1
|
{"args": {"mode": "dfa", "dataset": "cifar10", "depth": 20, "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": 1, "device": "cuda", "log_every": 500, "max_steps": 0, "probe_bs": 256, "outdir": "results", "tag": "scale_v3_cifar10_dfa_w64_d20_s1", "n_in": 3072, "n_out": 10}, "provenance": {"git_commit": "d4aef1a571c21e6e4e4125c90aa58da7f14e62e3", "git_dirty": false}, "steps": [{"step": 0, "epoch": 0, "train_loss": 2.8419671058654785, "cos_r_negg": [-0.06071232259273529, 0.005213736090809107, 0.0790824294090271, -0.018299266695976257, -0.052069876343011856, 0.004859775770455599, -0.04551982879638672, 0.026532437652349472, -0.06746740639209747, 0.009400063194334507, -0.028680970892310143, -0.05339047312736511, -0.01608198881149292, -0.016428595408797264, -0.026020267978310585, 0.01728573814034462, -0.08674904704093933, 0.07192159444093704, -0.04344324767589569, -0.02284659817814827], "cos_innovation_negg": [-0.06071232259273529, 0.005213736090809107, 0.0790824294090271, -0.018299266695976257, -0.052069876343011856, 0.004859775770455599, -0.04551982879638672, 0.026532437652349472, -0.06746740639209747, 0.009400063194334507, -0.028680970892310143, -0.05339047312736511, -0.01608198881149292, -0.016428595408797264, -0.026020267978310585, 0.01728573814034462, -0.08674904704093933, 0.07192159444093704, -0.04344324767589569, -0.02284659817814827], "cos_apical_negg": [-0.06071232259273529, 0.005213736090809107, 0.0790824294090271, -0.018299266695976257, -0.052069876343011856, 0.004859775770455599, -0.04551982879638672, 0.026532437652349472, -0.06746740639209747, 0.009400063194334507, -0.028680970892310143, -0.05339047312736511, -0.01608198881149292, -0.016428595408797264, -0.026020267978310585, 0.01728573814034462, -0.08674904704093933, 0.07192159444093704, -0.04344324767589569, -0.02284659817814827], "cos_Ac_negg": [-0.06071232259273529, 0.005213736090809107, 0.0790824294090271, -0.018299266695976257, -0.052069876343011856, 0.004859775770455599, -0.04551982879638672, 0.026532437652349472, -0.06746740639209747, 0.009400063194334507, -0.028680970892310143, -0.05339047312736511, -0.01608198881149292, -0.016428595408797264, -0.026020267978310585, 0.01728573814034462, -0.08674904704093933, 0.07192159444093704, -0.04344324767589569, -0.02284659817814827], "r_norm": [2.373753547668457, 2.381035804748535, 2.6179983615875244, 2.641928195953369, 2.546583652496338, 2.6552579402923584, 2.3505499362945557, 2.5264806747436523, 2.3543055057525635, 2.5372440814971924, 2.4776666164398193, 2.4098167419433594, 2.419410467147827, 2.5828075408935547, 2.4807233810424805, 2.4316391944885254, 2.407501220703125, 2.4628305435180664, 2.5737342834472656, 2.6307435035705566]}, {"epoch_end": 0, "step": 391, "test_acc": 0.3393, "test_loss": 1.8734920288085937}, {"step": 500, "epoch": 1, "train_loss": 1.8165531158447266, "cos_r_negg": [0.5002347230911255, 0.043858159333467484, 6.642681546509266e-05, 0.02587403729557991, -0.02261985093355179, -0.0590977668762207, 0.01530294306576252, -0.03123420476913452, 0.010971649549901485, -0.02743738703429699, -0.0003065895289182663, 0.062063269317150116, 0.008998382836580276, -0.03741520270705223, 0.01767190545797348, -0.0600443109869957, 0.08992031216621399, -0.06332942843437195, -0.046404529362916946, -0.041476354002952576], "cos_innovation_negg": [0.5002347230911255, 0.043858159333467484, 6.642681546509266e-05, 0.02587403729557991, -0.02261985093355179, -0.0590977668762207, 0.01530294306576252, -0.03123420476913452, 0.010971649549901485, -0.02743738703429699, -0.0003065895289182663, 0.062063269317150116, 0.008998382836580276, -0.03741520270705223, 0.01767190545797348, -0.0600443109869957, 0.08992031216621399, -0.06332942843437195, -0.046404529362916946, -0.041476354002952576], "cos_apical_negg": [0.5002347230911255, 0.043858159333467484, 6.642681546509266e-05, 0.02587403729557991, -0.02261985093355179, -0.0590977668762207, 0.01530294306576252, -0.03123420476913452, 0.010971649549901485, -0.02743738703429699, -0.0003065895289182663, 0.062063269317150116, 0.008998382836580276, -0.03741520270705223, 0.01767190545797348, -0.0600443109869957, 0.08992031216621399, -0.06332942843437195, -0.046404529362916946, -0.041476354002952576], "cos_Ac_negg": [0.5002347230911255, 0.043858159333467484, 6.642681546509266e-05, 0.02587403729557991, -0.02261985093355179, -0.0590977668762207, 0.01530294306576252, -0.03123420476913452, 0.010971649549901485, -0.02743738703429699, -0.0003065895289182663, 0.062063269317150116, 0.008998382836580276, -0.03741520270705223, 0.01767190545797348, -0.0600443109869957, 0.08992031216621399, -0.06332942843437195, -0.046404529362916946, -0.041476354002952576], "r_norm": [2.093061685562134, 2.1204607486724854, 2.3076467514038086, 2.323662757873535, 2.239992618560791, 2.3395605087280273, 2.0996146202087402, 2.237675428390503, 2.0879619121551514, 2.268934726715088, 2.232789993286133, 2.1217377185821533, 2.164804220199585, 2.3101634979248047, 2.1825289726257324, 2.1534056663513184, 2.124000310897827, 2.172222852706909, 2.29445219039917, 2.304210662841797]}, {"epoch_end": 1, "step": 782, "test_acc": 0.3498, "test_loss": 1.8694535278320312}, {"step": 1000, "epoch": 2, "train_loss": 1.8252761363983154, "cos_r_negg": [0.5522254705429077, 0.034440185874700546, -0.010455763898789883, 0.0917133018374443, -0.057586271315813065, -0.011674870736896992, 0.01938370056450367, -0.04388536885380745, 0.07874585688114166, 0.07022294402122498, 0.07727109640836716, 0.045922670513391495, 0.04192787781357765, -0.025899743661284447, -0.019367411732673645, -0.056662872433662415, 0.06541380286216736, 0.0024536545388400555, -0.02060539834201336, -0.05119837820529938], "cos_innovation_negg": [0.5522254705429077, 0.034440185874700546, -0.010455763898789883, 0.0917133018374443, -0.057586271315813065, -0.011674870736896992, 0.01938370056450367, -0.04388536885380745, 0.07874585688114166, 0.07022294402122498, 0.07727109640836716, 0.045922670513391495, 0.04192787781357765, -0.025899743661284447, -0.019367411732673645, -0.056662872433662415, 0.06541380286216736, 0.0024536545388400555, -0.02060539834201336, -0.05119837820529938], "cos_apical_negg": [0.5522254705429077, 0.034440185874700546, -0.010455763898789883, 0.0917133018374443, -0.057586271315813065, -0.011674870736896992, 0.01938370056450367, -0.04388536885380745, 0.07874585688114166, 0.07022294402122498, 0.07727109640836716, 0.045922670513391495, 0.04192787781357765, -0.025899743661284447, -0.019367411732673645, -0.056662872433662415, 0.06541380286216736, 0.0024536545388400555, -0.02060539834201336, -0.05119837820529938], "cos_Ac_negg": [0.5522254705429077, 0.034440185874700546, -0.010455763898789883, 0.0917133018374443, -0.057586271315813065, -0.011674870736896992, 0.01938370056450367, -0.04388536885380745, 0.07874585688114166, 0.07022294402122498, 0.07727109640836716, 0.045922670513391495, 0.04192787781357765, -0.025899743661284447, -0.019367411732673645, -0.056662872433662415, 0.06541380286216736, 0.0024536545388400555, -0.02060539834201336, -0.05119837820529938], "r_norm": [2.0586817264556885, 2.0759541988372803, 2.2594785690307617, 2.2859439849853516, 2.1925454139709473, 2.289053440093994, 2.065835475921631, 2.189729690551758, 2.0581436157226562, 2.2414002418518066, 2.1939404010772705, 2.075897216796875, 2.1219751834869385, 2.2505972385406494, 2.1458840370178223, 2.119065284729004, 2.100233316421509, 2.1362011432647705, 2.248685359954834, 2.276573896408081]}, {"epoch_end": 2, "step": 1173, "test_acc": 0.335, "test_loss": 1.878783251953125}, {"step": 1500, "epoch": 3, "train_loss": 1.663377285003662, "cos_r_negg": [0.5334740877151489, 0.047047413885593414, 0.07195411622524261, 0.04064621403813362, -0.03921854496002197, -0.040143974125385284, 0.05162792280316353, -0.03376632183790207, 0.06681475043296814, 0.020489215850830078, 0.0934518352150917, 0.02985125221312046, 0.0361470952630043, 0.018702486529946327, 0.02190229669213295, 0.0021633070427924395, 0.08257973939180374, 0.013935133814811707, 0.019158635288476944, -0.07838358730077744], "cos_innovation_negg": [0.5334740877151489, 0.047047413885593414, 0.07195411622524261, 0.04064621403813362, -0.03921854496002197, -0.040143974125385284, 0.05162792280316353, -0.03376632183790207, 0.06681475043296814, 0.020489215850830078, 0.0934518352150917, 0.02985125221312046, 0.0361470952630043, 0.018702486529946327, 0.02190229669213295, 0.0021633070427924395, 0.08257973939180374, 0.013935133814811707, 0.019158635288476944, -0.07838358730077744], "cos_apical_negg": [0.5334740877151489, 0.047047413885593414, 0.07195411622524261, 0.04064621403813362, -0.03921854496002197, -0.040143974125385284, 0.05162792280316353, -0.03376632183790207, 0.06681475043296814, 0.020489215850830078, 0.0934518352150917, 0.02985125221312046, 0.0361470952630043, 0.018702486529946327, 0.02190229669213295, 0.0021633070427924395, 0.08257973939180374, 0.013935133814811707, 0.019158635288476944, -0.07838358730077744], "cos_Ac_negg": [0.5334740877151489, 0.047047413885593414, 0.07195411622524261, 0.04064621403813362, -0.03921854496002197, -0.040143974125385284, 0.05162792280316353, -0.03376632183790207, 0.06681475043296814, 0.020489215850830078, 0.0934518352150917, 0.02985125221312046, 0.0361470952630043, 0.018702486529946327, 0.02190229669213295, 0.0021633070427924395, 0.08257973939180374, 0.013935133814811707, 0.019158635288476944, -0.07838358730077744], "r_norm": [2.050379514694214, 2.079763174057007, 2.256718158721924, 2.2905430793762207, 2.197795867919922, 2.303170680999756, 2.060720682144165, 2.1977427005767822, 2.056859016418457, 2.2393763065338135, 2.1910488605499268, 2.0860238075256348, 2.127885580062866, 2.2521402835845947, 2.1408917903900146, 2.116903781890869, 2.0888519287109375, 2.1458494663238525, 2.2517213821411133, 2.2701072692871094]}, {"epoch_end": 3, "step": 1564, "test_acc": 0.3674, "test_loss": 1.8051173828125}, {"epoch_end": 4, "step": 1955, "test_acc": 0.3685, "test_loss": 1.8002416870117188}], "final": {"test_acc": 0.3685, "test_loss": 1.8002416870117188, "wall_s": 15.431184530258179, "cos_r_negg": [0.47978752851486206, 0.06584946811199188, 0.04108799993991852, 0.06741911172866821, -0.003964925184845924, -0.04203829541802406, 0.014618560671806335, -0.018200188875198364, 0.1154070794582367, 0.057388801127672195, 0.10789145529270172, 0.02186506986618042, 0.05351800471544266, 0.041022367775440216, 0.02211340144276619, 0.013741509988904, 0.07854858040809631, 0.04901934042572975, 0.04385339841246605, -0.01647588610649109], "cos_innovation_negg": [0.47978752851486206, 0.06584946811199188, 0.04108799993991852, 0.06741911172866821, -0.003964925184845924, -0.04203829541802406, 0.014618560671806335, -0.018200188875198364, 0.1154070794582367, 0.057388801127672195, 0.10789145529270172, 0.02186506986618042, 0.05351800471544266, 0.041022367775440216, 0.02211340144276619, 0.013741509988904, 0.07854858040809631, 0.04901934042572975, 0.04385339841246605, -0.01647588610649109], "cos_apical_negg": [0.47978752851486206, 0.06584946811199188, 0.04108799993991852, 0.06741911172866821, -0.003964925184845924, -0.04203829541802406, 0.014618560671806335, -0.018200188875198364, 0.1154070794582367, 0.057388801127672195, 0.10789145529270172, 0.02186506986618042, 0.05351800471544266, 0.041022367775440216, 0.02211340144276619, 0.013741509988904, 0.07854858040809631, 0.04901934042572975, 0.04385339841246605, -0.01647588610649109], "cos_Ac_negg": [0.47978752851486206, 0.06584946811199188, 0.04108799993991852, 0.06741911172866821, -0.003964925184845924, -0.04203829541802406, 0.014618560671806335, -0.018200188875198364, 0.1154070794582367, 0.057388801127672195, 0.10789145529270172, 0.02186506986618042, 0.05351800471544266, 0.041022367775440216, 0.02211340144276619, 0.013741509988904, 0.07854858040809631, 0.04901934042572975, 0.04385339841246605, -0.01647588610649109]}}
|
