1
|
{"args": {"mode": "sdil", "dataset": "mnist", "depth": 3, "width": 256, "act": "tanh", "residual": 1, "epochs": 15, "batch_size": 128, "train_examples": 0, "val_examples": 5000, "split_seed": 2027, "eval_split": "validation", "task_seed": 0, "task_train_examples": 50000, "task_test_examples": 10000, "task_levels": 8, "task_n_in": 128, "task_classes": 10, "teacher_depth": 8, "teacher_width": 64, "teacher_residual": 1, "eta": 0.05, "eta_A": 0.02, "eta_P": 0.01, "momentum": 0.9, "w_scale": 1.0, "a_scale": 1.0, "feedback_scale": 1.0, "pert_sigma": 0.01, "pert_every": 4, "pert_ndirs": 16, "pert_mode": "simultaneous", "use_residual": 1, "raw_scale_control": "none", "learn_A": 1, "learn_P": 1, "p_neutral": 1, "p_warmup_steps": 200, "p_warmup_eta": 0.05, "nuis_rho": 0.5, "traffic_mode": "topdown", "predictor_mode": "diagonal", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 0, "device": "cuda", "log_every": 200, "diagnostics": "alignment", "eval_every": 1, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "traffic_time_dev_v1_mnist_topdown_rho0p5_ep0p01_wu200_d3_s0", "n_in": 784, "n_out": 10}, "split": {"dataset": "mnist", "split_seed": 2027, "validation_examples": 5000, "validation_index_sha256": "c4cec216363dc171cddb643b06414235ea84da4e5560059f6e10451287ef2e3d", "validation_class_counts": {"0": 500, "1": 500, "2": 500, "3": 500, "4": 500, "5": 500, "6": 500, "7": 500, "8": 500, "9": 500}, "train_examples": 55000, "test_examples": 10000, "split_from_training_only": true, "evaluation_split": "validation"}, "provenance": {"git_commit": "17bcd975b18d826010e2e4b83f009cbc3fb11e72", "git_dirty": false}, "steps": [{"step": 0, "epoch": 0, "train_loss": 2.6220171451568604, "cos_r_negg": [-0.010398169979453087, 0.027888260781764984, 0.03532874584197998], "cos_innovation_negg": [-0.010398169979453087, 0.027888260781764984, 0.03532874584197998], "cos_apical_negg": [-0.018074408173561096, 0.009428558871150017, 0.0030616112053394318], "cos_Ac_negg": [0.002011236734688282, 0.029851969331502914, 0.03533003106713295], "r_norm": [5.842154026031494, 5.354958534240723, 4.824590682983398], "traffic_norm": [6.859302520751953, 6.838433265686035, 6.696292877197266], "traffic_residual_norm": [3.1803629398345947, 2.325282573699951, 0.0010244514560326934], "traffic_r2": [0.7853798270225525, 0.8844690918922424, 1.0]}, {"step": 200, "epoch": 0, "train_loss": 1.9221855401992798, "cos_r_negg": [0.3072468936443329, 0.07879030704498291, 0.09927916526794434], "cos_innovation_negg": [0.3072468936443329, 0.07879030704498291, 0.09927916526794434], "cos_apical_negg": [0.12461313605308533, -0.34373581409454346, -0.4874652624130249], "cos_Ac_negg": [0.7383130788803101, 0.0981222465634346, 0.03428724408149719], "r_norm": [3.1366655826568604, 2.4359447956085205, 1.9612362384796143], "traffic_norm": [16.686153411865234, 16.617891311645508, 16.35941505432129], "traffic_residual_norm": [1.6023588180541992, 0.8013944029808044, 0.00016922412032727152], "traffic_r2": [0.9907783269882202, 0.9976743459701538, 1.0]}, {"step": 400, "epoch": 0, "train_loss": 0.8494653701782227, "cos_r_negg": [0.20249155163764954, -0.025379780679941177, 0.07149733603000641], "cos_innovation_negg": [0.20249155163764954, -0.025379780679941177, 0.07149733603000641], "cos_apical_negg": [0.002028997987508774, 0.013634475879371166, 0.03637028858065605], "cos_Ac_negg": [0.8233023881912231, -0.08157878369092941, 0.08842548727989197], "r_norm": [1.1659345626831055, 1.011657476425171, 0.9293937683105469], "traffic_norm": [16.733749389648438, 16.693849563598633, 16.416526794433594], "traffic_residual_norm": [0.2277946174144745, 0.11582747101783752, 3.093773193540983e-05], "traffic_r2": [0.9998146295547485, 0.9999518394470215, 1.0]}, {"epoch_end": 0, "step": 430, "eval_split": "validation", "eval_acc": 0.8224, "eval_loss": 0.853778466796875, "val_acc": 0.8224, "val_loss": 0.853778466796875}, {"step": 600, "epoch": 1, "train_loss": 0.9685262441635132, "cos_r_negg": [0.40377822518348694, -0.04111352190375328, 0.0827951654791832], "cos_innovation_negg": [0.40377822518348694, -0.04111352190375328, 0.0827951654791832], "cos_apical_negg": [0.015169039368629456, 0.01175125315785408, 0.0967833548784256], "cos_Ac_negg": [0.8095442652702332, 0.007084365002810955, 0.13092288374900818], "r_norm": [1.062812328338623, 0.9724699258804321, 0.922893762588501], "traffic_norm": [16.744239807128906, 16.700105667114258, 16.432209014892578], "traffic_residual_norm": [0.06855408847332001, 0.05987584590911865, 1.6088328266050667e-05], "traffic_r2": [0.9999832510948181, 0.9999871253967285, 1.0]}, {"step": 800, "epoch": 1, "train_loss": 0.6097962856292725, "cos_r_negg": [0.4509318470954895, 0.006314715836197138, 0.21590787172317505], "cos_innovation_negg": [0.4509318470954895, 0.006314715836197138, 0.21590787172317505], "cos_apical_negg": [0.05356534570455551, 0.013744549825787544, -0.09187033772468567], "cos_Ac_negg": [0.8785468935966492, 0.03470063582062721, 0.22199468314647675], "r_norm": [0.9921948909759521, 0.9037696719169617, 0.8696085214614868], "traffic_norm": [16.73683738708496, 16.69219207763672, 16.42508316040039], "traffic_residual_norm": [0.03699691593647003, 0.04156612604856491, 1.1835669283755124e-05], "traffic_r2": [0.9999951124191284, 0.9999938011169434, 1.0]}, {"epoch_end": 1, "step": 860, "eval_split": "validation", "eval_acc": 0.83, "eval_loss": 0.8371518188476562, "val_acc": 0.83, "val_loss": 0.8371518188476562}, {"step": 1000, "epoch": 2, "train_loss": 0.7001122236251831, "cos_r_negg": [0.5261197090148926, 0.029240474104881287, 0.19572460651397705], "cos_innovation_negg": [0.5261197090148926, 0.029240474104881287, 0.19572460651397705], "cos_apical_negg": [-0.01021679025143385, 0.033180609345436096, 0.031754400581121445], "cos_Ac_negg": [0.866756021976471, 0.16077709197998047, 0.21576404571533203], "r_norm": [0.6267708539962769, 0.5629653930664062, 0.5322524309158325], "traffic_norm": [16.731611251831055, 16.6871280670166, 16.413673400878906], "traffic_residual_norm": [0.03991291671991348, 0.038383208215236664, 1.1692450243572239e-05], "traffic_r2": [0.9999942779541016, 0.999994695186615, 1.0]}, {"step": 1200, "epoch": 2, "train_loss": 0.5328733921051025, "cos_r_negg": [0.6078847646713257, 0.02190019004046917, 0.1685364842414856], "cos_innovation_negg": [0.6078847646713257, 0.02190019004046917, 0.1685364842414856], "cos_apical_negg": [-0.022007619962096214, 0.008282653987407684, -0.11277654767036438], "cos_Ac_negg": [0.81191086769104, 0.15022166073322296, 0.17426347732543945], "r_norm": [0.6596190333366394, 0.5974870920181274, 0.5640003085136414], "traffic_norm": [16.733421325683594, 16.689090728759766, 16.416603088378906], "traffic_residual_norm": [0.026503168046474457, 0.0337568074464798, 1.0347976058255881e-05], "traffic_r2": [0.9999974966049194, 0.9999958872795105, 1.0]}, {"epoch_end": 2, "step": 1290, "eval_split": "validation", "eval_acc": 0.8912, "eval_loss": 0.46072123413085936, "val_acc": 0.8912, "val_loss": 0.46072123413085936}, {"step": 1400, "epoch": 3, "train_loss": 0.8486059904098511, "cos_r_negg": [0.6948257088661194, 0.06843973696231842, 0.12497122585773468], "cos_innovation_negg": [0.6948257088661194, 0.06843973696231842, 0.12497122585773468], "cos_apical_negg": [-0.013649115338921547, -0.0657200887799263, -0.14054501056671143], "cos_Ac_negg": [0.8433842062950134, 0.19475212693214417, 0.12914887070655823], "r_norm": [0.8028693199157715, 0.7074384689331055, 0.6794102191925049], "traffic_norm": [16.73027992248535, 16.68685531616211, 16.413774490356445], "traffic_residual_norm": [0.024954110383987427, 0.030749883502721786, 9.427447366761044e-06], "traffic_r2": [0.9999977946281433, 0.9999966025352478, 1.0]}, {"step": 1600, "epoch": 3, "train_loss": 0.8867263197898865, "cos_r_negg": [0.6347246170043945, 0.06891785562038422, 0.24629080295562744], "cos_innovation_negg": [0.6347246170043945, 0.06891785562038422, 0.24629080295562744], "cos_apical_negg": [0.04718630760908127, 0.036475956439971924, 0.015308016911149025], "cos_Ac_negg": [0.8818166255950928, 0.2460843324661255, 0.24673566222190857], "r_norm": [0.8717605471611023, 0.7782508730888367, 0.7441102266311646], "traffic_norm": [16.731840133666992, 16.6876163482666, 16.415563583374023], "traffic_residual_norm": [0.024579478427767754, 0.03511764854192734, 9.88561259873677e-06], "traffic_r2": [0.9999978542327881, 0.9999955892562866, 1.0]}, {"epoch_end": 3, "step": 1720, "eval_split": "validation", "eval_acc": 0.8908, "eval_loss": 0.5238083923339844, "val_acc": 0.8908, "val_loss": 0.5238083923339844}, {"step": 1800, "epoch": 4, "train_loss": 0.3615794777870178, "cos_r_negg": [0.635452389717102, 0.11842997372150421, 0.2014845311641693], "cos_innovation_negg": [0.635452389717102, 0.11842997372150421, 0.2014845311641693], "cos_apical_negg": [-0.020003575831651688, 0.054981134831905365, -0.04361838474869728], "cos_Ac_negg": [0.8399995565414429, 0.35311025381088257, 0.22789067029953003], "r_norm": [0.6610411405563354, 0.5820934772491455, 0.5577223300933838], "traffic_norm": [16.72909164428711, 16.682756423950195, 16.410274505615234], "traffic_residual_norm": [0.01932622864842415, 0.030415598303079605, 9.725232303026132e-06], "traffic_r2": [0.9999986886978149, 0.9999966621398926, 1.0]}, {"step": 2000, "epoch": 4, "train_loss": 0.40493154525756836, "cos_r_negg": [0.6167439818382263, 0.13029341399669647, 0.18877707421779633], "cos_innovation_negg": [0.6167439818382263, 0.13029341399669647, 0.18877707421779633], "cos_apical_negg": [0.02698972262442112, -0.11009132862091064, -0.010150730609893799], "cos_Ac_negg": [0.8274956941604614, 0.30399423837661743, 0.2141953706741333], "r_norm": [0.8167616128921509, 0.7183709740638733, 0.6818836331367493], "traffic_norm": [16.728187561035156, 16.684022903442383, 16.410045623779297], "traffic_residual_norm": [0.01928986981511116, 0.029661238193511963, 9.09292975848075e-06], "traffic_r2": [0.9999986886978149, 0.9999968409538269, 1.0]}, {"epoch_end": 4, "step": 2150, "eval_split": "validation", "eval_acc": 0.8872, "eval_loss": 0.4481373779296875, "val_acc": 0.8872, "val_loss": 0.4481373779296875}, {"step": 2200, "epoch": 5, "train_loss": 0.5866303443908691, "cos_r_negg": [0.6363464593887329, 0.1064923107624054, 0.2299477458000183], "cos_innovation_negg": [0.6363464593887329, 0.1064923107624054, 0.2299477458000183], "cos_apical_negg": [-0.046788282692432404, 0.033850230276584625, -0.021517496556043625], "cos_Ac_negg": [0.8726298213005066, 0.35873231291770935, 0.23314476013183594], "r_norm": [0.5317440032958984, 0.46822163462638855, 0.44360947608947754], "traffic_norm": [16.72458839416504, 16.67963218688965, 16.403730392456055], "traffic_residual_norm": [0.018238212913274765, 0.031426772475242615, 9.217044862452894e-06], "traffic_r2": [0.9999988079071045, 0.9999964237213135, 1.0]}, {"step": 2400, "epoch": 5, "train_loss": 0.5027966499328613, "cos_r_negg": [0.5772510170936584, 0.15082545578479767, 0.12909002602100372], "cos_innovation_negg": [0.5772510170936584, 0.15082545578479767, 0.12909002602100372], "cos_apical_negg": [-0.003627367317676544, -0.12207751721143723, -0.17891469597816467], "cos_Ac_negg": [0.8126503825187683, 0.38655731081962585, 0.13534078001976013], "r_norm": [0.6117122173309326, 0.5325509309768677, 0.5107419490814209], "traffic_norm": [16.720516204833984, 16.673954010009766, 16.399612426757812], "traffic_residual_norm": [0.017895709723234177, 0.03171016275882721, 9.207481525663752e-06], "traffic_r2": [0.9999988675117493, 0.9999963641166687, 1.0]}, {"epoch_end": 5, "step": 2580, "eval_split": "validation", "eval_acc": 0.9176, "eval_loss": 0.34603408203125, "val_acc": 0.9176, "val_loss": 0.34603408203125}, {"step": 2600, "epoch": 6, "train_loss": 0.4901365637779236, "cos_r_negg": [0.6239902377128601, 0.14972032606601715, 0.22296223044395447], "cos_innovation_negg": [0.6239902377128601, 0.14972032606601715, 0.22296223044395447], "cos_apical_negg": [-0.04814021289348602, 0.030415277928113937, -0.011384104378521442], "cos_Ac_negg": [0.8523377180099487, 0.4401775598526001, 0.2558102011680603], "r_norm": [0.5623122453689575, 0.4902656078338623, 0.45963406562805176], "traffic_norm": [16.72378158569336, 16.67740821838379, 16.403749465942383], "traffic_residual_norm": [0.020855816081166267, 0.03257087618112564, 9.123322342929896e-06], "traffic_r2": [0.9999984502792358, 0.9999961853027344, 1.0]}, {"step": 2800, "epoch": 6, "train_loss": 0.39365458488464355, "cos_r_negg": [0.558141827583313, 0.1182108148932457, 0.1649833768606186], "cos_innovation_negg": [0.558141827583313, 0.1182108148932457, 0.1649833768606186], "cos_apical_negg": [0.03147769346833229, -0.019167166203260422, -0.05596218258142471], "cos_Ac_negg": [0.8422704339027405, 0.3630308508872986, 0.2302132546901703], "r_norm": [0.5806503891944885, 0.5106035470962524, 0.4762130379676819], "traffic_norm": [16.721115112304688, 16.67425537109375, 16.39705467224121], "traffic_residual_norm": [0.01905844733119011, 0.031718503683805466, 9.080818927031942e-06], "traffic_r2": [0.9999986886978149, 0.9999963641166687, 1.0]}, {"step": 3000, "epoch": 6, "train_loss": 0.31749916076660156, "cos_r_negg": [0.5397504568099976, 0.1228315532207489, 0.2994537949562073], "cos_innovation_negg": [0.5397504568099976, 0.1228315532207489, 0.2994537949562073], "cos_apical_negg": [0.022167537361383438, 0.07470016926527023, -0.06126970052719116], "cos_Ac_negg": [0.8229678869247437, 0.41075003147125244, 0.3266908526420593], "r_norm": [0.4565797448158264, 0.40107476711273193, 0.3794265389442444], "traffic_norm": [16.720928192138672, 16.672405242919922, 16.399433135986328], "traffic_residual_norm": [0.015085864812135696, 0.024461939930915833, 8.943547982198652e-06], "traffic_r2": [0.9999991655349731, 0.9999978542327881, 1.0]}, {"epoch_end": 6, "step": 3010, "eval_split": "validation", "eval_acc": 0.899, "eval_loss": 0.44906309814453127, "val_acc": 0.899, "val_loss": 0.44906309814453127}, {"step": 3200, "epoch": 7, "train_loss": 0.2687760889530182, "cos_r_negg": [0.6195740699768066, 0.14605611562728882, 0.2972404956817627], "cos_innovation_negg": [0.6195740699768066, 0.14605611562728882, 0.2972404956817627], "cos_apical_negg": [-0.005979207344353199, 0.024305682629346848, -0.08173543959856033], "cos_Ac_negg": [0.8328645825386047, 0.38460204005241394, 0.3549844026565552], "r_norm": [0.6224360466003418, 0.5427509546279907, 0.5198906660079956], "traffic_norm": [16.719209671020508, 16.672996520996094, 16.396888732910156], "traffic_residual_norm": [0.01681511662900448, 0.02954423613846302, 9.012665941554587e-06], "traffic_r2": [0.9999989867210388, 0.9999968409538269, 1.0]}, {"step": 3400, "epoch": 7, "train_loss": 0.3107507526874542, "cos_r_negg": [0.5631675124168396, 0.15017078816890717, 0.25406792759895325], "cos_innovation_negg": [0.5631675124168396, 0.15017078816890717, 0.25406792759895325], "cos_apical_negg": [-0.0009092334657907486, -0.05509522184729576, -0.1405956745147705], "cos_Ac_negg": [0.839911937713623, 0.3893618583679199, 0.30697914958000183], "r_norm": [0.585620641708374, 0.5112837553024292, 0.48628076910972595], "traffic_norm": [16.718013763427734, 16.67198371887207, 16.3968505859375], "traffic_residual_norm": [0.020529458299279213, 0.03430723398923874, 8.911745680961758e-06], "traffic_r2": [0.9999985098838806, 0.999995768070221, 1.0]}, {"epoch_end": 7, "step": 3440, "eval_split": "validation", "eval_acc": 0.9056, "eval_loss": 0.4115598388671875, "val_acc": 0.9056, "val_loss": 0.4115598388671875}, {"step": 3600, "epoch": 8, "train_loss": 0.5112429857254028, "cos_r_negg": [0.6679482460021973, 0.13510537147521973, 0.3014112114906311], "cos_innovation_negg": [0.6679482460021973, 0.13510537147521973, 0.3014112114906311], "cos_apical_negg": [-0.02679060399532318, 0.0017720304895192385, -0.028320901095867157], "cos_Ac_negg": [0.912888765335083, 0.4572254419326782, 0.31956517696380615], "r_norm": [0.48635661602020264, 0.4296538829803467, 0.4025575518608093], "traffic_norm": [16.716827392578125, 16.667892456054688, 16.39114761352539], "traffic_residual_norm": [0.018593497574329376, 0.03704840689897537, 8.801255717116874e-06], "traffic_r2": [0.9999987483024597, 0.9999950528144836, 1.0]}, {"step": 3800, "epoch": 8, "train_loss": 0.22072608768939972, "cos_r_negg": [0.5619479417800903, 0.14734794199466705, 0.1942809373140335], "cos_innovation_negg": [0.5619479417800903, 0.14734794199466705, 0.1942809373140335], "cos_apical_negg": [-0.04591304063796997, -0.040516823530197144, -0.18995791673660278], "cos_Ac_negg": [0.8334862589836121, 0.39038369059562683, 0.23217618465423584], "r_norm": [0.5723936557769775, 0.5058073997497559, 0.4703829288482666], "traffic_norm": [16.720596313476562, 16.674240112304688, 16.398923873901367], "traffic_residual_norm": [0.02263394370675087, 0.04490997642278671, 8.838796020427253e-06], "traffic_r2": [0.999998152256012, 0.9999927282333374, 1.0]}, {"epoch_end": 8, "step": 3870, "eval_split": "validation", "eval_acc": 0.925, "eval_loss": 0.3377326110839844, "val_acc": 0.925, "val_loss": 0.3377326110839844}, {"step": 4000, "epoch": 9, "train_loss": 0.22121430933475494, "cos_r_negg": [0.6822465658187866, 0.1426977515220642, 0.23493865132331848], "cos_innovation_negg": [0.6822465658187866, 0.1426977515220642, 0.23493865132331848], "cos_apical_negg": [-0.05188564956188202, 0.003991218283772469, -0.07313698530197144], "cos_Ac_negg": [0.8752049803733826, 0.4215485751628876, 0.2050837278366089], "r_norm": [0.5389835834503174, 0.48024630546569824, 0.4473986029624939], "traffic_norm": [16.717716217041016, 16.668954849243164, 16.395851135253906], "traffic_residual_norm": [0.01840229332447052, 0.03956485912203789, 8.830062142806128e-06], "traffic_r2": [0.9999988079071045, 0.9999943375587463, 1.0]}, {"step": 4200, "epoch": 9, "train_loss": 0.6201872229576111, "cos_r_negg": [0.5655783414840698, 0.12774156033992767, 0.048115041106939316], "cos_innovation_negg": [0.5655783414840698, 0.12774156033992767, 0.048115041106939316], "cos_apical_negg": [-0.06598012149333954, -0.063723124563694, 0.04929042235016823], "cos_Ac_negg": [0.8401979207992554, 0.4298176169395447, 0.06379160284996033], "r_norm": [0.5274326801300049, 0.4944320321083069, 0.42953625321388245], "traffic_norm": [16.718875885009766, 16.671588897705078, 16.39642333984375], "traffic_residual_norm": [0.02813444286584854, 0.08034135401248932, 8.704378160473425e-06], "traffic_r2": [0.9999969005584717, 0.9999767541885376, 1.0]}, {"epoch_end": 9, "step": 4300, "eval_split": "validation", "eval_acc": 0.8624, "eval_loss": 0.6588449462890625, "val_acc": 0.8624, "val_loss": 0.6588449462890625}, {"step": 4400, "epoch": 10, "train_loss": 0.3123042583465576, "cos_r_negg": [0.5223839282989502, 0.12126774340867996, 0.12707433104515076], "cos_innovation_negg": [0.5223839282989502, 0.12126774340867996, 0.12707433104515076], "cos_apical_negg": [-0.05521392077207565, -0.050024814903736115, -0.04427860304713249], "cos_Ac_negg": [0.858045220375061, 0.4244263172149658, 0.15377357602119446], "r_norm": [0.544574499130249, 0.53046715259552, 0.4099833071231842], "traffic_norm": [16.71941375732422, 16.671649932861328, 16.396484375], "traffic_residual_norm": [0.07507802546024323, 0.1466761827468872, 8.794144378043711e-06], "traffic_r2": [0.99997878074646, 0.9999220967292786, 1.0]}, {"step": 4600, "epoch": 10, "train_loss": 0.2866203188896179, "cos_r_negg": [0.6251101493835449, 0.03315821290016174, 0.1269550621509552], "cos_innovation_negg": [0.6251101493835449, 0.03315821290016174, 0.1269550621509552], "cos_apical_negg": [-0.03232560306787491, -0.03663243353366852, -0.06252510845661163], "cos_Ac_negg": [0.8949145674705505, 0.41490668058395386, 0.14809121191501617], "r_norm": [0.45313745737075806, 0.4847583770751953, 0.35414427518844604], "traffic_norm": [16.716785430908203, 16.6694278717041, 16.394275665283203], "traffic_residual_norm": [0.04946214333176613, 0.15510648488998413, 8.656533282191958e-06], "traffic_r2": [0.9999904036521912, 0.9999130368232727, 1.0]}, {"epoch_end": 10, "step": 4730, "eval_split": "validation", "eval_acc": 0.9282, "eval_loss": 0.32804862060546874, "val_acc": 0.9282, "val_loss": 0.32804862060546874}, {"step": 4800, "epoch": 11, "train_loss": 0.42584359645843506, "cos_r_negg": [0.569387674331665, -0.021445777267217636, 0.07788755744695663], "cos_innovation_negg": [0.569387674331665, -0.021445777267217636, 0.07788755744695663], "cos_apical_negg": [0.0012139016762375832, -0.2737320065498352, -0.27947843074798584], "cos_Ac_negg": [0.8589891195297241, 0.13000330328941345, 0.08091104030609131], "r_norm": [0.9870644807815552, 0.6324470043182373, 0.5663155317306519], "traffic_norm": [16.70620346069336, 16.66411590576172, 16.378276824951172], "traffic_residual_norm": [0.39383891224861145, 0.08151396363973618, 8.756913302931935e-06], "traffic_r2": [0.9994359612464905, 0.9999682307243347, 1.0]}, {"step": 5000, "epoch": 11, "train_loss": 0.33878540992736816, "cos_r_negg": [0.5343164801597595, -0.12484867125749588, 0.25179845094680786], "cos_innovation_negg": [0.5343164801597595, -0.12484867125749588, 0.25179845094680786], "cos_apical_negg": [0.014747809618711472, -0.19522184133529663, -0.23055145144462585], "cos_Ac_negg": [0.8169348239898682, 0.10008575022220612, 0.27765029668807983], "r_norm": [0.6744849681854248, 0.5498247146606445, 0.5011183619499207], "traffic_norm": [16.70892333984375, 16.664722442626953, 16.38298797607422], "traffic_residual_norm": [0.11353477090597153, 0.07459615170955658, 8.729643013793975e-06], "traffic_r2": [0.9999514818191528, 0.9999773502349854, 1.0]}, {"epoch_end": 11, "step": 5160, "eval_split": "validation", "eval_acc": 0.9188, "eval_loss": 0.3829672546386719, "val_acc": 0.9188, "val_loss": 0.3829672546386719}, {"step": 5200, "epoch": 12, "train_loss": 0.33124202489852905, "cos_r_negg": [0.510591447353363, 0.14140227437019348, 0.15044483542442322], "cos_innovation_negg": [0.510591447353363, 0.14140227437019348, 0.15044483542442322], "cos_apical_negg": [-0.07288257777690887, -0.011183369904756546, -0.08381901681423187], "cos_Ac_negg": [0.8177251815795898, 0.4415138363838196, 0.21365901827812195], "r_norm": [0.49667710065841675, 0.4351290464401245, 0.3992788791656494], "traffic_norm": [16.71462059020996, 16.66646957397461, 16.392242431640625], "traffic_residual_norm": [0.030337601900100708, 0.040912412106990814, 8.646640708320774e-06], "traffic_r2": [0.9999964833259583, 0.9999938607215881, 1.0]}, {"step": 5400, "epoch": 12, "train_loss": 0.1448916792869568, "cos_r_negg": [0.6340402364730835, 0.1731949746608734, 0.05848720669746399], "cos_innovation_negg": [0.6340402364730835, 0.1731949746608734, 0.05848720669746399], "cos_apical_negg": [-0.05246572196483612, 0.09627346694469452, 0.07019852101802826], "cos_Ac_negg": [0.8617383241653442, 0.5490093231201172, 0.13326320052146912], "r_norm": [0.601049542427063, 0.5073962211608887, 0.47527796030044556], "traffic_norm": [16.71796417236328, 16.67278480529785, 16.396316528320312], "traffic_residual_norm": [0.05293964967131615, 0.038769181817770004, 8.563499250158202e-06], "traffic_r2": [0.9999880790710449, 0.9999944567680359, 1.0]}, {"epoch_end": 12, "step": 5590, "eval_split": "validation", "eval_acc": 0.9044, "eval_loss": 0.43743345947265627, "val_acc": 0.9044, "val_loss": 0.43743345947265627}, {"step": 5600, "epoch": 13, "train_loss": 0.36729857325553894, "cos_r_negg": [0.5521649122238159, 0.129394993185997, 0.15311720967292786], "cos_innovation_negg": [0.5521649122238159, 0.129394993185997, 0.15311720967292786], "cos_apical_negg": [-0.01950032263994217, -0.1222645491361618, -0.22166311740875244], "cos_Ac_negg": [0.8944004774093628, 0.4513343870639801, 0.21399372816085815], "r_norm": [0.4466858506202698, 0.4250529408454895, 0.36166253685951233], "traffic_norm": [16.713539123535156, 16.664043426513672, 16.38970184326172], "traffic_residual_norm": [0.03024229407310486, 0.0796641930937767, 8.685539796715602e-06], "traffic_r2": [0.999996542930603, 0.9999768137931824, 1.0]}, {"step": 5800, "epoch": 13, "train_loss": 0.32653433084487915, "cos_r_negg": [0.5370160937309265, 0.2405770719051361, 0.1910463124513626], "cos_innovation_negg": [0.5370160937309265, 0.2405770719051361, 0.1910463124513626], "cos_apical_negg": [-0.03104287199676037, -0.23904867470264435, -0.22981737554073334], "cos_Ac_negg": [0.8297132253646851, 0.4822441637516022, 0.2756957709789276], "r_norm": [0.471118688583374, 0.4458509683609009, 0.3807945251464844], "traffic_norm": [16.713542938232422, 16.665618896484375, 16.389698028564453], "traffic_residual_norm": [0.033846043050289154, 0.08371081948280334, 8.69196628627833e-06], "traffic_r2": [0.9999956488609314, 0.9999746680259705, 1.0]}, {"step": 6000, "epoch": 13, "train_loss": 0.2657964527606964, "cos_r_negg": [0.5824995636940002, 0.15381287038326263, 0.3176572918891907], "cos_innovation_negg": [0.5824995636940002, 0.15381287038326263, 0.3176572918891907], "cos_apical_negg": [-0.04274103790521622, 0.042859070003032684, 0.06740780174732208], "cos_Ac_negg": [0.849876344203949, 0.5670449137687683, 0.35979336500167847], "r_norm": [0.5796319246292114, 0.5166885256767273, 0.4652341604232788], "traffic_norm": [16.71086883544922, 16.659564971923828, 16.386951446533203], "traffic_residual_norm": [0.03602433204650879, 0.05913514643907547, 8.712356247997377e-06], "traffic_r2": [0.9999952912330627, 0.9999873638153076, 1.0]}, {"epoch_end": 13, "step": 6020, "eval_split": "validation", "eval_acc": 0.9294, "eval_loss": 0.3259512268066406, "val_acc": 0.9294, "val_loss": 0.3259512268066406}, {"step": 6200, "epoch": 14, "train_loss": 0.5345078706741333, "cos_r_negg": [0.5660679936408997, 0.30895131826400757, 0.25281786918640137], "cos_innovation_negg": [0.5660679936408997, 0.30895131826400757, 0.25281786918640137], "cos_apical_negg": [-0.038529977202415466, 0.2683370113372803, 0.19209490716457367], "cos_Ac_negg": [0.8785896897315979, 0.2654273509979248, 0.24289534986019135], "r_norm": [0.5788178443908691, 0.5129026174545288, 0.4635138511657715], "traffic_norm": [16.71424674987793, 16.665813446044922, 16.390087127685547], "traffic_residual_norm": [0.046288300305604935, 0.06189612299203873, 8.599348802817985e-06], "traffic_r2": [0.9999922513961792, 0.9999861717224121, 1.0]}, {"step": 6400, "epoch": 14, "train_loss": 0.48510420322418213, "cos_r_negg": [0.5758558511734009, 0.063137486577034, 0.32120901346206665], "cos_innovation_negg": [0.5758558511734009, 0.063137486577034, 0.32120901346206665], "cos_apical_negg": [-0.048102982342243195, -0.0930819883942604, -0.152420312166214], "cos_Ac_negg": [0.8744509220123291, 0.49594882130622864, 0.37930718064308167], "r_norm": [0.41220521926879883, 0.3659125566482544, 0.33370423316955566], "traffic_norm": [16.714954376220703, 16.66452407836914, 16.391374588012695], "traffic_residual_norm": [0.031139694154262543, 0.04626698046922684, 8.654622433823533e-06], "traffic_r2": [0.9999964237213135, 0.9999922513961792, 1.0]}, {"epoch_end": 14, "step": 6450, "eval_split": "validation", "eval_acc": 0.9112, "eval_loss": 0.4363379943847656, "val_acc": 0.9112, "val_loss": 0.4363379943847656}], "final": {"eval_split": "validation", "eval_acc": 0.9112, "eval_loss": 0.4363379943847656, "wall_s": 25.43530821800232, "val_acc": 0.9112, "val_loss": 0.4363379943847656, "cos_r_negg": [0.5759245753288269, 0.1112496554851532, 0.39244407415390015], "cos_innovation_negg": [0.5759245753288269, 0.1112496554851532, 0.39244407415390015], "cos_apical_negg": [-0.03328140825033188, -0.07143482565879822, -0.25836411118507385], "cos_Ac_negg": [0.8537358045578003, 0.5058251023292542, 0.4133973717689514], "traffic_norm": [16.710800170898438, 16.661441802978516, 16.38666343688965], "traffic_residual_norm": [0.02358362451195717, 0.03590838611125946, 8.845256161293946e-06], "traffic_r2": [0.9999977946281433, 0.9999953508377075, 1.0]}, "timing": {"warmup_wall_s": 0.4154942035675049, "training_loop_wall_s": 24.78669762611389, "diagnostics_wall_s": 0.5566790103912354, "inline_diagnostics_wall_s": 0.40077757835388184, "optimizer_wall_s_excluding_inline_diagnostics": 24.38592004776001, "evaluation_wall_s": 0.07347345352172852}, "cost": {"train_steps": 6450, "ordinary_training_forward_examples": 825000, "predictor_warmup_forward_examples": 25600, "perturbation_events": 1613, "calibration_batch_loss_evaluations": 51616, "calibration_example_loss_evaluations": 6606848, "calibration_forward_equivalent_examples": 6813312.0, "max_perturbation_batch_expansion": 32}}
|