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": 0, "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_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": [-0.01463812030851841, 0.017589397728443146, 0.02251502498984337], "cos_innovation_negg": [-0.01463812030851841, 0.017589397728443146, 0.02251502498984337], "cos_apical_negg": [-0.014638984575867653, 0.017571203410625458, 0.022482598200440407], "cos_Ac_negg": [-0.010890763252973557, 0.022868484258651733, 0.034097686409950256], "r_norm": [6.004908084869385, 5.742959976196289, 5.720208168029785], "traffic_norm": [3.153928518295288, 3.1129655838012695, 3.0246775150299072], "traffic_residual_norm": [3.147808790206909, 3.1060547828674316, 3.0169830322265625], "traffic_r2": [0.0038781166076660156, 0.0044364333152771, 0.005082309246063232]}, {"step": 200, "epoch": 0, "train_loss": 1.5508687496185303, "cos_r_negg": [0.07228126376867294, 0.053495120257139206, 0.032779667526483536], "cos_innovation_negg": [0.07228126376867294, 0.053495120257139206, 0.032779667526483536], "cos_apical_negg": [-0.15568238496780396, 0.0339798666536808, 0.02800610288977623], "cos_Ac_negg": [0.8266512751579285, 0.010320935398340225, 0.2914373278617859], "r_norm": [5.325007915496826, 5.01407527923584, 4.870715141296387], "traffic_norm": [7.263195037841797, 7.195355415344238, 7.1283979415893555], "traffic_residual_norm": [4.6634931564331055, 4.414658069610596, 4.197736740112305], "traffic_r2": [0.5877441167831421, 0.623568058013916, 0.6532285809516907]}, {"step": 400, "epoch": 0, "train_loss": 1.0616469383239746, "cos_r_negg": [0.17033344507217407, -0.01346967089921236, 0.19399161636829376], "cos_innovation_negg": [0.17033344507217407, -0.01346967089921236, 0.19399161636829376], "cos_apical_negg": [-0.051460810005664825, 0.03338727727532387, 0.17543232440948486], "cos_Ac_negg": [0.8281346559524536, -0.31631898880004883, 0.1701940894126892], "r_norm": [3.6373519897460938, 3.3676514625549316, 3.1885645389556885], "traffic_norm": [7.26746940612793, 7.20536470413208, 7.139492034912109], "traffic_residual_norm": [3.0619940757751465, 2.8298468589782715, 2.6530561447143555], "traffic_r2": [0.8224821090698242, 0.8457545042037964, 0.8619117140769958]}, {"epoch_end": 0, "step": 430, "eval_split": "validation", "eval_acc": 0.8486, "eval_loss": 0.8106024780273438, "val_acc": 0.8486, "val_loss": 0.8106024780273438}, {"step": 600, "epoch": 1, "train_loss": 1.0354970693588257, "cos_r_negg": [0.13826173543930054, 0.217519149184227, -0.17333851754665375], "cos_innovation_negg": [0.13826173543930054, 0.217519149184227, -0.17333851754665375], "cos_apical_negg": [0.01708313450217247, 0.35793426632881165, -0.20266306400299072], "cos_Ac_negg": [0.7329038381576538, 0.0792110487818718, 0.16209250688552856], "r_norm": [2.811591148376465, 2.527536630630493, 2.4366979598999023], "traffic_norm": [7.2631940841674805, 7.198980331420898, 7.1351118087768555], "traffic_residual_norm": [2.0135762691497803, 1.8294395208358765, 1.7120890617370605], "traffic_r2": [0.9231435060501099, 0.9354206919670105, 0.9424223899841309]}, {"step": 800, "epoch": 1, "train_loss": 3.982675552368164, "cos_r_negg": [0.22162571549415588, -0.15366297960281372, 0.2753320336341858], "cos_innovation_negg": [0.22162571549415588, -0.15366297960281372, 0.2753320336341858], "cos_apical_negg": [0.13582433760166168, -0.2062576711177826, 0.20396766066551208], "cos_Ac_negg": [0.6807984113693237, 0.6337816119194031, 0.3501881957054138], "r_norm": [3.102888345718384, 2.7690987586975098, 2.708944320678711], "traffic_norm": [7.24592399597168, 7.172102928161621, 7.108985424041748], "traffic_residual_norm": [1.3236088752746582, 1.1932039260864258, 1.1360384225845337], "traffic_r2": [0.9666319489479065, 0.9723213911056519, 0.9744608998298645]}, {"epoch_end": 1, "step": 860, "eval_split": "validation", "eval_acc": 0.8458, "eval_loss": 1.234580908203125, "val_acc": 0.8458, "val_loss": 1.234580908203125}, {"step": 1000, "epoch": 2, "train_loss": 0.5359265804290771, "cos_r_negg": [0.12372303009033203, 0.009569555521011353, 0.24362048506736755], "cos_innovation_negg": [0.12372303009033203, 0.009569555521011353, 0.24362048506736755], "cos_apical_negg": [-0.0007982142269611359, -0.020099354907870293, 0.17790642380714417], "cos_Ac_negg": [0.7210421562194824, 0.08570874482393265, 0.3373867869377136], "r_norm": [1.405590534210205, 1.2549983263015747, 1.293698787689209], "traffic_norm": [7.196039199829102, 7.106594085693359, 7.020566463470459], "traffic_residual_norm": [0.8809390068054199, 0.8133043050765991, 0.8619447946548462], "traffic_r2": [0.9850131869316101, 0.9868994951248169, 0.9849179983139038]}, {"step": 1200, "epoch": 2, "train_loss": 0.8847033977508545, "cos_r_negg": [0.21409422159194946, 0.1319613754749298, 0.23296773433685303], "cos_innovation_negg": [0.21409422159194946, 0.1319613754749298, 0.23296773433685303], "cos_apical_negg": [0.16573300957679749, 0.10986023396253586, 0.22249357402324677], "cos_Ac_negg": [0.8340822458267212, 0.059209637343883514, 0.3570753335952759], "r_norm": [1.4880552291870117, 1.3188538551330566, 1.3945643901824951], "traffic_norm": [7.169696807861328, 7.058465003967285, 6.9568095207214355], "traffic_residual_norm": [0.5808515548706055, 0.5531100034713745, 0.6467223763465881], "traffic_r2": [0.9934361577033997, 0.9938578605651855, 0.99135422706604]}, {"epoch_end": 2, "step": 1290, "eval_split": "validation", "eval_acc": 0.8872, "eval_loss": 0.6320192993164062, "val_acc": 0.8872, "val_loss": 0.6320192993164062}, {"step": 1400, "epoch": 3, "train_loss": 0.551588773727417, "cos_r_negg": [0.19596001505851746, 0.0704333558678627, 0.05814385414123535], "cos_innovation_negg": [0.19596001505851746, 0.0704333558678627, 0.05814385414123535], "cos_apical_negg": [0.02165362983942032, 0.0819244459271431, 0.03671260550618172], "cos_Ac_negg": [0.8291226625442505, 0.2539723813533783, 0.33063429594039917], "r_norm": [0.9528400897979736, 0.8659090995788574, 0.9427449703216553], "traffic_norm": [7.156251907348633, 7.041117191314697, 6.940891265869141], "traffic_residual_norm": [0.3783615231513977, 0.3773084878921509, 0.46533697843551636], "traffic_r2": [0.9971983432769775, 0.9971156716346741, 0.9955013394355774]}, {"step": 1600, "epoch": 3, "train_loss": 0.9900546073913574, "cos_r_negg": [0.12778544425964355, 0.08698683232069016, 0.02972552180290222], "cos_innovation_negg": [0.12778544425964355, 0.08698683232069016, 0.02972552180290222], "cos_apical_negg": [0.13360419869422913, 0.03751580789685249, 0.021252267062664032], "cos_Ac_negg": [0.8157199621200562, 0.09548740088939667, 0.19814276695251465], "r_norm": [0.877723753452301, 0.8437098264694214, 0.8777360320091248], "traffic_norm": [7.085708141326904, 6.950583457946777, 6.880371570587158], "traffic_residual_norm": [0.374426007270813, 0.4240994155406952, 0.4689444303512573], "traffic_r2": [0.9972051382064819, 0.9962748289108276, 0.9953493475914001]}, {"epoch_end": 3, "step": 1720, "eval_split": "validation", "eval_acc": 0.8868, "eval_loss": 0.5995822570800782, "val_acc": 0.8868, "val_loss": 0.5995822570800782}, {"step": 1800, "epoch": 4, "train_loss": 0.5943303108215332, "cos_r_negg": [0.11447359621524811, 0.13758036494255066, 0.10069657862186432], "cos_innovation_negg": [0.11447359621524811, 0.13758036494255066, 0.10069657862186432], "cos_apical_negg": [0.10095115005970001, -0.19248193502426147, -0.0038931332528591156], "cos_Ac_negg": [0.7611770033836365, 0.05476849526166916, 0.3354599177837372], "r_norm": [1.0108505487442017, 0.9698748588562012, 1.0884249210357666], "traffic_norm": [7.003476142883301, 6.865662574768066, 6.7824907302856445], "traffic_residual_norm": [0.5134180188179016, 0.5584638118743896, 0.6841188669204712], "traffic_r2": [0.9946166276931763, 0.9933614730834961, 0.9897682666778564]}, {"step": 2000, "epoch": 4, "train_loss": 0.5691625475883484, "cos_r_negg": [0.338296115398407, 0.02679941989481449, 0.054444968700408936], "cos_innovation_negg": [0.338296115398407, 0.02679941989481449, 0.054444968700408936], "cos_apical_negg": [-0.029294190928339958, 0.16796837747097015, 0.1109975054860115], "cos_Ac_negg": [0.8217772245407104, 0.39209651947021484, 0.47639331221580505], "r_norm": [0.9760621190071106, 0.8799145221710205, 1.0283770561218262], "traffic_norm": [6.977781772613525, 6.842925548553467, 6.735273361206055], "traffic_residual_norm": [0.446010947227478, 0.43516048789024353, 0.6062746047973633], "traffic_r2": [0.9959135055541992, 0.9959545135498047, 0.9918966293334961]}, {"epoch_end": 4, "step": 2150, "eval_split": "validation", "eval_acc": 0.884, "eval_loss": 0.7183745361328125, "val_acc": 0.884, "val_loss": 0.7183745361328125}, {"step": 2200, "epoch": 5, "train_loss": 0.3105159103870392, "cos_r_negg": [0.28424686193466187, 0.15930767357349396, 0.129553884267807], "cos_innovation_negg": [0.28424686193466187, 0.15930767357349396, 0.129553884267807], "cos_apical_negg": [0.09159699082374573, 0.04148861765861511, -0.008761502802371979], "cos_Ac_negg": [0.7986215353012085, 0.4152204394340515, 0.340450644493103], "r_norm": [0.8739050030708313, 0.7918081283569336, 0.9068015813827515], "traffic_norm": [6.972182273864746, 6.831573963165283, 6.729757785797119], "traffic_residual_norm": [0.2971452474594116, 0.3141511380672455, 0.44768524169921875], "traffic_r2": [0.9981803894042969, 0.9978797435760498, 0.9955704212188721]}, {"step": 2400, "epoch": 5, "train_loss": 0.9285709857940674, "cos_r_negg": [0.32201722264289856, 0.11343815177679062, 0.0033456701785326004], "cos_innovation_negg": [0.32201722264289856, 0.11343815177679062, 0.0033456701785326004], "cos_apical_negg": [0.08244777470827103, -0.03869384899735451, 0.12109512090682983], "cos_Ac_negg": [0.8274989128112793, 0.2648058831691742, 0.39783352613449097], "r_norm": [0.8560277223587036, 0.8250517845153809, 0.9184103012084961], "traffic_norm": [6.945221900939941, 6.799282073974609, 6.696070194244385], "traffic_residual_norm": [0.27373796701431274, 0.3515825569629669, 0.4580916166305542], "traffic_r2": [0.9984279274940491, 0.9972954988479614, 0.9953052997589111]}, {"epoch_end": 5, "step": 2580, "eval_split": "validation", "eval_acc": 0.8998, "eval_loss": 0.57167587890625, "val_acc": 0.8998, "val_loss": 0.57167587890625}, {"step": 2600, "epoch": 6, "train_loss": 0.6966447830200195, "cos_r_negg": [0.21739566326141357, 0.151336669921875, 0.0411028154194355], "cos_innovation_negg": [0.21739566326141357, 0.151336669921875, 0.0411028154194355], "cos_apical_negg": [0.07139886915683746, 0.17279762029647827, 0.11283290386199951], "cos_Ac_negg": [0.764687180519104, 0.5626650452613831, 0.5846080780029297], "r_norm": [1.1176000833511353, 1.0937459468841553, 1.1384482383728027], "traffic_norm": [6.8367919921875, 6.671014785766602, 6.600320816040039], "traffic_residual_norm": [0.5032526254653931, 0.6023764610290527, 0.6459999680519104], "traffic_r2": [0.9945781230926514, 0.9918421506881714, 0.9904189109802246]}, {"step": 2800, "epoch": 6, "train_loss": 0.25261190533638, "cos_r_negg": [0.2910735011100769, 0.13151437044143677, 0.1675054132938385], "cos_innovation_negg": [0.2910735011100769, 0.13151437044143677, 0.1675054132938385], "cos_apical_negg": [0.07826834172010422, 0.17271894216537476, -0.01945815607905388], "cos_Ac_negg": [0.7982432246208191, 0.5282488465309143, 0.574988603591919], "r_norm": [1.1142640113830566, 1.0749049186706543, 1.1106771230697632], "traffic_norm": [6.812582969665527, 6.6329345703125, 6.581552982330322], "traffic_residual_norm": [0.4092510938644409, 0.5014917850494385, 0.539832592010498], "traffic_r2": [0.9963819980621338, 0.994276225566864, 0.9932637810707092]}, {"step": 3000, "epoch": 6, "train_loss": 0.5639420747756958, "cos_r_negg": [0.32765448093414307, 0.21147888898849487, -0.15396878123283386], "cos_innovation_negg": [0.32765448093414307, 0.21147888898849487, -0.15396878123283386], "cos_apical_negg": [0.0315646193921566, 0.05242437869310379, 0.3159368932247162], "cos_Ac_negg": [0.8308868408203125, 0.4556877613067627, 0.6610067486763], "r_norm": [0.8922561407089233, 0.8950119018554688, 0.9707379341125488], "traffic_norm": [6.773487091064453, 6.579384803771973, 6.519960403442383], "traffic_residual_norm": [0.3677035868167877, 0.46822622418403625, 0.5547679662704468], "traffic_r2": [0.997049868106842, 0.9949305057525635, 0.9927533268928528]}, {"epoch_end": 6, "step": 3010, "eval_split": "validation", "eval_acc": 0.913, "eval_loss": 0.45731302490234377, "val_acc": 0.913, "val_loss": 0.45731302490234377}, {"step": 3200, "epoch": 7, "train_loss": 0.3463449478149414, "cos_r_negg": [0.26451486349105835, 0.1616654098033905, 0.10552624613046646], "cos_innovation_negg": [0.26451486349105835, 0.1616654098033905, 0.10552624613046646], "cos_apical_negg": [0.09542642533779144, 0.0309453085064888, 0.15314142405986786], "cos_Ac_negg": [0.7925080060958862, 0.5350390076637268, 0.6269146800041199], "r_norm": [0.7570170164108276, 0.7622001767158508, 0.8234515190124512], "traffic_norm": [6.756630897521973, 6.554869651794434, 6.493659496307373], "traffic_residual_norm": [0.29279136657714844, 0.3836429715156555, 0.4536961317062378], "traffic_r2": [0.9981054067611694, 0.99653559923172, 0.9951028823852539]}, {"step": 3400, "epoch": 7, "train_loss": 0.816376805305481, "cos_r_negg": [0.3080472946166992, 0.24988366663455963, 0.1405317485332489], "cos_innovation_negg": [0.3080472946166992, 0.24988366663455963, 0.1405317485332489], "cos_apical_negg": [0.08412647247314453, 0.014220265671610832, 0.11704225838184357], "cos_Ac_negg": [0.8150067329406738, 0.659591555595398, 0.6062873601913452], "r_norm": [0.8758167624473572, 0.9565331339836121, 1.0414761304855347], "traffic_norm": [6.715968132019043, 6.473941802978516, 6.406770706176758], "traffic_residual_norm": [0.3495597839355469, 0.5385509133338928, 0.6362298130989075], "traffic_r2": [0.9972625970840454, 0.9930420517921448, 0.9900953769683838]}, {"epoch_end": 7, "step": 3440, "eval_split": "validation", "eval_acc": 0.9086, "eval_loss": 0.449579443359375, "val_acc": 0.9086, "val_loss": 0.449579443359375}, {"step": 3600, "epoch": 8, "train_loss": 0.49933868646621704, "cos_r_negg": [0.42373815178871155, 0.09484478831291199, 0.173893541097641], "cos_innovation_negg": [0.42373815178871155, 0.09484478831291199, 0.173893541097641], "cos_apical_negg": [0.055165790021419525, 0.1699301302433014, 0.0831654816865921], "cos_Ac_negg": [0.8366053104400635, 0.6093257069587708, 0.6128814816474915], "r_norm": [0.8561396598815918, 0.8950052261352539, 0.9880412220954895], "traffic_norm": [6.685146808624268, 6.4396071434021, 6.338421821594238], "traffic_residual_norm": [0.30797821283340454, 0.4523734450340271, 0.5570394992828369], "traffic_r2": [0.9978747963905334, 0.9950633645057678, 0.9922757148742676]}, {"step": 3800, "epoch": 8, "train_loss": 0.6081211566925049, "cos_r_negg": [0.40183359384536743, 0.1580265611410141, 0.22251546382904053], "cos_innovation_negg": [0.40183359384536743, 0.1580265611410141, 0.22251546382904053], "cos_apical_negg": [0.1328115165233612, 0.0969180315732956, 0.18744200468063354], "cos_Ac_negg": [0.8258247375488281, 0.4634578227996826, 0.5871021747589111], "r_norm": [0.9141162633895874, 0.9057572484016418, 0.9898527264595032], "traffic_norm": [6.672176361083984, 6.424254417419434, 6.325205326080322], "traffic_residual_norm": [0.2865261435508728, 0.39016637206077576, 0.49217671155929565], "traffic_r2": [0.9981459975242615, 0.9963072538375854, 0.9939413070678711]}, {"epoch_end": 8, "step": 3870, "eval_split": "validation", "eval_acc": 0.8978, "eval_loss": 0.5712450866699219, "val_acc": 0.8978, "val_loss": 0.5712450866699219}, {"step": 4000, "epoch": 9, "train_loss": 0.32439661026000977, "cos_r_negg": [0.2686125636100769, 0.11267878115177155, 0.01995042711496353], "cos_innovation_negg": [0.2686125636100769, 0.11267878115177155, 0.01995042711496353], "cos_apical_negg": [0.009143242612481117, 0.07863607257604599, 0.10745428502559662], "cos_Ac_negg": [0.7915383577346802, 0.4967081546783447, 0.36312493681907654], "r_norm": [0.9756293296813965, 0.892959475517273, 0.9966975450515747], "traffic_norm": [6.582466125488281, 6.370530128479004, 6.251289367675781], "traffic_residual_norm": [0.46137306094169617, 0.4583509564399719, 0.5795546770095825], "traffic_r2": [0.9950730204582214, 0.9948176741600037, 0.9914010763168335]}, {"step": 4200, "epoch": 9, "train_loss": 0.46830981969833374, "cos_r_negg": [0.493760883808136, 0.19700747728347778, 0.2461394965648651], "cos_innovation_negg": [0.493760883808136, 0.19700747728347778, 0.2461394965648651], "cos_apical_negg": [0.03864724561572075, -0.010183300822973251, 0.10342079401016235], "cos_Ac_negg": [0.8227550983428955, 0.6218793988227844, 0.5516298413276672], "r_norm": [1.0196621417999268, 0.8858048915863037, 0.9543749094009399], "traffic_norm": [6.5331220626831055, 6.334781646728516, 6.221335411071777], "traffic_residual_norm": [0.49812519550323486, 0.4374581575393677, 0.5213135480880737], "traffic_r2": [0.994179368019104, 0.9952237010002136, 0.9929742217063904]}, {"epoch_end": 9, "step": 4300, "eval_split": "validation", "eval_acc": 0.9136, "eval_loss": 0.3987699951171875, "val_acc": 0.9136, "val_loss": 0.3987699951171875}, {"step": 4400, "epoch": 10, "train_loss": 0.3301096558570862, "cos_r_negg": [0.4546046257019043, 0.21519474685192108, 0.16234444081783295], "cos_innovation_negg": [0.4546046257019043, 0.21519474685192108, 0.16234444081783295], "cos_apical_negg": [0.19426754117012024, 0.19386982917785645, 0.4661407768726349], "cos_Ac_negg": [0.8350080251693726, 0.4292094111442566, 0.30175599455833435], "r_norm": [1.3920034170150757, 1.2485058307647705, 1.2554576396942139], "traffic_norm": [6.473640441894531, 6.280667304992676, 6.1932373046875], "traffic_residual_norm": [0.5613882541656494, 0.550182044506073, 0.5756829977035522], "traffic_r2": [0.9924544095993042, 0.9923142194747925, 0.9913357496261597]}, {"step": 4600, "epoch": 10, "train_loss": 0.2177892029285431, "cos_r_negg": [0.43049901723861694, 0.09367294609546661, 0.3068027198314667], "cos_innovation_negg": [0.43049901723861694, 0.09367294609546661, 0.3068027198314667], "cos_apical_negg": [0.03575428947806358, 0.18670102953910828, -0.30254918336868286], "cos_Ac_negg": [0.831981897354126, 0.6193559765815735, 0.6878318786621094], "r_norm": [1.1869381666183472, 1.191392183303833, 1.1144152879714966], "traffic_norm": [6.398405075073242, 6.189604759216309, 6.138981819152832], "traffic_residual_norm": [0.611641526222229, 0.7173423767089844, 0.6420388221740723], "traffic_r2": [0.9908438324928284, 0.9865489602088928, 0.9890547394752502]}, {"epoch_end": 10, "step": 4730, "eval_split": "validation", "eval_acc": 0.9212, "eval_loss": 0.34688682861328124, "val_acc": 0.9212, "val_loss": 0.34688682861328124}, {"step": 4800, "epoch": 11, "train_loss": 0.4509929120540619, "cos_r_negg": [0.4069407880306244, 0.19624674320220947, 0.1709219217300415], "cos_innovation_negg": [0.4069407880306244, 0.19624674320220947, 0.1709219217300415], "cos_apical_negg": [0.06834903359413147, 5.653686821460724e-05, 0.2694782614707947], "cos_Ac_negg": [0.8009541630744934, 0.4917171597480774, 0.4546341001987457], "r_norm": [1.1119400262832642, 1.188272476196289, 1.010162115097046], "traffic_norm": [6.331110000610352, 6.0775146484375, 6.09089469909668], "traffic_residual_norm": [0.5341922044754028, 0.7213125824928284, 0.5285744667053223], "traffic_r2": [0.9928804636001587, 0.9859123826026917, 0.9924682378768921]}, {"step": 5000, "epoch": 11, "train_loss": 0.3528995215892792, "cos_r_negg": [0.49249017238616943, 0.14166752994060516, 0.23823389410972595], "cos_innovation_negg": [0.49249017238616943, 0.14166752994060516, 0.23823389410972595], "cos_apical_negg": [0.05663485452532768, 0.13191692531108856, 0.0035408069379627705], "cos_Ac_negg": [0.8609826564788818, 0.6951887607574463, 0.7451362609863281], "r_norm": [1.027008056640625, 1.0008066892623901, 0.9066281318664551], "traffic_norm": [6.299563884735107, 6.057082176208496, 6.066628932952881], "traffic_residual_norm": [0.4799019694328308, 0.5410213470458984, 0.44213032722473145], "traffic_r2": [0.9941964149475098, 0.9920216798782349, 0.9946864247322083]}, {"epoch_end": 11, "step": 5160, "eval_split": "validation", "eval_acc": 0.9156, "eval_loss": 0.42706893920898437, "val_acc": 0.9156, "val_loss": 0.42706893920898437}, {"step": 5200, "epoch": 12, "train_loss": 0.2754926085472107, "cos_r_negg": [0.47034838795661926, 0.17571698129177094, 0.22291897237300873], "cos_innovation_negg": [0.47034838795661926, 0.17571698129177094, 0.22291897237300873], "cos_apical_negg": [0.04967786371707916, 0.14751748740673065, 0.1265617161989212], "cos_Ac_negg": [0.794133186340332, 0.7314112186431885, 0.5936719179153442], "r_norm": [0.9099260568618774, 0.8166990280151367, 0.8028416633605957], "traffic_norm": [6.272687911987305, 6.041177749633789, 6.029552459716797], "traffic_residual_norm": [0.40995603799819946, 0.3898113965988159, 0.37430065870285034], "traffic_r2": [0.9957283735275269, 0.995836079120636, 0.9961429238319397]}, {"step": 5400, "epoch": 12, "train_loss": 0.2582134008407593, "cos_r_negg": [0.4793291687965393, 0.1726164072751999, 0.2218720018863678], "cos_innovation_negg": [0.4793291687965393, 0.1726164072751999, 0.2218720018863678], "cos_apical_negg": [-0.014152652584016323, 0.11757677793502808, 0.13132703304290771], "cos_Ac_negg": [0.8267862796783447, 0.7123868465423584, 0.5983591079711914], "r_norm": [0.8260752558708191, 0.743263840675354, 0.7598657011985779], "traffic_norm": [6.260848045349121, 6.028813362121582, 6.002175331115723], "traffic_residual_norm": [0.29243385791778564, 0.2867127060890198, 0.3112105429172516], "traffic_r2": [0.9978112578392029, 0.9977290034294128, 0.9973055124282837]}, {"epoch_end": 12, "step": 5590, "eval_split": "validation", "eval_acc": 0.9236, "eval_loss": 0.3905919494628906, "val_acc": 0.9236, "val_loss": 0.3905919494628906}, {"step": 5600, "epoch": 13, "train_loss": 0.34401923418045044, "cos_r_negg": [0.5005125999450684, 0.17475861310958862, 0.2482946813106537], "cos_innovation_negg": [0.5005125999450684, 0.17475861310958862, 0.2482946813106537], "cos_apical_negg": [0.12962734699249268, 0.142244353890419, 0.2219916135072708], "cos_Ac_negg": [0.7689594030380249, 0.6959550976753235, 0.6565076112747192], "r_norm": [1.0125975608825684, 0.9138861894607544, 0.8565850257873535], "traffic_norm": [6.198980808258057, 5.967781066894531, 5.944883823394775], "traffic_residual_norm": [0.4122450649738312, 0.39587152004241943, 0.3392345905303955], "traffic_r2": [0.9955766201019287, 0.9955990314483643, 0.996743381023407]}, {"step": 5800, "epoch": 13, "train_loss": 0.27206873893737793, "cos_r_negg": [0.4102433919906616, 0.18244415521621704, 0.2359255701303482], "cos_innovation_negg": [0.4102433919906616, 0.18244415521621704, 0.2359255701303482], "cos_apical_negg": [0.10015351325273514, 0.10827833414077759, 0.035604916512966156], "cos_Ac_negg": [0.822207510471344, 0.7231507897377014, 0.759455680847168], "r_norm": [1.0721704959869385, 0.8982479572296143, 0.8794547319412231], "traffic_norm": [6.220501899719238, 5.981916427612305, 5.928659439086914], "traffic_residual_norm": [0.4696327745914459, 0.37033283710479736, 0.34932613372802734], "traffic_r2": [0.9942964911460876, 0.9961659908294678, 0.9965271949768066]}, {"step": 6000, "epoch": 13, "train_loss": 0.36051613092422485, "cos_r_negg": [0.45392414927482605, 0.2472323775291443, 0.19182196259498596], "cos_innovation_negg": [0.45392414927482605, 0.2472323775291443, 0.19182196259498596], "cos_apical_negg": [0.027701396495103836, 0.16805726289749146, 0.12057417631149292], "cos_Ac_negg": [0.8448233008384705, 0.7209128141403198, 0.7393832802772522], "r_norm": [1.034424066543579, 0.8247294425964355, 0.8287858963012695], "traffic_norm": [6.151833534240723, 5.947857856750488, 5.895804405212402], "traffic_residual_norm": [0.5493430495262146, 0.3990558981895447, 0.4061514437198639], "traffic_r2": [0.9920240640640259, 0.9954931735992432, 0.9952459335327148]}, {"epoch_end": 13, "step": 6020, "eval_split": "validation", "eval_acc": 0.9178, "eval_loss": 0.37428397216796877, "val_acc": 0.9178, "val_loss": 0.37428397216796877}, {"step": 6200, "epoch": 14, "train_loss": 0.17265759408473969, "cos_r_negg": [0.456135094165802, 0.21867743134498596, 0.15639203786849976], "cos_innovation_negg": [0.456135094165802, 0.21867743134498596, 0.15639203786849976], "cos_apical_negg": [0.08768941462039948, 0.01694190315902233, 0.16049444675445557], "cos_Ac_negg": [0.8375599384307861, 0.7194112539291382, 0.6980650424957275], "r_norm": [0.8958156108856201, 0.7566332817077637, 0.8599330186843872], "traffic_norm": [6.093433380126953, 5.887142181396484, 5.820982933044434], "traffic_residual_norm": [0.4681093692779541, 0.3818332552909851, 0.4964945912361145], "traffic_r2": [0.9940983057022095, 0.9957934021949768, 0.9927249550819397]}, {"step": 6400, "epoch": 14, "train_loss": 0.49075835943222046, "cos_r_negg": [0.5150485038757324, 0.18084299564361572, 0.18392734229564667], "cos_innovation_negg": [0.5150485038757324, 0.18084299564361572, 0.18392734229564667], "cos_apical_negg": [0.10971377789974213, 0.1601342260837555, 0.26839298009872437], "cos_Ac_negg": [0.8201638460159302, 0.6996762752532959, 0.5704896450042725], "r_norm": [1.3213073015213013, 0.9568098783493042, 1.1107807159423828], "traffic_norm": [6.004549980163574, 5.817584991455078, 5.732410430908203], "traffic_residual_norm": [0.7672569751739502, 0.4483259320259094, 0.625511646270752], "traffic_r2": [0.9836724996566772, 0.9940611124038696, 0.988092839717865]}, {"epoch_end": 14, "step": 6450, "eval_split": "validation", "eval_acc": 0.915, "eval_loss": 0.37882452392578125, "val_acc": 0.915, "val_loss": 0.37882452392578125}], "final": {"eval_split": "validation", "eval_acc": 0.915, "eval_loss": 0.37882452392578125, "wall_s": 22.04463005065918, "val_acc": 0.915, "val_loss": 0.37882452392578125, "cos_r_negg": [0.4430997371673584, 0.10377670079469681, 0.21606779098510742], "cos_innovation_negg": [0.4430997371673584, 0.10377670079469681, 0.21606779098510742], "cos_apical_negg": [0.14113739132881165, 0.06739220023155212, 0.1979568898677826], "cos_Ac_negg": [0.791334867477417, 0.637399435043335, 0.6359128952026367], "traffic_norm": [5.999402046203613, 5.809797286987305, 5.719898223876953], "traffic_residual_norm": [0.698007345199585, 0.41442620754241943, 0.5782809257507324], "traffic_r2": [0.9864634275436401, 0.9949111342430115, 0.9897767901420593]}, "timing": {"warmup_wall_s": 0.0, "training_loop_wall_s": 21.794094562530518, "diagnostics_wall_s": 0.5068919658660889, "inline_diagnostics_wall_s": 0.3647339344024658, "optimizer_wall_s_excluding_inline_diagnostics": 21.42936062812805, "evaluation_wall_s": 0.061923980712890625}, "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}}
|