{ "123": { "dfa": { "log": { "train_loss": [ 1.9917970180892943, 1.9455738663482667, 1.9412584506225585, 1.9381972328948975, 1.935788204345703, 1.9355586415863038, 1.9291299639892578, 1.9304945249938965, 1.9279169077301026, 1.9247258889389038, 1.9270652098083496, 1.9239173442077637, 1.920924050216675, 1.9179519243621825, 1.9187614895629883, 1.9149749173736572, 1.9101889783477783, 1.9113776821517945, 1.9123343227767944, 1.9115107091522217, 1.9105930194091796, 1.9092110122680663, 1.9085102457046508, 1.9054700579071044, 1.906116495628357, 1.9055400652313232, 1.9055342751312256, 1.9067684815216064, 1.9050640679168702, 1.9041575216674804 ], "train_acc": [ 0.2828, 0.3048, 0.30986, 0.31168, 0.31358, 0.31522, 0.31574, 0.317, 0.31914, 0.31836, 0.31874, 0.31798, 0.32148, 0.32276, 0.32502, 0.3259, 0.32756, 0.32626, 0.32706, 0.32784, 0.329, 0.32942, 0.32888, 0.33042, 0.33034, 0.33086, 0.33476, 0.33214, 0.33166, 0.33136 ], "test_acc": [ 0.3309, 0.3491, 0.3382, 0.3519, 0.3375, 0.3487, 0.3436, 0.3431, 0.3632, 0.341, 0.3632, 0.3623, 0.3531, 0.3536, 0.3476, 0.3485, 0.3662, 0.3539, 0.3453, 0.36, 0.3532, 0.3566, 0.3562, 0.3639, 0.3622, 0.3623, 0.3579, 0.3541, 0.3576, 0.3582 ] }, "diagnostics": { "bp_cosine": [ 0.3322567343711853, 0.10321325808763504, 0.15636520087718964, 0.19004486501216888 ], "perturbation_rho": [ 0.18363387882709503, 0.04898637533187866, 0.09784461557865143, 0.09405896812677383 ], "nudging": { "0.001": [ -1.0297924745827913e-05, -3.0525843612849712e-06, -4.650210030376911e-06, -5.671870894730091e-06 ], "0.003": [ -3.089715028181672e-05, -9.163166396319866e-06, -1.3922981452196836e-05, -1.7057405784726143e-05 ], "0.01": [ -0.00010294892126694322, -3.057112917304039e-05, -4.6447094064205885e-05, -5.68098621442914e-05 ] }, "hidden_norms_per_layer": [ 13127.3251953125, 13128.8193359375, 13141.9501953125, 13135.3408203125, 13141.1806640625 ], "bp_grad_norms_per_layer": [ 1.0922197361651342e-05, 1.074585088645108e-05, 1.0719732017605565e-05, 1.0678052603907418e-05, 1.0736169315350708e-05 ] }, "drift": { "embed.weight": 95.06253847427116, "embed.bias": 69.56343057263858, "blocks.0.ln.weight": 0.35518038272857666, "blocks.0.w1.weight": 3.5318213424051828, "blocks.0.w1.bias": 6.6251757142063346, "blocks.0.w2.weight": 18.1649292755703, "blocks.1.ln.weight": 0.37566548585891724, "blocks.1.w1.weight": 3.577683688515039, "blocks.1.w1.bias": 6.4976786510039295, "blocks.1.w2.weight": 18.132601820579644, "blocks.2.ln.weight": 0.39002206921577454, "blocks.2.w1.weight": 3.7319215256746268, "blocks.2.w1.bias": 6.266003010226504, "blocks.2.w2.weight": 18.870330386369503, "blocks.3.ln.weight": 0.35860922932624817, "blocks.3.w1.weight": 3.790655898639472, "blocks.3.w1.bias": 5.892665738859735, "blocks.3.w2.weight": 20.602751226104075, "out_ln.weight": 0.15972787141799927, "out_head.weight": 2.3651003853179144, "out_head.bias": 1.3627504013501806 } } }, "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": [ 123 ], "gpu": 0, "output_dir": "results/round41_dfa_penalty_30ep_s123", "methods": [ "dfa" ], "random_targets": false, "penalty_lam": 0.01, "num_classes": 10 } }