summaryrefslogtreecommitdiff
path: root/results/scale_v3_cifar10_dfa_w64_d10_s4.json
blob: 01d98413b7849472e56bfde0431873dde654c1d4 (plain)
1
{"args": {"mode": "dfa", "dataset": "cifar10", "depth": 10, "width": 64, "act": "tanh", "residual": 1, "epochs": 5, "batch_size": 128, "train_examples": 0, "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": 0.002, "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": 0.0, "predictor_mode": "diagonal", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 4, "device": "cuda", "log_every": 500, "max_steps": 0, "probe_bs": 256, "outdir": "results", "tag": "scale_v3_cifar10_dfa_w64_d10_s4", "n_in": 3072, "n_out": 10}, "provenance": {"git_commit": "d4aef1a571c21e6e4e4125c90aa58da7f14e62e3", "git_dirty": false}, "steps": [{"step": 0, "epoch": 0, "train_loss": 2.522786855697632, "cos_r_negg": [-0.007739442400634289, -0.006245890632271767, -0.08203615248203278, 0.024547457695007324, 0.03436309099197388, -0.029096223413944244, -0.052443165332078934, 0.01502314768731594, 0.07241200655698776, 0.029875531792640686], "cos_innovation_negg": [-0.007739442400634289, -0.006245890632271767, -0.08203615248203278, 0.024547457695007324, 0.03436309099197388, -0.029096223413944244, -0.052443165332078934, 0.01502314768731594, 0.07241200655698776, 0.029875531792640686], "cos_apical_negg": [-0.007739442400634289, -0.006245890632271767, -0.08203615248203278, 0.024547457695007324, 0.03436309099197388, -0.029096223413944244, -0.052443165332078934, 0.01502314768731594, 0.07241200655698776, 0.029875531792640686], "cos_Ac_negg": [-0.007739442400634289, -0.006245890632271767, -0.08203615248203278, 0.024547457695007324, 0.03436309099197388, -0.029096223413944244, -0.052443165332078934, 0.01502314768731594, 0.07241200655698776, 0.029875531792640686], "r_norm": [2.479470729827881, 2.485487461090088, 2.4711484909057617, 2.364689350128174, 2.4592936038970947, 2.557290554046631, 2.431260824203491, 2.344165563583374, 2.3765830993652344, 2.43996524810791]}, {"epoch_end": 0, "step": 391, "test_acc": 0.3316, "test_loss": 1.886237939453125}, {"step": 500, "epoch": 1, "train_loss": 1.7576546669006348, "cos_r_negg": [0.5612596273422241, 0.02252304181456566, -0.015010729432106018, 0.0160028338432312, -0.09989511221647263, 0.026551667600870132, -0.021338706836104393, -0.01848692074418068, 0.01972203515470028, 0.09419336915016174], "cos_innovation_negg": [0.5612596273422241, 0.02252304181456566, -0.015010729432106018, 0.0160028338432312, -0.09989511221647263, 0.026551667600870132, -0.021338706836104393, -0.01848692074418068, 0.01972203515470028, 0.09419336915016174], "cos_apical_negg": [0.5612596273422241, 0.02252304181456566, -0.015010729432106018, 0.0160028338432312, -0.09989511221647263, 0.026551667600870132, -0.021338706836104393, -0.01848692074418068, 0.01972203515470028, 0.09419336915016174], "cos_Ac_negg": [0.5612596273422241, 0.02252304181456566, -0.015010729432106018, 0.0160028338432312, -0.09989511221647263, 0.026551667600870132, -0.021338706836104393, -0.01848692074418068, 0.01972203515470028, 0.09419336915016174], "r_norm": [2.2240962982177734, 2.248375654220581, 2.2176074981689453, 2.1457266807556152, 2.2318925857543945, 2.3284144401550293, 2.2114317417144775, 2.1355457305908203, 2.1657586097717285, 2.226271629333496]}, {"epoch_end": 1, "step": 782, "test_acc": 0.3621, "test_loss": 1.8385595825195313}, {"step": 1000, "epoch": 2, "train_loss": 1.8353302478790283, "cos_r_negg": [0.5641099214553833, -0.02232620120048523, 0.025593329221010208, 0.07588539272546768, -0.04040656238794327, 0.05010192096233368, -0.015542181208729744, 0.03003668412566185, 0.06691727042198181, 0.10104583203792572], "cos_innovation_negg": [0.5641099214553833, -0.02232620120048523, 0.025593329221010208, 0.07588539272546768, -0.04040656238794327, 0.05010192096233368, -0.015542181208729744, 0.03003668412566185, 0.06691727042198181, 0.10104583203792572], "cos_apical_negg": [0.5641099214553833, -0.02232620120048523, 0.025593329221010208, 0.07588539272546768, -0.04040656238794327, 0.05010192096233368, -0.015542181208729744, 0.03003668412566185, 0.06691727042198181, 0.10104583203792572], "cos_Ac_negg": [0.5641099214553833, -0.02232620120048523, 0.025593329221010208, 0.07588539272546768, -0.04040656238794327, 0.05010192096233368, -0.015542181208729744, 0.03003668412566185, 0.06691727042198181, 0.10104583203792572], "r_norm": [2.2162680625915527, 2.237281322479248, 2.2048563957214355, 2.1317412853240967, 2.217128038406372, 2.3222227096557617, 2.2041666507720947, 2.1271893978118896, 2.1549363136291504, 2.2138476371765137]}, {"epoch_end": 2, "step": 1173, "test_acc": 0.3394, "test_loss": 1.8579822875976562}, {"step": 1500, "epoch": 3, "train_loss": 1.7180376052856445, "cos_r_negg": [0.5600875616073608, 0.039104580879211426, 0.054650839418172836, 0.12868060171604156, 0.02561519481241703, 0.0921848714351654, 0.062459275126457214, 0.04157208651304245, 0.12034167349338531, 0.11256880313158035], "cos_innovation_negg": [0.5600875616073608, 0.039104580879211426, 0.054650839418172836, 0.12868060171604156, 0.02561519481241703, 0.0921848714351654, 0.062459275126457214, 0.04157208651304245, 0.12034167349338531, 0.11256880313158035], "cos_apical_negg": [0.5600875616073608, 0.039104580879211426, 0.054650839418172836, 0.12868060171604156, 0.02561519481241703, 0.0921848714351654, 0.062459275126457214, 0.04157208651304245, 0.12034167349338531, 0.11256880313158035], "cos_Ac_negg": [0.5600875616073608, 0.039104580879211426, 0.054650839418172836, 0.12868060171604156, 0.02561519481241703, 0.0921848714351654, 0.062459275126457214, 0.04157208651304245, 0.12034167349338531, 0.11256880313158035], "r_norm": [2.1913352012634277, 2.210455894470215, 2.1745495796203613, 2.107302665710449, 2.1984729766845703, 2.292149305343628, 2.179086685180664, 2.102180004119873, 2.1253716945648193, 2.1851916313171387]}, {"epoch_end": 3, "step": 1564, "test_acc": 0.3527, "test_loss": 1.8559609375}, {"epoch_end": 4, "step": 1955, "test_acc": 0.3641, "test_loss": 1.812884619140625}], "final": {"test_acc": 0.3641, "test_loss": 1.812884619140625, "wall_s": 8.841753482818604, "cos_r_negg": [0.5059987306594849, 0.013865919783711433, 0.10778409242630005, 0.13851752877235413, 0.06517338752746582, 0.07959023863077164, 0.09674673527479172, 0.08164706826210022, 0.06737769395112991, 0.17072650790214539], "cos_innovation_negg": [0.5059987306594849, 0.013865919783711433, 0.10778409242630005, 0.13851752877235413, 0.06517338752746582, 0.07959023863077164, 0.09674673527479172, 0.08164706826210022, 0.06737769395112991, 0.17072650790214539], "cos_apical_negg": [0.5059987306594849, 0.013865919783711433, 0.10778409242630005, 0.13851752877235413, 0.06517338752746582, 0.07959023863077164, 0.09674673527479172, 0.08164706826210022, 0.06737769395112991, 0.17072650790214539], "cos_Ac_negg": [0.5059987306594849, 0.013865919783711433, 0.10778409242630005, 0.13851752877235413, 0.06517338752746582, 0.07959023863077164, 0.09674673527479172, 0.08164706826210022, 0.06737769395112991, 0.17072650790214539]}}