1
|
{"args": {"mode": "sdil", "dataset": "mnist", "depth": 3, "width": 256, "act": "tanh", "residual": 1, "epochs": 15, "batch_size": 128, "train_examples": 0, "val_examples": 5000, "split_seed": 2027, "eval_split": "validation", "task_seed": 0, "task_train_examples": 50000, "task_test_examples": 10000, "task_levels": 8, "task_n_in": 128, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.05, "eta_A": 0.02, "eta_P": 0.002, "momentum": 0.9, "w_scale": 1.0, "a_scale": 1.0, "feedback_scale": 1.0, "pert_sigma": 0.01, "pert_every": 4, "pert_ndirs": 16, "pert_mode": "simultaneous", "use_residual": 1, "raw_scale_control": "none", "learn_A": 1, "learn_P": 1, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.2, "traffic_mode": "topdown", "predictor_mode": "diagonal", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 0, "device": "cuda", "log_every": 200, "diagnostics": "alignment", "eval_every": 1, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "traffic_time_dev_v1_mnist_topdown_rho0p2_ep0p002_wu200_d3_s0", "n_in": 784, "n_out": 10}, "split": {"dataset": "mnist", "split_seed": 2027, "validation_examples": 5000, "validation_index_sha256": "c4cec216363dc171cddb643b06414235ea84da4e5560059f6e10451287ef2e3d", "validation_class_counts": {"0": 500, "1": 500, "2": 500, "3": 500, "4": 500, "5": 500, "6": 500, "7": 500, "8": 500, "9": 500}, "train_examples": 55000, "test_examples": 10000, "split_from_training_only": true, "evaluation_split": "validation"}, "provenance": {"git_commit": "17bcd975b18d826010e2e4b83f009cbc3fb11e72", "git_dirty": false}, "steps": [{"step": 0, "epoch": 0, "train_loss": 2.6220171451568604, "cos_r_negg": [-0.00404270738363266, 0.027453625574707985, 0.034645840525627136], "cos_innovation_negg": [-0.00404270738363266, 0.027453625574707985, 0.034645840525627136], "cos_apical_negg": [-0.008226173929870129, 0.02111753262579441, 0.02196228876709938], "cos_Ac_negg": [0.0006024977192282677, 0.026888281106948853, 0.03464633971452713], "r_norm": [5.039546012878418, 4.895608425140381, 4.822951316833496], "traffic_norm": [2.6605329513549805, 2.6591362953186035, 2.5999975204467773], "traffic_residual_norm": [1.253713846206665, 0.916963517665863, 0.000413300673244521], "traffic_r2": [0.7782078385353088, 0.8811529278755188, 1.0]}, {"step": 200, "epoch": 0, "train_loss": 0.9323784112930298, "cos_r_negg": [0.15900751948356628, 0.06436875462532043, 0.1054995134472847], "cos_innovation_negg": [0.15900751948356628, 0.06436875462532043, 0.1054995134472847], "cos_apical_negg": [0.0897379070520401, 0.22078362107276917, 0.022702867165207863], "cos_Ac_negg": [0.48548609018325806, 0.2826274633407593, 0.09806336462497711], "r_norm": [3.2938714027404785, 2.1983866691589355, 1.2354724407196045], "traffic_norm": [6.447277069091797, 6.227522850036621, 6.333403587341309], "traffic_residual_norm": [2.617978572845459, 1.3813244104385376, 0.00033421983243897557], "traffic_r2": [0.835115373134613, 0.9507999420166016, 1.0]}, {"step": 400, "epoch": 0, "train_loss": 1.0052947998046875, "cos_r_negg": [0.2408212423324585, 0.07033445686101913, 0.08413510769605637], "cos_innovation_negg": [0.2408212423324585, 0.07033445686101913, 0.08413510769605637], "cos_apical_negg": [0.19003161787986755, 0.06510424613952637, 0.13223904371261597], "cos_Ac_negg": [0.7699878215789795, 0.0063654398545622826, 0.05393844097852707], "r_norm": [2.861599922180176, 1.991804838180542, 1.3370634317398071], "traffic_norm": [6.567568778991699, 6.375127792358398, 6.4317216873168945], "traffic_residual_norm": [1.896318793296814, 0.9292928576469421, 0.00022234722564462572], "traffic_r2": [0.9166294932365417, 0.9787514805793762, 1.0]}, {"epoch_end": 0, "step": 430, "eval_split": "validation", "eval_acc": 0.7962, "eval_loss": 0.8800016479492188, "val_acc": 0.7962, "val_loss": 0.8800016479492188}, {"step": 600, "epoch": 1, "train_loss": 0.6466549634933472, "cos_r_negg": [0.07757936418056488, -0.5330348014831543, 0.25673621892929077], "cos_innovation_negg": [0.07757936418056488, -0.5330348014831543, 0.25673621892929077], "cos_apical_negg": [0.023835286498069763, -0.2580062448978424, 0.13618531823158264], "cos_Ac_negg": [0.7958760261535645, -0.6512874364852905, 0.24777838587760925], "r_norm": [2.1829214096069336, 1.4850492477416992, 1.0576488971710205], "traffic_norm": [6.584615707397461, 6.437387466430664, 6.464559078216553], "traffic_residual_norm": [1.334553599357605, 0.61241614818573, 0.0001487152185291052], "traffic_r2": [0.9589218497276306, 0.9909491539001465, 1.0]}, {"step": 800, "epoch": 1, "train_loss": 1.2018479108810425, "cos_r_negg": [0.10527151077985764, 0.08393409103155136, 0.2706998884677887], "cos_innovation_negg": [0.10527151077985764, 0.08393409103155136, 0.2706998884677887], "cos_apical_negg": [-0.07224473357200623, 0.09687083959579468, 0.060407862067222595], "cos_Ac_negg": [0.8463864326477051, 0.2784714698791504, 0.28527072072029114], "r_norm": [1.7230737209320068, 1.1727310419082642, 0.8497105836868286], "traffic_norm": [6.639394760131836, 6.466908931732178, 6.477208137512207], "traffic_residual_norm": [0.9702131152153015, 0.4597437381744385, 0.00010013816063292325], "traffic_r2": [0.9786460995674133, 0.9949455857276917, 1.0]}, {"epoch_end": 1, "step": 860, "eval_split": "validation", "eval_acc": 0.779, "eval_loss": 1.0340379516601563, "val_acc": 0.779, "val_loss": 1.0340379516601563}, {"step": 1000, "epoch": 2, "train_loss": 1.0072448253631592, "cos_r_negg": [0.2548796534538269, 0.1735328584909439, 0.3421844244003296], "cos_innovation_negg": [0.2548796534538269, 0.1735328584909439, 0.3421844244003296], "cos_apical_negg": [0.0025696922093629837, 0.2173481583595276, -0.07620467245578766], "cos_Ac_negg": [0.8877443075180054, 0.5683277249336243, 0.37328025698661804], "r_norm": [1.4383184909820557, 0.9769741296768188, 0.7567906379699707], "traffic_norm": [6.661558628082275, 6.506039142608643, 6.535297393798828], "traffic_residual_norm": [0.722628116607666, 0.3142322301864624, 7.04212870914489e-05], "traffic_r2": [0.9882326722145081, 0.9976662993431091, 1.0]}, {"step": 1200, "epoch": 2, "train_loss": 0.7817351818084717, "cos_r_negg": [0.053765833377838135, -0.026373252272605896, 0.3707562983036041], "cos_innovation_negg": [0.053765833377838135, -0.026373252272605896, 0.3707562983036041], "cos_apical_negg": [-0.10476191341876984, -0.015862243250012398, -0.0033468701876699924], "cos_Ac_negg": [0.792730450630188, -0.00957019068300724, 0.5608258247375488], "r_norm": [1.2712841033935547, 0.8836820125579834, 0.671884298324585], "traffic_norm": [6.648972988128662, 6.506423473358154, 6.524881362915039], "traffic_residual_norm": [0.6094027161598206, 0.2809427082538605, 4.982022801414132e-05], "traffic_r2": [0.991594672203064, 0.9981145262718201, 1.0]}, {"epoch_end": 2, "step": 1290, "eval_split": "validation", "eval_acc": 0.855, "eval_loss": 0.7150523803710938, "val_acc": 0.855, "val_loss": 0.7150523803710938}, {"step": 1400, "epoch": 3, "train_loss": 0.8158656358718872, "cos_r_negg": [0.22867053747177124, 0.0033523906022310257, 0.29663553833961487], "cos_innovation_negg": [0.22867053747177124, 0.0033523906022310257, 0.29663553833961487], "cos_apical_negg": [-0.02038625255227089, 0.19146336615085602, 0.0264933742582798], "cos_Ac_negg": [0.7998595237731934, 0.4184272587299347, 0.347024142742157], "r_norm": [1.4465525150299072, 0.9641604423522949, 0.7921051979064941], "traffic_norm": [6.641875267028809, 6.511187553405762, 6.505472183227539], "traffic_residual_norm": [0.6903664469718933, 0.272748738527298, 3.8722919271094725e-05], "traffic_r2": [0.9891961812973022, 0.998138427734375, 1.0]}, {"step": 1600, "epoch": 3, "train_loss": 0.6467530131340027, "cos_r_negg": [0.1311189979314804, -0.06312969326972961, 0.3566320538520813], "cos_innovation_negg": [0.1311189979314804, -0.06312969326972961, 0.3566320538520813], "cos_apical_negg": [0.00603673979640007, 0.006540990434587002, -0.14807547628879547], "cos_Ac_negg": [0.6081014275550842, -0.10753148049116135, 0.5209767818450928], "r_norm": [1.3712655305862427, 0.9633283615112305, 0.8244985342025757], "traffic_norm": [6.646936416625977, 6.51093864440918, 6.504302978515625], "traffic_residual_norm": [0.5286931991577148, 0.19916217029094696, 3.0081122531555593e-05], "traffic_r2": [0.993670642375946, 0.9990202784538269, 1.0]}, {"epoch_end": 3, "step": 1720, "eval_split": "validation", "eval_acc": 0.8994, "eval_loss": 0.5213863647460938, "val_acc": 0.8994, "val_loss": 0.5213863647460938}, {"step": 1800, "epoch": 4, "train_loss": 0.25079163908958435, "cos_r_negg": [0.16934075951576233, -0.12234373390674591, 0.4128657877445221], "cos_innovation_negg": [0.16934075951576233, -0.12234373390674591, 0.4128657877445221], "cos_apical_negg": [0.049495629966259, 0.10958608239889145, -0.06298723816871643], "cos_Ac_negg": [0.7153122425079346, -0.04362776130437851, 0.5871828198432922], "r_norm": [1.2281830310821533, 0.8363975286483765, 0.7306379675865173], "traffic_norm": [6.644603252410889, 6.5157856941223145, 6.497584819793701], "traffic_residual_norm": [0.4494900107383728, 0.14898130297660828, 2.4800576284178533e-05], "traffic_r2": [0.9954238533973694, 0.9994626045227051, 1.0]}, {"step": 2000, "epoch": 4, "train_loss": 0.5647637844085693, "cos_r_negg": [0.14806000888347626, -0.08192439377307892, 0.40098893642425537], "cos_innovation_negg": [0.14806000888347626, -0.08192439377307892, 0.40098893642425537], "cos_apical_negg": [0.03396216407418251, 0.11489957571029663, -0.012363939546048641], "cos_Ac_negg": [0.8366168141365051, 0.1032198965549469, 0.5039911270141602], "r_norm": [1.0812678337097168, 0.6927978992462158, 0.5918296575546265], "traffic_norm": [6.611415863037109, 6.4945549964904785, 6.464434623718262], "traffic_residual_norm": [0.4439111649990082, 0.1335110068321228, 2.26448719331529e-05], "traffic_r2": [0.9954918026924133, 0.9995710849761963, 1.0]}, {"epoch_end": 4, "step": 2150, "eval_split": "validation", "eval_acc": 0.8992, "eval_loss": 0.524011962890625, "val_acc": 0.8992, "val_loss": 0.524011962890625}, {"step": 2200, "epoch": 5, "train_loss": 0.4558678865432739, "cos_r_negg": [0.24158497154712677, 0.16789352893829346, 0.2964246869087219], "cos_innovation_negg": [0.24158497154712677, 0.16789352893829346, 0.2964246869087219], "cos_apical_negg": [0.11693146079778671, 0.23016008734703064, 0.044748470187187195], "cos_Ac_negg": [0.8380728363990784, 0.6975541114807129, 0.3378947377204895], "r_norm": [1.1138290166854858, 0.7596010565757751, 0.6957237720489502], "traffic_norm": [6.618823051452637, 6.510370254516602, 6.471869945526123], "traffic_residual_norm": [0.34623605012893677, 0.0972936749458313, 1.9628909285529517e-05], "traffic_r2": [0.9972633123397827, 0.9997743964195251, 1.0]}, {"step": 2400, "epoch": 5, "train_loss": 0.5152024030685425, "cos_r_negg": [0.27635976672172546, 0.2611757516860962, 0.41243746876716614], "cos_innovation_negg": [0.27635976672172546, 0.2611757516860962, 0.41243746876716614], "cos_apical_negg": [0.02436543069779873, -0.22097794711589813, -0.08287247270345688], "cos_Ac_negg": [0.8567706942558289, 0.2569397985935211, 0.437755286693573], "r_norm": [0.9567050933837891, 0.6938559412956238, 0.6550405025482178], "traffic_norm": [6.62771463394165, 6.525442123413086, 6.486946105957031], "traffic_residual_norm": [0.23435774445533752, 0.07607661932706833, 1.782511390047148e-05], "traffic_r2": [0.9987495541572571, 0.9998630285263062, 1.0]}, {"epoch_end": 5, "step": 2580, "eval_split": "validation", "eval_acc": 0.8418, "eval_loss": 0.822113525390625, "val_acc": 0.8418, "val_loss": 0.822113525390625}, {"step": 2600, "epoch": 6, "train_loss": 0.7646856307983398, "cos_r_negg": [0.19304046034812927, -0.030286453664302826, 0.535049557685852], "cos_innovation_negg": [0.19304046034812927, -0.030286453664302826, 0.535049557685852], "cos_apical_negg": [0.017224125564098358, 0.29080185294151306, 0.041409000754356384], "cos_Ac_negg": [0.8627030253410339, 0.31638720631599426, 0.6023257374763489], "r_norm": [1.0323424339294434, 0.5836083889007568, 0.5410177111625671], "traffic_norm": [6.572677135467529, 6.4835052490234375, 6.454897880554199], "traffic_residual_norm": [0.45585140585899353, 0.06531824171543121, 1.8679700588108972e-05], "traffic_r2": [0.9951882362365723, 0.999897837638855, 1.0]}, {"step": 2800, "epoch": 6, "train_loss": 0.34986612200737, "cos_r_negg": [0.238646000623703, 0.09582706540822983, 0.3571614623069763], "cos_innovation_negg": [0.238646000623703, 0.09582706540822983, 0.3571614623069763], "cos_apical_negg": [0.019167905673384666, -0.37626954913139343, -0.017931761220097542], "cos_Ac_negg": [0.8106469511985779, -0.17905893921852112, 0.365262895822525], "r_norm": [0.9606461524963379, 0.6232796907424927, 0.5898219347000122], "traffic_norm": [6.567977428436279, 6.482635498046875, 6.450570106506348], "traffic_residual_norm": [0.32216596603393555, 0.057510942220687866, 1.8172240743297152e-05], "traffic_r2": [0.9975935220718384, 0.999920666217804, 1.0]}, {"step": 3000, "epoch": 6, "train_loss": 0.34643036127090454, "cos_r_negg": [0.2967384457588196, -0.06492570787668228, 0.47990965843200684], "cos_innovation_negg": [0.2967384457588196, -0.06492570787668228, 0.47990965843200684], "cos_apical_negg": [0.023263178765773773, 0.2754817008972168, 0.1565592885017395], "cos_Ac_negg": [0.8209381103515625, 0.4828030467033386, 0.5042271614074707], "r_norm": [0.9964900612831116, 0.7482192516326904, 0.6975353360176086], "traffic_norm": [6.574790000915527, 6.488768577575684, 6.45911169052124], "traffic_residual_norm": [0.21803441643714905, 0.08151912689208984, 1.752164462232031e-05], "traffic_r2": [0.9989002346992493, 0.9998418092727661, 1.0]}, {"epoch_end": 6, "step": 3010, "eval_split": "validation", "eval_acc": 0.876, "eval_loss": 0.5910611755371094, "val_acc": 0.876, "val_loss": 0.5910611755371094}, {"step": 3200, "epoch": 7, "train_loss": 0.19205665588378906, "cos_r_negg": [0.30593669414520264, -0.018011488020420074, 0.44686198234558105], "cos_innovation_negg": [0.30593669414520264, -0.018011488020420074, 0.44686198234558105], "cos_apical_negg": [0.06976453959941864, 0.15019521117210388, 0.058800503611564636], "cos_Ac_negg": [0.8789693117141724, 0.5767152905464172, 0.4445362091064453], "r_norm": [0.8533759117126465, 0.6467377543449402, 0.6022179126739502], "traffic_norm": [6.570281982421875, 6.4827775955200195, 6.453278064727783], "traffic_residual_norm": [0.17200830578804016, 0.06487096846103668, 1.7483018382336013e-05], "traffic_r2": [0.9993143081665039, 0.9998993873596191, 1.0]}, {"step": 3400, "epoch": 7, "train_loss": 0.4280366599559784, "cos_r_negg": [0.3646857738494873, 0.15213046967983246, 0.4184509515762329], "cos_innovation_negg": [0.3646857738494873, 0.15213046967983246, 0.4184509515762329], "cos_apical_negg": [-0.026498379185795784, -0.03023921512067318, -0.09083180129528046], "cos_Ac_negg": [0.8499939441680908, 0.33405399322509766, 0.4621775150299072], "r_norm": [0.9575053453445435, 0.732962965965271, 0.690421462059021], "traffic_norm": [6.569457530975342, 6.482351303100586, 6.446322441101074], "traffic_residual_norm": [0.1746671497821808, 0.06439901888370514, 1.7109510736190714e-05], "traffic_r2": [0.9992920160293579, 0.9999009370803833, 1.0]}, {"epoch_end": 7, "step": 3440, "eval_split": "validation", "eval_acc": 0.8926, "eval_loss": 0.5215070617675781, "val_acc": 0.8926, "val_loss": 0.5215070617675781}, {"step": 3600, "epoch": 8, "train_loss": 0.47636735439300537, "cos_r_negg": [0.31760790944099426, -0.04389245808124542, 0.4331001043319702], "cos_innovation_negg": [0.31760790944099426, -0.04389245808124542, 0.4331001043319702], "cos_apical_negg": [-0.011962512508034706, 0.12403607368469238, -0.01789310947060585], "cos_Ac_negg": [0.9051072001457214, 0.34350335597991943, 0.49222269654273987], "r_norm": [0.8024435043334961, 0.5449646711349487, 0.5046321749687195], "traffic_norm": [6.570004463195801, 6.480433464050293, 6.44782829284668], "traffic_residual_norm": [0.25037136673927307, 0.061289265751838684, 1.6783316823421046e-05], "traffic_r2": [0.9985439777374268, 0.9999102354049683, 1.0]}, {"step": 3800, "epoch": 8, "train_loss": 0.18262054026126862, "cos_r_negg": [0.35584455728530884, 0.014419511891901493, 0.5247029066085815], "cos_innovation_negg": [0.35584455728530884, 0.014419511891901493, 0.5247029066085815], "cos_apical_negg": [-0.001069638878107071, 0.15030980110168457, 0.10269508510828018], "cos_Ac_negg": [0.867664635181427, 0.3285750448703766, 0.5335276126861572], "r_norm": [0.8590066432952881, 0.6109097003936768, 0.5726625919342041], "traffic_norm": [6.568493366241455, 6.482247829437256, 6.448148727416992], "traffic_residual_norm": [0.21016916632652283, 0.0565204843878746, 1.661202622926794e-05], "traffic_r2": [0.9989755749702454, 0.9999237656593323, 1.0]}, {"epoch_end": 8, "step": 3870, "eval_split": "validation", "eval_acc": 0.9134, "eval_loss": 0.4131411315917969, "val_acc": 0.9134, "val_loss": 0.4131411315917969}, {"step": 4000, "epoch": 9, "train_loss": 0.18865084648132324, "cos_r_negg": [0.40255671739578247, 0.14300325512886047, 0.5051268935203552], "cos_innovation_negg": [0.40255671739578247, 0.14300325512886047, 0.5051268935203552], "cos_apical_negg": [-0.07140658795833588, -0.02357449382543564, -0.02428489550948143], "cos_Ac_negg": [0.9051531553268433, 0.4883214831352234, 0.5318533182144165], "r_norm": [0.8481141924858093, 0.5949658751487732, 0.5647392272949219], "traffic_norm": [6.5579986572265625, 6.461441993713379, 6.410675048828125], "traffic_residual_norm": [0.22684991359710693, 0.05403994023799896, 1.8298314898856916e-05], "traffic_r2": [0.9988027215003967, 0.9999299049377441, 1.0]}, {"step": 4200, "epoch": 9, "train_loss": 0.6902337074279785, "cos_r_negg": [0.36350005865097046, 0.12422843277454376, 0.44337642192840576], "cos_innovation_negg": [0.36350005865097046, 0.12422843277454376, 0.44337642192840576], "cos_apical_negg": [-0.03095867484807968, 0.0187084823846817, -0.06186681240797043], "cos_Ac_negg": [0.8658245205879211, 0.2324446439743042, 0.5317682027816772], "r_norm": [0.8238120079040527, 0.565418004989624, 0.5319932699203491], "traffic_norm": [6.5584940910339355, 6.463529586791992, 6.417040824890137], "traffic_residual_norm": [0.234411358833313, 0.051461003720760345, 1.8000077034230344e-05], "traffic_r2": [0.9987178444862366, 0.9999364614486694, 1.0]}, {"epoch_end": 9, "step": 4300, "eval_split": "validation", "eval_acc": 0.8856, "eval_loss": 0.5286158630371094, "val_acc": 0.8856, "val_loss": 0.5286158630371094}, {"step": 4400, "epoch": 10, "train_loss": 0.39770856499671936, "cos_r_negg": [0.4373714327812195, 0.20169192552566528, 0.4584939479827881], "cos_innovation_negg": [0.4373714327812195, 0.20169192552566528, 0.4584939479827881], "cos_apical_negg": [-0.04434895142912865, -0.060282494872808456, -0.09462476521730423], "cos_Ac_negg": [0.8599269390106201, 0.2117495834827423, 0.550071656703949], "r_norm": [0.976725697517395, 0.6634008884429932, 0.6277979612350464], "traffic_norm": [6.5558013916015625, 6.458782196044922, 6.419185638427734], "traffic_residual_norm": [0.2798205316066742, 0.05244946479797363, 1.7677964933682233e-05], "traffic_r2": [0.9981618523597717, 0.9999337792396545, 1.0]}, {"step": 4600, "epoch": 10, "train_loss": 0.29895836114883423, "cos_r_negg": [0.4542761445045471, 0.14774686098098755, 0.510329008102417], "cos_innovation_negg": [0.4542761445045471, 0.14774686098098755, 0.510329008102417], "cos_apical_negg": [-0.05376717448234558, 0.23060739040374756, 0.006061492953449488], "cos_Ac_negg": [0.8864158391952515, 0.37675535678863525, 0.5123541355133057], "r_norm": [0.8558309078216553, 0.616879940032959, 0.5803134441375732], "traffic_norm": [6.561469078063965, 6.469771862030029, 6.4231276512146], "traffic_residual_norm": [0.20962420105934143, 0.05760180205106735, 1.696227081993129e-05], "traffic_r2": [0.9989789724349976, 0.9999205470085144, 1.0]}, {"epoch_end": 10, "step": 4730, "eval_split": "validation", "eval_acc": 0.9184, "eval_loss": 0.4081424560546875, "val_acc": 0.9184, "val_loss": 0.4081424560546875}, {"step": 4800, "epoch": 11, "train_loss": 0.2085874378681183, "cos_r_negg": [0.35002633929252625, 0.20767536759376526, 0.5125534534454346], "cos_innovation_negg": [0.35002633929252625, 0.20767536759376526, 0.5125534534454346], "cos_apical_negg": [0.017445158213377, -0.07995351403951645, -0.06757029891014099], "cos_Ac_negg": [0.8769694566726685, 0.6419504880905151, 0.5647639036178589], "r_norm": [0.7328353524208069, 0.5639629364013672, 0.5135319828987122], "traffic_norm": [6.554441452026367, 6.461040496826172, 6.41654634475708], "traffic_residual_norm": [0.1484275460243225, 0.06819857656955719, 2.2149732103571296e-05], "traffic_r2": [0.9994866847991943, 0.9998884797096252, 1.0]}, {"step": 5000, "epoch": 11, "train_loss": 0.2899661064147949, "cos_r_negg": [0.4067528545856476, 0.16630961000919342, 0.5248256325721741], "cos_innovation_negg": [0.4067528545856476, 0.16630961000919342, 0.5248256325721741], "cos_apical_negg": [0.0430239699780941, 0.14899417757987976, -0.09406904131174088], "cos_Ac_negg": [0.8782027959823608, 0.7379088401794434, 0.6074784398078918], "r_norm": [0.8162586092948914, 0.6158162355422974, 0.559954047203064], "traffic_norm": [6.546053886413574, 6.448751449584961, 6.404865264892578], "traffic_residual_norm": [0.19295933842658997, 0.08314387500286102, 1.9600072846515104e-05], "traffic_r2": [0.9991273880004883, 0.9998336434364319, 1.0]}, {"epoch_end": 11, "step": 5160, "eval_split": "validation", "eval_acc": 0.9082, "eval_loss": 0.4540973754882813, "val_acc": 0.9082, "val_loss": 0.4540973754882813}, {"step": 5200, "epoch": 12, "train_loss": 0.12979966402053833, "cos_r_negg": [0.39793169498443604, 0.12062463164329529, 0.4670393764972687], "cos_innovation_negg": [0.39793169498443604, 0.12062463164329529, 0.4670393764972687], "cos_apical_negg": [-0.04132218658924103, 0.12708118557929993, -0.004672444425523281], "cos_Ac_negg": [0.855082631111145, 0.5940340161323547, 0.5267490148544312], "r_norm": [0.7426825761795044, 0.550107479095459, 0.490190327167511], "traffic_norm": [6.5427446365356445, 6.459900856018066, 6.4198198318481445], "traffic_residual_norm": [0.19776907563209534, 0.0799434632062912, 1.954737126652617e-05], "traffic_r2": [0.999082088470459, 0.9998466372489929, 1.0]}, {"step": 5400, "epoch": 12, "train_loss": 0.42847853899002075, "cos_r_negg": [0.4481130838394165, 0.08582563698291779, 0.4659475088119507], "cos_innovation_negg": [0.4481130838394165, 0.08582563698291779, 0.4659475088119507], "cos_apical_negg": [-0.029627054929733276, 0.24605202674865723, 0.004714989103376865], "cos_Ac_negg": [0.8742615580558777, 0.5904257297515869, 0.585253894329071], "r_norm": [0.8454612493515015, 0.6486009359359741, 0.5971398949623108], "traffic_norm": [6.503408432006836, 6.440482139587402, 6.40101432800293], "traffic_residual_norm": [0.18629483878612518, 0.0643930584192276, 1.837385934777558e-05], "traffic_r2": [0.9991716146469116, 0.9998998641967773, 1.0]}, {"epoch_end": 12, "step": 5590, "eval_split": "validation", "eval_acc": 0.9186, "eval_loss": 0.3700715393066406, "val_acc": 0.9186, "val_loss": 0.3700715393066406}, {"step": 5600, "epoch": 13, "train_loss": 0.37909072637557983, "cos_r_negg": [0.4440132677555084, 0.18727529048919678, 0.4742017984390259], "cos_innovation_negg": [0.4440132677555084, 0.18727529048919678, 0.4742017984390259], "cos_apical_negg": [-0.09622073173522949, 0.038316622376441956, 0.024648919701576233], "cos_Ac_negg": [0.8702258467674255, 0.5884068012237549, 0.5366061925888062], "r_norm": [0.5913292169570923, 0.4601777195930481, 0.4178823232650757], "traffic_norm": [6.509119033813477, 6.441884994506836, 6.409041881561279], "traffic_residual_norm": [0.1249019056558609, 0.05930052697658539, 1.785897256922908e-05], "traffic_r2": [0.9996291995048523, 0.9999151825904846, 1.0]}, {"step": 5800, "epoch": 13, "train_loss": 0.3182448744773865, "cos_r_negg": [0.4086453914642334, 0.1944352388381958, 0.35972142219543457], "cos_innovation_negg": [0.4086453914642334, 0.1944352388381958, 0.35972142219543457], "cos_apical_negg": [-0.0021681729704141617, -0.045527853071689606, -0.025201747193932533], "cos_Ac_negg": [0.8650181889533997, 0.5023183226585388, 0.48673540353775024], "r_norm": [0.5573519468307495, 0.45436909794807434, 0.41208726167678833], "traffic_norm": [6.506284713745117, 6.442291259765625, 6.402486324310303], "traffic_residual_norm": [0.09327206760644913, 0.05648960918188095, 1.773228359525092e-05], "traffic_r2": [0.9997841119766235, 0.9999229907989502, 1.0]}, {"step": 6000, "epoch": 13, "train_loss": 0.24404039978981018, "cos_r_negg": [0.403228759765625, 0.12163899838924408, 0.49020886421203613], "cos_innovation_negg": [0.403228759765625, 0.12163899838924408, 0.49020886421203613], "cos_apical_negg": [0.020540934056043625, 0.21965646743774414, 0.08739084750413895], "cos_Ac_negg": [0.884434700012207, 0.5297906398773193, 0.6067909002304077], "r_norm": [0.8567236661911011, 0.7286198735237122, 0.6945141553878784], "traffic_norm": [6.508857727050781, 6.443134784698486, 6.404284477233887], "traffic_residual_norm": [0.06572970747947693, 0.05437226593494415, 1.7682628822512925e-05], "traffic_r2": [0.9998962879180908, 0.9999286532402039, 1.0]}, {"epoch_end": 13, "step": 6020, "eval_split": "validation", "eval_acc": 0.9028, "eval_loss": 0.41332857055664063, "val_acc": 0.9028, "val_loss": 0.41332857055664063}, {"step": 6200, "epoch": 14, "train_loss": 0.48969805240631104, "cos_r_negg": [0.45092302560806274, 0.05935497209429741, 0.4385910630226135], "cos_innovation_negg": [0.45092302560806274, 0.05935497209429741, 0.4385910630226135], "cos_apical_negg": [-0.05572124570608139, -0.0490303598344326, -0.016167981550097466], "cos_Ac_negg": [0.8636496067047119, 0.45878660678863525, 0.49328699707984924], "r_norm": [0.4937863051891327, 0.4082013666629791, 0.36768242716789246], "traffic_norm": [6.510239601135254, 6.445257663726807, 6.4085516929626465], "traffic_residual_norm": [0.07448782026767731, 0.05033930763602257, 1.6917601897148415e-05], "traffic_r2": [0.9998679757118225, 0.9999388456344604, 1.0]}, {"step": 6400, "epoch": 14, "train_loss": 0.4201534390449524, "cos_r_negg": [0.4524872899055481, 0.17321112751960754, 0.5375621914863586], "cos_innovation_negg": [0.4524872899055481, 0.17321112751960754, 0.5375621914863586], "cos_apical_negg": [-0.07248817384243011, -0.032181236892938614, -0.05088210850954056], "cos_Ac_negg": [0.8831174373626709, 0.5241477489471436, 0.5856068134307861], "r_norm": [0.6124165058135986, 0.48295438289642334, 0.45006701350212097], "traffic_norm": [6.510244369506836, 6.4430975914001465, 6.407101154327393], "traffic_residual_norm": [0.10714198648929596, 0.04747368022799492, 1.7400911019649357e-05], "traffic_r2": [0.999725341796875, 0.9999454021453857, 1.0]}, {"epoch_end": 14, "step": 6450, "eval_split": "validation", "eval_acc": 0.9102, "eval_loss": 0.42569903564453127, "val_acc": 0.9102, "val_loss": 0.42569903564453127}], "final": {"eval_split": "validation", "eval_acc": 0.9102, "eval_loss": 0.42569903564453127, "wall_s": 21.2925546169281, "val_acc": 0.9102, "val_loss": 0.42569903564453127, "cos_r_negg": [0.40888869762420654, 0.23761305212974548, 0.3860348165035248], "cos_innovation_negg": [0.40888869762420654, 0.23761305212974548, 0.3860348165035248], "cos_apical_negg": [-0.09437962621450424, 0.033791325986385345, -0.020948387682437897], "cos_Ac_negg": [0.8731504082679749, 0.6088327169418335, 0.5036921501159668], "traffic_norm": [6.509462356567383, 6.440916061401367, 6.4056925773620605], "traffic_residual_norm": [0.1483639031648636, 0.047352567315101624, 1.6414793208241463e-05], "traffic_r2": [0.999471127986908, 0.9999456405639648, 1.0]}, "timing": {"warmup_wall_s": 0.35772156715393066, "training_loop_wall_s": 20.747973442077637, "diagnostics_wall_s": 0.5052170753479004, "inline_diagnostics_wall_s": 0.3813605308532715, "optimizer_wall_s_excluding_inline_diagnostics": 20.366612911224365, "evaluation_wall_s": 0.05990457534790039}, "cost": {"train_steps": 6450, "ordinary_training_forward_examples": 825000, "predictor_warmup_forward_examples": 25600, "perturbation_events": 1613, "calibration_batch_loss_evaluations": 51616, "calibration_example_loss_evaluations": 6606848, "calibration_forward_equivalent_examples": 6813312.0, "max_perturbation_batch_expansion": 32}}
|