{"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": "none", "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": 1, "device": "cuda", "log_every": 100, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "nuis_ctrl_v1_rho0p05_raw_s1", "n_in": 784}, "provenance": {"git_commit": "c8a372cae2612839bdbdafcf9c114f52cb4a6de3", "git_dirty": false}, "steps": [{"step": 0, "epoch": 0, "train_loss": 2.327335834503174, "cos_r_negg": [0.013905191794037819, -0.013632269576191902, -0.006711989641189575], "cos_innovation_negg": [0.014363561756908894, -0.013818755745887756, -0.0063642337918281555], "cos_apical_negg": [0.013905191794037819, -0.013632269576191902, -0.006711989641189575], "cos_Ac_negg": [0.014363492839038372, -0.0138187725096941, -0.0063642594031989574], "r_norm": [4.815765857696533, 4.82281494140625, 4.805631637573242], "ldr": {"dL_sdil": -0.0417790412902832, "dL_bp": -0.21432781219482422, "ratio": 0.19493056389856678}}, {"step": 100, "epoch": 0, "train_loss": 0.3111538290977478, "cos_r_negg": [0.22585797309875488, 0.2289954423904419, 0.22102117538452148], "cos_innovation_negg": [0.47757360339164734, 0.41555705666542053, 0.31677794456481934], "cos_apical_negg": [0.22585797309875488, 0.2289954423904419, 0.22102117538452148], "cos_Ac_negg": [0.4775737524032593, 0.4155714511871338, 0.31678393483161926], "r_norm": [1.507110357284546, 1.5036137104034424, 1.5179942846298218]}, {"step": 200, "epoch": 0, "train_loss": 0.31935569643974304, "cos_r_negg": [0.21915887296199799, 0.2109759896993637, 0.24562522768974304], "cos_innovation_negg": [0.45117610692977905, 0.3945627808570862, 0.34682363271713257], "cos_apical_negg": [0.21915887296199799, 0.2109759896993637, 0.24562522768974304], "cos_Ac_negg": [0.45115044713020325, 0.3945918381214142, 0.3468201756477356], "r_norm": [1.3864176273345947, 1.3908636569976807, 1.404003620147705]}, {"step": 300, "epoch": 0, "train_loss": 0.2104194015264511, "cos_r_negg": [0.16776320338249207, 0.19028759002685547, 0.23572096228599548], "cos_innovation_negg": [0.4569571614265442, 0.4386172890663147, 0.4506438076496124], "cos_apical_negg": [0.16776320338249207, 0.19028759002685547, 0.23572096228599548], "cos_Ac_negg": [0.45695632696151733, 0.43851780891418457, 0.4506588280200958], "r_norm": [1.2572340965270996, 1.2549448013305664, 1.2671852111816406]}, {"step": 400, "epoch": 0, "train_loss": 0.18861497938632965, "cos_r_negg": [0.16798993945121765, 0.2042803317308426, 0.27786386013031006], "cos_innovation_negg": [0.5486142635345459, 0.5082403421401978, 0.4526609778404236], "cos_apical_negg": [0.16798993945121765, 0.2042803317308426, 0.27786386013031006], "cos_Ac_negg": [0.5486135482788086, 0.5081335306167603, 0.4527098834514618], "r_norm": [1.2299907207489014, 1.2317063808441162, 1.2414379119873047]}, {"epoch_end": 0, "step": 469, "test_acc": 0.9422, "test_loss": 0.20943933486938476}, {"step": 500, "epoch": 1, "train_loss": 0.19546829164028168, "cos_r_negg": [0.12399314343929291, 0.19445103406906128, 0.2384604513645172], "cos_innovation_negg": [0.49418824911117554, 0.486931174993515, 0.4326460361480713], "cos_apical_negg": [0.12399314343929291, 0.19445103406906128, 0.2384604513645172], "cos_Ac_negg": [0.494186669588089, 0.4869386553764343, 0.43236231803894043], "r_norm": [1.1982481479644775, 1.1957364082336426, 1.206697940826416], "ldr": {"dL_sdil": -0.00836414098739624, "dL_bp": -0.006997942924499512, "ratio": 1.1952285232441273}}, {"step": 600, "epoch": 1, "train_loss": 0.10189378261566162, "cos_r_negg": [0.12711183726787567, 0.21554352343082428, 0.2960144877433777], "cos_innovation_negg": [0.49261781573295593, 0.517648458480835, 0.4768499433994293], "cos_apical_negg": [0.12711183726787567, 0.21554352343082428, 0.2960144877433777], "cos_Ac_negg": [0.49261361360549927, 0.5176858901977539, 0.4766700267791748], "r_norm": [1.1861131191253662, 1.1880137920379639, 1.1966147422790527]}, {"step": 700, "epoch": 1, "train_loss": 0.1639467477798462, "cos_r_negg": [0.16854949295520782, 0.21466293931007385, 0.405530720949173], "cos_innovation_negg": [0.5290984511375427, 0.5688223838806152, 0.6095681190490723], "cos_apical_negg": [0.16854949295520782, 0.21466293931007385, 0.405530720949173], "cos_Ac_negg": [0.5290928483009338, 0.5687499046325684, 0.6095190048217773], "r_norm": [1.2222474813461304, 1.225698471069336, 1.2332630157470703]}, {"step": 800, "epoch": 1, "train_loss": 0.1482793390750885, "cos_r_negg": [0.14247897267341614, 0.20840352773666382, 0.2946995496749878], "cos_innovation_negg": [0.49229753017425537, 0.5285365581512451, 0.5543482303619385], "cos_apical_negg": [0.14247897267341614, 0.20840352773666382, 0.2946995496749878], "cos_Ac_negg": [0.4923862814903259, 0.5285008549690247, 0.5543349981307983], "r_norm": [1.1485624313354492, 1.151463508605957, 1.1598212718963623]}, {"step": 900, "epoch": 1, "train_loss": 0.24120372533798218, "cos_r_negg": [0.15464136004447937, 0.22585511207580566, 0.2914172410964966], "cos_innovation_negg": [0.526918888092041, 0.5539251565933228, 0.5657846927642822], "cos_apical_negg": [0.15464136004447937, 0.22585511207580566, 0.2914172410964966], "cos_Ac_negg": [0.5269187092781067, 0.5538864135742188, 0.5658643245697021], "r_norm": [1.1452395915985107, 1.1458568572998047, 1.1549901962280273]}, {"epoch_end": 1, "step": 938, "test_acc": 0.9489, "test_loss": 0.19619377784729003}, {"step": 1000, "epoch": 2, "train_loss": 0.05176971107721329, "cos_r_negg": [0.11876438558101654, 0.20083294808864594, 0.28751203417778015], "cos_innovation_negg": [0.45458948612213135, 0.49184131622314453, 0.49922239780426025], "cos_apical_negg": [0.11876438558101654, 0.20083294808864594, 0.28751203417778015], "cos_Ac_negg": [0.4545850157737732, 0.4917981028556824, 0.4991696774959564], "r_norm": [1.1385282278060913, 1.1386324167251587, 1.14955735206604], "ldr": {"dL_sdil": -0.006788343191146851, "dL_bp": -0.006677001714706421, "ratio": 1.0166753703530127}}, {"step": 1100, "epoch": 2, "train_loss": 0.16464105248451233, "cos_r_negg": [0.11798076331615448, 0.23670539259910583, 0.293226420879364], "cos_innovation_negg": [0.5064412355422974, 0.5673417448997498, 0.5176621675491333], "cos_apical_negg": [0.11798076331615448, 0.23670539259910583, 0.293226420879364], "cos_Ac_negg": [0.5064129829406738, 0.5672239065170288, 0.5174068212509155], "r_norm": [1.1557585000991821, 1.1562490463256836, 1.162174940109253]}, {"step": 1200, "epoch": 2, "train_loss": 0.22122842073440552, "cos_r_negg": [0.11416029930114746, 0.23655816912651062, 0.34759199619293213], "cos_innovation_negg": [0.5368348360061646, 0.5717867612838745, 0.5662579536437988], "cos_apical_negg": [0.11416029930114746, 0.23655816912651062, 0.34759199619293213], "cos_Ac_negg": [0.5368198752403259, 0.5716375112533569, 0.5660883188247681], "r_norm": [1.1090397834777832, 1.1073741912841797, 1.1163573265075684]}, {"step": 1300, "epoch": 2, "train_loss": 0.16772988438606262, "cos_r_negg": [0.1405804306268692, 0.2335662841796875, 0.3327704668045044], "cos_innovation_negg": [0.5201165676116943, 0.570197582244873, 0.5598894953727722], "cos_apical_negg": [0.1405804306268692, 0.2335662841796875, 0.3327704668045044], "cos_Ac_negg": [0.5201260447502136, 0.5701526403427124, 0.5599179267883301], "r_norm": [1.1047096252441406, 1.1055306196212769, 1.1118117570877075]}, {"step": 1400, "epoch": 2, "train_loss": 0.13406571745872498, "cos_r_negg": [0.13619279861450195, 0.20941613614559174, 0.3439789116382599], "cos_innovation_negg": [0.5422729253768921, 0.591759979724884, 0.6170070171356201], "cos_apical_negg": [0.13619279861450195, 0.20941613614559174, 0.3439789116382599], "cos_Ac_negg": [0.5422627329826355, 0.5916441679000854, 0.6169291734695435], "r_norm": [1.0734968185424805, 1.0720207691192627, 1.0773144960403442]}, {"epoch_end": 2, "step": 1407, "test_acc": 0.9577, "test_loss": 0.1640996253967285}, {"step": 1500, "epoch": 3, "train_loss": 0.10332301259040833, "cos_r_negg": [0.11957291513681412, 0.20736415684223175, 0.30939364433288574], "cos_innovation_negg": [0.5052887201309204, 0.5775151252746582, 0.5726076364517212], "cos_apical_negg": [0.11957291513681412, 0.20736415684223175, 0.30939364433288574], "cos_Ac_negg": [0.5052045583724976, 0.5773980617523193, 0.5725935697555542], "r_norm": [1.0954232215881348, 1.096743106842041, 1.1033844947814941], "ldr": {"dL_sdil": -0.004838660359382629, "dL_bp": -0.004917040467262268, "ratio": 0.984059495055716}}, {"step": 1600, "epoch": 3, "train_loss": 0.105115607380867, "cos_r_negg": [0.10056674480438232, 0.17087051272392273, 0.33277279138565063], "cos_innovation_negg": [0.46659916639328003, 0.5546510219573975, 0.5421147346496582], "cos_apical_negg": [0.10056674480438232, 0.17087051272392273, 0.33277279138565063], "cos_Ac_negg": [0.4664914608001709, 0.5545365810394287, 0.5414087772369385], "r_norm": [1.0722692012786865, 1.0694310665130615, 1.0746160745620728]}, {"step": 1700, "epoch": 3, "train_loss": 0.11515215039253235, "cos_r_negg": [0.09613379091024399, 0.22954893112182617, 0.36259105801582336], "cos_innovation_negg": [0.45003971457481384, 0.566016674041748, 0.6306129097938538], "cos_apical_negg": [0.09613379091024399, 0.22954893112182617, 0.36259105801582336], "cos_Ac_negg": [0.4499720335006714, 0.565842866897583, 0.6303963661193848], "r_norm": [1.1150307655334473, 1.116746425628662, 1.1212819814682007]}, {"step": 1800, "epoch": 3, "train_loss": 0.02130863070487976, "cos_r_negg": [0.10759153962135315, 0.216238334774971, 0.3577558994293213], "cos_innovation_negg": [0.5138797760009766, 0.5725353956222534, 0.6255130767822266], "cos_apical_negg": [0.10759153962135315, 0.216238334774971, 0.3577558994293213], "cos_Ac_negg": [0.5136306881904602, 0.5721944570541382, 0.6250292062759399], "r_norm": [1.049177646636963, 1.0476912260055542, 1.055837869644165]}, {"epoch_end": 3, "step": 1876, "test_acc": 0.9571, "test_loss": 0.16132894821166993}, {"step": 1900, "epoch": 4, "train_loss": 0.08336880803108215, "cos_r_negg": [0.11278876662254333, 0.18105575442314148, 0.34281179308891296], "cos_innovation_negg": [0.48743218183517456, 0.5750161409378052, 0.6512165665626526], "cos_apical_negg": [0.11278876662254333, 0.18105575442314148, 0.34281179308891296], "cos_Ac_negg": [0.48735135793685913, 0.5750393271446228, 0.6512870788574219], "r_norm": [1.0934102535247803, 1.0947078466415405, 1.1014515161514282]}, {"step": 2000, "epoch": 4, "train_loss": 0.14404454827308655, "cos_r_negg": [0.09042025357484818, 0.18417401611804962, 0.3266880512237549], "cos_innovation_negg": [0.46254539489746094, 0.5401381850242615, 0.5898675918579102], "cos_apical_negg": [0.09042025357484818, 0.18417401611804962, 0.3266880512237549], "cos_Ac_negg": [0.4624313712120056, 0.5399748086929321, 0.5898659229278564], "r_norm": [1.0635663270950317, 1.0584098100662231, 1.0648940801620483], "ldr": {"dL_sdil": -0.005668938159942627, "dL_bp": -0.0066918134689331055, "ratio": 0.8471452747840028}}, {"step": 2100, "epoch": 4, "train_loss": 0.06375595927238464, "cos_r_negg": [0.12302787601947784, 0.2330954372882843, 0.4018273949623108], "cos_innovation_negg": [0.4991646111011505, 0.5582159757614136, 0.6183664202690125], "cos_apical_negg": [0.12302787601947784, 0.2330954372882843, 0.4018273949623108], "cos_Ac_negg": [0.4991351366043091, 0.5577704906463623, 0.6179418563842773], "r_norm": [1.0973044633865356, 1.100116491317749, 1.1074888706207275]}, {"step": 2200, "epoch": 4, "train_loss": 0.03386557847261429, "cos_r_negg": [0.08939173817634583, 0.17289167642593384, 0.2986927330493927], "cos_innovation_negg": [0.48034992814064026, 0.5516642332077026, 0.5839909315109253], "cos_apical_negg": [0.08939173817634583, 0.17289167642593384, 0.2986927330493927], "cos_Ac_negg": [0.48025286197662354, 0.5514553785324097, 0.5833542346954346], "r_norm": [1.0268346071243286, 1.0239297151565552, 1.0311834812164307]}, {"step": 2300, "epoch": 4, "train_loss": 0.18919767439365387, "cos_r_negg": [0.11006394028663635, 0.22932633757591248, 0.3612495958805084], "cos_innovation_negg": [0.5155396461486816, 0.6342329978942871, 0.686772882938385], "cos_apical_negg": [0.11006394028663635, 0.22932633757591248, 0.3612495958805084], "cos_Ac_negg": [0.5153635740280151, 0.6339044570922852, 0.6864478588104248], "r_norm": [1.0438575744628906, 1.0415865182876587, 1.0491456985473633]}, {"epoch_end": 4, "step": 2345, "test_acc": 0.9591, "test_loss": 0.16723030738830566}, {"step": 2400, "epoch": 5, "train_loss": 0.12079156190156937, "cos_r_negg": [0.11126136034727097, 0.21823373436927795, 0.3169762194156647], "cos_innovation_negg": [0.4963069260120392, 0.5969429612159729, 0.6623464226722717], "cos_apical_negg": [0.11126136034727097, 0.21823373436927795, 0.3169762194156647], "cos_Ac_negg": [0.49633920192718506, 0.5970313549041748, 0.6623939275741577], "r_norm": [1.0386252403259277, 1.0362870693206787, 1.0430126190185547]}, {"step": 2500, "epoch": 5, "train_loss": 0.05364352464675903, "cos_r_negg": [0.09973915666341782, 0.2178991287946701, 0.3339419662952423], "cos_innovation_negg": [0.5003910064697266, 0.5921323299407959, 0.7173104286193848], "cos_apical_negg": [0.09973915666341782, 0.2178991287946701, 0.3339419662952423], "cos_Ac_negg": [0.500408947467804, 0.5916909575462341, 0.7167575359344482], "r_norm": [1.0779132843017578, 1.071363925933838, 1.0775971412658691], "ldr": {"dL_sdil": -0.009411841630935669, "dL_bp": -0.011350810527801514, "ratio": 0.829177934728385}}, {"step": 2600, "epoch": 5, "train_loss": 0.0950925424695015, "cos_r_negg": [0.07227948307991028, 0.21443215012550354, 0.3914218544960022], "cos_innovation_negg": [0.484333872795105, 0.6018844842910767, 0.6960376501083374], "cos_apical_negg": [0.07227948307991028, 0.21443215012550354, 0.3914218544960022], "cos_Ac_negg": [0.48413121700286865, 0.6007460951805115, 0.695088803768158], "r_norm": [1.037591576576233, 1.0353960990905762, 1.0433489084243774]}, {"step": 2700, "epoch": 5, "train_loss": 0.07270301878452301, "cos_r_negg": [0.07606332004070282, 0.17456433176994324, 0.32549673318862915], "cos_innovation_negg": [0.48064371943473816, 0.5868778228759766, 0.6518148183822632], "cos_apical_negg": [0.07606332004070282, 0.17456433176994324, 0.32549673318862915], "cos_Ac_negg": [0.48056966066360474, 0.5866889953613281, 0.6516287326812744], "r_norm": [1.0150339603424072, 1.0124701261520386, 1.0199952125549316]}, {"step": 2800, "epoch": 5, "train_loss": 0.18392163515090942, "cos_r_negg": [0.07067474722862244, 0.16180387139320374, 0.36166679859161377], "cos_innovation_negg": [0.4727996587753296, 0.5990469455718994, 0.6918765306472778], "cos_apical_negg": [0.07067474722862244, 0.16180387139320374, 0.36166679859161377], "cos_Ac_negg": [0.4728286862373352, 0.599349856376648, 0.6918573379516602], "r_norm": [1.0224902629852295, 1.0210386514663696, 1.0280964374542236]}, {"epoch_end": 5, "step": 2814, "test_acc": 0.9622, "test_loss": 0.1542672260284424}, {"step": 2900, "epoch": 6, "train_loss": 0.02426699735224247, "cos_r_negg": [0.08332119882106781, 0.1748197078704834, 0.32856476306915283], "cos_innovation_negg": [0.4771268367767334, 0.610244631767273, 0.6739753484725952], "cos_apical_negg": [0.08332119882106781, 0.1748197078704834, 0.32856476306915283], "cos_Ac_negg": [0.47707369923591614, 0.6102348566055298, 0.6737914085388184], "r_norm": [1.0442066192626953, 1.0395574569702148, 1.047477126121521]}, {"step": 3000, "epoch": 6, "train_loss": 0.3423105478286743, "cos_r_negg": [0.06307509541511536, 0.15948311984539032, 0.30363959074020386], "cos_innovation_negg": [0.46689513325691223, 0.5805035829544067, 0.6361100673675537], "cos_apical_negg": [0.06307509541511536, 0.15948311984539032, 0.30363959074020386], "cos_Ac_negg": [0.4668571949005127, 0.5801676511764526, 0.6358035802841187], "r_norm": [1.0098156929016113, 1.006965160369873, 1.0127813816070557], "ldr": {"dL_sdil": -0.003946930170059204, "dL_bp": -0.0042160749435424805, "ratio": 0.9361622416376848}}, {"step": 3100, "epoch": 6, "train_loss": 0.0975814163684845, "cos_r_negg": [0.07857103645801544, 0.1796930581331253, 0.36494001746177673], "cos_innovation_negg": [0.45325079560279846, 0.5922108888626099, 0.6971831917762756], "cos_apical_negg": [0.07857103645801544, 0.1796930581331253, 0.36494001746177673], "cos_Ac_negg": [0.4532625675201416, 0.5921370983123779, 0.6970545649528503], "r_norm": [1.0387482643127441, 1.0345592498779297, 1.0429034233093262]}, {"step": 3200, "epoch": 6, "train_loss": 0.10800129920244217, "cos_r_negg": [0.03326963260769844, 0.17290058732032776, 0.3843672275543213], "cos_innovation_negg": [0.4869436025619507, 0.5830050706863403, 0.7204266786575317], "cos_apical_negg": [0.03326963260769844, 0.17290058732032776, 0.3843672275543213], "cos_Ac_negg": [0.4869372546672821, 0.582355797290802, 0.7196407318115234], "r_norm": [1.0017832517623901, 0.997281551361084, 1.004207730293274]}, {"epoch_end": 6, "step": 3283, "test_acc": 0.9623, "test_loss": 0.1552559108734131}, {"step": 3300, "epoch": 7, "train_loss": 0.05600239336490631, "cos_r_negg": [0.06956698000431061, 0.138072669506073, 0.3201645612716675], "cos_innovation_negg": [0.4550621509552002, 0.5418501496315002, 0.6678740382194519], "cos_apical_negg": [0.06956698000431061, 0.138072669506073, 0.3201645612716675], "cos_Ac_negg": [0.4549746811389923, 0.5419297218322754, 0.6677284240722656], "r_norm": [1.0094234943389893, 1.006321907043457, 1.012253761291504]}, {"step": 3400, "epoch": 7, "train_loss": 0.1245359480381012, "cos_r_negg": [0.0801684632897377, 0.19823044538497925, 0.37976396083831787], "cos_innovation_negg": [0.4689311385154724, 0.605712890625, 0.6607257723808289], "cos_apical_negg": [0.0801684632897377, 0.19823044538497925, 0.37976396083831787], "cos_Ac_negg": [0.46858859062194824, 0.6056188941001892, 0.659482479095459], "r_norm": [1.0204098224639893, 1.0160210132598877, 1.0230128765106201]}, {"step": 3500, "epoch": 7, "train_loss": 0.11146698892116547, "cos_r_negg": [0.07624058425426483, 0.15644685924053192, 0.2621228098869324], "cos_innovation_negg": [0.44706106185913086, 0.49836763739585876, 0.5497217178344727], "cos_apical_negg": [0.07624058425426483, 0.15644685924053192, 0.2621228098869324], "cos_Ac_negg": [0.4469067454338074, 0.4980616867542267, 0.5484806299209595], "r_norm": [1.00234854221344, 0.9983804225921631, 1.0046262741088867], "ldr": {"dL_sdil": -0.003928542137145996, "dL_bp": -0.005212306976318359, "ratio": 0.7537050590064953}}, {"step": 3600, "epoch": 7, "train_loss": 0.02609572932124138, "cos_r_negg": [0.08282971382141113, 0.17203842103481293, 0.3280511498451233], "cos_innovation_negg": [0.477922260761261, 0.5371098518371582, 0.6868182420730591], "cos_apical_negg": [0.08282971382141113, 0.17203842103481293, 0.3280511498451233], "cos_Ac_negg": [0.4778933525085449, 0.5369381308555603, 0.6866243481636047], "r_norm": [1.0103245973587036, 1.0051637887954712, 1.0108013153076172]}, {"step": 3700, "epoch": 7, "train_loss": 0.06168299913406372, "cos_r_negg": [0.07405601441860199, 0.14879007637500763, 0.38227328658103943], "cos_innovation_negg": [0.49088534712791443, 0.5814383029937744, 0.7205300331115723], "cos_apical_negg": [0.07405601441860199, 0.14879007637500763, 0.38227328658103943], "cos_Ac_negg": [0.490772008895874, 0.5812431573867798, 0.7198448181152344], "r_norm": [1.0193647146224976, 1.0127074718475342, 1.01832115650177]}, {"epoch_end": 7, "step": 3752, "test_acc": 0.9634, "test_loss": 0.15631410789489747}, {"step": 3800, "epoch": 8, "train_loss": 0.04240557178854942, "cos_r_negg": [0.058022718876600266, 0.1690751612186432, 0.31578749418258667], "cos_innovation_negg": [0.46928760409355164, 0.5878943204879761, 0.6567918658256531], "cos_apical_negg": [0.058022718876600266, 0.1690751612186432, 0.31578749418258667], "cos_Ac_negg": [0.4692709743976593, 0.5876166820526123, 0.6564027070999146], "r_norm": [1.0154097080230713, 1.010310411453247, 1.0156447887420654]}, {"step": 3900, "epoch": 8, "train_loss": 0.13345283269882202, "cos_r_negg": [0.07689585536718369, 0.18118694424629211, 0.2707849442958832], "cos_innovation_negg": [0.4771782159805298, 0.5578058362007141, 0.6736498475074768], "cos_apical_negg": [0.07689585536718369, 0.18118694424629211, 0.2707849442958832], "cos_Ac_negg": [0.4770645797252655, 0.5567095279693604, 0.6727590560913086], "r_norm": [1.02763032913208, 1.0211572647094727, 1.026594638824463]}, {"step": 4000, "epoch": 8, "train_loss": 0.10269910097122192, "cos_r_negg": [0.0932910293340683, 0.16571977734565735, 0.3383203148841858], "cos_innovation_negg": [0.4715930223464966, 0.5827735066413879, 0.679978609085083], "cos_apical_negg": [0.0932910293340683, 0.16571977734565735, 0.3383203148841858], "cos_Ac_negg": [0.47118639945983887, 0.5813443660736084, 0.6783742308616638], "r_norm": [1.0120360851287842, 1.0063525438308716, 1.0133607387542725], "ldr": {"dL_sdil": -0.005438610911369324, "dL_bp": -0.007597222924232483, "ratio": 0.7158682805031372}}, {"step": 4100, "epoch": 8, "train_loss": 0.023356664925813675, "cos_r_negg": [0.0712575763463974, 0.14939527213573456, 0.27680835127830505], "cos_innovation_negg": [0.46732020378112793, 0.6015224456787109, 0.6808948516845703], "cos_apical_negg": [0.0712575763463974, 0.14939527213573456, 0.27680835127830505], "cos_Ac_negg": [0.46705782413482666, 0.6018350720405579, 0.6822611689567566], "r_norm": [0.9958678483963013, 0.9913784861564636, 0.9969359636306763]}, {"step": 4200, "epoch": 8, "train_loss": 0.1036907359957695, "cos_r_negg": [0.07991500198841095, 0.166538268327713, 0.3339913487434387], "cos_innovation_negg": [0.4539138674736023, 0.5708163976669312, 0.6744144558906555], "cos_apical_negg": [0.07991500198841095, 0.166538268327713, 0.3339913487434387], "cos_Ac_negg": [0.4537237584590912, 0.5714869499206543, 0.6745802164077759], "r_norm": [0.9879276156425476, 0.9835188388824463, 0.9894178509712219]}, {"epoch_end": 8, "step": 4221, "test_acc": 0.9601, "test_loss": 0.17026060676574706}, {"step": 4300, "epoch": 9, "train_loss": 0.03131965920329094, "cos_r_negg": [0.06058957427740097, 0.15393486618995667, 0.39697808027267456], "cos_innovation_negg": [0.4793262183666229, 0.6009911298751831, 0.7217565774917603], "cos_apical_negg": [0.06058957427740097, 0.15393486618995667, 0.39697808027267456], "cos_Ac_negg": [0.4792332649230957, 0.6009107828140259, 0.7214959859848022], "r_norm": [1.0002045631408691, 0.9942552447319031, 0.9997059106826782]}, {"step": 4400, "epoch": 9, "train_loss": 0.04176610708236694, "cos_r_negg": [0.07677163183689117, 0.146663099527359, 0.3605833053588867], "cos_innovation_negg": [0.44574159383773804, 0.5401622653007507, 0.6500576734542847], "cos_apical_negg": [0.07677163183689117, 0.146663099527359, 0.3605833053588867], "cos_Ac_negg": [0.445831835269928, 0.5404576063156128, 0.647425651550293], "r_norm": [1.017586350440979, 1.0118733644485474, 1.0189217329025269]}, {"step": 4500, "epoch": 9, "train_loss": 0.13476458191871643, "cos_r_negg": [0.054127536714076996, 0.16489824652671814, 0.3267046809196472], "cos_innovation_negg": [0.46263542771339417, 0.5874254703521729, 0.673530101776123], "cos_apical_negg": [0.054127536714076996, 0.16489824652671814, 0.3267046809196472], "cos_Ac_negg": [0.4628351628780365, 0.5875599384307861, 0.6728527545928955], "r_norm": [1.0008635520935059, 0.99664306640625, 1.0026768445968628], "ldr": {"dL_sdil": -0.004442289471626282, "dL_bp": -0.007203981280326843, "ratio": 0.6166436722646143}}, {"step": 4600, "epoch": 9, "train_loss": 0.026353374123573303, "cos_r_negg": [0.06898064911365509, 0.17992939054965973, 0.33015739917755127], "cos_innovation_negg": [0.4801621437072754, 0.6100026965141296, 0.726508617401123], "cos_apical_negg": [0.06898064911365509, 0.17992939054965973, 0.33015739917755127], "cos_Ac_negg": [0.4800562858581543, 0.6101363897323608, 0.7259714603424072], "r_norm": [1.0048885345458984, 0.9985988140106201, 1.003064513206482]}, {"epoch_end": 9, "step": 4690, "test_acc": 0.9629, "test_loss": 0.16283142318725585}, {"step": 4700, "epoch": 10, "train_loss": 0.07447795569896698, "cos_r_negg": [0.04082931578159332, 0.13814829289913177, 0.29323896765708923], "cos_innovation_negg": [0.4813762605190277, 0.5998628735542297, 0.706299901008606], "cos_apical_negg": [0.04082931578159332, 0.13814829289913177, 0.29323896765708923], "cos_Ac_negg": [0.48167258501052856, 0.6006890535354614, 0.7043177485466003], "r_norm": [1.0003453493118286, 0.9935581088066101, 0.9974279403686523]}, {"step": 4800, "epoch": 10, "train_loss": 0.0702333077788353, "cos_r_negg": [0.06491591036319733, 0.12049730122089386, 0.2656426429748535], "cos_innovation_negg": [0.4316897690296173, 0.5156323313713074, 0.6949378848075867], "cos_apical_negg": [0.06491591036319733, 0.12049730122089386, 0.2656426429748535], "cos_Ac_negg": [0.43133971095085144, 0.5148912668228149, 0.6932864189147949], "r_norm": [0.9971155524253845, 0.9910149574279785, 0.9951145648956299]}, {"step": 4900, "epoch": 10, "train_loss": 0.07635124027729034, "cos_r_negg": [0.06220356002449989, 0.1753993034362793, 0.26844465732574463], "cos_innovation_negg": [0.47314804792404175, 0.5789248943328857, 0.7186846137046814], "cos_apical_negg": [0.06220356002449989, 0.1753993034362793, 0.26844465732574463], "cos_Ac_negg": [0.4726075530052185, 0.5764424800872803, 0.7174698114395142], "r_norm": [1.0121158361434937, 1.0073261260986328, 1.0124716758728027]}, {"step": 5000, "epoch": 10, "train_loss": 0.032667431980371475, "cos_r_negg": [0.06057484447956085, 0.13061538338661194, 0.36302196979522705], "cos_innovation_negg": [0.44369348883628845, 0.5756018161773682, 0.7298516035079956], "cos_apical_negg": [0.06057484447956085, 0.13061538338661194, 0.36302196979522705], "cos_Ac_negg": [0.44343769550323486, 0.5754837989807129, 0.7288978099822998], "r_norm": [1.0030885934829712, 0.9972087144851685, 1.0033472776412964], "ldr": {"dL_sdil": -0.004772692918777466, "dL_bp": -0.007688939571380615, "ratio": 0.6207218660609772}}, {"step": 5100, "epoch": 10, "train_loss": 0.06812694668769836, "cos_r_negg": [0.040807634592056274, 0.146819606423378, 0.3157557249069214], "cos_innovation_negg": [0.4270401895046234, 0.562659740447998, 0.6880521774291992], "cos_apical_negg": [0.040807634592056274, 0.146819606423378, 0.3157557249069214], "cos_Ac_negg": [0.42722243070602417, 0.5621775984764099, 0.6871858239173889], "r_norm": [0.9940504431724548, 0.9884777069091797, 0.991992712020874]}, {"epoch_end": 10, "step": 5159, "test_acc": 0.9637, "test_loss": 0.16795218696594238}, {"step": 5200, "epoch": 11, "train_loss": 0.005242047365754843, "cos_r_negg": [0.04723719134926796, 0.168500155210495, 0.31768786907196045], "cos_innovation_negg": [0.4486674666404724, 0.5615569353103638, 0.6777231097221375], "cos_apical_negg": [0.04723719134926796, 0.168500155210495, 0.31768786907196045], "cos_Ac_negg": [0.4490351974964142, 0.5618119835853577, 0.6759191751480103], "r_norm": [1.0135362148284912, 1.006046175956726, 1.0099883079528809]}, {"step": 5300, "epoch": 11, "train_loss": 0.04645835980772972, "cos_r_negg": [0.02789916656911373, 0.1501055508852005, 0.2888845205307007], "cos_innovation_negg": [0.39689427614212036, 0.4925505816936493, 0.6453686356544495], "cos_apical_negg": [0.02789916656911373, 0.1501055508852005, 0.2888845205307007], "cos_Ac_negg": [0.39727583527565, 0.49246904253959656, 0.6451837420463562], "r_norm": [1.0048892498016357, 0.9984314441680908, 1.0019359588623047]}, {"step": 5400, "epoch": 11, "train_loss": 0.08015645295381546, "cos_r_negg": [0.036679863929748535, 0.16330870985984802, 0.3469422459602356], "cos_innovation_negg": [0.4299176335334778, 0.5475703477859497, 0.7247656583786011], "cos_apical_negg": [0.036679863929748535, 0.16330870985984802, 0.3469422459602356], "cos_Ac_negg": [0.4303250312805176, 0.5479323863983154, 0.7244442701339722], "r_norm": [0.986892819404602, 0.9811404943466187, 0.9852945804595947]}, {"step": 5500, "epoch": 11, "train_loss": 0.10337873548269272, "cos_r_negg": [0.04536229372024536, 0.12637490034103394, 0.3078010082244873], "cos_innovation_negg": [0.4053611755371094, 0.5329481959342957, 0.7376441955566406], "cos_apical_negg": [0.04536229372024536, 0.12637490034103394, 0.3078010082244873], "cos_Ac_negg": [0.4052439332008362, 0.5332862138748169, 0.7368306517601013], "r_norm": [0.9978095293045044, 0.9923269748687744, 0.9953716993331909], "ldr": {"dL_sdil": -0.005255058407783508, "dL_bp": -0.009846597909927368, "ratio": 0.5336927998740909}}, {"step": 5600, "epoch": 11, "train_loss": 0.03063536249101162, "cos_r_negg": [0.05646223947405815, 0.16315333545207977, 0.34648674726486206], "cos_innovation_negg": [0.41272374987602234, 0.5321862101554871, 0.7480096817016602], "cos_apical_negg": [0.05646223947405815, 0.16315333545207977, 0.34648674726486206], "cos_Ac_negg": [0.4125385880470276, 0.5328807830810547, 0.7469910979270935], "r_norm": [0.9943451881408691, 0.987838625907898, 0.9920538663864136]}, {"epoch_end": 11, "step": 5628, "test_acc": 0.964, "test_loss": 0.16722139854431153}, {"step": 5700, "epoch": 12, "train_loss": 0.01672079786658287, "cos_r_negg": [0.049960989505052567, 0.18621963262557983, 0.3282274603843689], "cos_innovation_negg": [0.41187095642089844, 0.5453330278396606, 0.7323659658432007], "cos_apical_negg": [0.049960989505052567, 0.18621963262557983, 0.3282274603843689], "cos_Ac_negg": [0.41169118881225586, 0.5457819104194641, 0.7328809499740601], "r_norm": [1.0035459995269775, 0.9972206950187683, 1.0026814937591553]}, {"step": 5800, "epoch": 12, "train_loss": 0.018970094621181488, "cos_r_negg": [0.04647641256451607, 0.14888887107372284, 0.33303457498550415], "cos_innovation_negg": [0.3900400996208191, 0.5297600626945496, 0.720571756362915], "cos_apical_negg": [0.04647641256451607, 0.14888887107372284, 0.33303457498550415], "cos_Ac_negg": [0.39014071226119995, 0.5302168130874634, 0.7207975387573242], "r_norm": [0.9923798441886902, 0.9846975803375244, 0.9881832599639893]}, {"step": 5900, "epoch": 12, "train_loss": 0.026083137840032578, "cos_r_negg": [0.04985422641038895, 0.1442699432373047, 0.31639280915260315], "cos_innovation_negg": [0.4107506275177002, 0.4937441945075989, 0.6798127889633179], "cos_apical_negg": [0.04985422641038895, 0.1442699432373047, 0.31639280915260315], "cos_Ac_negg": [0.40956953167915344, 0.49411192536354065, 0.676051914691925], "r_norm": [1.010257601737976, 1.0013244152069092, 1.0046613216400146]}, {"step": 6000, "epoch": 12, "train_loss": 0.008461544290184975, "cos_r_negg": [0.029255229979753494, 0.18780213594436646, 0.2699863910675049], "cos_innovation_negg": [0.43998152017593384, 0.5333098769187927, 0.6794975996017456], "cos_apical_negg": [0.029255229979753494, 0.18780213594436646, 0.2699863910675049], "cos_Ac_negg": [0.44005224108695984, 0.5306196808815002, 0.6777482032775879], "r_norm": [1.002807378768921, 0.9942639470100403, 0.9987096786499023], "ldr": {"dL_sdil": -0.005444705486297607, "dL_bp": -0.011139541864395142, "ratio": 0.48877283757066303}}, {"epoch_end": 12, "step": 6097, "test_acc": 0.9647, "test_loss": 0.17159284400939942}, {"step": 6100, "epoch": 13, "train_loss": 0.024622051045298576, "cos_r_negg": [0.034463364630937576, 0.15558412671089172, 0.22659939527511597], "cos_innovation_negg": [0.3993053436279297, 0.5189934968948364, 0.7001391053199768], "cos_apical_negg": [0.034463364630937576, 0.15558412671089172, 0.22659939527511597], "cos_Ac_negg": [0.39973369240760803, 0.5202004909515381, 0.7011544704437256], "r_norm": [0.9926055669784546, 0.9858974814414978, 0.9902071952819824]}, {"step": 6200, "epoch": 13, "train_loss": 0.013772055506706238, "cos_r_negg": [0.05448794364929199, 0.16238385438919067, 0.2878323197364807], "cos_innovation_negg": [0.39762502908706665, 0.5268219709396362, 0.7149753570556641], "cos_apical_negg": [0.05448794364929199, 0.16238385438919067, 0.2878323197364807], "cos_Ac_negg": [0.39821648597717285, 0.5261987447738647, 0.7132536172866821], "r_norm": [1.0000630617141724, 0.9914873242378235, 0.9952001571655273]}, {"step": 6300, "epoch": 13, "train_loss": 0.009136243723332882, "cos_r_negg": [0.054326675832271576, 0.17687208950519562, 0.30371466279029846], "cos_innovation_negg": [0.40638431906700134, 0.5090117454528809, 0.6876112222671509], "cos_apical_negg": [0.054326675832271576, 0.17687208950519562, 0.30371466279029846], "cos_Ac_negg": [0.40657466650009155, 0.5081708431243896, 0.6861210465431213], "r_norm": [1.0032062530517578, 0.9955238103866577, 0.9993560314178467]}, {"step": 6400, "epoch": 13, "train_loss": 0.01309981755912304, "cos_r_negg": [0.04377492889761925, 0.13936050236225128, 0.2471282035112381], "cos_innovation_negg": [0.40372008085250854, 0.5213161706924438, 0.6748590469360352], "cos_apical_negg": [0.04377492889761925, 0.13936050236225128, 0.2471282035112381], "cos_Ac_negg": [0.40482962131500244, 0.5230716466903687, 0.6761823892593384], "r_norm": [1.0090835094451904, 1.000382661819458, 1.0045630931854248]}, {"step": 6500, "epoch": 13, "train_loss": 0.005054917186498642, "cos_r_negg": [0.03787251189351082, 0.1691415011882782, 0.27734020352363586], "cos_innovation_negg": [0.39364761114120483, 0.4857965111732483, 0.7172523140907288], "cos_apical_negg": [0.03787251189351082, 0.1691415011882782, 0.27734020352363586], "cos_Ac_negg": [0.394325852394104, 0.48641687631607056, 0.7171387672424316], "r_norm": [1.0038480758666992, 0.9952888488769531, 0.9986175298690796], "ldr": {"dL_sdil": -0.0053085386753082275, "dL_bp": -0.009034797549247742, "ratio": 0.5875658692263922}}, {"epoch_end": 13, "step": 6566, "test_acc": 0.965, "test_loss": 0.16982969284057617}, {"step": 6600, "epoch": 14, "train_loss": 0.01103228610008955, "cos_r_negg": [0.0637044683098793, 0.16204126179218292, 0.2569505274295807], "cos_innovation_negg": [0.4007999897003174, 0.4942467212677002, 0.7148151993751526], "cos_apical_negg": [0.0637044683098793, 0.16204126179218292, 0.2569505274295807], "cos_Ac_negg": [0.40037351846694946, 0.4896727204322815, 0.7124040126800537], "r_norm": [0.9995841383934021, 0.9915225505828857, 0.9953269362449646]}, {"step": 6700, "epoch": 14, "train_loss": 0.07044786959886551, "cos_r_negg": [0.0554986298084259, 0.1700323522090912, 0.24690739810466766], "cos_innovation_negg": [0.3683430850505829, 0.4520756006240845, 0.6897143125534058], "cos_apical_negg": [0.0554986298084259, 0.1700323522090912, 0.24690739810466766], "cos_Ac_negg": [0.3696437180042267, 0.44101646542549133, 0.6870521306991577], "r_norm": [1.0063462257385254, 0.9971489906311035, 0.9997471570968628]}, {"step": 6800, "epoch": 14, "train_loss": 0.07112330943346024, "cos_r_negg": [0.030334604904055595, 0.16273115575313568, 0.28592705726623535], "cos_innovation_negg": [0.41890501976013184, 0.5037792325019836, 0.7524046897888184], "cos_apical_negg": [0.030334604904055595, 0.16273115575313568, 0.28592705726623535], "cos_Ac_negg": [0.4216950237751007, 0.5074898600578308, 0.7542818784713745], "r_norm": [0.99994957447052, 0.992084801197052, 0.9965077638626099]}, {"step": 6900, "epoch": 14, "train_loss": 0.03472434729337692, "cos_r_negg": [0.03598427027463913, 0.19057166576385498, 0.24942214787006378], "cos_innovation_negg": [0.4031265377998352, 0.4925141930580139, 0.7411197423934937], "cos_apical_negg": [0.03598427027463913, 0.19057166576385498, 0.24942214787006378], "cos_Ac_negg": [0.4042908847332001, 0.48933911323547363, 0.7436213493347168], "r_norm": [0.9963934421539307, 0.9881312847137451, 0.990851640701294]}, {"step": 7000, "epoch": 14, "train_loss": 0.028984980657696724, "cos_r_negg": [0.045679546892642975, 0.1608377993106842, 0.24076461791992188], "cos_innovation_negg": [0.43705978989601135, 0.5236760973930359, 0.6773004531860352], "cos_apical_negg": [0.045679546892642975, 0.1608377993106842, 0.24076461791992188], "cos_Ac_negg": [0.43844664096832275, 0.5239940881729126, 0.6781637668609619], "r_norm": [1.0056698322296143, 0.9973837733268738, 1.0014251470565796], "ldr": {"dL_sdil": -0.005319669842720032, "dL_bp": -0.00823262333869934, "ratio": 0.6461694679645672}}, {"epoch_end": 14, "step": 7035, "test_acc": 0.9653, "test_loss": 0.17632402153015136}], "final": {"test_acc": 0.9653, "test_loss": 0.17632402153015136, "wall_s": 22.891586780548096, "cos_r_negg": [0.05158198997378349, 0.16588646173477173, 0.24333959817886353], "cos_innovation_negg": [0.43237197399139404, 0.5271012783050537, 0.6925674080848694], "cos_apical_negg": [0.05158198997378349, 0.16588646173477173, 0.24333959817886353], "cos_Ac_negg": [0.43461257219314575, 0.5270849466323853, 0.6901035308837891]}}