{ "42": { "state_bridge": { "log": { "train_loss": [ 2.1615624513244627, 2.201089162750244, 2.182833108444214, 2.1709032915496826, 2.1161820389556887, 2.2302013398742675, 2.1329537266540526, 2.10827269241333, 2.0998276640319826, 2.083012378692627, 2.102786566772461, 2.106615118637085, 2.0964282553863525, 2.0857217935943604, 2.0992775797271728, 2.0625892084503175, 2.0862051232147216, 2.1001052409362795, 2.094951697998047, 2.0933474546813966, 2.0933452713775633, 2.08693749671936, 2.0883480770874026, 2.086935211029053, 2.0861202153015137, 2.0841029391479493, 2.082290132446289, 2.0792110958099363, 2.0804510121154784, 2.081232653427124 ], "train_acc": [ 0.19528, 0.1606, 0.17582, 0.16834, 0.18852, 0.1556, 0.17988, 0.18368, 0.19014, 0.19926, 0.19686, 0.1923, 0.1981, 0.20012, 0.19082, 0.2146, 0.20116, 0.18934, 0.19132, 0.19428, 0.18736, 0.1931, 0.19568, 0.19958, 0.20248, 0.20444, 0.2046, 0.2062, 0.20122, 0.20856 ], "test_acc": [ 0.2011, 0.1577, 0.176, 0.1994, 0.1896, 0.1661, 0.1688, 0.181, 0.1939, 0.1873, 0.1878, 0.2058, 0.2082, 0.1967, 0.2019, 0.2005, 0.1949, 0.1854, 0.1759, 0.1734, 0.1868, 0.1995, 0.1949, 0.2007, 0.2077, 0.2085, 0.2109, 0.2079, 0.2132, 0.2126 ], "state_pred_error": [ 0.3881664722061157, 0.5467375702285766, 0.4459853865337372, 0.21270744318962098, 0.2084006814146042, 0.1407120229911804, 0.19075847319602968, 0.1764928385257721, 0.10211428045988083, 0.07563618842363358, 0.05846105343937874, 0.03846761897563934, 0.03369148607134819, 0.022906817595362665, 0.015594833841919899, 0.007780850533694029, 0.005383170440495014, 0.00387467138633132, 0.0025940868555754423, 0.0021150133569538595, 0.001786851437687874, 0.0014581093015894294, 0.0012195721989311278, 0.001029884922001511, 0.0009008185034245252, 0.0008114725712127983, 0.0007245114441029728, 0.0006559462931565941, 0.0006020296776108444, 0.0005471098060533404 ] }, "diagnostics": { "bp_cosine": [ 0.2372891902923584, 0.12591305375099182, 0.1843540072441101, 0.28460749983787537 ], "perturbation_rho": [ 0.5516304969787598, 0.003948327153921127, -0.007016957737505436, 0.0 ], "nudging": { "0.001": [ -1.981109380722046e-05, -8.381903171539307e-09, 0.0, 0.0 ], "0.003": [ -5.944725126028061e-05, -1.862645149230957e-09, 0.0, 0.0 ], "0.01": [ -0.00019809138029813766, -1.862645149230957e-08, 3.725290298461914e-09, 0.0 ] }, "hidden_norms_per_layer": [ 695.4109497070312, 561166.6875, 5258578.5, 12430880.0, 9851606.0 ], "bp_grad_norms_per_layer": [ 1.0803730219777208e-05, 1.284981632210247e-08, 9.346035056978508e-09, 9.320208604890468e-09, 9.30139343324754e-09 ] }, "drift": { "embed.weight": 10.683279498765518, "embed.bias": 16.186979717078597, "blocks.0.ln.weight": 1.8698891401290894, "blocks.0.w1.weight": 34.012515230994815, "blocks.0.w1.bias": 46.85058602976643, "blocks.0.w2.weight": 124.44356621058887, "blocks.1.ln.weight": 1.7833274602890015, "blocks.1.w1.weight": 48.03603598766101, "blocks.1.w1.bias": 58.602063412195584, "blocks.1.w2.weight": 123.88140584500592, "blocks.2.ln.weight": 2.2544941902160645, "blocks.2.w1.weight": 66.40401175683013, "blocks.2.w1.bias": 63.25326651988191, "blocks.2.w2.weight": 127.01673478324264, "blocks.3.ln.weight": 1.7423062324523926, "blocks.3.w1.weight": 37.248001175030076, "blocks.3.w1.bias": 38.949358252268084, "blocks.3.w2.weight": 101.10015971156598, "out_ln.weight": 0.1668696403503418, "out_head.weight": 1.4905390969722283, "out_head.bias": 1.6423172242266484 } }, "credit_bridge": { "log": { "train_loss": [ 2.039658761978149, 2.0134409933471678, 2.00583283241272, 2.009486458129883, 2.008044157333374, 2.0073662790679934, 2.0122061133575437, 2.040649677429199, 2.0651579305267336, 2.0854766065979002, 2.099567792701721, 2.1041016172790528, 2.111146937561035, 2.0936275885772706, 2.100649955825806, 2.1145339432525634, 2.1261457512664794, 2.134411395187378, 2.138650359802246, 2.1414046758270264, 2.1410846240997317, 2.140825333404541, 2.1406545039367675, 2.1377127909851072, 2.138584699020386, 2.1371214043426514, 2.137926218109131, 2.134101504898071, 2.1338819580078123, 2.134741836090088 ], "train_acc": [ 0.25218, 0.26698, 0.27004, 0.26836, 0.26838, 0.26756, 0.26486, 0.2526, 0.24742, 0.23242, 0.22372, 0.2182, 0.2176, 0.21986, 0.21266, 0.2051, 0.19802, 0.1941, 0.1906, 0.18834, 0.19216, 0.19062, 0.1907, 0.19222, 0.19358, 0.19584, 0.1945, 0.19384, 0.19432, 0.19732 ], "test_acc": [ 0.2777, 0.2762, 0.2822, 0.3004, 0.3017, 0.2837, 0.2964, 0.2601, 0.2555, 0.2308, 0.2401, 0.2235, 0.1973, 0.2062, 0.1917, 0.232, 0.2137, 0.19, 0.182, 0.1824, 0.1929, 0.191, 0.1938, 0.2015, 0.2114, 0.2172, 0.2056, 0.2037, 0.2205, 0.2105 ], "value_loss": [ 0.4988088236522675, 0.10717420438051224, 0.08406411410808563, 0.06901271151542664, 0.05796313747167587, 0.04655733848810196, 0.043197430835962294, 0.04390601473927498, 0.036221856105923655, 0.03994749122738838, 0.038669472066164015, 0.03699167508423328, 0.0456404883146286, 0.026927970384061337, 0.02069880782365799, 0.01933776231765747, 0.02080854148507118, 0.016443329193592073, 0.013170968658924103, 0.012027110563218594, 0.00889115786999464, 0.0077953302666544915, 0.005822759210169315, 0.004206860280707479, 0.003205367206633091, 0.0018577339798584581, 0.0014363039807975292, 0.001063295077085495, 0.0008753935668803751, 0.000745832763761282 ] }, "diagnostics": { "bp_cosine": [ 0.17070117592811584, 0.12735798954963684, 0.21822263300418854, 0.2762623727321625 ], "perturbation_rho": [ 0.021132776513695717, -0.006791723892092705, 0.0, 0.023015428334474564 ], "nudging": { "0.001": [ -7.441267371177673e-07, -1.5925616025924683e-07, 0.0, 0.0 ], "0.003": [ -2.280808985233307e-06, -5.373731255531311e-07, 0.0, 0.0 ], "0.01": [ -7.669441401958466e-06, -1.7955899238586426e-06, 0.0, 1.862645149230957e-09 ] }, "hidden_norms_per_layer": [ 7008.69921875, 7480314.5, 66006552.0, 65819076.0, 67016280.0 ], "bp_grad_norms_per_layer": [ 9.282441055802337e-07, 5.1517807797551995e-09, 4.764760141995339e-09, 4.764532768319896e-09, 4.767270578298621e-09 ] }, "drift": { "embed.weight": 55.317448573929425, "embed.bias": 47.56034585553186, "blocks.0.ln.weight": 4.106407642364502, "blocks.0.w1.weight": 67.89937919497739, "blocks.0.w1.bias": 55.34124320193868, "blocks.0.w2.weight": 132.61316975770185, "blocks.1.ln.weight": 3.4333691596984863, "blocks.1.w1.weight": 109.26717483313107, "blocks.1.w1.bias": 120.3421088832556, "blocks.1.w2.weight": 196.69358325668478, "blocks.2.ln.weight": 2.4557623863220215, "blocks.2.w1.weight": 55.033274573984386, "blocks.2.w1.bias": 54.84317146947108, "blocks.2.w2.weight": 98.576185000917, "blocks.3.ln.weight": 2.2460579872131348, "blocks.3.w1.weight": 51.81531440048247, "blocks.3.w1.bias": 53.49338639187709, "blocks.3.w2.weight": 92.14827023111492, "out_ln.weight": 0.22502917051315308, "out_head.weight": 2.8997412886471703, "out_head.bias": 0.4396077262245561 } } }, "config": { "dataset": "cifar10", "d_hidden": 256, "num_blocks": 4, "batch_size": 128, "epochs": 30, "lr": 0.001, "lr_fb": 0.001, "wd": 0.01, "lam": 0.1, "K": 4, "sigma_bridge": 0.05, "ema_momentum": 0.995, "term_grad_weight": 1.0, "seeds": [ 42 ], "gpu": 0, "output_dir": "results/round38_sbcb_baseline_30ep", "methods": [ "state_bridge", "credit_bridge" ], "random_targets": false, "penalty_lam": 0.0, "num_classes": 10 } }