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": 0.01, "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": 0, "p_warmup_eta": 0.05, "nuis_rho": 0.5, "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_rho0p5_ep0p01_wu0_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": 2.5729050636291504, "cos_r_negg": [-0.02926800213754177, 0.0021967983338981867, 0.008305815979838371], "cos_innovation_negg": [-0.02926800213754177, 0.0021967983338981867, 0.008305815979838371], "cos_apical_negg": [-0.029235415160655975, 0.002071091439574957, 0.008074866607785225], "cos_Ac_negg": [-0.026053454726934433, 0.019999800249934196, 0.03730686753988266], "r_norm": [11.040082931518555, 10.752248764038086, 10.448749542236328], "traffic_norm": [9.704059600830078, 9.645772933959961, 9.290628433227539], "traffic_residual_norm": [9.616304397583008, 9.54352855682373, 9.173799514770508], "traffic_r2": [0.017995774745941162, 0.021080315113067627, 0.02498716115951538]}, {"step": 200, "epoch": 0, "train_loss": 9.191450119018555, "cos_r_negg": [0.40730932354927063, 0.15799131989479065, -0.27001264691352844], "cos_innovation_negg": [0.40730932354927063, 0.15799131989479065, -0.27001264691352844], "cos_apical_negg": [0.014963524416089058, 0.14739948511123657, -0.0589916855096817], "cos_Ac_negg": [0.5841583013534546, 0.16743749380111694, -0.5644031763076782], "r_norm": [4.367799758911133, 4.345888137817383, 4.210847854614258], "traffic_norm": [18.072242736816406, 17.88100814819336, 17.761192321777344], "traffic_residual_norm": [2.314831495285034, 2.2667341232299805, 2.2602319717407227], "traffic_r2": [0.9835935235023499, 0.9839280843734741, 0.9838019013404846]}, {"step": 400, "epoch": 0, "train_loss": 1.347914457321167, "cos_r_negg": [0.45114192366600037, -0.16096068918704987, -0.07092725485563278], "cos_innovation_negg": [0.45114192366600037, -0.16096068918704987, -0.07092725485563278], "cos_apical_negg": [0.16552570462226868, -0.12974558770656586, -0.018916495144367218], "cos_Ac_negg": [0.7254474759101868, 0.015854045748710632, -0.30139559507369995], "r_norm": [1.619210124015808, 1.6007684469223022, 1.7011075019836426], "traffic_norm": [18.01609992980957, 17.795795440673828, 17.658527374267578], "traffic_residual_norm": [0.31395453214645386, 0.38028180599212646, 0.5400084257125854], "traffic_r2": [0.9996907711029053, 0.9995370507240295, 0.9990512728691101]}, {"epoch_end": 0, "step": 430, "eval_split": "validation", "eval_acc": 0.7668, "eval_loss": 1.5390849365234376, "val_acc": 0.7668, "val_loss": 1.5390849365234376}, {"step": 600, "epoch": 1, "train_loss": 0.8249157667160034, "cos_r_negg": [0.4736281931400299, -0.061475250869989395, 0.06920460611581802], "cos_innovation_negg": [0.4736281931400299, -0.061475250869989395, 0.06920460611581802], "cos_apical_negg": [0.07623498141765594, -0.30132344365119934, -0.01397777907550335], "cos_Ac_negg": [0.7289226651191711, 0.044154614210128784, 0.20300045609474182], "r_norm": [1.609207272529602, 1.6728483438491821, 1.8820196390151978], "traffic_norm": [17.87793731689453, 17.658977508544922, 17.42089080810547], "traffic_residual_norm": [0.439175009727478, 0.6322450041770935, 0.8910300731658936], "traffic_r2": [0.9993893504142761, 0.9987073540687561, 0.9973691701889038]}, {"step": 800, "epoch": 1, "train_loss": 1.4595685005187988, "cos_r_negg": [0.4198853373527527, 0.15266019105911255, 0.2499999850988388], "cos_innovation_negg": [0.4198853373527527, 0.15266019105911255, 0.2499999850988388], "cos_apical_negg": [0.0475311279296875, -0.311387836933136, -0.1132812350988388], "cos_Ac_negg": [0.765862226486206, -0.04170502722263336, 0.4609375], "r_norm": [1.1332871913909912, 1.215933084487915, 1.2968943119049072], "traffic_norm": [17.840717315673828, 17.624311447143555, 17.32931137084961], "traffic_residual_norm": [0.2108323574066162, 0.39368104934692383, 0.492044061422348], "traffic_r2": [0.9998527765274048, 0.9994875192642212, 0.9991875290870667]}, {"epoch_end": 1, "step": 860, "eval_split": "validation", "eval_acc": 0.8542, "eval_loss": 0.8723792114257812, "val_acc": 0.8542, "val_loss": 0.8723792114257812}, {"step": 1000, "epoch": 2, "train_loss": 0.6080271005630493, "cos_r_negg": [0.44007113575935364, 0.18410684168338776, 0.09438187628984451], "cos_innovation_negg": [0.44007113575935364, 0.18410684168338776, 0.09438187628984451], "cos_apical_negg": [-0.026839682832360268, -0.17591054737567902, 0.07490946352481842], "cos_Ac_negg": [0.7832607626914978, -0.044654183089733124, 0.4526536762714386], "r_norm": [1.0002599954605103, 1.1045507192611694, 1.1939061880111694], "traffic_norm": [17.732723236083984, 17.451656341552734, 17.188201904296875], "traffic_residual_norm": [0.3045201301574707, 0.4999510943889618, 0.5920867919921875], "traffic_r2": [0.9997046589851379, 0.9991785287857056, 0.9988123774528503]}, {"step": 1200, "epoch": 2, "train_loss": 0.7668049931526184, "cos_r_negg": [0.4118879735469818, -0.013927379623055458, 0.054875269532203674], "cos_innovation_negg": [0.4118879735469818, -0.013927379623055458, 0.054875269532203674], "cos_apical_negg": [0.009184055030345917, -0.3046608865261078, 0.1843404322862625], "cos_Ac_negg": [0.8101706504821777, -0.08023817092180252, 0.21173949539661407], "r_norm": [0.7963392734527588, 0.814545750617981, 0.9036046266555786], "traffic_norm": [17.7015323638916, 17.394668579101562, 17.143735885620117], "traffic_residual_norm": [0.1438058316707611, 0.24261045455932617, 0.33308207988739014], "traffic_r2": [0.999933123588562, 0.999804675579071, 0.9996219873428345]}, {"epoch_end": 2, "step": 1290, "eval_split": "validation", "eval_acc": 0.8566, "eval_loss": 0.810403955078125, "val_acc": 0.8566, "val_loss": 0.810403955078125}, {"step": 1400, "epoch": 3, "train_loss": 0.8306488990783691, "cos_r_negg": [0.48032891750335693, 0.149397075176239, 0.407828152179718], "cos_innovation_negg": [0.48032891750335693, 0.149397075176239, 0.407828152179718], "cos_apical_negg": [0.09116197377443314, -0.13329827785491943, 0.015882721170783043], "cos_Ac_negg": [0.789527177810669, -0.07864009588956833, 0.4784314036369324], "r_norm": [0.7738761901855469, 0.7383730411529541, 0.8524385094642639], "traffic_norm": [17.576778411865234, 17.303678512573242, 17.046344757080078], "traffic_residual_norm": [0.22236676514148712, 0.24269691109657288, 0.3681308329105377], "traffic_r2": [0.9998392462730408, 0.9998022317886353, 0.9995316863059998]}, {"step": 1600, "epoch": 3, "train_loss": 1.080175757408142, "cos_r_negg": [0.5206195712089539, 0.0865127369761467, 0.08477165549993515], "cos_innovation_negg": [0.5206195712089539, 0.0865127369761467, 0.08477165549993515], "cos_apical_negg": [0.07654829323291779, -0.07915733754634857, 0.1580560803413391], "cos_Ac_negg": [0.7570786476135254, 0.0015652076108381152, 0.09588814526796341], "r_norm": [0.7063867449760437, 0.7308411598205566, 0.85150146484375], "traffic_norm": [17.560834884643555, 17.284021377563477, 17.022701263427734], "traffic_residual_norm": [0.13526718318462372, 0.2305343598127365, 0.37227991223335266], "traffic_r2": [0.9999375939369202, 0.9998189210891724, 0.9995144009590149]}, {"epoch_end": 3, "step": 1720, "eval_split": "validation", "eval_acc": 0.8772, "eval_loss": 0.6858205322265625, "val_acc": 0.8772, "val_loss": 0.6858205322265625}, {"step": 1800, "epoch": 4, "train_loss": 0.6304429769515991, "cos_r_negg": [0.559229850769043, 0.04153801128268242, 0.21139416098594666], "cos_innovation_negg": [0.559229850769043, 0.04153801128268242, 0.21139416098594666], "cos_apical_negg": [0.09306858479976654, -0.1025671437382698, -0.03491509333252907], "cos_Ac_negg": [0.8428690433502197, 0.013539859093725681, 0.4374999403953552], "r_norm": [0.5729638338088989, 0.5650748014450073, 0.6146010756492615], "traffic_norm": [17.538471221923828, 17.262178421020508, 16.957855224609375], "traffic_residual_norm": [0.07069917023181915, 0.12488162517547607, 0.16651007533073425], "traffic_r2": [0.9999837279319763, 0.9999476671218872, 0.9999035596847534]}, {"step": 2000, "epoch": 4, "train_loss": 0.5708114504814148, "cos_r_negg": [0.5382818579673767, 0.191399946808815, 0.448140949010849], "cos_innovation_negg": [0.5382818579673767, 0.191399946808815, 0.448140949010849], "cos_apical_negg": [0.0575249046087265, -0.013625058345496655, 0.07240699231624603], "cos_Ac_negg": [0.7988659143447876, -0.1553386002779007, 0.6360079050064087], "r_norm": [0.6886090040206909, 0.6542329788208008, 0.7084299325942993], "traffic_norm": [17.545825958251953, 17.27718162536621, 16.980056762695312], "traffic_residual_norm": [0.09410500526428223, 0.13568679988384247, 0.18809956312179565], "traffic_r2": [0.9999712109565735, 0.9999383091926575, 0.9998772740364075]}, {"epoch_end": 4, "step": 2150, "eval_split": "validation", "eval_acc": 0.8776, "eval_loss": 0.6313579467773438, "val_acc": 0.8776, "val_loss": 0.6313579467773438}, {"step": 2200, "epoch": 5, "train_loss": 0.6068122982978821, "cos_r_negg": [0.6301424503326416, -0.008663814514875412, 0.4102945923805237], "cos_innovation_negg": [0.6301424503326416, -0.008663814514875412, 0.4102945923805237], "cos_apical_negg": [0.11072802543640137, -0.09650864452123642, 0.3780255913734436], "cos_Ac_negg": [0.7854105234146118, 0.10123223811388016, 0.5234253406524658], "r_norm": [0.8700582981109619, 0.7974772453308105, 0.8917244076728821], "traffic_norm": [17.479738235473633, 17.235427856445312, 16.9188289642334], "traffic_residual_norm": [0.18759998679161072, 0.20007967948913574, 0.28618717193603516], "traffic_r2": [0.9998847842216492, 0.9998652338981628, 0.9997138381004333]}, {"step": 2400, "epoch": 5, "train_loss": 0.6308832168579102, "cos_r_negg": [0.5650438070297241, 0.23663656413555145, 0.3755865693092346], "cos_innovation_negg": [0.5650438070297241, 0.23663656413555145, 0.3755865693092346], "cos_apical_negg": [0.050648417323827744, -0.39601802825927734, 2.01107013708679e-05], "cos_Ac_negg": [0.8331104516983032, 0.05629042908549309, 0.676701545715332], "r_norm": [0.7259911894798279, 0.6775956749916077, 0.7095768451690674], "traffic_norm": [17.48702621459961, 17.24958038330078, 16.934560775756836], "traffic_residual_norm": [0.043732210993766785, 0.07303883880376816, 0.10350840538740158], "traffic_r2": [0.999993622303009, 0.9999818801879883, 0.999962329864502]}, {"epoch_end": 5, "step": 2580, "eval_split": "validation", "eval_acc": 0.837, "eval_loss": 0.8042372314453125, "val_acc": 0.837, "val_loss": 0.8042372314453125}, {"step": 2600, "epoch": 6, "train_loss": 0.6540078520774841, "cos_r_negg": [0.5595158934593201, 0.02034205198287964, 0.29021185636520386], "cos_innovation_negg": [0.5595158934593201, 0.02034205198287964, 0.29021185636520386], "cos_apical_negg": [0.10577590018510818, -0.11867126077413559, 0.4242912232875824], "cos_Ac_negg": [0.7797212600708008, 0.10789945721626282, 0.5316884517669678], "r_norm": [0.5895704030990601, 0.5737064480781555, 0.6276414394378662], "traffic_norm": [17.487407684326172, 17.247398376464844, 16.936782836914062], "traffic_residual_norm": [0.0702124536037445, 0.11976464837789536, 0.1678057760000229], "traffic_r2": [0.9999833703041077, 0.9999510049819946, 0.9999006986618042]}, {"step": 2800, "epoch": 6, "train_loss": 0.3609694540500641, "cos_r_negg": [0.6106383800506592, 0.17043796181678772, 0.4355754256248474], "cos_innovation_negg": [0.6106383800506592, 0.17043796181678772, 0.4355754256248474], "cos_apical_negg": [-0.006261806935071945, 0.05776357650756836, 0.13708925247192383], "cos_Ac_negg": [0.8172305822372437, -0.04436216503381729, 0.47267499566078186], "r_norm": [0.6861857175827026, 0.68879234790802, 0.7961775064468384], "traffic_norm": [17.477018356323242, 17.240161895751953, 16.927045822143555], "traffic_residual_norm": [0.11959812045097351, 0.2072981745004654, 0.31765633821487427], "traffic_r2": [0.9999496340751648, 0.9998506307601929, 0.999640166759491]}, {"step": 3000, "epoch": 6, "train_loss": 0.6547532081604004, "cos_r_negg": [0.45445752143859863, 0.18406185507774353, 0.2742880582809448], "cos_innovation_negg": [0.45445752143859863, 0.18406185507774353, 0.2742880582809448], "cos_apical_negg": [-0.021003995090723038, 0.03730618208646774, -0.25263118743896484], "cos_Ac_negg": [0.810817301273346, 0.10658902674913406, 0.5444826483726501], "r_norm": [0.6248596906661987, 0.6036475896835327, 0.6530754566192627], "traffic_norm": [17.43729019165039, 17.185104370117188, 16.837642669677734], "traffic_residual_norm": [0.0774548277258873, 0.12743833661079407, 0.17501482367515564], "traffic_r2": [0.9999800324440002, 0.999944806098938, 0.9998918771743774]}, {"epoch_end": 6, "step": 3010, "eval_split": "validation", "eval_acc": 0.886, "eval_loss": 0.5806802734375, "val_acc": 0.886, "val_loss": 0.5806802734375}, {"step": 3200, "epoch": 7, "train_loss": 0.5762102603912354, "cos_r_negg": [0.4895506799221039, 0.18989309668540955, 0.48095130920410156], "cos_innovation_negg": [0.4895506799221039, 0.18989309668540955, 0.48095130920410156], "cos_apical_negg": [0.14099016785621643, -0.1820731908082962, 0.22787925601005554], "cos_Ac_negg": [0.8152327537536621, 0.06635388731956482, 0.6878893971443176], "r_norm": [0.5823264718055725, 0.54439777135849, 0.5774194002151489], "traffic_norm": [17.420066833496094, 17.167076110839844, 16.814498901367188], "traffic_residual_norm": [0.044465746730566025, 0.072133868932724, 0.10263136029243469], "traffic_r2": [0.9999935030937195, 0.9999823570251465, 0.9999627470970154]}, {"step": 3400, "epoch": 7, "train_loss": 0.6916836500167847, "cos_r_negg": [0.502947211265564, 0.13125552237033844, 0.46474841237068176], "cos_innovation_negg": [0.502947211265564, 0.13125552237033844, 0.46474841237068176], "cos_apical_negg": [0.026874210685491562, -0.18883080780506134, 0.16868281364440918], "cos_Ac_negg": [0.7748224139213562, 0.041467100381851196, 0.6990178227424622], "r_norm": [0.6004093885421753, 0.5600197315216064, 0.5936304330825806], "traffic_norm": [17.41574478149414, 17.1646785736084, 16.81524085998535], "traffic_residual_norm": [0.04988721013069153, 0.07664892077445984, 0.10309219360351562], "traffic_r2": [0.9999915361404419, 0.9999799728393555, 0.999962329864502]}, {"epoch_end": 7, "step": 3440, "eval_split": "validation", "eval_acc": 0.8982, "eval_loss": 0.5022089965820312, "val_acc": 0.8982, "val_loss": 0.5022089965820312}, {"step": 3600, "epoch": 8, "train_loss": 0.5976980924606323, "cos_r_negg": [0.5800508856773376, 0.12778928875923157, 0.4498378038406372], "cos_innovation_negg": [0.5800508856773376, 0.12778928875923157, 0.4498378038406372], "cos_apical_negg": [-0.012420812621712685, -0.03236222639679909, 0.24603310227394104], "cos_Ac_negg": [0.8600389957427979, 0.10952381044626236, 0.5945179462432861], "r_norm": [0.5586740970611572, 0.5209232568740845, 0.5484635829925537], "traffic_norm": [17.41336441040039, 17.15850257873535, 16.807130813598633], "traffic_residual_norm": [0.058861710131168365, 0.07791759073734283, 0.10280801355838776], "traffic_r2": [0.9999880790710449, 0.9999791979789734, 0.9999623894691467]}, {"step": 3800, "epoch": 8, "train_loss": 0.4620821177959442, "cos_r_negg": [0.6197863817214966, 0.2110237032175064, 0.622911810874939], "cos_innovation_negg": [0.6197863817214966, 0.2110237032175064, 0.622911810874939], "cos_apical_negg": [0.008259712718427181, -0.0663338154554367, 0.1671108603477478], "cos_Ac_negg": [0.8325635194778442, 0.025438254699110985, 0.7641738653182983], "r_norm": [0.580440878868103, 0.5276294946670532, 0.5516676902770996], "traffic_norm": [17.408632278442383, 17.15414047241211, 16.812026977539062], "traffic_residual_norm": [0.027406208217144012, 0.04242695868015289, 0.05264660343527794], "traffic_r2": [0.9999974966049194, 0.9999938607215881, 0.9999901652336121]}, {"epoch_end": 8, "step": 3870, "eval_split": "validation", "eval_acc": 0.9146, "eval_loss": 0.4460726867675781, "val_acc": 0.9146, "val_loss": 0.4460726867675781}, {"step": 4000, "epoch": 9, "train_loss": 0.7522757649421692, "cos_r_negg": [0.5793463587760925, 0.14711643755435944, 0.4812980592250824], "cos_innovation_negg": [0.5793463587760925, 0.14711643755435944, 0.4812980592250824], "cos_apical_negg": [0.06694866716861725, -0.02450501173734665, 0.20494775474071503], "cos_Ac_negg": [0.8495094180107117, 0.023249749094247818, 0.7183051705360413], "r_norm": [0.6857647895812988, 0.61922287940979, 0.6380773782730103], "traffic_norm": [17.409805297851562, 17.150482177734375, 16.806718826293945], "traffic_residual_norm": [0.018770046532154083, 0.03651503473520279, 0.04284099489450455], "traffic_r2": [0.9999988079071045, 0.9999954700469971, 0.9999935030937195]}, {"step": 4200, "epoch": 9, "train_loss": 0.25757378339767456, "cos_r_negg": [0.5706931948661804, 0.098822683095932, 0.42369234561920166], "cos_innovation_negg": [0.5706931948661804, 0.098822683095932, 0.42369234561920166], "cos_apical_negg": [0.12467954307794571, -0.23523901402950287, 0.45012128353118896], "cos_Ac_negg": [0.8595685958862305, 0.28387612104415894, 0.7369866371154785], "r_norm": [0.49969565868377686, 0.4620383679866791, 0.4772936701774597], "traffic_norm": [17.41164779663086, 17.153854370117188, 16.81137466430664], "traffic_residual_norm": [0.022504528984427452, 0.041739024221897125, 0.04604519158601761], "traffic_r2": [0.9999983310699463, 0.9999940991401672, 0.9999924898147583]}, {"epoch_end": 9, "step": 4300, "eval_split": "validation", "eval_acc": 0.8964, "eval_loss": 0.5536068420410156, "val_acc": 0.8964, "val_loss": 0.5536068420410156}, {"step": 4400, "epoch": 10, "train_loss": 0.501835823059082, "cos_r_negg": [0.49429064989089966, 0.2713915705680847, 0.5112995505332947], "cos_innovation_negg": [0.49429064989089966, 0.2713915705680847, 0.5112995505332947], "cos_apical_negg": [0.05058717727661133, -0.09609510749578476, 0.11945267021656036], "cos_Ac_negg": [0.7847882509231567, 0.11881616711616516, 0.7791651487350464], "r_norm": [0.4414997696876526, 0.4373307228088379, 0.44659000635147095], "traffic_norm": [17.410324096679688, 17.1561336517334, 16.812313079833984], "traffic_residual_norm": [0.04976958781480789, 0.09722775220870972, 0.09924943745136261], "traffic_r2": [0.9999918341636658, 0.9999678730964661, 0.9999651312828064]}, {"step": 4600, "epoch": 10, "train_loss": 0.30725884437561035, "cos_r_negg": [0.5565490126609802, 0.1994166374206543, 0.503318190574646], "cos_innovation_negg": [0.5565490126609802, 0.1994166374206543, 0.503318190574646], "cos_apical_negg": [0.04052366688847542, -0.07296238094568253, 0.1657426655292511], "cos_Ac_negg": [0.8320354223251343, 0.0888294205069542, 0.7558749914169312], "r_norm": [0.451987624168396, 0.4213142991065979, 0.43114355206489563], "traffic_norm": [17.414949417114258, 17.159423828125, 16.811906814575195], "traffic_residual_norm": [0.02187616191804409, 0.043929122388362885, 0.04504164308309555], "traffic_r2": [0.9999984502792358, 0.9999934434890747, 0.999992847442627]}, {"epoch_end": 10, "step": 4730, "eval_split": "validation", "eval_acc": 0.903, "eval_loss": 0.4980585144042969, "val_acc": 0.903, "val_loss": 0.4980585144042969}, {"step": 4800, "epoch": 11, "train_loss": 0.5252959132194519, "cos_r_negg": [0.5531732439994812, 0.09254436194896698, 0.5908687710762024], "cos_innovation_negg": [0.5531732439994812, 0.09254436194896698, 0.5908687710762024], "cos_apical_negg": [0.05509180575609207, 0.10173593461513519, 0.12224455922842026], "cos_Ac_negg": [0.8321958780288696, -0.022472640499472618, 0.7750086188316345], "r_norm": [0.5463076829910278, 0.5059767961502075, 0.5150413513183594], "traffic_norm": [17.410667419433594, 17.154705047607422, 16.806259155273438], "traffic_residual_norm": [0.02188124880194664, 0.04549102485179901, 0.04509928077459335], "traffic_r2": [0.9999983906745911, 0.9999929666519165, 0.9999927878379822]}, {"step": 5000, "epoch": 11, "train_loss": 0.4965777099132538, "cos_r_negg": [0.5020623207092285, 0.18193310499191284, 0.48738983273506165], "cos_innovation_negg": [0.5020623207092285, 0.18193310499191284, 0.48738983273506165], "cos_apical_negg": [0.09963833540678024, -0.23525866866111755, 0.18715454638004303], "cos_Ac_negg": [0.8630633354187012, 0.1479453146457672, 0.86327064037323], "r_norm": [0.5626107454299927, 0.5215040445327759, 0.5332711935043335], "traffic_norm": [17.418121337890625, 17.158828735351562, 16.81444549560547], "traffic_residual_norm": [0.021786585450172424, 0.04307079315185547, 0.04569429159164429], "traffic_r2": [0.9999984502792358, 0.9999936819076538, 0.9999926090240479]}, {"epoch_end": 11, "step": 5160, "eval_split": "validation", "eval_acc": 0.9284, "eval_loss": 0.35788525390625, "val_acc": 0.9284, "val_loss": 0.35788525390625}, {"step": 5200, "epoch": 12, "train_loss": 0.4684188961982727, "cos_r_negg": [0.6042249798774719, 0.13318586349487305, 0.5071821212768555], "cos_innovation_negg": [0.6042249798774719, 0.13318586349487305, 0.5071821212768555], "cos_apical_negg": [0.12943145632743835, 0.1890847086906433, 0.08558309823274612], "cos_Ac_negg": [0.866034984588623, 0.09617049992084503, 0.8305250406265259], "r_norm": [0.5909778475761414, 0.5637969970703125, 0.570620059967041], "traffic_norm": [17.40816879272461, 17.145950317382812, 16.803604125976562], "traffic_residual_norm": [0.034846238791942596, 0.07698634266853333, 0.07510589063167572], "traffic_r2": [0.9999960064888, 0.9999798536300659, 0.9999800324440002]}, {"step": 5400, "epoch": 12, "train_loss": 0.3720622658729553, "cos_r_negg": [0.6675350666046143, 0.1522524207830429, 0.540471613407135], "cos_innovation_negg": [0.6675350666046143, 0.1522524207830429, 0.540471613407135], "cos_apical_negg": [-0.005114714149385691, 0.0018856582464650273, -0.12498566508293152], "cos_Ac_negg": [0.8735774159431458, 0.15558862686157227, 0.795048713684082], "r_norm": [0.6438705325126648, 0.6051763296127319, 0.6094135046005249], "traffic_norm": [17.416885375976562, 17.156497955322266, 16.814006805419922], "traffic_residual_norm": [0.025340620428323746, 0.06207484379410744, 0.052292875945568085], "traffic_r2": [0.9999978542327881, 0.9999868869781494, 0.9999903440475464]}, {"epoch_end": 12, "step": 5590, "eval_split": "validation", "eval_acc": 0.9174, "eval_loss": 0.48078128662109376, "val_acc": 0.9174, "val_loss": 0.48078128662109376}, {"step": 5600, "epoch": 13, "train_loss": 0.34502092003822327, "cos_r_negg": [0.5777524709701538, 0.1718326359987259, 0.6020432114601135], "cos_innovation_negg": [0.5777524709701538, 0.1718326359987259, 0.6020432114601135], "cos_apical_negg": [0.0827050730586052, 0.08354153484106064, 0.31681710481643677], "cos_Ac_negg": [0.8358394503593445, 0.1796082705259323, 0.8264672160148621], "r_norm": [0.5701993703842163, 0.5481967926025391, 0.5530691146850586], "traffic_norm": [17.414005279541016, 17.15876007080078, 16.814056396484375], "traffic_residual_norm": [0.03890334442257881, 0.07976066321134567, 0.08165612071752548], "traffic_r2": [0.9999949932098389, 0.9999783635139465, 0.999976396560669]}, {"step": 5800, "epoch": 13, "train_loss": 0.45985403656959534, "cos_r_negg": [0.6803562045097351, 0.16359266638755798, 0.6556211709976196], "cos_innovation_negg": [0.6803562045097351, 0.16359266638755798, 0.6556211709976196], "cos_apical_negg": [0.16848795115947723, -0.13539376854896545, 0.3218902051448822], "cos_Ac_negg": [0.8722413778305054, 0.17943820357322693, 0.863325834274292], "r_norm": [0.6869567036628723, 0.6612423658370972, 0.6705272197723389], "traffic_norm": [17.40721893310547, 17.138935089111328, 16.796554565429688], "traffic_residual_norm": [0.05388820543885231, 0.11214785277843475, 0.10554473102092743], "traffic_r2": [0.9999904036521912, 0.9999572038650513, 0.9999605417251587]}, {"step": 6000, "epoch": 13, "train_loss": 0.428764283657074, "cos_r_negg": [0.5672508478164673, 0.21874532103538513, 0.4673140347003937], "cos_innovation_negg": [0.5672508478164673, 0.21874532103538513, 0.4673140347003937], "cos_apical_negg": [0.05228808522224426, -0.14458677172660828, 0.17067287862300873], "cos_Ac_negg": [0.8553906083106995, 0.29516440629959106, 0.8008654713630676], "r_norm": [0.4754239320755005, 0.46945974230766296, 0.4648358225822449], "traffic_norm": [17.412996292114258, 17.15575408935547, 16.81039810180664], "traffic_residual_norm": [0.04055044800043106, 0.08553362637758255, 0.07173678278923035], "traffic_r2": [0.9999945759773254, 0.9999751448631287, 0.9999817609786987]}, {"epoch_end": 13, "step": 6020, "eval_split": "validation", "eval_acc": 0.9028, "eval_loss": 0.43316475219726563, "val_acc": 0.9028, "val_loss": 0.43316475219726563}, {"step": 6200, "epoch": 14, "train_loss": 0.1387675404548645, "cos_r_negg": [0.5873503088951111, 0.21353112161159515, 0.5083860754966736], "cos_innovation_negg": [0.5873503088951111, 0.21353112161159515, 0.5083860754966736], "cos_apical_negg": [-0.0035195620730519295, 0.1403542011976242, -0.09898240119218826], "cos_Ac_negg": [0.8445125818252563, 0.19162985682487488, 0.8598094582557678], "r_norm": [0.48700636625289917, 0.4878460168838501, 0.5007750988006592], "traffic_norm": [17.411991119384766, 17.15509796142578, 16.809833526611328], "traffic_residual_norm": [0.05385788902640343, 0.10869280993938446, 0.11299280822277069], "traffic_r2": [0.9999904036521912, 0.9999598860740662, 0.9999548196792603]}, {"step": 6400, "epoch": 14, "train_loss": 0.540920078754425, "cos_r_negg": [0.6114809513092041, 0.27895480394363403, 0.5980128049850464], "cos_innovation_negg": [0.6114809513092041, 0.27895480394363403, 0.5980128049850464], "cos_apical_negg": [0.06280726194381714, -0.1629868447780609, 0.08268605172634125], "cos_Ac_negg": [0.8584685921669006, 0.22659403085708618, 0.9008587002754211], "r_norm": [0.4299374222755432, 0.428289532661438, 0.4333003759384155], "traffic_norm": [17.417116165161133, 17.16429901123047, 16.814380645751953], "traffic_residual_norm": [0.04511423408985138, 0.08848651498556137, 0.0920269638299942], "traffic_r2": [0.9999932646751404, 0.9999734163284302, 0.999970018863678]}, {"epoch_end": 14, "step": 6450, "eval_split": "validation", "eval_acc": 0.924, "eval_loss": 0.36928642578125, "val_acc": 0.924, "val_loss": 0.36928642578125}], "final": {"eval_split": "validation", "eval_acc": 0.924, "eval_loss": 0.36928642578125, "wall_s": 22.91165566444397, "val_acc": 0.924, "val_loss": 0.36928642578125, "cos_r_negg": [0.69914710521698, 0.2873772382736206, 0.6352859139442444], "cos_innovation_negg": [0.69914710521698, 0.2873772382736206, 0.6352859139442444], "cos_apical_negg": [0.003709297627210617, -0.0834898054599762, 0.06427844613790512], "cos_Ac_negg": [0.8809698820114136, 0.28216326236724854, 0.9182184934616089], "traffic_norm": [17.414810180664062, 17.157764434814453, 16.815170288085938], "traffic_residual_norm": [0.04845082014799118, 0.0937943235039711, 0.11335664987564087], "traffic_r2": [0.9999921321868896, 0.9999698996543884, 0.9999542832374573]}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 22.696565628051758, "diagnostics_wall_s": 0.5023319721221924, "inline_diagnostics_wall_s": 0.36158204078674316, "optimizer_wall_s_excluding_inline_diagnostics": 22.334983587265015, "evaluation_wall_s": 0.07082080841064453}, "cost": {"train_steps": 6450, "ordinary_training_forward_examples": 825000, "predictor_warmup_forward_examples": 0, "perturbation_events": 1613, "calibration_batch_loss_evaluations": 51616, "calibration_example_loss_evaluations": 6606848, "calibration_forward_equivalent_examples": 6813312.0, "max_perturbation_batch_expansion": 32}}
|