{ "9": { "dfa": { "log": { "train_loss": [ 2.073234083404541, 2.0497977373504637, 2.0341352867126465, 2.031250879135132, 2.026975922088623, 2.021300171508789, 2.0191297291564942, 2.0188821714782716, 2.017022749557495, 2.016341851234436, 2.0123678436279295, 2.010649602279663, 2.0090650662994385, 2.010900563201904, 2.0095366540527344, 2.0077656902313232, 2.0068030431365966, 2.008629825515747, 2.003841092376709, 2.005812280654907, 2.0057511726379396, 2.004702599334717, 2.0014493046569823, 2.004868564796448, 2.001082904891968, 1.9998853713989257, 2.0003265306854248, 2.0005562553405762, 1.9986425912094117, 2.0000717253875733, 1.9990827807617189, 1.9992364362335204, 1.9985607556152343, 1.996207259979248, 1.9981809964370727, 1.9958161113739015, 1.9976731842422486, 1.9966179808807374, 1.9954889965057374, 1.9954629190826416, 1.9958389172363282, 1.994535606842041, 1.9974303654479981, 1.9959670455932617, 1.9949795317840575, 1.995265456085205, 1.9942493894195557, 1.995007307357788, 1.996318332901001, 1.9937139911651611, 1.994312198791504, 1.9913738401031493, 1.9951153566741944, 1.9924673150253296, 1.9923780603790284, 1.9935618125152588, 1.992571726989746, 1.9926865383911132, 1.992229995956421, 1.993404683456421, 1.9912095419311524, 1.991911597442627, 1.9921949435806274, 1.9902286859512328, 1.9926082902526856, 1.9909677504730225, 1.9914979708099365, 1.9912472879791259, 1.990876022491455, 1.9902124596405029, 1.9935787561035156, 1.991565812225342, 1.9911484326171875, 1.9914972548675538, 1.990185121498108, 1.9900826383972168, 1.9883790439605713, 1.9883859337997436, 1.9883267873382569, 1.9883251065444947, 1.9901439308166504, 1.9902003237915038, 1.9887106338500977, 1.9913360192871095, 1.9886777478027344, 1.99161105342865, 1.9905986673736573, 1.9889384605407714, 1.990373504333496, 1.988581312599182, 1.9881270119476318, 1.986687230491638, 1.9884908292007446, 1.9869463809204102, 1.9887974228668213, 1.986062038192749, 1.9904356650543213, 1.9901889600372313, 1.989139567527771, 1.9876907534790038 ], "train_acc": [ 0.23456, 0.24532, 0.2504, 0.25112, 0.25282, 0.26, 0.2606, 0.2593, 0.25506, 0.2613, 0.26512, 0.26252, 0.26454, 0.2637, 0.26436, 0.26586, 0.26598, 0.26426, 0.26796, 0.26508, 0.26656, 0.26618, 0.26934, 0.26756, 0.26878, 0.2704, 0.27164, 0.2692, 0.27146, 0.27, 0.27024, 0.2712, 0.27242, 0.27516, 0.27246, 0.27374, 0.27374, 0.2729, 0.27492, 0.27392, 0.2749, 0.27492, 0.2729, 0.2735, 0.27376, 0.2749, 0.27598, 0.27244, 0.27474, 0.27518, 0.27482, 0.27532, 0.27598, 0.277, 0.27772, 0.27702, 0.27816, 0.2744, 0.27456, 0.27834, 0.27612, 0.27584, 0.27394, 0.27708, 0.27626, 0.27684, 0.27762, 0.27826, 0.2762, 0.27832, 0.277, 0.27744, 0.27744, 0.27842, 0.27896, 0.27732, 0.27968, 0.28094, 0.27758, 0.28126, 0.28152, 0.27998, 0.2792, 0.27652, 0.27846, 0.27788, 0.27668, 0.28002, 0.27868, 0.27792, 0.2785, 0.28064, 0.27878, 0.28056, 0.279, 0.28108, 0.2801, 0.27866, 0.27774, 0.27842 ], "test_acc": [ 0.2428, 0.2739, 0.2627, 0.2812, 0.2892, 0.2843, 0.2907, 0.2804, 0.276, 0.3009, 0.2792, 0.2917, 0.2992, 0.2872, 0.2869, 0.2731, 0.2863, 0.3004, 0.2738, 0.2964, 0.2885, 0.2907, 0.2934, 0.2869, 0.283, 0.295, 0.2853, 0.2785, 0.2809, 0.3069, 0.2927, 0.2929, 0.3024, 0.296, 0.2988, 0.2998, 0.3001, 0.3009, 0.2946, 0.3043, 0.3071, 0.2996, 0.308, 0.2894, 0.2951, 0.292, 0.286, 0.295, 0.2997, 0.2967, 0.2942, 0.2974, 0.2921, 0.2946, 0.3006, 0.303, 0.303, 0.3102, 0.3042, 0.31, 0.3053, 0.294, 0.2993, 0.3065, 0.2955, 0.2908, 0.2943, 0.3014, 0.3029, 0.3044, 0.3066, 0.2909, 0.3097, 0.3031, 0.3, 0.3012, 0.2993, 0.2967, 0.3001, 0.3051, 0.2993, 0.3091, 0.3013, 0.3013, 0.3007, 0.3031, 0.3047, 0.3032, 0.3053, 0.3041, 0.3041, 0.3041, 0.3037, 0.3036, 0.3046, 0.3036, 0.3044, 0.3041, 0.3042, 0.3041 ] }, "diagnostics": { "bp_cosine": [ 0.3925938308238983, 0.000656805990729481, 0.0001544215774629265, -0.0004282527952454984, 0.0003690449520945549, -0.00013921636855229735 ], "perturbation_rho": [ 0.005831995978951454, 0.0, 0.0, 0.0, 0.0, 0.0 ], "nudging": { "0.001": [ -3.3574178814888e-07, 0.0, 0.0, 0.0, 0.0, 0.0 ], "0.003": [ -1.0384246706962585e-06, 0.0, 0.0, 0.0, 0.0, 0.0 ], "0.01": [ -3.6191195249557495e-06, 1.862645149230957e-09, 0.0, 0.0, 0.0, 0.0 ] }, "hidden_norms_per_layer": [ 54551.35546875, 1099375360.0, 3004748800.0, 5508783616.0, 6090545664.0, 6586548736.0, 9585897472.0 ], "bp_grad_norms_per_layer": [ 2.399317509116372e-07, 1.9445174637144902e-10, 1.943899208267652e-10, 1.9486748326080772e-10, 1.9483341329173953e-10, 1.9483435698131046e-10, 1.9498901104864075e-10 ] }, "drift": { "embed.weight": 328.6355950373237, "embed.bias": 242.36454815135576, "blocks.0.ln.weight": 10.003397541810306, "blocks.0.w1.weight": 280.20952373585186, "blocks.0.w1.bias": 245.5434949270726, "blocks.0.w2.weight": 491.79926117548297, "blocks.1.ln.weight": 8.729605792932698, "blocks.1.w1.weight": 325.11308424934106, "blocks.1.w1.bias": 311.5625818317441, "blocks.1.w2.weight": 334.530484796484, "blocks.2.ln.weight": 9.417363473500547, "blocks.2.w1.weight": 404.6194334537652, "blocks.2.w1.bias": 374.09843900052965, "blocks.2.w2.weight": 394.7925902511279, "blocks.3.ln.weight": 8.451752857956773, "blocks.3.w1.weight": 326.2362218443471, "blocks.3.w1.bias": 306.77785286979747, "blocks.3.w2.weight": 303.662545367935, "blocks.4.ln.weight": 8.451660472063256, "blocks.4.w1.weight": 344.2742835972674, "blocks.4.w1.bias": 326.1264884495552, "blocks.4.w2.weight": 324.1161613020223, "blocks.5.ln.weight": 11.13712283522192, "blocks.5.w1.weight": 457.04149442154977, "blocks.5.w1.bias": 422.6989592235929, "blocks.5.w2.weight": 446.6716640278749, "out_ln.weight": 0.5876580707043741, "out_head.weight": 9.25778651436399, "out_head.bias": 0.5027234759174332 } }, "fa": { "log": { "train_loss": [ 2.0332628253936766, 1.9553094621276856, 1.9304233611679078, 1.9176126587677003, 1.9049958919906615, 1.8898755237197875, 1.8779738095474243, 1.87334542842865, 1.8662727130126953, 1.8637196960830689, 1.858073745689392, 1.8548571952819823, 1.8493252722930908, 1.851236528892517, 1.848223992576599, 1.8478172241210937, 1.8419879977416993, 1.8413512692260743, 1.8363069415664672, 1.8369076172256469, 1.8356854833984375, 1.8313965267562866, 1.826233511695862, 1.8260947198867798, 1.8246617177581788, 1.8188705309677125, 1.8175858071899413, 1.817788968887329, 1.8145698094940186, 1.811288226928711, 1.8075631722259522, 1.8072492791366577, 1.8021571556854248, 1.8001763897705079, 1.8002913983535767, 1.7960401634979248, 1.7964820532226562, 1.7898353637695312, 1.7879999541854859, 1.789700000267029, 1.7842918268585206, 1.7852493328475951, 1.7841659473037719, 1.7837900876617432, 1.7787483280181884, 1.7782914597320556, 1.7754856842422486, 1.7753860149765015, 1.7756032290267945, 1.7689688110351562, 1.7688034854507446, 1.7665690698623657, 1.767441110267639, 1.7627149563598632, 1.7611339742279053, 1.7607593900299072, 1.7611287790679933, 1.7574685613250733, 1.754263912963867, 1.7551698053741456, 1.7525737002563477, 1.7504821019744874, 1.7550537628936766, 1.749086598892212, 1.746996110610962, 1.7465003091812135, 1.7470869120025634, 1.7463814974975587, 1.7397632767486573, 1.7418430507659912, 1.7430473554229737, 1.741324652786255, 1.7399064194488525, 1.7389730927276612, 1.7398483694076539, 1.7381602671051026, 1.7356981131362914, 1.7315244303131103, 1.7377736062622071, 1.7310680517196655, 1.7351790799713134, 1.733889009361267, 1.73604018699646, 1.7337274952697754, 1.7322516064453124, 1.7330827197647094, 1.7328004323577881, 1.7354140316772462, 1.7350243264007568, 1.7329780157470702, 1.732491442489624, 1.7300388946151732, 1.7312415341949463, 1.7289181127929687, 1.7315491919326782, 1.7262304349517823, 1.733718437461853, 1.7327616833877564, 1.732974825668335, 1.7304670404434204 ], "train_acc": [ 0.25128, 0.2863, 0.29668, 0.30188, 0.30926, 0.31842, 0.32158, 0.32524, 0.326, 0.32964, 0.33238, 0.33504, 0.33644, 0.33418, 0.33848, 0.33768, 0.34274, 0.34018, 0.34388, 0.34162, 0.34312, 0.3448, 0.34788, 0.34724, 0.34808, 0.34792, 0.35052, 0.3484, 0.34834, 0.35138, 0.35202, 0.3548, 0.35396, 0.35878, 0.35564, 0.35712, 0.35584, 0.36054, 0.36252, 0.3624, 0.36288, 0.36378, 0.36368, 0.36364, 0.36672, 0.36452, 0.36916, 0.36916, 0.36646, 0.36894, 0.37018, 0.37108, 0.36976, 0.3718, 0.37092, 0.37276, 0.37434, 0.37378, 0.37382, 0.37552, 0.37628, 0.37336, 0.37356, 0.37596, 0.37708, 0.37716, 0.37758, 0.37918, 0.37974, 0.37924, 0.3793, 0.38072, 0.3823, 0.38326, 0.3807, 0.3811, 0.38186, 0.3819, 0.37932, 0.38418, 0.38282, 0.38236, 0.38112, 0.38594, 0.38248, 0.38184, 0.38116, 0.3824, 0.38242, 0.38396, 0.3811, 0.38402, 0.38436, 0.38572, 0.38324, 0.38604, 0.38198, 0.38274, 0.3843, 0.38394 ], "test_acc": [ 0.2784, 0.3102, 0.3139, 0.3338, 0.3473, 0.3407, 0.3546, 0.341, 0.3527, 0.3631, 0.3575, 0.361, 0.3661, 0.3657, 0.3671, 0.3545, 0.3752, 0.3558, 0.3637, 0.3696, 0.3788, 0.3719, 0.3783, 0.3713, 0.3801, 0.3776, 0.3826, 0.3803, 0.3804, 0.38, 0.3788, 0.3834, 0.3783, 0.385, 0.3843, 0.3798, 0.3854, 0.3913, 0.3868, 0.3793, 0.3823, 0.387, 0.3862, 0.3885, 0.393, 0.3914, 0.3913, 0.388, 0.395, 0.3924, 0.3895, 0.3884, 0.3872, 0.3949, 0.3852, 0.3969, 0.3964, 0.3952, 0.4008, 0.3959, 0.3952, 0.3928, 0.3987, 0.3957, 0.3957, 0.4041, 0.4015, 0.4019, 0.4027, 0.4016, 0.4043, 0.3973, 0.4001, 0.3968, 0.4013, 0.4044, 0.4001, 0.4014, 0.4009, 0.4, 0.3997, 0.3998, 0.4038, 0.3999, 0.402, 0.4017, 0.4004, 0.402, 0.4016, 0.4016, 0.4032, 0.4014, 0.4028, 0.4035, 0.4031, 0.4026, 0.403, 0.4021, 0.4026, 0.4025 ] }, "diagnostics": { "bp_cosine": [ 0.04259001836180687, 0.06560904532670975, -0.04747920483350754, -0.030196242034435272, -0.011228787712752819, 0.9897887706756592 ], "perturbation_rho": [ 0.027144353836774826, 0.02096467837691307, -0.03647351637482643, 0.004128730855882168, 0.035667650401592255, 0.0024879188276827335 ], "nudging": { "0.001": [ -3.065855707973242e-06, -3.6228448152542114e-07, -7.62520357966423e-09, 4.0279701352119446e-08, -3.4924596548080444e-09, -1.401233021169901e-06 ], "0.003": [ -9.421346476301551e-06, -1.0454095900058746e-06, 1.4924444258213043e-07, 1.5087425708770752e-07, 2.1478626877069473e-08, -4.659174010157585e-06 ], "0.01": [ -3.1415780540555716e-05, -3.479945007711649e-06, 5.55417500436306e-07, 4.3446198105812073e-07, 1.6833655536174774e-07, -1.60514609888196e-05 ] }, "hidden_norms_per_layer": [ 8874.3623046875, 79337.484375, 1180381.75, 1586145.875, 1886674.0, 1911188.375, 1308062.125 ], "bp_grad_norms_per_layer": [ 2.749456871242728e-05, 1.8702693296290818e-06, 6.362390649883309e-07, 6.333205533337605e-07, 6.349519026116468e-07, 6.351035040097486e-07, 6.164591468404979e-07 ] }, "drift": { "embed.weight": 53.51461020835447, "embed.bias": 12.540707882150965, "blocks.0.ln.weight": 1.1919744306866826, "blocks.0.w1.weight": 17.44693961444689, "blocks.0.w1.bias": 11.761966180483919, "blocks.0.w2.weight": 54.05207721791165, "blocks.1.ln.weight": 1.2851156800356192, "blocks.1.w1.weight": 27.13129237243264, "blocks.1.w1.bias": 21.945665839852136, "blocks.1.w2.weight": 45.75957303201888, "blocks.2.ln.weight": 0.765602321043846, "blocks.2.w1.weight": 21.68864597033105, "blocks.2.w1.bias": 20.84889617417975, "blocks.2.w2.weight": 36.85421923299837, "blocks.3.ln.weight": 0.7387797290176757, "blocks.3.w1.weight": 21.304625635934347, "blocks.3.w1.bias": 21.302804910719708, "blocks.3.w2.weight": 42.158761396062424, "blocks.4.ln.weight": 0.5093491405372358, "blocks.4.w1.weight": 16.62521527836923, "blocks.4.w1.bias": 14.569854707495779, "blocks.4.w2.weight": 46.72830505948521, "blocks.5.ln.weight": 0.569690236673946, "blocks.5.w1.weight": 18.425801313816322, "blocks.5.w1.bias": 15.707781883071574, "blocks.5.w2.weight": 65.35620786546646, "out_ln.weight": 0.42037958632530303, "out_head.weight": 6.6986797412678145, "out_head.bias": 0.660644788913502 } } }, "config": { "dataset": "cifar10", "d_hidden": 512, "num_blocks": 6, "batch_size": 128, "epochs": 100, "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": [ 9 ], "gpu": 0, "output_dir": "results/fa_dfa_d512_L6_seed9", "methods": [ "fa", "dfa" ], "random_targets": false, "penalty_lam": 0.0, "num_classes": 10 } }