{"args": {"mode": "sdil", "dataset": "mnist", "depth": 3, "width": 256, "act": "tanh", "residual": 0, "epochs": 15, "batch_size": 128, "eta": 0.05, "eta_A": 0.02, "eta_P": 0.002, "momentum": 0.9, "w_scale": 1.0, "a_scale": 1.0, "pert_sigma": 0.01, "pert_every": 4, "pert_ndirs": 16, "pert_mode": "simultaneous", "use_residual": 0, "raw_scale_control": "match_innovation_norm", "learn_A": 1, "learn_P": 1, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.05, "predictor_mode": "diagonal", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 0, "device": "cuda", "log_every": 100, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "nuis_ctrl_v1_rho0p05_matched_s0", "n_in": 784}, "provenance": {"git_commit": "c8a372cae2612839bdbdafcf9c114f52cb4a6de3", "git_dirty": false}, "steps": [{"step": 0, "epoch": 0, "train_loss": 2.3603768348693848, "cos_r_negg": [-0.018888577818870544, -0.025971781462430954, 0.032935731112957], "cos_innovation_negg": [-0.01886148937046528, -0.025820758193731308, 0.03358188271522522], "cos_apical_negg": [-0.018888579681515694, -0.025971779599785805, 0.0329357348382473], "cos_Ac_negg": [-0.01886145956814289, -0.025820761919021606, 0.033581867814064026], "r_norm": [4.897843360900879, 4.826535701751709, 4.823820114135742], "ldr": {"dL_sdil": -0.0094451904296875, "dL_bp": -0.23505353927612305, "ratio": 0.04018314490722051}}, {"step": 100, "epoch": 0, "train_loss": 0.2925134003162384, "cos_r_negg": [0.2600802779197693, 0.26295357942581177, 0.26131099462509155], "cos_innovation_negg": [0.5080794095993042, 0.47609972953796387, 0.4329450726509094], "cos_apical_negg": [0.2600802779197693, 0.26295357942581177, 0.26131099462509155], "cos_Ac_negg": [0.5081974267959595, 0.47605615854263306, 0.4330165386199951], "r_norm": [1.0393927097320557, 1.017633080482483, 1.0119723081588745]}, {"step": 200, "epoch": 0, "train_loss": 0.35974711179733276, "cos_r_negg": [0.21381282806396484, 0.21494406461715698, 0.2353992760181427], "cos_innovation_negg": [0.5576655864715576, 0.5311211347579956, 0.4876919090747833], "cos_apical_negg": [0.21381282806396484, 0.21494406461715698, 0.2353992760181427], "cos_Ac_negg": [0.5579656362533569, 0.5309754014015198, 0.4882020056247711], "r_norm": [0.7699924111366272, 0.7572832107543945, 0.7494314312934875]}, {"step": 300, "epoch": 0, "train_loss": 0.2397359162569046, "cos_r_negg": [0.20823854207992554, 0.21287313103675842, 0.21960140764713287], "cos_innovation_negg": [0.5615941286087036, 0.5571704506874084, 0.51097571849823], "cos_apical_negg": [0.20823854207992554, 0.21287314593791962, 0.21960140764713287], "cos_Ac_negg": [0.5611709356307983, 0.5571225881576538, 0.5108929872512817], "r_norm": [0.636157751083374, 0.622569739818573, 0.6193011999130249]}, {"step": 400, "epoch": 0, "train_loss": 0.16822239756584167, "cos_r_negg": [0.24075540900230408, 0.2516685128211975, 0.23215335607528687], "cos_innovation_negg": [0.5859761238098145, 0.5680060386657715, 0.573615550994873], "cos_apical_negg": [0.24075540900230408, 0.2516685128211975, 0.23215335607528687], "cos_Ac_negg": [0.585658848285675, 0.5683999061584473, 0.5734416246414185], "r_norm": [0.7184232473373413, 0.7055515050888062, 0.7036710977554321]}, {"epoch_end": 0, "step": 469, "test_acc": 0.9367, "test_loss": 0.20731257934570313}, {"step": 500, "epoch": 1, "train_loss": 0.2124519646167755, "cos_r_negg": [0.19893747568130493, 0.21118637919425964, 0.21870911121368408], "cos_innovation_negg": [0.5929887294769287, 0.5930097103118896, 0.5744439363479614], "cos_apical_negg": [0.19893747568130493, 0.21118637919425964, 0.21870911121368408], "cos_Ac_negg": [0.5926508903503418, 0.5930609703063965, 0.5743594169616699], "r_norm": [0.5818136930465698, 0.5715469121932983, 0.569089412689209], "ldr": {"dL_sdil": -0.01115541160106659, "dL_bp": -0.010585024952888489, "ratio": 1.0538861883384083}}, {"step": 600, "epoch": 1, "train_loss": 0.09760530292987823, "cos_r_negg": [0.1724637746810913, 0.1980946958065033, 0.21229907870292664], "cos_innovation_negg": [0.5967092514038086, 0.5914995670318604, 0.6301769018173218], "cos_apical_negg": [0.17246375977993011, 0.1980946958065033, 0.21229907870292664], "cos_Ac_negg": [0.5966199636459351, 0.5912399291992188, 0.6300961971282959], "r_norm": [0.5214858651161194, 0.5144153237342834, 0.5116932392120361]}, {"step": 700, "epoch": 1, "train_loss": 0.2011653482913971, "cos_r_negg": [0.18453750014305115, 0.2045569121837616, 0.21993455290794373], "cos_innovation_negg": [0.5839407444000244, 0.5973491668701172, 0.6437468528747559], "cos_apical_negg": [0.18453750014305115, 0.2045568972826004, 0.21993455290794373], "cos_Ac_negg": [0.5839197039604187, 0.5971819162368774, 0.643746018409729], "r_norm": [0.47424277663230896, 0.46459078788757324, 0.4637201726436615]}, {"step": 800, "epoch": 1, "train_loss": 0.17559540271759033, "cos_r_negg": [0.1664370447397232, 0.1795715093612671, 0.18634530901908875], "cos_innovation_negg": [0.5826842188835144, 0.6021206378936768, 0.6356761455535889], "cos_apical_negg": [0.1664370447397232, 0.17957152426242828, 0.18634530901908875], "cos_Ac_negg": [0.5825594067573547, 0.6016643643379211, 0.6353830099105835], "r_norm": [0.49542903900146484, 0.4850253462791443, 0.48303377628326416]}, {"step": 900, "epoch": 1, "train_loss": 0.16637590527534485, "cos_r_negg": [0.16264620423316956, 0.1967795044183731, 0.2149035483598709], "cos_innovation_negg": [0.5857820510864258, 0.5955974459648132, 0.6671602725982666], "cos_apical_negg": [0.16264620423316956, 0.1967795193195343, 0.2149035632610321], "cos_Ac_negg": [0.5857768654823303, 0.5954445600509644, 0.6669906377792358], "r_norm": [0.4447886347770691, 0.43750229477882385, 0.4374445974826813]}, {"epoch_end": 1, "step": 938, "test_acc": 0.9507, "test_loss": 0.18049224014282225}, {"step": 1000, "epoch": 2, "train_loss": 0.07176452875137329, "cos_r_negg": [0.19881696999073029, 0.23217740654945374, 0.23776540160179138], "cos_innovation_negg": [0.5895395278930664, 0.6104155778884888, 0.6908552646636963], "cos_apical_negg": [0.19881696999073029, 0.23217740654945374, 0.23776540160179138], "cos_Ac_negg": [0.5894551277160645, 0.6103118062019348, 0.690676212310791], "r_norm": [0.5243955850601196, 0.518196702003479, 0.5109337568283081], "ldr": {"dL_sdil": -0.024999797344207764, "dL_bp": -0.024990737438201904, "ratio": 1.0003625305587025}}, {"step": 1100, "epoch": 2, "train_loss": 0.21467101573944092, "cos_r_negg": [0.14204896986484528, 0.175796777009964, 0.20578661561012268], "cos_innovation_negg": [0.5654441714286804, 0.5900775194168091, 0.668527364730835], "cos_apical_negg": [0.14204896986484528, 0.175796777009964, 0.20578661561012268], "cos_Ac_negg": [0.5649679899215698, 0.5897566080093384, 0.6681588292121887], "r_norm": [0.40868109464645386, 0.40163397789001465, 0.40070417523384094]}, {"step": 1200, "epoch": 2, "train_loss": 0.23202364146709442, "cos_r_negg": [0.14741116762161255, 0.1883571445941925, 0.21747739613056183], "cos_innovation_negg": [0.573001503944397, 0.6010042428970337, 0.6759384870529175], "cos_apical_negg": [0.14741116762161255, 0.1883571445941925, 0.21747739613056183], "cos_Ac_negg": [0.5726856589317322, 0.6007051467895508, 0.675615668296814], "r_norm": [0.40492138266563416, 0.3971380889415741, 0.3964496850967407]}, {"step": 1300, "epoch": 2, "train_loss": 0.14806914329528809, "cos_r_negg": [0.12889954447746277, 0.1617678552865982, 0.2525172233581543], "cos_innovation_negg": [0.5734646320343018, 0.5951402187347412, 0.6983329653739929], "cos_apical_negg": [0.12889954447746277, 0.1617678552865982, 0.2525172233581543], "cos_Ac_negg": [0.573162317276001, 0.5947151780128479, 0.697583794593811], "r_norm": [0.4237639307975769, 0.41418731212615967, 0.4155843257904053]}, {"step": 1400, "epoch": 2, "train_loss": 0.17553207278251648, "cos_r_negg": [0.13460661470890045, 0.15178261697292328, 0.2361353039741516], "cos_innovation_negg": [0.5718982815742493, 0.6067863702774048, 0.7094591856002808], "cos_apical_negg": [0.13460661470890045, 0.15178261697292328, 0.23613528907299042], "cos_Ac_negg": [0.5713673830032349, 0.6060904860496521, 0.708818256855011], "r_norm": [0.36141514778137207, 0.3557671308517456, 0.3550999164581299]}, {"epoch_end": 2, "step": 1407, "test_acc": 0.9538, "test_loss": 0.1557269275665283}, {"step": 1500, "epoch": 3, "train_loss": 0.13190995156764984, "cos_r_negg": [0.13948999345302582, 0.15122002363204956, 0.17426201701164246], "cos_innovation_negg": [0.5648857355117798, 0.6017195582389832, 0.7207717895507812], "cos_apical_negg": [0.13948999345302582, 0.15122002363204956, 0.17426201701164246], "cos_Ac_negg": [0.5642306804656982, 0.6009235382080078, 0.7202455401420593], "r_norm": [0.3556489050388336, 0.3492661416530609, 0.34925156831741333], "ldr": {"dL_sdil": -0.006761610507965088, "dL_bp": -0.007895633578300476, "ratio": 0.8563733918134173}}, {"step": 1600, "epoch": 3, "train_loss": 0.12020806968212128, "cos_r_negg": [0.1394944041967392, 0.1747950315475464, 0.22166553139686584], "cos_innovation_negg": [0.5503301024436951, 0.6051430106163025, 0.7060390114784241], "cos_apical_negg": [0.1394944190979004, 0.1747950315475464, 0.22166553139686584], "cos_Ac_negg": [0.5502285361289978, 0.6051437854766846, 0.7059417963027954], "r_norm": [0.3799726665019989, 0.37381744384765625, 0.37025994062423706]}, {"step": 1700, "epoch": 3, "train_loss": 0.12640194594860077, "cos_r_negg": [0.13853128254413605, 0.17026695609092712, 0.192499577999115], "cos_innovation_negg": [0.5643026828765869, 0.614161491394043, 0.7227206230163574], "cos_apical_negg": [0.13853129744529724, 0.17026695609092712, 0.192499577999115], "cos_Ac_negg": [0.564063549041748, 0.6139358282089233, 0.722440242767334], "r_norm": [0.41324523091316223, 0.40703123807907104, 0.4055345356464386]}, {"step": 1800, "epoch": 3, "train_loss": 0.021324677392840385, "cos_r_negg": [0.11448002606630325, 0.1622004359960556, 0.15379531681537628], "cos_innovation_negg": [0.5607688426971436, 0.5947904586791992, 0.7067182064056396], "cos_apical_negg": [0.11448002606630325, 0.1622004210948944, 0.15379533171653748], "cos_Ac_negg": [0.560477614402771, 0.594211220741272, 0.7057375907897949], "r_norm": [0.34976544976234436, 0.34635043144226074, 0.3432033061981201]}, {"epoch_end": 3, "step": 1876, "test_acc": 0.9589, "test_loss": 0.13953808059692382}, {"step": 1900, "epoch": 4, "train_loss": 0.08204656839370728, "cos_r_negg": [0.11766695976257324, 0.15146785974502563, 0.16843679547309875], "cos_innovation_negg": [0.5684402585029602, 0.5953737497329712, 0.7274236679077148], "cos_apical_negg": [0.11766695976257324, 0.15146785974502563, 0.16843679547309875], "cos_Ac_negg": [0.5680156350135803, 0.5950276851654053, 0.7267047166824341], "r_norm": [0.3084789514541626, 0.30436310172080994, 0.3013650178909302]}, {"step": 2000, "epoch": 4, "train_loss": 0.16747495532035828, "cos_r_negg": [0.09503549337387085, 0.13075682520866394, 0.19327878952026367], "cos_innovation_negg": [0.5502690076828003, 0.5977407693862915, 0.7345064878463745], "cos_apical_negg": [0.09503549337387085, 0.13075682520866394, 0.19327880442142487], "cos_Ac_negg": [0.5494052171707153, 0.597416877746582, 0.7338563799858093], "r_norm": [0.33181628584861755, 0.3239547610282898, 0.32429227232933044], "ldr": {"dL_sdil": -0.011301577091217041, "dL_bp": -0.014433056116104126, "ratio": 0.7830342375380194}}, {"step": 2100, "epoch": 4, "train_loss": 0.0479811355471611, "cos_r_negg": [0.1210709661245346, 0.16421130299568176, 0.21043890714645386], "cos_innovation_negg": [0.5522477626800537, 0.5935909748077393, 0.7170997262001038], "cos_apical_negg": [0.1210709661245346, 0.16421130299568176, 0.21043890714645386], "cos_Ac_negg": [0.5515713691711426, 0.5923702716827393, 0.7165814638137817], "r_norm": [0.4709198772907257, 0.46099433302879333, 0.4572564959526062]}, {"step": 2200, "epoch": 4, "train_loss": 0.030225181952118874, "cos_r_negg": [0.10094594210386276, 0.15073946118354797, 0.1727372705936432], "cos_innovation_negg": [0.5547534227371216, 0.6040734648704529, 0.7391708493232727], "cos_apical_negg": [0.10094594210386276, 0.15073946118354797, 0.172737255692482], "cos_Ac_negg": [0.5542435050010681, 0.6037734746932983, 0.7387421131134033], "r_norm": [0.30935361981391907, 0.3030080497264862, 0.3018103241920471]}, {"step": 2300, "epoch": 4, "train_loss": 0.22975580394268036, "cos_r_negg": [0.11159854382276535, 0.14088667929172516, 0.16555047035217285], "cos_innovation_negg": [0.5509880185127258, 0.5912683606147766, 0.7330929040908813], "cos_apical_negg": [0.11159853637218475, 0.14088669419288635, 0.16555047035217285], "cos_Ac_negg": [0.550560474395752, 0.5900709629058838, 0.7318021059036255], "r_norm": [0.2574431002140045, 0.25378215312957764, 0.2512046694755554]}, {"epoch_end": 4, "step": 2345, "test_acc": 0.9608, "test_loss": 0.13752566566467286}, {"step": 2400, "epoch": 5, "train_loss": 0.13832847774028778, "cos_r_negg": [0.09531106054782867, 0.11300396919250488, 0.17430379986763], "cos_innovation_negg": [0.5580445528030396, 0.6096996665000916, 0.732465386390686], "cos_apical_negg": [0.09531106799840927, 0.11300396919250488, 0.17430379986763], "cos_Ac_negg": [0.5576131343841553, 0.6092826128005981, 0.731804370880127], "r_norm": [0.29090115427970886, 0.286607027053833, 0.284349262714386]}, {"step": 2500, "epoch": 5, "train_loss": 0.03487436845898628, "cos_r_negg": [0.08470235764980316, 0.09950131922960281, 0.13054385781288147], "cos_innovation_negg": [0.5414856672286987, 0.5867946147918701, 0.7511945962905884], "cos_apical_negg": [0.08470235764980316, 0.09950131922960281, 0.13054384291172028], "cos_Ac_negg": [0.5413762927055359, 0.5864846706390381, 0.7509655952453613], "r_norm": [0.27695605158805847, 0.27181476354599, 0.27067533135414124], "ldr": {"dL_sdil": -0.00807391107082367, "dL_bp": -0.010298319160938263, "ratio": 0.7840028012967573}}, {"step": 2600, "epoch": 5, "train_loss": 0.11033789813518524, "cos_r_negg": [0.0956190824508667, 0.12613743543624878, 0.135524183511734], "cos_innovation_negg": [0.5575433969497681, 0.587949275970459, 0.7775989770889282], "cos_apical_negg": [0.0956190824508667, 0.12613743543624878, 0.135524183511734], "cos_Ac_negg": [0.5576225519180298, 0.5881427526473999, 0.7769110798835754], "r_norm": [0.34104347229003906, 0.3375483453273773, 0.331754595041275]}, {"step": 2700, "epoch": 5, "train_loss": 0.04781732335686684, "cos_r_negg": [0.08584916591644287, 0.11772655695676804, 0.16713127493858337], "cos_innovation_negg": [0.5652093291282654, 0.5992367267608643, 0.7422999739646912], "cos_apical_negg": [0.08584915846586227, 0.11772656440734863, 0.16713127493858337], "cos_Ac_negg": [0.5655986070632935, 0.5995399951934814, 0.741814136505127], "r_norm": [0.28275835514068604, 0.2767255902290344, 0.2750614881515503]}, {"step": 2800, "epoch": 5, "train_loss": 0.1706671565771103, "cos_r_negg": [0.09122958779335022, 0.11384502053260803, 0.17169907689094543], "cos_innovation_negg": [0.5433820486068726, 0.5899790525436401, 0.761295735836029], "cos_apical_negg": [0.09122960269451141, 0.11384502798318863, 0.17169909179210663], "cos_Ac_negg": [0.5424270629882812, 0.58916836977005, 0.7602348327636719], "r_norm": [0.30017024278640747, 0.2951509356498718, 0.29237961769104004]}, {"epoch_end": 5, "step": 2814, "test_acc": 0.9637, "test_loss": 0.13457500228881836}, {"step": 2900, "epoch": 6, "train_loss": 0.019902069121599197, "cos_r_negg": [0.08643680810928345, 0.10017009824514389, 0.16301289200782776], "cos_innovation_negg": [0.5555968880653381, 0.6130666732788086, 0.743310809135437], "cos_apical_negg": [0.08643680065870285, 0.10017010569572449, 0.16301289200782776], "cos_Ac_negg": [0.5552680492401123, 0.6136595010757446, 0.744665265083313], "r_norm": [0.29115843772888184, 0.285186231136322, 0.2850620150566101]}, {"step": 3000, "epoch": 6, "train_loss": 0.1896212100982666, "cos_r_negg": [0.05150953307747841, 0.07955276966094971, 0.12302252650260925], "cos_innovation_negg": [0.556449294090271, 0.6050234436988831, 0.7544301748275757], "cos_apical_negg": [0.05150953680276871, 0.07955276966094971, 0.12302251905202866], "cos_Ac_negg": [0.5569143891334534, 0.606092095375061, 0.7551864385604858], "r_norm": [0.25071385502815247, 0.2459258735179901, 0.2457805871963501], "ldr": {"dL_sdil": -0.0064574480056762695, "dL_bp": -0.00949336588382721, "ratio": 0.6802063761891981}}, {"step": 3100, "epoch": 6, "train_loss": 0.07059909403324127, "cos_r_negg": [0.07147012650966644, 0.10582158714532852, 0.13621792197227478], "cos_innovation_negg": [0.5145161151885986, 0.5923010110855103, 0.7556737661361694], "cos_apical_negg": [0.07147012650966644, 0.10582158714532852, 0.1362179070711136], "cos_Ac_negg": [0.5149269104003906, 0.5934772491455078, 0.7549319267272949], "r_norm": [0.26133793592453003, 0.2572306990623474, 0.2575467824935913]}, {"step": 3200, "epoch": 6, "train_loss": 0.07592140883207321, "cos_r_negg": [0.07781057804822922, 0.09806972742080688, 0.16834115982055664], "cos_innovation_negg": [0.5632877349853516, 0.6036882400512695, 0.7605648040771484], "cos_apical_negg": [0.07781057804822922, 0.09806972742080688, 0.16834115982055664], "cos_Ac_negg": [0.5641144514083862, 0.6079229116439819, 0.7628395557403564], "r_norm": [0.24588164687156677, 0.24154649674892426, 0.23966088891029358]}, {"epoch_end": 6, "step": 3283, "test_acc": 0.9658, "test_loss": 0.12357935066223144}, {"step": 3300, "epoch": 7, "train_loss": 0.028484942391514778, "cos_r_negg": [0.06780457496643066, 0.10129158943891525, 0.13399964570999146], "cos_innovation_negg": [0.5625133514404297, 0.5992419719696045, 0.7527547478675842], "cos_apical_negg": [0.06780457496643066, 0.10129158943891525, 0.13399964570999146], "cos_Ac_negg": [0.56358802318573, 0.6007165908813477, 0.753697395324707], "r_norm": [0.22360146045684814, 0.21901917457580566, 0.218156635761261]}, {"step": 3400, "epoch": 7, "train_loss": 0.1063101589679718, "cos_r_negg": [0.07290801405906677, 0.09812893718481064, 0.11037027835845947], "cos_innovation_negg": [0.5527506470680237, 0.6091886758804321, 0.7597736716270447], "cos_apical_negg": [0.07290801405906677, 0.09812894463539124, 0.11037027835845947], "cos_Ac_negg": [0.5524070858955383, 0.6110753417015076, 0.7611789703369141], "r_norm": [0.2745347023010254, 0.2698018550872803, 0.2688489556312561]}, {"step": 3500, "epoch": 7, "train_loss": 0.037357836961746216, "cos_r_negg": [0.06743920594453812, 0.08692482113838196, 0.13602200150489807], "cos_innovation_negg": [0.5601925849914551, 0.604422926902771, 0.7443182468414307], "cos_apical_negg": [0.06743919849395752, 0.08692482113838196, 0.13602200150489807], "cos_Ac_negg": [0.5611312985420227, 0.6055264472961426, 0.7438371181488037], "r_norm": [0.2416018843650818, 0.23779501020908356, 0.23534145951271057], "ldr": {"dL_sdil": -0.006105475127696991, "dL_bp": -0.009596720337867737, "ratio": 0.6362043398936377}}, {"step": 3600, "epoch": 7, "train_loss": 0.029826544225215912, "cos_r_negg": [0.08168534189462662, 0.10181734710931778, 0.1450318992137909], "cos_innovation_negg": [0.5346965789794922, 0.6047930717468262, 0.7627032399177551], "cos_apical_negg": [0.08168533444404602, 0.10181735455989838, 0.1450318992137909], "cos_Ac_negg": [0.5337558388710022, 0.6045243144035339, 0.7588868141174316], "r_norm": [0.26648473739624023, 0.26119711995124817, 0.2588616609573364]}, {"step": 3700, "epoch": 7, "train_loss": 0.031768057495355606, "cos_r_negg": [0.08394169807434082, 0.10271388292312622, 0.14723770320415497], "cos_innovation_negg": [0.5592381954193115, 0.609021008014679, 0.7928593158721924], "cos_apical_negg": [0.08394169807434082, 0.10271389037370682, 0.14723770320415497], "cos_Ac_negg": [0.5604240894317627, 0.6106314659118652, 0.7925615906715393], "r_norm": [0.2781471908092499, 0.2734353542327881, 0.27141472697257996]}, {"epoch_end": 7, "step": 3752, "test_acc": 0.9639, "test_loss": 0.13228435287475587}, {"step": 3800, "epoch": 8, "train_loss": 0.05007845163345337, "cos_r_negg": [0.0717189684510231, 0.08457625657320023, 0.10764709860086441], "cos_innovation_negg": [0.5546162128448486, 0.5975524187088013, 0.7884159088134766], "cos_apical_negg": [0.0717189610004425, 0.08457625657320023, 0.10764710605144501], "cos_Ac_negg": [0.5563030242919922, 0.6000141501426697, 0.789344310760498], "r_norm": [0.24482181668281555, 0.24063773453235626, 0.23882639408111572]}, {"step": 3900, "epoch": 8, "train_loss": 0.08228941261768341, "cos_r_negg": [0.06533907353878021, 0.07992380857467651, 0.147702157497406], "cos_innovation_negg": [0.5418471097946167, 0.5887052416801453, 0.7823539972305298], "cos_apical_negg": [0.06533907353878021, 0.07992380857467651, 0.147702157497406], "cos_Ac_negg": [0.5450972318649292, 0.5908803939819336, 0.7821483612060547], "r_norm": [0.25028395652770996, 0.24401697516441345, 0.24258068203926086]}, {"step": 4000, "epoch": 8, "train_loss": 0.05164622142910957, "cos_r_negg": [0.06902293860912323, 0.08827603608369827, 0.1320556104183197], "cos_innovation_negg": [0.5270317196846008, 0.5888156294822693, 0.7667114734649658], "cos_apical_negg": [0.06902293115854263, 0.08827603608369827, 0.1320556104183197], "cos_Ac_negg": [0.5293920040130615, 0.591464638710022, 0.7683452367782593], "r_norm": [0.2521548867225647, 0.2463391125202179, 0.24513624608516693], "ldr": {"dL_sdil": -0.009338408708572388, "dL_bp": -0.015852496027946472, "ratio": 0.5890812836104575}}, {"step": 4100, "epoch": 8, "train_loss": 0.02004571259021759, "cos_r_negg": [0.0625336766242981, 0.08250875771045685, 0.1356448531150818], "cos_innovation_negg": [0.5627686977386475, 0.5938019752502441, 0.7631560564041138], "cos_apical_negg": [0.0625336766242981, 0.08250876516103745, 0.1356448382139206], "cos_Ac_negg": [0.5648341178894043, 0.5962578058242798, 0.7650654315948486], "r_norm": [0.24856719374656677, 0.24405287206172943, 0.24061977863311768]}, {"step": 4200, "epoch": 8, "train_loss": 0.09874685108661652, "cos_r_negg": [0.07091013342142105, 0.07042770832777023, 0.13188888132572174], "cos_innovation_negg": [0.5426219701766968, 0.6038415431976318, 0.780397355556488], "cos_apical_negg": [0.07091013342142105, 0.07042770832777023, 0.13188888132572174], "cos_Ac_negg": [0.5441992282867432, 0.6068688631057739, 0.78309166431427], "r_norm": [0.2244134247303009, 0.2188449203968048, 0.21817874908447266]}, {"epoch_end": 8, "step": 4221, "test_acc": 0.9655, "test_loss": 0.13557769584655763}, {"step": 4300, "epoch": 9, "train_loss": 0.02034987322986126, "cos_r_negg": [0.06329119205474854, 0.07320599257946014, 0.10350732505321503], "cos_innovation_negg": [0.5358413457870483, 0.593376636505127, 0.7736502885818481], "cos_apical_negg": [0.06329119205474854, 0.07320599257946014, 0.10350732505321503], "cos_Ac_negg": [0.5360220074653625, 0.5959383845329285, 0.7705438137054443], "r_norm": [0.2687934637069702, 0.26312386989593506, 0.26151275634765625]}, {"step": 4400, "epoch": 9, "train_loss": 0.01585252955555916, "cos_r_negg": [0.0704190656542778, 0.08715783804655075, 0.12539517879486084], "cos_innovation_negg": [0.5446029901504517, 0.602871298789978, 0.7813178896903992], "cos_apical_negg": [0.0704190656542778, 0.08715783804655075, 0.12539517879486084], "cos_Ac_negg": [0.5481078624725342, 0.6076574325561523, 0.78309166431427], "r_norm": [0.23737162351608276, 0.2324787676334381, 0.22979232668876648]}, {"step": 4500, "epoch": 9, "train_loss": 0.19144883751869202, "cos_r_negg": [0.05317222326993942, 0.06681753695011139, 0.14209970831871033], "cos_innovation_negg": [0.5579216480255127, 0.6123360395431519, 0.7830038070678711], "cos_apical_negg": [0.05317222326993942, 0.06681753695011139, 0.14209970831871033], "cos_Ac_negg": [0.560858964920044, 0.6170240044593811, 0.7845879793167114], "r_norm": [0.22783441841602325, 0.2232532948255539, 0.2201940417289734], "ldr": {"dL_sdil": -0.007011055946350098, "dL_bp": -0.010684967041015625, "ratio": 0.656160746162085}}, {"step": 4600, "epoch": 9, "train_loss": 0.03670297563076019, "cos_r_negg": [0.0664428174495697, 0.07982797175645828, 0.13727284967899323], "cos_innovation_negg": [0.5728162527084351, 0.62354975938797, 0.7998867034912109], "cos_apical_negg": [0.0664428174495697, 0.07982797175645828, 0.13727283477783203], "cos_Ac_negg": [0.5757042169570923, 0.6258006691932678, 0.8005550503730774], "r_norm": [0.283415824174881, 0.2762840688228607, 0.2732480466365814]}, {"epoch_end": 9, "step": 4690, "test_acc": 0.9698, "test_loss": 0.1227607042312622}, {"step": 4700, "epoch": 10, "train_loss": 0.04839279130101204, "cos_r_negg": [0.06364335119724274, 0.07567989826202393, 0.1274554431438446], "cos_innovation_negg": [0.5271023511886597, 0.5936386585235596, 0.7894782423973083], "cos_apical_negg": [0.06364335119724274, 0.07567991316318512, 0.1274554431438446], "cos_Ac_negg": [0.532446026802063, 0.5963029861450195, 0.7881503701210022], "r_norm": [0.2292705476284027, 0.22390416264533997, 0.2221139371395111]}, {"step": 4800, "epoch": 10, "train_loss": 0.03952376917004585, "cos_r_negg": [0.046887755393981934, 0.06370607018470764, 0.12445493787527084], "cos_innovation_negg": [0.5435724854469299, 0.6021091341972351, 0.797446072101593], "cos_apical_negg": [0.046887755393981934, 0.06370607018470764, 0.12445493787527084], "cos_Ac_negg": [0.5482916235923767, 0.6075999736785889, 0.7988876104354858], "r_norm": [0.17225481569766998, 0.16954630613327026, 0.16898219287395477]}, {"step": 4900, "epoch": 10, "train_loss": 0.03187403082847595, "cos_r_negg": [0.0643419474363327, 0.08255770057439804, 0.13570451736450195], "cos_innovation_negg": [0.5599974393844604, 0.6049967408180237, 0.7847260236740112], "cos_apical_negg": [0.0643419474363327, 0.08255770057439804, 0.13570451736450195], "cos_Ac_negg": [0.5625724792480469, 0.6064321994781494, 0.7859264016151428], "r_norm": [0.23878632485866547, 0.2340824455022812, 0.23296594619750977]}, {"step": 5000, "epoch": 10, "train_loss": 0.02667912095785141, "cos_r_negg": [0.06717013567686081, 0.06284972280263901, 0.1056877002120018], "cos_innovation_negg": [0.552981436252594, 0.6072264909744263, 0.7859450578689575], "cos_apical_negg": [0.06717013567686081, 0.06284972280263901, 0.1056877002120018], "cos_Ac_negg": [0.557763934135437, 0.6128823757171631, 0.788381814956665], "r_norm": [0.2379167377948761, 0.2322022020816803, 0.23069347441196442], "ldr": {"dL_sdil": -0.00891779363155365, "dL_bp": -0.015383400022983551, "ratio": 0.5797023816731042}}, {"step": 5100, "epoch": 10, "train_loss": 0.0677889883518219, "cos_r_negg": [0.04698483645915985, 0.07755410671234131, 0.1409112960100174], "cos_innovation_negg": [0.5482721924781799, 0.620044469833374, 0.7865081429481506], "cos_apical_negg": [0.04698483645915985, 0.07755410671234131, 0.1409112960100174], "cos_Ac_negg": [0.5589918494224548, 0.6287689805030823, 0.7922115325927734], "r_norm": [0.2318282574415207, 0.22801828384399414, 0.22497932612895966]}, {"epoch_end": 10, "step": 5159, "test_acc": 0.9652, "test_loss": 0.13374782333374025}, {"step": 5200, "epoch": 11, "train_loss": 0.0038604428991675377, "cos_r_negg": [0.058167487382888794, 0.07114555686712265, 0.11589882522821426], "cos_innovation_negg": [0.5481350421905518, 0.6034847497940063, 0.7792742848396301], "cos_apical_negg": [0.058167487382888794, 0.07114555686712265, 0.11589881777763367], "cos_Ac_negg": [0.5543985366821289, 0.6114673614501953, 0.7864778637886047], "r_norm": [0.22882503271102905, 0.2241784632205963, 0.22140666842460632]}, {"step": 5300, "epoch": 11, "train_loss": 0.07033146917819977, "cos_r_negg": [0.04950704425573349, 0.0705556720495224, 0.10611505806446075], "cos_innovation_negg": [0.5287195444107056, 0.5920501351356506, 0.7762441635131836], "cos_apical_negg": [0.04950704053044319, 0.0705556720495224, 0.10611505806446075], "cos_Ac_negg": [0.5348626375198364, 0.5972569584846497, 0.7768690586090088], "r_norm": [0.23408210277557373, 0.23052264750003815, 0.2266005277633667]}, {"step": 5400, "epoch": 11, "train_loss": 0.08506366610527039, "cos_r_negg": [0.05567062646150589, 0.07443711906671524, 0.1290496587753296], "cos_innovation_negg": [0.5281569957733154, 0.5864236354827881, 0.7631567120552063], "cos_apical_negg": [0.05567063018679619, 0.07443712651729584, 0.1290496587753296], "cos_Ac_negg": [0.5394855737686157, 0.5949428081512451, 0.7680245041847229], "r_norm": [0.2382836788892746, 0.23389361798763275, 0.2303137481212616]}, {"step": 5500, "epoch": 11, "train_loss": 0.012573130428791046, "cos_r_negg": [0.06361153721809387, 0.06807530671358109, 0.12357844412326813], "cos_innovation_negg": [0.5378892421722412, 0.6000767350196838, 0.7792697548866272], "cos_apical_negg": [0.06361154466867447, 0.06807530671358109, 0.12357844412326813], "cos_Ac_negg": [0.5450212955474854, 0.6080093383789062, 0.7858061194419861], "r_norm": [0.21391001343727112, 0.20936641097068787, 0.20761336386203766], "ldr": {"dL_sdil": -0.006099686026573181, "dL_bp": -0.011147961020469666, "ratio": 0.5471571003318956}}, {"step": 5600, "epoch": 11, "train_loss": 0.009609967470169067, "cos_r_negg": [0.06594008207321167, 0.06625820696353912, 0.07251353561878204], "cos_innovation_negg": [0.527391791343689, 0.5948604345321655, 0.7574151754379272], "cos_apical_negg": [0.06594006717205048, 0.06625820696353912, 0.07251353561878204], "cos_Ac_negg": [0.5348454713821411, 0.6017602682113647, 0.7670135498046875], "r_norm": [0.22620606422424316, 0.2204313576221466, 0.21838465332984924]}, {"epoch_end": 11, "step": 5628, "test_acc": 0.9696, "test_loss": 0.12623830223083496}, {"step": 5700, "epoch": 12, "train_loss": 0.016506992280483246, "cos_r_negg": [0.0547497496008873, 0.07152171432971954, 0.09657566994428635], "cos_innovation_negg": [0.5452012419700623, 0.5976607799530029, 0.7862253785133362], "cos_apical_negg": [0.0547497496008873, 0.07152171432971954, 0.09657566994428635], "cos_Ac_negg": [0.5535536408424377, 0.6069254875183105, 0.7960631847381592], "r_norm": [0.24224236607551575, 0.23622053861618042, 0.23375500738620758]}, {"step": 5800, "epoch": 12, "train_loss": 0.0026011494919657707, "cos_r_negg": [0.05949844419956207, 0.06380505114793777, 0.11441205441951752], "cos_innovation_negg": [0.5408796668052673, 0.5976915955543518, 0.7925673723220825], "cos_apical_negg": [0.05949845165014267, 0.06380504369735718, 0.11441206932067871], "cos_Ac_negg": [0.5505724549293518, 0.6095927953720093, 0.8027863502502441], "r_norm": [0.2224837690591812, 0.217766672372818, 0.2157045602798462]}, {"step": 5900, "epoch": 12, "train_loss": 0.021796507760882378, "cos_r_negg": [0.054697148501873016, 0.07463476061820984, 0.0868355929851532], "cos_innovation_negg": [0.5256735682487488, 0.5992733240127563, 0.7848494052886963], "cos_apical_negg": [0.054697152227163315, 0.07463476061820984, 0.0868355929851532], "cos_Ac_negg": [0.5358738899230957, 0.6056530475616455, 0.7925519943237305], "r_norm": [0.21211518347263336, 0.2071298062801361, 0.2064683437347412]}, {"step": 6000, "epoch": 12, "train_loss": 0.026512151584029198, "cos_r_negg": [0.057589564472436905, 0.06602048873901367, 0.07767926156520844], "cos_innovation_negg": [0.5288076400756836, 0.6061176061630249, 0.7742867469787598], "cos_apical_negg": [0.0575895681977272, 0.06602048873901367, 0.07767926156520844], "cos_Ac_negg": [0.5394042730331421, 0.6129321455955505, 0.7777043581008911], "r_norm": [0.21243371069431305, 0.20848143100738525, 0.20531213283538818], "ldr": {"dL_sdil": -0.009594418108463287, "dL_bp": -0.020150206983089447, "ratio": 0.4761448910433109}}, {"epoch_end": 12, "step": 6097, "test_acc": 0.9672, "test_loss": 0.13024798736572266}, {"step": 6100, "epoch": 13, "train_loss": 0.009351560845971107, "cos_r_negg": [0.052101340144872665, 0.058585524559020996, 0.06036121025681496], "cos_innovation_negg": [0.5419372320175171, 0.6053557395935059, 0.783679187297821], "cos_apical_negg": [0.052101343870162964, 0.058585528284311295, 0.06036120653152466], "cos_Ac_negg": [0.5503400564193726, 0.6160815954208374, 0.7936914563179016], "r_norm": [0.21938712894916534, 0.21519601345062256, 0.2139711081981659]}, {"step": 6200, "epoch": 13, "train_loss": 0.006052996497601271, "cos_r_negg": [0.053515542298555374, 0.055314261466264725, 0.05995969474315643], "cos_innovation_negg": [0.5447695255279541, 0.6130589842796326, 0.7757765054702759], "cos_apical_negg": [0.053515542298555374, 0.055314261466264725, 0.05995970219373703], "cos_Ac_negg": [0.5529961585998535, 0.6255960464477539, 0.7874360084533691], "r_norm": [0.20926839113235474, 0.20431242883205414, 0.20193041861057281]}, {"step": 6300, "epoch": 13, "train_loss": 0.018606022000312805, "cos_r_negg": [0.05628553405404091, 0.05468010902404785, 0.057541463524103165], "cos_innovation_negg": [0.5407936573028564, 0.5999512672424316, 0.7780349254608154], "cos_apical_negg": [0.05628553777933121, 0.05468010902404785, 0.057541463524103165], "cos_Ac_negg": [0.550012469291687, 0.6131508350372314, 0.7917049527168274], "r_norm": [0.2285028100013733, 0.22168998420238495, 0.22110216319561005]}, {"step": 6400, "epoch": 13, "train_loss": 0.004905556328594685, "cos_r_negg": [0.050309933722019196, 0.06324630230665207, 0.08564186096191406], "cos_innovation_negg": [0.5209953784942627, 0.5941855907440186, 0.7652787566184998], "cos_apical_negg": [0.0503099262714386, 0.06324630230665207, 0.08564185351133347], "cos_Ac_negg": [0.5333911180496216, 0.6052359342575073, 0.771978497505188], "r_norm": [0.2235257625579834, 0.2172807902097702, 0.21643313765525818]}, {"step": 6500, "epoch": 13, "train_loss": 0.0013855393044650555, "cos_r_negg": [0.05341285839676857, 0.06854744255542755, 0.1042645275592804], "cos_innovation_negg": [0.5446828007698059, 0.6036902666091919, 0.7699993848800659], "cos_apical_negg": [0.05341285467147827, 0.06854744255542755, 0.10426453500986099], "cos_Ac_negg": [0.5551351308822632, 0.6163210272789001, 0.7806403636932373], "r_norm": [0.22807006537914276, 0.2236315906047821, 0.22066211700439453], "ldr": {"dL_sdil": -0.008510112762451172, "dL_bp": -0.016563981771469116, "ratio": 0.513772164197231}}, {"epoch_end": 13, "step": 6566, "test_acc": 0.9695, "test_loss": 0.12776789474487305}, {"step": 6600, "epoch": 14, "train_loss": 0.0038456113543361425, "cos_r_negg": [0.05062942951917648, 0.07090011239051819, 0.1098744124174118], "cos_innovation_negg": [0.5379418730735779, 0.5994153022766113, 0.7756435871124268], "cos_apical_negg": [0.05062942951917648, 0.07090011984109879, 0.1098744198679924], "cos_Ac_negg": [0.5495676398277283, 0.6087256669998169, 0.7847073674201965], "r_norm": [0.20909591019153595, 0.20454992353916168, 0.2028166502714157]}, {"step": 6700, "epoch": 14, "train_loss": 0.006470425985753536, "cos_r_negg": [0.05011637881398201, 0.05745463818311691, 0.10158899426460266], "cos_innovation_negg": [0.524763286113739, 0.5984794497489929, 0.7564308047294617], "cos_apical_negg": [0.05011637881398201, 0.05745463818311691, 0.10158898681402206], "cos_Ac_negg": [0.5375996828079224, 0.6064271926879883, 0.7629046440124512], "r_norm": [0.2094632089138031, 0.20427276194095612, 0.2030295431613922]}, {"step": 6800, "epoch": 14, "train_loss": 0.02861160971224308, "cos_r_negg": [0.0517265610396862, 0.062266431748867035, 0.06925496459007263], "cos_innovation_negg": [0.5419613122940063, 0.6001811027526855, 0.7763111591339111], "cos_apical_negg": [0.0517265647649765, 0.06226643547415733, 0.06925496459007263], "cos_Ac_negg": [0.5514795780181885, 0.6109452843666077, 0.7812302112579346], "r_norm": [0.20821134746074677, 0.20343603193759918, 0.20183655619621277]}, {"step": 6900, "epoch": 14, "train_loss": 0.012429247610270977, "cos_r_negg": [0.05247191712260246, 0.060937635600566864, 0.06693011522293091], "cos_innovation_negg": [0.5486399531364441, 0.6003872156143188, 0.7809118032455444], "cos_apical_negg": [0.05247192084789276, 0.060937635600566864, 0.06693011522293091], "cos_Ac_negg": [0.5589513182640076, 0.6118178367614746, 0.7880287170410156], "r_norm": [0.212667778134346, 0.2083602100610733, 0.20657868683338165]}, {"step": 7000, "epoch": 14, "train_loss": 0.004722911864519119, "cos_r_negg": [0.0473896786570549, 0.06399355083703995, 0.06589870899915695], "cos_innovation_negg": [0.5357134938240051, 0.5969662666320801, 0.7830662727355957], "cos_apical_negg": [0.0473896823823452, 0.06399355083703995, 0.06589871644973755], "cos_Ac_negg": [0.5482646226882935, 0.6102514266967773, 0.7924281358718872], "r_norm": [0.20618706941604614, 0.20211279392242432, 0.2001674473285675], "ldr": {"dL_sdil": -0.00756506621837616, "dL_bp": -0.014824427664279938, "ratio": 0.5103108456999318}}, {"epoch_end": 14, "step": 7035, "test_acc": 0.971, "test_loss": 0.13526744537353516}], "final": {"test_acc": 0.971, "test_loss": 0.13526744537353516, "wall_s": 25.47541618347168, "cos_r_negg": [0.04763709008693695, 0.05940413102507591, 0.05433851107954979], "cos_innovation_negg": [0.5352548360824585, 0.5923671722412109, 0.7811884880065918], "cos_apical_negg": [0.04763708636164665, 0.059404127299785614, 0.05433851107954979], "cos_Ac_negg": [0.5465336441993713, 0.6060895919799805, 0.7910354137420654]}}