.436337,0.09656384838916639,0.4918032786885246,-391.33742460463645,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0007-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0007-development.json
+iteration-0007-development,development,20260906,mnist_wide,G5,44.936667,40.666667,-4.270000000000003,1.665007,1.764515,-0.059764313303187405,0.5,-391.33742460463645,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0007-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0007-development.json
+iteration-0007-development,development,20260906,fashion_compact,G8,44.193333,48.753333,4.559999999999995,1.553097,1.375751,0.11418861796784104,0.4918032786885246,-391.33742460463645,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0007-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0007-development.json
+iteration-0007-development,development,20260906,fashion_wide,G5,48.37,39.916667,-8.453333,1.490657,1.609595,-0.07978897895357565,0.5,-391.33742460463645,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0007-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0007-development.json
+iteration-0008-development,development,20260905,mnist_compact,G8,50.713333,50.723333,0.00999999999999801,1.534298,1.492805,0.0270436381980554,0.4918032786885246,-390.14811518493707,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0008-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0008-development.json
+iteration-0008-development,development,20260905,mnist_wide,G5,46.656667,41.14,-5.516666999999998,1.677872,1.748292,-0.041969828449369154,0.5,-390.14811518493707,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0008-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0008-development.json
+iteration-0008-development,development,20260905,fashion_compact,G8,46.366667,49.576667,3.210000000000001,1.497844,1.400574,0.06494000710354347,0.4918032786885246,-390.14811518493707,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0008-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0008-development.json
+iteration-0008-development,development,20260905,fashion_wide,G5,48.516667,45.726667,-2.789999999999999,1.490304,1.558163,-0.04553366293051612,0.5,-390.14811518493707,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0008-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0008-development.json
+iteration-0008-development,development,20260906,mnist_compact,G8,46.81,51.663333,4.853332999999999,1.58986,1.436337,0.09656384838916639,0.4918032786885246,-390.14811518493707,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0008-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0008-development.json
+iteration-0008-development,development,20260906,mnist_wide,G5,44.936667,39.596667,-5.340000000000003,1.665007,1.763671,-0.059257408527411654,0.5,-390.14811518493707,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0008-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0008-development.json
+iteration-0008-development,development,20260906,fashion_compact,G8,44.193333,48.753333,4.559999999999995,1.553097,1.375751,0.11418861796784104,0.4918032786885246,-390.14811518493707,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0008-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0008-development.json
+iteration-0008-development,development,20260906,fashion_wide,G5,48.37,43.246667,-5.123332999999995,1.490657,1.649349,-0.10645775654627461,0.5,-390.14811518493707,False,False,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/candidates/iteration-0008-development.json,.omc/autoresearch/heavy-pso-cross-split-robustness/runs/20260903T162504Z/evaluations/iteration-0008-development.json

benchmark_results/pso_v8_post_training_ensemble.csv

@@ -0,0 +1,29 @@
+Workload,Phase,Method,Accuracy,NLL,Brier,ECE,Margin,WallTimeSeconds,ParameterMultiplier,InferenceMultiplier
+mnist,validation,reference_single_10e,98.2300,0.060105,0.027982,0.003458,0.969858,0.0000,1.0,1.0
+mnist,validation,best_single_10e,98.2900,0.054029,0.025526,0.002795,0.971419,0.0000,1.0,1.0
+mnist,validation,single_50e,98.5200,0.073988,0.024922,0.009463,0.989411,0.0000,1.0,1.0
+mnist,validation,uniform_ensemble,98.6300,0.046385,0.021527,0.006228,0.965339,0.0000,5.0,5.0
+mnist,validation,uniform_temperature,98.6300,0.045355,0.021099,0.002518,0.972273,0.0412,5.0,5.0
+mnist,validation,slsqp_weights,98.6000,0.045902,0.021307,0.005994,0.965581,0.0097,5.0,5.0
+mnist,validation,pso_weights,98.6100,0.045902,0.021307,0.005893,0.965582,1.8925,5.0,5.0
+mnist,official_test,reference_single_10e,98.4700,0.044991,0.022366,0.003516,0.970575,0.0000,1.0,1.0
+mnist,official_test,best_single_10e,98.5500,0.044944,0.022052,0.002515,0.974650,0.0000,1.0,1.0
+mnist,official_test,single_50e,98.6000,0.062102,0.022912,0.008283,0.988829,0.0000,1.0,1.0
+mnist,official_test,uniform_ensemble,98.8600,0.036184,0.018010,0.006311,0.968415,0.0000,5.0,5.0
+mnist,official_test,uniform_temperature,98.8600,0.034129,0.017658,0.003340,0.974869,0.0000,5.0,5.0
+mnist,official_test,slsqp_weights,98.8300,0.036179,0.017989,0.005818,0.969195,0.0000,5.0,5.0
+mnist,official_test,pso_weights,98.8300,0.036178,0.017989,0.005817,0.969196,0.0000,5.0,5.0
+fashion_mnist,validation,reference_single_10e,89.4700,0.303799,0.150671,0.012197,0.808017,0.0000,1.0,1.0
+fashion_mnist,validation,best_single_10e,89.4700,0.303799,0.150671,0.012197,0.808017,0.0000,1.0,1.0
+fashion_mnist,validation,single_50e,90.3200,0.289660,0.142115,0.024556,0.869484,0.0000,1.0,1.0
+fashion_mnist,validation,uniform_ensemble,90.2800,0.286751,0.142881,0.023702,0.795588,0.0000,5.0,5.0
+fashion_mnist,validation,uniform_temperature,90.2800,0.285048,0.141557,0.011886,0.814383,0.0414,5.0,5.0
+fashion_mnist,validation,slsqp_weights,90.4200,0.285338,0.142289,0.023705,0.798413,0.0104,5.0,5.0
+fashion_mnist,validation,pso_weights,90.4200,0.285338,0.142289,0.023661,0.798397,1.8963,5.0,5.0
+fashion_mnist,official_test,reference_single_10e,88.9000,0.314516,0.161465,0.005320,0.805307,0.0000,1.0,1.0
+fashion_mnist,official_test,best_single_10e,88.9000,0.314516,0.161465,0.005320,0.805307,0.0000,1.0,1.0
+fashion_mnist,official_test,single_50e,89.9300,0.302348,0.148289,0.026401,0.865728,0.0000,1.0,1.0
+fashion_mnist,official_test,uniform_ensemble,89.6500,0.293522,0.151400,0.019111,0.791698,0.0000,5.0,5.0
+fashion_mnist,official_test,uniform_temperature,89.6500,0.291996,0.150581,0.007597,0.810594,0.0000,5.0,5.0
+fashion_mnist,official_test,slsqp_weights,89.5400,0.291696,0.150558,0.017186,0.794293,0.0000,5.0,5.0
+fashion_mnist,official_test,pso_weights,89.5400,0.291700,0.150559,0.017378,0.794274,0.0000,5.0,5.0

benchmark_results/pso_v8_post_training_ensemble.json

@@ -0,0 +1,683 @@
+{
+  "protocol_version": "POST-TRAINING-PSO-ENSEMBLE 1.1.0",
+  "config": {
+    "iteration": 1,
+    "archived_iteration0_reference": {
+      "epochs": 50,
+      "queries_per_seed": 1500,
+      "sample_evaluations_per_seed": 15000000,
+      "reason": "wall_time_ratio_gate_exceeded"
+    },
+    "datasets": [
+      "mnist",
+      "fashion_mnist"
+    ],
+    "split_seed": 20260904,
+    "search_samples": 50000,
+    "validation_samples": 10000,
+    "pool_seeds": [
+      201,
+      202,
+      203,
+      204,
+      205
+    ],
+    "reference_single_seed": 201,
+    "equal_budget_single_epochs": 50,
+    "adam_lr": 0.001,
+    "adam_batch_size": 256,
+    "pso": {
+      "method": "constriction",
+      "evaluation": "full",
+      "renewal": "loss",
+      "particles": 30,
+      "epochs": 30,
+      "swarm_seeds": [
+        301,
+        302,
+        303
+      ],
+      "queries_per_seed": 900,
+      "sample_evaluations_per_seed": 9000000,
+      "particle_bounds": [
+        -4.0,
+        4.0
+      ],
+      "boundary_strategy": "reflect",
+      "velocity_limit_ratio": 0.1,
+      "initial_position_noise": 0.0
+    },
+    "device": "mps"
+  },
+  "workloads": {
+    "mnist": {
+      "provenance": {
+        "dataset_name": "MNIST",
+        "split_seed": 20260904,
+        "search_samples": 50000,
+        "validation_samples": 10000,
+        "normalization": {
+          "mean": 0.1307128667831421,
+          "std": 0.3081730008125305
+        },
+        "data_fingerprint": "861b07884cffefb9",
+        "split_fingerprint": "258b42dde7da6324"
+      },
+      "training": {
+        "architecture": "CompactCNN",
+        "parameters": 9098,
+        "pool_seeds": [
+          201,
+          202,
+          203,
+          204,
+          205
+        ],
+        "pool_epochs_each": 10,
+        "adam_pool_epochs": 50,
+        "adam_lr": 0.001,
+        "adam_batch_size": 256,
+        "adam_pool_wall_time_seconds": 22.098886500985827,
+        "single_50e_epochs": 50,
+        "single_50e_wall_time_seconds": 22.06671220799035,
+        "model_fingerprints": {
+          "201": "046605462c4fc294",
+          "single_50e": "53ff4c61db919c1e",
+          "202": "cd3150799e0db43e",
+          "203": "1feb0f4dd7630ec2",
+          "204": "9dc0606cd8076cb2",
+          "205": "5bc16dc8f86cc869"
+        }
+      },
+      "validation_cache": {
+        "valid": true,
+        "pool_forward_passes": 5,
+        "long_single_forward_passes": 1,
+        "base_cnn_forward_passes_during_optimization": 0,
+        "shape": [
+          5,
+          10000,
+          10
+        ],
+        "memory_bytes": 2400000,
+        "wall_time_seconds": 0.09316845799912699
+      },
+      "validation": {
+        "methods": {
+          "reference_single_10e": {
+            "accuracy": 98.23,
+            "nll": 0.060105,
+            "brier": 0.027982,
+            "ece": 0.003458,
+            "margin": 0.969858
+          },
+          "best_single_10e": {
+            "accuracy": 98.29,
+            "nll": 0.054029,
+            "brier": 0.025526,
+            "ece": 0.002795,
+            "margin": 0.971419,
+            "selected_seed": 205
+          },
+          "single_50e": {
+            "accuracy": 98.52,
+            "nll": 0.073988,
+            "brier": 0.024922,
+            "ece": 0.009463,
+            "margin": 0.989411
+          },
+          "uniform_ensemble": {
+            "accuracy": 98.63,
+            "nll": 0.046385,
+            "brier": 0.021527,
+            "ece": 0.006228,
+            "margin": 0.965339
+          },
+          "uniform_temperature": {
+            "fitted_temperature": 0.854572,
+            "wall_time_seconds": 0.04119049999280833,
+            "evaluations": 23,
+            "metrics": {
+              "accuracy": 98.63,
+              "nll": 0.045355,
+              "brier": 0.021099,
+              "ece": 0.002518,
+              "margin": 0.972273
+            }
+          },
+          "slsqp_weights": {
+            "weights": [
+              0.05767137130295378,
+              0.2585033373652759,
+              0.10841769748910005,
+              0.25254869759276993,
+              0.3228588962499004
+            ],
+            "evaluations": 23,
+            "wall_time_seconds": 0.009686834004241973,
+            "success": true,
+            "message": "Optimization terminated successfully",
+            "metrics": {
+              "accuracy": 98.6,
+              "nll": 0.045902,
+              "brier": 0.021307,
+              "ece": 0.005994,
+              "margin": 0.965581
+            }
+          },
+          "pso_weights": {
+            "per_seed_runs": [
+              {
+                "seed": 301,
+                "queries": 900,
+                "sample_evaluations": 9000000,
+                "wall_time_seconds": 1.8924723340023775,
+                "metrics": {
+                  "accuracy": 98.61,
+                  "nll": 0.045902,
+                  "brier": 0.021307,
+                  "ece": 0.005893,
+                  "margin": 0.965582
+                },
+                "weights": [
+                  0.057233214378356934,
+                  0.2590548098087311,
+                  0.10862385481595993,
+                  0.25228697061538696,
+                  0.3228012025356293
+                ]
+              },
+              {
+                "seed": 302,
+                "queries": 900,
+                "sample_evaluations": 9000000,
+                "wall_time_seconds": 1.7105932499980554,
+                "metrics": {
+                  "accuracy": 98.6,
+                  "nll": 0.045902,
+                  "brier": 0.021306,
+                  "ece": 0.006183,
+                  "margin": 0.965578
+                },
+                "weights": [
+                  0.058317527174949646,
+                  0.25828817486763,
+                  0.10924994200468063,
+                  0.252756267786026,
+                  0.3213881254196167
+                ]
+              },
+              {
+                "seed": 303,
+                "queries": 900,
+                "sample_evaluations": 9000000,
+                "wall_time_seconds": 1.8959678749961313,
+                "metrics": {
+                  "accuracy": 98.61,
+                  "nll": 0.045902,
+                  "brier": 0.021307,
+                  "ece": 0.005833,
+                  "margin": 0.965583
+                },
+                "weights": [
+                  0.056311722844839096,
+                  0.2563577890396118,
+                  0.11076968908309937,
+                  0.25350210070610046,
+                  0.3230587840080261
+                ]
+              }
+            ],
+            "selected_seed": 301,
+            "selected_weights": [
+              0.057233214378356934,
+              0.2590548098087311,
+              0.10862385481595993,
+              0.25228697061538696,
+              0.3228012025356293
+            ],
+            "metrics": {
+              "accuracy": 98.61,
+              "nll": 0.045902,
+              "brier": 0.021307,
+              "ece": 0.005893,
+              "margin": 0.965582
+            },
+            "queries_per_seed": 900,
+            "sample_evaluations_per_seed": 9000000,
+            "total_queries": 2700,
+            "total_sample_evaluations": 27000000,
+            "median_one_seed_wall_time_seconds": 1.8924723340023775,
+            "total_wall_time_seconds": 5.499033458996564
+          }
+        }
+      },
+      "official_test_data_loaded_before_freeze": false,
+      "official_test_evaluations_before_freeze": 0,
+      "confirmation": {
+        "official_test_data_loaded": true,
+        "test_cache_counts": {
+          "dataset_loads": 1,
+          "pool_forward_passes": 5,
+          "long_single_forward_passes": 1,
+          "base_cnn_forward_passes_during_optimization": 0,
+          "memory_bytes": 2400000,
+          "wall_time_seconds": 0.11975424998672679
+        },
+        "frozen_methods": {
+          "selected_pso_seed": 301,
+          "selected_pso_weights": [
+            0.057233214378356934,
+            0.2590548098087311,
+            0.10862385481595993,
+            0.25228697061538696,
+            0.3228012025356293
+          ],
+          "slsqp_weights": [
+            0.05767137130295378,
+            0.2585033373652759,
+            0.10841769748910005,
+            0.25254869759276993,
+            0.3228588962499004
+          ],
+          "fitted_temperature": 0.854572
+        },
+        "methods": {
+          "reference_single_10e": {
+            "accuracy": 98.47,
+            "nll": 0.044991,
+            "brier": 0.022366,
+            "ece": 0.003516,
+            "margin": 0.970575
+          },
+          "best_single_10e": {
+            "accuracy": 98.55,
+            "nll": 0.044944,
+            "brier": 0.022052,
+            "ece": 0.002515,
+            "margin": 0.97465
+          },
+          "single_50e": {
+            "accuracy": 98.6,
+            "nll": 0.062102,
+            "brier": 0.022912,
+            "ece": 0.008283,
+            "margin": 0.988829
+          },
+          "uniform_ensemble": {
+            "accuracy": 98.86,
+            "nll": 0.036184,
+            "brier": 0.01801,
+            "ece": 0.006311,
+            "margin": 0.968415
+          },
+          "uniform_temperature": {
+            "accuracy": 98.86,
+            "nll": 0.034129,
+            "brier": 0.017658,
+            "ece": 0.00334,
+            "margin": 0.974869
+          },
+          "slsqp_weights": {
+            "accuracy": 98.83,
+            "nll": 0.036179,
+            "brier": 0.017989,
+            "ece": 0.005818,
+            "margin": 0.969195
+          },
+          "pso_weights": {
+            "accuracy": 98.83,
+            "nll": 0.036178,
+            "brier": 0.017989,
+            "ece": 0.005817,
+            "margin": 0.969196
+          }
+        },
+        "confirmation_gates": {
+          "all_values_finite": true,
+          "official_test_dataset_loads": 1,
+          "official_test_pool_forward_passes": 5,
+          "official_test_long_single_forward_passes": 1,
+          "maximum_pso_accuracy_regression_vs_uniform_pp": true,
+          "pso_nll_below_reference_single": true,
+          "maximum_pso_nll_regression_vs_equal_budget_single": true,
+          "pass": true
+        }
+      }
+    },
+    "fashion_mnist": {
+      "provenance": {
+        "dataset_name": "FashionMNIST",
+        "split_seed": 20260904,
+        "search_samples": 50000,
+        "validation_samples": 10000,
+        "normalization": {
+          "mean": 0.28573352098464966,
+          "std": 0.35278451442718506
+        },
+        "data_fingerprint": "b56a39e2abf7d9ad",
+        "split_fingerprint": "489c1a04fb44b59a"
+      },
+      "training": {
+        "architecture": "CompactCNN",
+        "parameters": 9098,
+        "pool_seeds": [
+          201,
+          202,
+          203,
+          204,
+          205
+        ],
+        "pool_epochs_each": 10,
+        "adam_pool_epochs": 50,
+        "adam_lr": 0.001,
+        "adam_batch_size": 256,
+        "adam_pool_wall_time_seconds": 22.363562541009742,
+        "single_50e_epochs": 50,
+        "single_50e_wall_time_seconds": 22.406685083013144,
+        "model_fingerprints": {
+          "201": "8189d6d2cce8bde8",
+          "single_50e": "5a5b8444219bf301",
+          "202": "a5492b64bfda73cd",
+          "203": "664e02f11e62031f",
+          "204": "be2a23fca4153703",
+          "205": "d3a37cfa6ee482bf"
+        }
+      },
+      "validation_cache": {
+        "valid": true,
+        "pool_forward_passes": 5,
+        "long_single_forward_passes": 1,
+        "base_cnn_forward_passes_during_optimization": 0,
+        "shape": [
+          5,
+          10000,
+          10
+        ],
+        "memory_bytes": 2400000,
+        "wall_time_seconds": 0.08030625000537839
+      },
+      "validation": {
+        "methods": {
+          "reference_single_10e": {
+            "accuracy": 89.47,
+            "nll": 0.303799,
+            "brier": 0.150671,
+            "ece": 0.012197,
+            "margin": 0.808017
+          },
+          "best_single_10e": {
+            "accuracy": 89.47,
+            "nll": 0.303799,
+            "brier": 0.150671,
+            "ece": 0.012197,
+            "margin": 0.808017,
+            "selected_seed": 201
+          },
+          "single_50e": {
+            "accuracy": 90.32,
+            "nll": 0.28966,
+            "brier": 0.142115,
+            "ece": 0.024556,
+            "margin": 0.869484
+          },
+          "uniform_ensemble": {
+            "accuracy": 90.28,
+            "nll": 0.286751,
+            "brier": 0.142881,
+            "ece": 0.023702,
+            "margin": 0.795588
+          },
+          "uniform_temperature": {
+            "fitted_temperature": 0.906309,
+            "wall_time_seconds": 0.04135037500236649,
+            "evaluations": 21,
+            "metrics": {
+              "accuracy": 90.28,
+              "nll": 0.285048,
+              "brier": 0.141557,
+              "ece": 0.011886,
+              "margin": 0.814383
+            }
+          },
+          "slsqp_weights": {
+            "weights": [
+              0.23910131729416598,
+              0.2493341982620314,
+              0.18054455195387528,
+              0.021147090032828317,
+              0.30987284245709906
+            ],
+            "evaluations": 23,
+            "wall_time_seconds": 0.010389625007519498,
+            "success": true,
+            "message": "Optimization terminated successfully",
+            "metrics": {
+              "accuracy": 90.42,
+              "nll": 0.285338,
+              "brier": 0.142289,
+              "ece": 0.023705,
+              "margin": 0.798413
+            }
+          },
+          "pso_weights": {
+            "per_seed_runs": [
+              {
+                "seed": 301,
+                "queries": 900,
+                "sample_evaluations": 9000000,
+                "wall_time_seconds": 1.8963403329980792,
+                "metrics": {
+                  "accuracy": 90.42,
+                  "nll": 0.285338,
+                  "brier": 0.142289,
+                  "ece": 0.023661,
+                  "margin": 0.798397
+                },
+                "weights": [
+                  0.2385721057653427,
+                  0.24956248700618744,
+                  0.1805897355079651,
+                  0.021864112466573715,
+                  0.30941155552864075
+                ]
+              },
+              {
+                "seed": 302,
+                "queries": 900,
+                "sample_evaluations": 9000000,
+                "wall_time_seconds": 1.7477522080007475,
+                "metrics": {
+                  "accuracy": 90.43,
+                  "nll": 0.285338,
+                  "brier": 0.142286,
+                  "ece": 0.023912,
+                  "margin": 0.79841
+                },
+                "weights": [
+                  0.2415684461593628,
+                  0.2483692765235901,
+                  0.18002673983573914,
+                  0.02121216244995594,
+                  0.30882343649864197
+                ]
+              },
+              {
+                "seed": 303,
+                "queries": 900,
+                "sample_evaluations": 9000000,
+                "wall_time_seconds": 2.0081020840007113,
+                "metrics": {
+                  "accuracy": 90.41,
+                  "nll": 0.285338,
+                  "brier": 0.14229,
+                  "ece": 0.023601,
+                  "margin": 0.798444
+                },
+                "weights": [
+                  0.23893120884895325,
+                  0.24851834774017334,
+                  0.18254171311855316,
+                  0.019977180287241936,
+                  0.31003159284591675
+                ]
+              }
+            ],
+            "selected_seed": 301,
+            "selected_weights": [
+              0.2385721057653427,
+              0.24956248700618744,
+              0.1805897355079651,
+              0.021864112466573715,
+              0.30941155552864075
+            ],
+            "metrics": {
+              "accuracy": 90.42,
+              "nll": 0.285338,
+              "brier": 0.142289,
+              "ece": 0.023661,
+              "margin": 0.798397
+            },
+            "queries_per_seed": 900,
+            "sample_evaluations_per_seed": 9000000,
+            "total_queries": 2700,
+            "total_sample_evaluations": 27000000,
+            "median_one_seed_wall_time_seconds": 1.8963403329980792,
+            "total_wall_time_seconds": 5.652194624999538
+          }
+        }
+      },
+      "official_test_data_loaded_before_freeze": false,
+      "official_test_evaluations_before_freeze": 0,
+      "confirmation": {
+        "official_test_data_loaded": true,
+        "test_cache_counts": {
+          "dataset_loads": 1,
+          "pool_forward_passes": 5,
+          "long_single_forward_passes": 1,
+          "base_cnn_forward_passes_during_optimization": 0,
+          "memory_bytes": 2400000,
+          "wall_time_seconds": 0.11122124998655636
+        },
+        "frozen_methods": {
+          "selected_pso_seed": 301,
+          "selected_pso_weights": [
+            0.2385721057653427,
+            0.24956248700618744,
+            0.1805897355079651,
+            0.021864112466573715,
+            0.30941155552864075
+          ],
+          "slsqp_weights": [
+            0.23910131729416598,
+            0.2493341982620314,
+            0.18054455195387528,
+            0.021147090032828317,
+            0.30987284245709906
+          ],
+          "fitted_temperature": 0.906309
+        },
+        "methods": {
+          "reference_single_10e": {
+            "accuracy": 88.9,
+            "nll": 0.314516,
+            "brier": 0.161465,
+            "ece": 0.00532,
+            "margin": 0.805307
+          },
+          "best_single_10e": {
+            "accuracy": 88.9,
+            "nll": 0.314516,
+            "brier": 0.161465,
+            "ece": 0.00532,
+            "margin": 0.805307
+          },
+          "single_50e": {
+            "accuracy": 89.93,
+            "nll": 0.302348,
+            "brier": 0.148289,
+            "ece": 0.026401,
+            "margin": 0.865728
+          },
+          "uniform_ensemble": {
+            "accuracy": 89.65,
+            "nll": 0.293522,
+            "brier": 0.1514,
+            "ece": 0.019111,
+            "margin": 0.791698
+          },
+          "uniform_temperature": {
+            "accuracy": 89.65,
+            "nll": 0.291996,
+            "brier": 0.150581,
+            "ece": 0.007597,
+            "margin": 0.810594
+          },
+          "slsqp_weights": {
+            "accuracy": 89.54,
+            "nll": 0.291696,
+            "brier": 0.150558,
+            "ece": 0.017186,
+            "margin": 0.794293
+          },
+          "pso_weights": {
+            "accuracy": 89.54,
+            "nll": 0.2917,
+            "brier": 0.150559,
+            "ece": 0.017378,
+            "margin": 0.794274
+          }
+        },
+        "confirmation_gates": {
+          "all_values_finite": true,
+          "official_test_dataset_loads": 1,
+          "official_test_pool_forward_passes": 5,
+          "official_test_long_single_forward_passes": 1,
+          "maximum_pso_accuracy_regression_vs_uniform_pp": true,
+          "pso_nll_below_reference_single": true,
+          "maximum_pso_nll_regression_vs_equal_budget_single": true,
+          "pass": true
+        }
+      }
+    }
+  },
+  "development_pass": true,
+  "development_gates": {
+    "pass": true,
+    "failed_hard_gate_count": 0,
+    "gate_results": {
+      "all_values_finite": true,
+      "validation_pool_forward_passes_exact": true,
+      "optimization_base_model_forward_passes": true,
+      "official_test_data_loaded_before_freeze": true,
+      "slsqp_solver_success": true,
+      "query_and_sample_accounting_exact": true,
+      "maximum_pso_nll_regression_vs_uniform": true,
+      "maximum_pso_accuracy_regression_vs_uniform_pp": true,
+      "pso_nll_below_reference_single": true,
+      "maximum_pso_nll_regression_vs_equal_budget_single": true,
+      "maximum_relative_pso_nll_gap_vs_slsqp": true,
+      "cross_dataset_mean_relative_pso_nll_reduction_vs_uniform_minimum": true,
+      "maximum_median_one_seed_pso_to_pool_training_wall_ratio": true
+    },
+    "issues": []
+  },
+  "policy_frozen": true,
+  "official_test_data_loaded": true,
+  "official_test_evaluations_before_freeze": 0,
+  "post_test_tuning_or_reruns": 0,
+  "resource_totals": {
+    "adam_pool_epochs": 100,
+    "adam_pool_wall_time_seconds": 44.46244904199557,
+    "single_50e_wall_time_seconds": 44.473397291003494,
+    "validation_cache_forward_passes": 12,
+    "pso_total_queries": 5400,
+    "pso_total_sample_evaluations": 54000000,
+    "pso_research_wall_time_seconds": 11.151228083996102,
+    "pso_production_wall_time_seconds": 3.7888126670004567,
+    "pso_to_pool_wall_ratio": 0.0852162664640409,
+    "slsqp_total_evaluations": 46,
+    "slsqp_total_wall_time_seconds": 0.02007645901176147,
+    "official_test_cache_forward_passes": 12
+  }
+}

benchmark_results/pso_v8_post_training_ensemble_evaluation.json

@@ -0,0 +1,53 @@
+{
+  "evaluator_version": "POST-TRAINING-PSO-ENSEMBLE-EVALUATOR 1.2.0",
+  "pass": true,
+  "score": 22.55299842776338,
+  "development_pass": true,
+  "confirmation_pass": true,
+  "failed_hard_gate_count": 0,
+  "issues": {
+    "schema": [],
+    "config": [],
+    "finite": [],
+    "weights": [],
+    "accounting": [],
+    "leakage": [],
+    "tuning": [],
+    "slsqp": [],
+    "consistency": [],
+    "gates": []
+  },
+  "development_gates": {
+    "all_values_finite": true,
+    "simplex_tolerance": true,
+    "validation_pool_forward_passes_each_dataset": true,
+    "optimization_base_model_forward_passes": true,
+    "official_test_data_loaded_before_freeze": true,
+    "official_test_evaluations_before_freeze": true,
+    "query_and_sample_accounting_exact": true,
+    "maximum_pso_nll_regression_vs_uniform": true,
+    "maximum_pso_accuracy_regression_vs_uniform_pp": true,
+    "pso_nll_below_reference_single": true,
+    "maximum_pso_nll_regression_vs_equal_budget_single": true,
+    "maximum_relative_pso_nll_gap_vs_slsqp": true,
+    "cross_dataset_mean_relative_pso_nll_reduction_vs_uniform_minimum": true,
+    "maximum_median_one_seed_pso_to_pool_training_wall_ratio": true
+  },
+  "confirmation_gates": {
+    "all_values_finite": true,
+    "official_test_dataset_loads_each_dataset": true,
+    "official_test_pool_forward_passes_each_dataset": true,
+    "official_test_long_single_forward_passes_each_dataset": true,
+    "frozen_policy_consistency": true,
+    "maximum_pso_accuracy_regression_vs_uniform_pp": true,
+    "pso_nll_below_reference_single": true,
+    "maximum_pso_nll_regression_vs_equal_budget_single": true,
+    "post_test_tuning_or_reruns": true
+  },
+  "metrics": {
+    "mean_val_relative_nll_reduction_vs_equal_budget_single": 0.197261519715641,
+    "mean_val_accuracy_gain_vs_equal_budget_single_pp": 0.09500000000000597,
+    "mean_test_relative_nll_reduction_vs_equal_budget_single": 0.2263299842776338,
+    "mean_test_accuracy_gain_vs_equal_budget_single_pp": -0.0799999999999983
+  }
+}