1
|
{"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": 8, "use_residual": 1, "learn_A": 1, "learn_P": 1, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 2.0, "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": "diag_nuis_v2_rho2_res1_s0", "n_in": 784}, "provenance": {"git_commit": "54633b2e0f51a8f48d4ba627d778c87927f54ca1", "git_dirty": false}, "steps": [{"step": 0, "epoch": 0, "train_loss": 2.3603768348693848, "cos_r_negg": [-0.019525395706295967, -0.02552098035812378, 0.03377208486199379], "cos_apical_negg": [-0.008897066116333008, -0.013789067044854164, 0.004767646547406912], "cos_Ac_negg": [-0.019523516297340393, -0.02552095055580139, 0.033771760761737823], "r_norm": [4.8982954025268555, 4.826826095581055, 4.823868751525879], "ldr": {"dL_sdil": -0.009236812591552734, "dL_bp": -0.24081993103027344, "ratio": 0.03835568157517484}}, {"step": 100, "epoch": 0, "train_loss": 0.2104795277118683, "cos_r_negg": [0.4429137110710144, 0.3625786602497101, 0.3501937985420227], "cos_apical_negg": [0.02913225069642067, 0.0413375049829483, 0.09548983722925186], "cos_Ac_negg": [0.4604405462741852, 0.3800952434539795, 0.36436647176742554], "r_norm": [0.8362362384796143, 0.8200146555900574, 0.817680299282074]}, {"step": 200, "epoch": 0, "train_loss": 0.2786995768547058, "cos_r_negg": [0.46058589220046997, 0.38203614950180054, 0.3677726984024048], "cos_apical_negg": [0.014530155807733536, 0.041846051812171936, 0.09950847923755646], "cos_Ac_negg": [0.49076348543167114, 0.4162990152835846, 0.3995876908302307], "r_norm": [0.7659614086151123, 0.7472445964813232, 0.7463458776473999]}, {"step": 300, "epoch": 0, "train_loss": 0.1681954711675644, "cos_r_negg": [0.5001245737075806, 0.4624118208885193, 0.4346352815628052], "cos_apical_negg": [0.02112080529332161, 0.03908928483724594, 0.08193816989660263], "cos_Ac_negg": [0.5147418975830078, 0.4728078842163086, 0.43891194462776184], "r_norm": [0.5594145059585571, 0.5477821826934814, 0.5476260781288147]}, {"step": 400, "epoch": 0, "train_loss": 0.1623169183731079, "cos_r_negg": [0.5148628950119019, 0.48024970293045044, 0.430259644985199], "cos_apical_negg": [0.019099585711956024, 0.029904361814260483, 0.07807426899671555], "cos_Ac_negg": [0.5364471077919006, 0.4963504672050476, 0.4369904696941376], "r_norm": [0.5310893654823303, 0.5209400653839111, 0.5205697417259216]}, {"epoch_end": 0, "step": 469, "test_acc": 0.9452, "test_loss": 0.18473090362548827}, {"step": 500, "epoch": 1, "train_loss": 0.2528912127017975, "cos_r_negg": [0.4793739914894104, 0.46812039613723755, 0.4326707124710083], "cos_apical_negg": [0.014721769839525223, 0.04298064857721329, 0.09922119975090027], "cos_Ac_negg": [0.5140164494514465, 0.4957870841026306, 0.45290637016296387], "r_norm": [0.534934401512146, 0.5158824920654297, 0.5166623592376709], "ldr": {"dL_sdil": -0.018173053860664368, "dL_bp": -0.016316741704940796, "ratio": 1.1137673310818832}}, {"step": 600, "epoch": 1, "train_loss": 0.09457680583000183, "cos_r_negg": [0.48122870922088623, 0.4665687680244446, 0.43470096588134766], "cos_apical_negg": [0.00775455217808485, 0.030592598021030426, 0.09659455716609955], "cos_Ac_negg": [0.5211817026138306, 0.49690839648246765, 0.46077242493629456], "r_norm": [0.4056410491466522, 0.39857375621795654, 0.39543288946151733]}, {"step": 700, "epoch": 1, "train_loss": 0.18847103416919708, "cos_r_negg": [0.465390682220459, 0.4821118116378784, 0.45879414677619934], "cos_apical_negg": [0.003821765072643757, 0.04252293333411217, 0.1065407395362854], "cos_Ac_negg": [0.515001654624939, 0.5152771472930908, 0.4849601984024048], "r_norm": [0.4375430941581726, 0.42695391178131104, 0.4247203469276428]}, {"step": 800, "epoch": 1, "train_loss": 0.16579097509384155, "cos_r_negg": [0.508460283279419, 0.5228742361068726, 0.47062426805496216], "cos_apical_negg": [0.01265665516257286, 0.04129909351468086, 0.10414852201938629], "cos_Ac_negg": [0.541027843952179, 0.5396636724472046, 0.48390957713127136], "r_norm": [0.3352852463722229, 0.3273393213748932, 0.32507359981536865]}, {"step": 900, "epoch": 1, "train_loss": 0.18073377013206482, "cos_r_negg": [0.5507323741912842, 0.5638676881790161, 0.48661351203918457], "cos_apical_negg": [0.003939080983400345, 0.0321526899933815, 0.09565563499927521], "cos_Ac_negg": [0.5651798248291016, 0.5708035230636597, 0.48854005336761475], "r_norm": [0.3477749824523926, 0.3410939872264862, 0.33893710374832153]}, {"epoch_end": 1, "step": 938, "test_acc": 0.9545, "test_loss": 0.14937169456481933}, {"step": 1000, "epoch": 2, "train_loss": 0.05696079134941101, "cos_r_negg": [0.503943681716919, 0.5188122987747192, 0.47791409492492676], "cos_apical_negg": [0.005241388455033302, 0.03323132544755936, 0.09120991826057434], "cos_Ac_negg": [0.5342360138893127, 0.5386021137237549, 0.48692798614501953], "r_norm": [0.30671465396881104, 0.29946863651275635, 0.29653069376945496], "ldr": {"dL_sdil": -0.008440732955932617, "dL_bp": -0.007455542683601379, "ratio": 1.132141993432374}}, {"step": 1100, "epoch": 2, "train_loss": 0.09730264544487, "cos_r_negg": [0.4988236427307129, 0.5263614058494568, 0.47083091735839844], "cos_apical_negg": [0.01208480540663004, 0.04290391132235527, 0.08295844495296478], "cos_Ac_negg": [0.5341017246246338, 0.5504165887832642, 0.4866880178451538], "r_norm": [0.35640671849250793, 0.3450624346733093, 0.3451017439365387]}, {"step": 1200, "epoch": 2, "train_loss": 0.1298961490392685, "cos_r_negg": [0.5133548974990845, 0.5411757826805115, 0.4817139804363251], "cos_apical_negg": [0.0056552765890955925, 0.03371363878250122, 0.08027148246765137], "cos_Ac_negg": [0.5373932123184204, 0.5554259419441223, 0.4888988137245178], "r_norm": [0.3141748905181885, 0.3061453700065613, 0.3053479790687561]}, {"step": 1300, "epoch": 2, "train_loss": 0.14578621089458466, "cos_r_negg": [0.5057678818702698, 0.528052806854248, 0.4672167897224426], "cos_apical_negg": [0.003955305553972721, 0.039266444742679596, 0.09113456308841705], "cos_Ac_negg": [0.535873532295227, 0.5495326519012451, 0.4734228253364563], "r_norm": [0.32081305980682373, 0.3140150308609009, 0.311356782913208]}, {"step": 1400, "epoch": 2, "train_loss": 0.15383687615394592, "cos_r_negg": [0.47028887271881104, 0.5024353861808777, 0.45877429842948914], "cos_apical_negg": [-0.000482750590890646, 0.043554145842790604, 0.09052295982837677], "cos_Ac_negg": [0.4933069348335266, 0.5139294266700745, 0.46155816316604614], "r_norm": [0.28868618607521057, 0.28035980463027954, 0.2792937159538269]}, {"epoch_end": 2, "step": 1407, "test_acc": 0.9641, "test_loss": 0.12704081268310546}, {"step": 1500, "epoch": 3, "train_loss": 0.07152651250362396, "cos_r_negg": [0.46427416801452637, 0.5024504661560059, 0.4708280861377716], "cos_apical_negg": [0.005512584000825882, 0.04187152534723282, 0.09414386749267578], "cos_Ac_negg": [0.5033773183822632, 0.5319558382034302, 0.4813491702079773], "r_norm": [0.23926027119159698, 0.23365998268127441, 0.2324751913547516], "ldr": {"dL_sdil": -0.004995793104171753, "dL_bp": -0.004729658365249634, "ratio": 1.0562693366771476}}, {"step": 1600, "epoch": 3, "train_loss": 0.0974859744310379, "cos_r_negg": [0.4652981460094452, 0.5063573122024536, 0.48069697618484497], "cos_apical_negg": [-0.008264572359621525, 0.03116120584309101, 0.09904827177524567], "cos_Ac_negg": [0.5057864189147949, 0.541628360748291, 0.49891218543052673], "r_norm": [0.2457239031791687, 0.23975686728954315, 0.23840028047561646]}, {"step": 1700, "epoch": 3, "train_loss": 0.0348355695605278, "cos_r_negg": [0.47419023513793945, 0.513866662979126, 0.48939383029937744], "cos_apical_negg": [-0.009268120862543583, 0.03703146427869797, 0.10172654688358307], "cos_Ac_negg": [0.5040693879127502, 0.5393393039703369, 0.49916988611221313], "r_norm": [0.2812294661998749, 0.27222853899002075, 0.2717058062553406]}, {"step": 1800, "epoch": 3, "train_loss": 0.021023377776145935, "cos_r_negg": [0.465406596660614, 0.5011543035507202, 0.4807687997817993], "cos_apical_negg": [-0.002072057453915477, 0.028440648689866066, 0.08577912300825119], "cos_Ac_negg": [0.5046777725219727, 0.5338564515113831, 0.49394911527633667], "r_norm": [0.21957169473171234, 0.21516944468021393, 0.2121545374393463]}, {"epoch_end": 3, "step": 1876, "test_acc": 0.9674, "test_loss": 0.1202873867034912}, {"step": 1900, "epoch": 4, "train_loss": 0.02093462646007538, "cos_r_negg": [0.4647122025489807, 0.5101388096809387, 0.4865616261959076], "cos_apical_negg": [-0.005798174999654293, 0.03767603635787964, 0.08937495946884155], "cos_Ac_negg": [0.5050925612449646, 0.5463800430297852, 0.5008324384689331], "r_norm": [0.2384132444858551, 0.2347446233034134, 0.23052723705768585]}, {"step": 2000, "epoch": 4, "train_loss": 0.10467495024204254, "cos_r_negg": [0.3863235116004944, 0.4237010180950165, 0.4237331748008728], "cos_apical_negg": [-0.01224249042570591, 0.03634224086999893, 0.09541654586791992], "cos_Ac_negg": [0.4352112412452698, 0.4720419943332672, 0.44034767150878906], "r_norm": [0.16932746767997742, 0.16426919400691986, 0.16319605708122253], "ldr": {"dL_sdil": -0.003473937511444092, "dL_bp": -0.0037461817264556885, "ratio": 0.9273275471157747}}, {"step": 2100, "epoch": 4, "train_loss": 0.017748109996318817, "cos_r_negg": [0.4375082850456238, 0.48254895210266113, 0.47639352083206177], "cos_apical_negg": [0.010370379313826561, 0.03625763952732086, 0.09920184314250946], "cos_Ac_negg": [0.4788301885128021, 0.5247420072555542, 0.494262158870697], "r_norm": [0.20781828463077545, 0.20187830924987793, 0.20186066627502441]}, {"step": 2200, "epoch": 4, "train_loss": 0.012651554308831692, "cos_r_negg": [0.40401971340179443, 0.4583597183227539, 0.4443907141685486], "cos_apical_negg": [0.001474043820053339, 0.028314366936683655, 0.08980541676282883], "cos_Ac_negg": [0.44599056243896484, 0.4941726624965668, 0.4544076919555664], "r_norm": [0.19201666116714478, 0.1869010627269745, 0.1835378110408783]}, {"step": 2300, "epoch": 4, "train_loss": 0.12230519950389862, "cos_r_negg": [0.39164015650749207, 0.43627384305000305, 0.4347958564758301], "cos_apical_negg": [-0.0016493774019181728, 0.0343293771147728, 0.10509329289197922], "cos_Ac_negg": [0.44150510430336, 0.48020583391189575, 0.4471698999404907], "r_norm": [0.19779588282108307, 0.19356980919837952, 0.1928449273109436]}, {"epoch_end": 4, "step": 2345, "test_acc": 0.9671, "test_loss": 0.12299122066497803}, {"step": 2400, "epoch": 5, "train_loss": 0.07835916429758072, "cos_r_negg": [0.39116325974464417, 0.44303035736083984, 0.43706852197647095], "cos_apical_negg": [0.0005689701065421104, 0.03678145259618759, 0.10175281018018723], "cos_Ac_negg": [0.43070727586746216, 0.4839155972003937, 0.44670569896698], "r_norm": [0.17620541155338287, 0.17209164798259735, 0.16851550340652466]}, {"step": 2500, "epoch": 5, "train_loss": 0.03489034250378609, "cos_r_negg": [0.3287951350212097, 0.3683970868587494, 0.37697267532348633], "cos_apical_negg": [-0.0003596874885261059, 0.032686635851860046, 0.09533073753118515], "cos_Ac_negg": [0.3681502342224121, 0.4131554365158081, 0.38685232400894165], "r_norm": [0.21968555450439453, 0.21451908349990845, 0.21199902892112732], "ldr": {"dL_sdil": -0.00855495035648346, "dL_bp": -0.009747728705406189, "ratio": 0.8776352538144396}}, {"step": 2600, "epoch": 5, "train_loss": 0.022090446203947067, "cos_r_negg": [0.35706132650375366, 0.39211851358413696, 0.40173202753067017], "cos_apical_negg": [-0.003336722496896982, 0.029070865362882614, 0.09573492407798767], "cos_Ac_negg": [0.3907826244831085, 0.43139955401420593, 0.4079221189022064], "r_norm": [0.1914403736591339, 0.1862468719482422, 0.18351085484027863]}, {"step": 2700, "epoch": 5, "train_loss": 0.04956228286027908, "cos_r_negg": [0.3962959051132202, 0.42606425285339355, 0.42945998907089233], "cos_apical_negg": [-0.006446274928748608, 0.027332115918397903, 0.09695732593536377], "cos_Ac_negg": [0.43585681915283203, 0.4715687036514282, 0.44163092970848083], "r_norm": [0.1681450456380844, 0.16476768255233765, 0.16256943345069885]}, {"step": 2800, "epoch": 5, "train_loss": 0.061803221702575684, "cos_r_negg": [0.3516920804977417, 0.3995523154735565, 0.406349778175354], "cos_apical_negg": [0.010522197932004929, 0.03801455348730087, 0.1026088148355484], "cos_Ac_negg": [0.39506417512893677, 0.44050419330596924, 0.41427141427993774], "r_norm": [0.18657535314559937, 0.18346653878688812, 0.18025299906730652]}, {"epoch_end": 5, "step": 2814, "test_acc": 0.9712, "test_loss": 0.11963936405181885}, {"step": 2900, "epoch": 6, "train_loss": 0.0029091197066009045, "cos_r_negg": [0.3168780207633972, 0.3566356301307678, 0.36284130811691284], "cos_apical_negg": [-0.002426113933324814, 0.03104003146290779, 0.10274471342563629], "cos_Ac_negg": [0.35930225253105164, 0.3978833556175232, 0.37234923243522644], "r_norm": [0.15753984451293945, 0.15396586060523987, 0.15256044268608093]}, {"step": 3000, "epoch": 6, "train_loss": 0.06555234640836716, "cos_r_negg": [0.29954034090042114, 0.3406285345554352, 0.35170841217041016], "cos_apical_negg": [0.004214042332023382, 0.03327927738428116, 0.09258485585451126], "cos_Ac_negg": [0.34664613008499146, 0.3861979842185974, 0.36547720432281494], "r_norm": [0.15212282538414001, 0.14929822087287903, 0.14506179094314575], "ldr": {"dL_sdil": -0.0056327953934669495, "dL_bp": -0.007527798414230347, "ratio": 0.7482659714716676}}, {"step": 3100, "epoch": 6, "train_loss": 0.005165440030395985, "cos_r_negg": [0.2905978262424469, 0.3293302655220032, 0.33835867047309875], "cos_apical_negg": [0.007865103892982006, 0.03524419665336609, 0.0903964713215828], "cos_Ac_negg": [0.3323310315608978, 0.37043750286102295, 0.34681978821754456], "r_norm": [0.1576562076807022, 0.1533031165599823, 0.15176625549793243]}, {"step": 3200, "epoch": 6, "train_loss": 0.024154968559741974, "cos_r_negg": [0.2837607264518738, 0.3160426914691925, 0.3240395188331604], "cos_apical_negg": [0.007447120267897844, 0.02975824847817421, 0.0925460159778595], "cos_Ac_negg": [0.32272806763648987, 0.3581736087799072, 0.33435964584350586], "r_norm": [0.1338508129119873, 0.13185735046863556, 0.12934786081314087]}, {"epoch_end": 6, "step": 3283, "test_acc": 0.972, "test_loss": 0.11885602874755859}, {"step": 3300, "epoch": 7, "train_loss": 0.011400345712900162, "cos_r_negg": [0.2728060185909271, 0.3027855455875397, 0.319587767124176], "cos_apical_negg": [-0.013917852193117142, 0.02657162956893444, 0.10002721101045609], "cos_Ac_negg": [0.31768205761909485, 0.34560680389404297, 0.3265172243118286], "r_norm": [0.1522861123085022, 0.14947503805160522, 0.14627283811569214]}, {"step": 3400, "epoch": 7, "train_loss": 0.01935477927327156, "cos_r_negg": [0.24118351936340332, 0.2715790569782257, 0.2915768623352051], "cos_apical_negg": [-0.00106140598654747, 0.022906463593244553, 0.10799513757228851], "cos_Ac_negg": [0.2775115370750427, 0.3115522265434265, 0.2949219048023224], "r_norm": [0.16454564034938812, 0.16119705140590668, 0.15943534672260284]}, {"step": 3500, "epoch": 7, "train_loss": 0.009277483448386192, "cos_r_negg": [0.22907055914402008, 0.2603045701980591, 0.2829299569129944], "cos_apical_negg": [0.002978227101266384, 0.030053822323679924, 0.1073094829916954], "cos_Ac_negg": [0.2745738625526428, 0.3033802807331085, 0.29376792907714844], "r_norm": [0.14773321151733398, 0.1443905234336853, 0.14236226677894592], "ldr": {"dL_sdil": -0.005328170955181122, "dL_bp": -0.007607020437717438, "ratio": 0.7004281109542927}}, {"step": 3600, "epoch": 7, "train_loss": 0.0019572130404412746, "cos_r_negg": [0.22610728442668915, 0.2564864456653595, 0.2687748372554779], "cos_apical_negg": [-0.0074714538641273975, 0.03298413008451462, 0.1017625704407692], "cos_Ac_negg": [0.2628031075000763, 0.2908572554588318, 0.27389365434646606], "r_norm": [0.16391804814338684, 0.16055110096931458, 0.15846961736679077]}, {"step": 3700, "epoch": 7, "train_loss": 0.006348735652863979, "cos_r_negg": [0.2245323210954666, 0.26149487495422363, 0.27766892313957214], "cos_apical_negg": [-0.003066269913688302, 0.035145848989486694, 0.09773439168930054], "cos_Ac_negg": [0.2678852677345276, 0.30052393674850464, 0.2874765694141388], "r_norm": [0.1471036672592163, 0.14351904392242432, 0.14086958765983582]}, {"epoch_end": 7, "step": 3752, "test_acc": 0.9727, "test_loss": 0.11842634115219117}, {"step": 3800, "epoch": 8, "train_loss": 0.008779313415288925, "cos_r_negg": [0.20920786261558533, 0.24275532364845276, 0.2592512369155884], "cos_apical_negg": [-0.005042918957769871, 0.0364946611225605, 0.10146982222795486], "cos_Ac_negg": [0.25226157903671265, 0.2811870574951172, 0.27063748240470886], "r_norm": [0.15609592199325562, 0.15288537740707397, 0.15019282698631287]}, {"step": 3900, "epoch": 8, "train_loss": 0.029239613562822342, "cos_r_negg": [0.1989031434059143, 0.2301427274942398, 0.24724996089935303], "cos_apical_negg": [-0.00978272408246994, 0.03374071791768074, 0.09561595320701599], "cos_Ac_negg": [0.24267317354679108, 0.2690669596195221, 0.258054256439209], "r_norm": [0.15312577784061432, 0.15000101923942566, 0.1476624608039856]}, {"step": 4000, "epoch": 8, "train_loss": 0.017709914594888687, "cos_r_negg": [0.19030193984508514, 0.2226574569940567, 0.23706385493278503], "cos_apical_negg": [-0.0130704864859581, 0.02808399870991707, 0.09148284792900085], "cos_Ac_negg": [0.2278730422258377, 0.2558026909828186, 0.246942937374115], "r_norm": [0.13309969007968903, 0.13005393743515015, 0.12923449277877808], "ldr": {"dL_sdil": -0.005619361996650696, "dL_bp": -0.0077913179993629456, "ratio": 0.721233813985023}}, {"step": 4100, "epoch": 8, "train_loss": 0.009126199409365654, "cos_r_negg": [0.19683533906936646, 0.2265663743019104, 0.24318739771842957], "cos_apical_negg": [0.0003269389271736145, 0.03210022300481796, 0.08994849026203156], "cos_Ac_negg": [0.2389659285545349, 0.2672910690307617, 0.2573346197605133], "r_norm": [0.13370434939861298, 0.13035769760608673, 0.12766152620315552]}, {"step": 4200, "epoch": 8, "train_loss": 0.016402723267674446, "cos_r_negg": [0.1817178726196289, 0.211750328540802, 0.22712285816669464], "cos_apical_negg": [0.004159575328230858, 0.03204348310828209, 0.09628812223672867], "cos_Ac_negg": [0.22042880952358246, 0.24788206815719604, 0.23804104328155518], "r_norm": [0.15196307003498077, 0.14829201996326447, 0.14516544342041016]}, {"epoch_end": 8, "step": 4221, "test_acc": 0.972, "test_loss": 0.12942726612091066}, {"step": 4300, "epoch": 9, "train_loss": 0.0011772941797971725, "cos_r_negg": [0.171378031373024, 0.2018832415342331, 0.21645718812942505], "cos_apical_negg": [0.0004408275708556175, 0.03153624385595322, 0.09550978988409042], "cos_Ac_negg": [0.21185803413391113, 0.24009981751441956, 0.2312624454498291], "r_norm": [0.16429108381271362, 0.16050873696804047, 0.1585191786289215]}, {"step": 4400, "epoch": 9, "train_loss": 0.0017589031485840678, "cos_r_negg": [0.18395528197288513, 0.21181973814964294, 0.22694475948810577], "cos_apical_negg": [0.004927505739033222, 0.03397145867347717, 0.09346376359462738], "cos_Ac_negg": [0.22359634935855865, 0.25286808609962463, 0.2441624402999878], "r_norm": [0.15238656103610992, 0.14788872003555298, 0.1455930471420288]}, {"step": 4500, "epoch": 9, "train_loss": 0.0499572828412056, "cos_r_negg": [0.16712220013141632, 0.19889728724956512, 0.21286870539188385], "cos_apical_negg": [0.000697003211826086, 0.03215166926383972, 0.0884338840842247], "cos_Ac_negg": [0.20319169759750366, 0.227967768907547, 0.22162587940692902], "r_norm": [0.1630411297082901, 0.15995687246322632, 0.15871787071228027], "ldr": {"dL_sdil": -0.007485546171665192, "dL_bp": -0.011915862560272217, "ratio": 0.6282001100467699}}, {"step": 4600, "epoch": 9, "train_loss": 0.00047172076301649213, "cos_r_negg": [0.1720544397830963, 0.19713377952575684, 0.20793026685714722], "cos_apical_negg": [0.0006287680007517338, 0.03027612343430519, 0.08023521304130554], "cos_Ac_negg": [0.2048678696155548, 0.22890585660934448, 0.2191743701696396], "r_norm": [0.14321473240852356, 0.13957840204238892, 0.13830171525478363]}, {"epoch_end": 9, "step": 4690, "test_acc": 0.9717, "test_loss": 0.1321223569869995}, {"step": 4700, "epoch": 10, "train_loss": 0.002440714742988348, "cos_r_negg": [0.15346384048461914, 0.17721645534038544, 0.19002754986286163], "cos_apical_negg": [-0.007044458761811256, 0.030264757573604584, 0.0846354067325592], "cos_Ac_negg": [0.18284618854522705, 0.20435011386871338, 0.1971147656440735], "r_norm": [0.15336430072784424, 0.14960631728172302, 0.14814691245555878]}, {"step": 4800, "epoch": 10, "train_loss": 0.004119536839425564, "cos_r_negg": [0.14772538840770721, 0.17196312546730042, 0.18439564108848572], "cos_apical_negg": [-0.00809923093765974, 0.030700212344527245, 0.08027894794940948], "cos_Ac_negg": [0.17864659428596497, 0.19959218800067902, 0.19346624612808228], "r_norm": [0.1517089605331421, 0.14809250831604004, 0.14698970317840576]}, {"step": 4900, "epoch": 10, "train_loss": 0.0019432123517617583, "cos_r_negg": [0.1436895728111267, 0.16759449243545532, 0.18327967822551727], "cos_apical_negg": [-0.004903081804513931, 0.029023364186286926, 0.08085942268371582], "cos_Ac_negg": [0.17696651816368103, 0.19628387689590454, 0.19116395711898804], "r_norm": [0.1441417783498764, 0.14046290516853333, 0.13939125835895538]}, {"step": 5000, "epoch": 10, "train_loss": 0.0011039283126592636, "cos_r_negg": [0.1456233710050583, 0.16664552688598633, 0.18366661667823792], "cos_apical_negg": [-0.00496734818443656, 0.027270881459116936, 0.07906773686408997], "cos_Ac_negg": [0.177052304148674, 0.19654516875743866, 0.192416250705719], "r_norm": [0.153082475066185, 0.1500624716281891, 0.1487290859222412], "ldr": {"dL_sdil": -0.00773930549621582, "dL_bp": -0.012882724404335022, "ratio": 0.6007506838857434}}, {"step": 5100, "epoch": 10, "train_loss": 0.0019905571825802326, "cos_r_negg": [0.14017845690250397, 0.1623867303133011, 0.1790478378534317], "cos_apical_negg": [-0.006340588442981243, 0.02820134535431862, 0.07454545050859451], "cos_Ac_negg": [0.17037701606750488, 0.18867020308971405, 0.18366976082324982], "r_norm": [0.1452169120311737, 0.1420271098613739, 0.14085450768470764]}, {"epoch_end": 10, "step": 5159, "test_acc": 0.9727, "test_loss": 0.13825255422592164}, {"step": 5200, "epoch": 11, "train_loss": 0.00032047624699771404, "cos_r_negg": [0.13907811045646667, 0.16218888759613037, 0.17825764417648315], "cos_apical_negg": [-0.004437034949660301, 0.027674730867147446, 0.07651957124471664], "cos_Ac_negg": [0.16903731226921082, 0.18682733178138733, 0.18353214859962463], "r_norm": [0.14994730055332184, 0.14747193455696106, 0.1453215479850769]}, {"step": 5300, "epoch": 11, "train_loss": 0.001317075453698635, "cos_r_negg": [0.1397765725851059, 0.15966805815696716, 0.17645254731178284], "cos_apical_negg": [-0.007834138348698616, 0.027934767305850983, 0.0747075229883194], "cos_Ac_negg": [0.17215052247047424, 0.19023580849170685, 0.185514435172081], "r_norm": [0.15098218619823456, 0.14723503589630127, 0.14609037339687347]}, {"step": 5400, "epoch": 11, "train_loss": 0.0019159776857122779, "cos_r_negg": [0.13290445506572723, 0.1524316817522049, 0.16961708664894104], "cos_apical_negg": [-0.011777791194617748, 0.027024639770388603, 0.07163079082965851], "cos_Ac_negg": [0.16187609732151031, 0.1794547736644745, 0.1759195327758789], "r_norm": [0.14193996787071228, 0.13908618688583374, 0.1379406601190567]}, {"step": 5500, "epoch": 11, "train_loss": 0.0018864399753510952, "cos_r_negg": [0.1298910677433014, 0.15068814158439636, 0.16763979196548462], "cos_apical_negg": [-0.01511571928858757, 0.02759353071451187, 0.07189469039440155], "cos_Ac_negg": [0.15988746285438538, 0.17757219076156616, 0.17409983277320862], "r_norm": [0.14620211720466614, 0.14322549104690552, 0.14220669865608215], "ldr": {"dL_sdil": -0.007692717015743256, "dL_bp": -0.01239149272441864, "ratio": 0.6208063214679543}}, {"step": 5600, "epoch": 11, "train_loss": 0.00045360150397755206, "cos_r_negg": [0.12779346108436584, 0.14734984934329987, 0.1627187728881836], "cos_apical_negg": [-0.0132527407258749, 0.02890801802277565, 0.07118037343025208], "cos_Ac_negg": [0.15754546225070953, 0.17433032393455505, 0.1712392419576645], "r_norm": [0.1398567259311676, 0.1368670016527176, 0.1359477937221527]}, {"epoch_end": 11, "step": 5628, "test_acc": 0.9727, "test_loss": 0.1380262689590454}, {"step": 5700, "epoch": 12, "train_loss": 0.0010436055017635226, "cos_r_negg": [0.12593592703342438, 0.14469364285469055, 0.16063137352466583], "cos_apical_negg": [-0.014148636721074581, 0.02715049311518669, 0.06879954785108566], "cos_Ac_negg": [0.15484890341758728, 0.17134228348731995, 0.16855867207050323], "r_norm": [0.13897186517715454, 0.13628479838371277, 0.1349405199289322]}, {"step": 5800, "epoch": 12, "train_loss": 0.000346480926964432, "cos_r_negg": [0.12032702565193176, 0.13968771696090698, 0.15657800436019897], "cos_apical_negg": [-0.013153620064258575, 0.026260606944561005, 0.06714522838592529], "cos_Ac_negg": [0.15047548711299896, 0.1682988554239273, 0.1652776002883911], "r_norm": [0.14279009401798248, 0.13989438116550446, 0.1387310028076172]}, {"step": 5900, "epoch": 12, "train_loss": 0.001038936898112297, "cos_r_negg": [0.12642058730125427, 0.1462206393480301, 0.1628953218460083], "cos_apical_negg": [-0.014939341694116592, 0.027533335611224174, 0.06816059350967407], "cos_Ac_negg": [0.15559664368629456, 0.17304499447345734, 0.17043958604335785], "r_norm": [0.14304625988006592, 0.1399763524532318, 0.13858327269554138]}, {"step": 6000, "epoch": 12, "train_loss": 0.0006938993465155363, "cos_r_negg": [0.12021420150995255, 0.13889288902282715, 0.15497863292694092], "cos_apical_negg": [-0.01536024920642376, 0.027249818667769432, 0.06561246514320374], "cos_Ac_negg": [0.14953388273715973, 0.16581204533576965, 0.16294600069522858], "r_norm": [0.14249981939792633, 0.13942919671535492, 0.13825908303260803], "ldr": {"dL_sdil": -0.0077394768595695496, "dL_bp": -0.013016998767852783, "ratio": 0.5945669195792829}}, {"epoch_end": 12, "step": 6097, "test_acc": 0.9728, "test_loss": 0.14158571338653564}, {"step": 6100, "epoch": 13, "train_loss": 0.0006675304030068219, "cos_r_negg": [0.12018933892250061, 0.13733035326004028, 0.15323753654956818], "cos_apical_negg": [-0.01626214012503624, 0.02698269486427307, 0.06725136935710907], "cos_Ac_negg": [0.14689643681049347, 0.16226354241371155, 0.15944406390190125], "r_norm": [0.14575155079364777, 0.1428396850824356, 0.1414227932691574]}, {"step": 6200, "epoch": 13, "train_loss": 0.0004157752846367657, "cos_r_negg": [0.11543910205364227, 0.1320739984512329, 0.14808131754398346], "cos_apical_negg": [-0.016483131796121597, 0.02685311622917652, 0.06537206470966339], "cos_Ac_negg": [0.1437090039253235, 0.15889741480350494, 0.15584781765937805], "r_norm": [0.14185965061187744, 0.1389819085597992, 0.13789618015289307]}, {"step": 6300, "epoch": 13, "train_loss": 0.0011423379182815552, "cos_r_negg": [0.1137993186712265, 0.1301238238811493, 0.14570337533950806], "cos_apical_negg": [-0.017200684174895287, 0.026560001075267792, 0.06553532183170319], "cos_Ac_negg": [0.14120899140834808, 0.15593808889389038, 0.15304967761039734], "r_norm": [0.14338231086730957, 0.14059233665466309, 0.1393817961215973]}, {"step": 6400, "epoch": 13, "train_loss": 0.0003152909630443901, "cos_r_negg": [0.11512938141822815, 0.13242846727371216, 0.14829787611961365], "cos_apical_negg": [-0.015821989625692368, 0.02621270716190338, 0.06602505594491959], "cos_Ac_negg": [0.14223501086235046, 0.15746155381202698, 0.15523487329483032], "r_norm": [0.1417124718427658, 0.13895007967948914, 0.13768382370471954]}, {"step": 6500, "epoch": 13, "train_loss": 0.00015983209596015513, "cos_r_negg": [0.11377174407243729, 0.1315324753522873, 0.1473967730998993], "cos_apical_negg": [-0.01693193055689335, 0.02656492590904236, 0.06760591268539429], "cos_Ac_negg": [0.14286907017230988, 0.15832945704460144, 0.15576104819774628], "r_norm": [0.14350740611553192, 0.14046916365623474, 0.1392725259065628], "ldr": {"dL_sdil": -0.00824255496263504, "dL_bp": -0.014194592833518982, "ratio": 0.5806827331581605}}, {"epoch_end": 13, "step": 6566, "test_acc": 0.9727, "test_loss": 0.14322918624877928}, {"step": 6600, "epoch": 14, "train_loss": 0.0003591316635720432, "cos_r_negg": [0.11212986707687378, 0.1291949599981308, 0.14487352967262268], "cos_apical_negg": [-0.014751208014786243, 0.02600904554128647, 0.06539197266101837], "cos_Ac_negg": [0.1404675394296646, 0.15541696548461914, 0.15315845608711243], "r_norm": [0.13971495628356934, 0.13707971572875977, 0.13579395413398743]}, {"step": 6700, "epoch": 14, "train_loss": 0.0004953234456479549, "cos_r_negg": [0.11256467550992966, 0.12910374999046326, 0.1448487937450409], "cos_apical_negg": [-0.014729049056768417, 0.025265991687774658, 0.0659550279378891], "cos_Ac_negg": [0.14007018506526947, 0.155273899435997, 0.15296408534049988], "r_norm": [0.1439666450023651, 0.14105452597141266, 0.13976356387138367]}, {"step": 6800, "epoch": 14, "train_loss": 0.0005221293540671468, "cos_r_negg": [0.1103338971734047, 0.12624657154083252, 0.1417093127965927], "cos_apical_negg": [-0.015459051355719566, 0.025464942678809166, 0.06509223580360413], "cos_Ac_negg": [0.13784286379814148, 0.15203189849853516, 0.15006974339485168], "r_norm": [0.1405353844165802, 0.13785463571548462, 0.13657337427139282]}, {"step": 6900, "epoch": 14, "train_loss": 0.0006800569244660437, "cos_r_negg": [0.10567227005958557, 0.12139198929071426, 0.13738387823104858], "cos_apical_negg": [-0.017395276576280594, 0.025409504771232605, 0.064261794090271], "cos_Ac_negg": [0.1346426159143448, 0.14878396689891815, 0.14626097679138184], "r_norm": [0.13921457529067993, 0.13651230931282043, 0.13532032072544098]}, {"step": 7000, "epoch": 14, "train_loss": 0.000475144712254405, "cos_r_negg": [0.1054995208978653, 0.12289255857467651, 0.13774460554122925], "cos_apical_negg": [-0.01858312450349331, 0.026010826230049133, 0.06366252899169922], "cos_Ac_negg": [0.13563184440135956, 0.1502842903137207, 0.14762620627880096], "r_norm": [0.14071890711784363, 0.13789233565330505, 0.1369301676750183], "ldr": {"dL_sdil": -0.008366659283638, "dL_bp": -0.014673717319965363, "ratio": 0.57017994153766}}, {"epoch_end": 14, "step": 7035, "test_acc": 0.9731, "test_loss": 0.14614157485961915}], "final": {"test_acc": 0.9731, "test_loss": 0.14614157485961915, "wall_s": 41.07883429527283, "cos_r_negg": [0.10460954904556274, 0.12169840931892395, 0.13700059056282043], "cos_apical_negg": [-0.01877596229314804, 0.026302088052034378, 0.06400805711746216], "cos_Ac_negg": [0.13483735918998718, 0.14949391782283783, 0.14689740538597107]}}
|