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": 0, "device": "cuda", "log_every": 500, "max_steps": 0, "probe_bs": 256, "outdir": "results", "tag": "scale_v3_cifar10_dfa_w64_d10_s0", "n_in": 3072, "n_out": 10}, "provenance": {"git_commit": "d4aef1a571c21e6e4e4125c90aa58da7f14e62e3", "git_dirty": false}, "steps": [{"step": 0, "epoch": 0, "train_loss": 2.783294200897217, "cos_r_negg": [-0.01275080069899559, -0.08000953495502472, 0.045286111533641815, -0.11770164966583252, -0.016248609870672226, 0.042709290981292725, 0.03047596663236618, -0.05089080333709717, 0.003128529991954565, -0.009440174326300621], "cos_innovation_negg": [-0.01275080069899559, -0.08000953495502472, 0.045286111533641815, -0.11770164966583252, -0.016248609870672226, 0.042709290981292725, 0.03047596663236618, -0.05089080333709717, 0.003128529991954565, -0.009440174326300621], "cos_apical_negg": [-0.01275080069899559, -0.08000953495502472, 0.045286111533641815, -0.11770164966583252, -0.016248609870672226, 0.042709290981292725, 0.03047596663236618, -0.05089080333709717, 0.003128529991954565, -0.009440174326300621], "cos_Ac_negg": [-0.01275080069899559, -0.08000953495502472, 0.045286111533641815, -0.11770164966583252, -0.016248609870672226, 0.042709290981292725, 0.03047596663236618, -0.05089080333709717, 0.003128529991954565, -0.009440174326300621], "r_norm": [2.4784693717956543, 2.4636573791503906, 2.5729591846466064, 2.5139784812927246, 2.6724207401275635, 2.5304789543151855, 2.720974922180176, 2.47133731842041, 2.5015954971313477, 2.359222173690796]}, {"epoch_end": 0, "step": 391, "test_acc": 0.342, "test_loss": 1.8846063232421875}, {"step": 500, "epoch": 1, "train_loss": 1.9188976287841797, "cos_r_negg": [0.40976661443710327, 0.0823037400841713, -0.04269060865044594, 0.05714689940214157, 0.043102577328681946, 0.07622861862182617, 0.03270609304308891, -0.014595740474760532, -0.011077770963311195, 0.02199837937951088], "cos_innovation_negg": [0.40976661443710327, 0.0823037400841713, -0.04269060865044594, 0.05714689940214157, 0.043102577328681946, 0.07622861862182617, 0.03270609304308891, -0.014595740474760532, -0.011077770963311195, 0.02199837937951088], "cos_apical_negg": [0.40976661443710327, 0.0823037400841713, -0.04269060865044594, 0.05714689940214157, 0.043102577328681946, 0.07622861862182617, 0.03270609304308891, -0.014595740474760532, -0.011077770963311195, 0.02199837937951088], "cos_Ac_negg": [0.40976661443710327, 0.0823037400841713, -0.04269060865044594, 0.05714689940214157, 0.043102577328681946, 0.07622861862182617, 0.03270609304308891, -0.014595740474760532, -0.011077770963311195, 0.02199837937951088], "r_norm": [2.1915183067321777, 2.128800630569458, 2.2551941871643066, 2.152132511138916, 2.292327880859375, 2.1761560440063477, 2.334346055984497, 2.1396353244781494, 2.1809182167053223, 2.0230562686920166]}, {"epoch_end": 1, "step": 782, "test_acc": 0.3478, "test_loss": 1.8655614135742187}, {"step": 1000, "epoch": 2, "train_loss": 1.8765400648117065, "cos_r_negg": [0.5463718771934509, 0.0869361013174057, 0.014505192637443542, 0.011697749607264996, 0.049766309559345245, 0.030424848198890686, 0.03534165769815445, 0.04410041123628616, -0.001267608255147934, 0.057372160255908966], "cos_innovation_negg": [0.5463718771934509, 0.0869361013174057, 0.014505192637443542, 0.011697749607264996, 0.049766309559345245, 0.030424848198890686, 0.03534165769815445, 0.04410041123628616, -0.001267608255147934, 0.057372160255908966], "cos_apical_negg": [0.5463718771934509, 0.0869361013174057, 0.014505192637443542, 0.011697749607264996, 0.049766309559345245, 0.030424848198890686, 0.03534165769815445, 0.04410041123628616, -0.001267608255147934, 0.057372160255908966], "cos_Ac_negg": [0.5463718771934509, 0.0869361013174057, 0.014505192637443542, 0.011697749607264996, 0.049766309559345245, 0.030424848198890686, 0.03534165769815445, 0.04410041123628616, -0.001267608255147934, 0.057372160255908966], "r_norm": [2.18318772315979, 2.121843099594116, 2.2471535205841064, 2.1383323669433594, 2.264104127883911, 2.1673600673675537, 2.320222854614258, 2.1218836307525635, 2.159454345703125, 2.0140483379364014]}, {"epoch_end": 2, "step": 1173, "test_acc": 0.3536, "test_loss": 1.84021171875}, {"step": 1500, "epoch": 3, "train_loss": 1.6959843635559082, "cos_r_negg": [0.5065834522247314, 0.09959973394870758, 0.014033414423465729, 0.04542933776974678, 0.06719458103179932, 0.12632529437541962, 0.044706087559461594, -0.004179208539426327, 0.07284034788608551, 0.08661135286092758], "cos_innovation_negg": [0.5065834522247314, 0.09959973394870758, 0.014033414423465729, 0.04542933776974678, 0.06719458103179932, 0.12632529437541962, 0.044706087559461594, -0.004179208539426327, 0.07284034788608551, 0.08661135286092758], "cos_apical_negg": [0.5065834522247314, 0.09959973394870758, 0.014033414423465729, 0.04542933776974678, 0.06719458103179932, 0.12632529437541962, 0.044706087559461594, -0.004179208539426327, 0.07284034788608551, 0.08661135286092758], "cos_Ac_negg": [0.5065834522247314, 0.09959973394870758, 0.014033414423465729, 0.04542933776974678, 0.06719458103179932, 0.12632529437541962, 0.044706087559461594, -0.004179208539426327, 0.07284034788608551, 0.08661135286092758], "r_norm": [2.1689391136169434, 2.110792875289917, 2.238413095474243, 2.1340909004211426, 2.2669715881347656, 2.167285203933716, 2.3172264099121094, 2.1193315982818604, 2.161379814147949, 2.009406089782715]}, {"epoch_end": 3, "step": 1564, "test_acc": 0.3507, "test_loss": 1.850153955078125}, {"epoch_end": 4, "step": 1955, "test_acc": 0.3685, "test_loss": 1.8021558959960937}], "final": {"test_acc": 0.3685, "test_loss": 1.8021558959960937, "wall_s": 12.518126726150513, "cos_r_negg": [0.4909083843231201, 0.14414826035499573, 0.05920688062906265, 0.08365839719772339, 0.07772965729236603, 0.1297677755355835, 0.07531890273094177, 0.07135607302188873, 0.07624480128288269, 0.059515342116355896], "cos_innovation_negg": [0.4909083843231201, 0.14414826035499573, 0.05920688062906265, 0.08365839719772339, 0.07772965729236603, 0.1297677755355835, 0.07531890273094177, 0.07135607302188873, 0.07624480128288269, 0.059515342116355896], "cos_apical_negg": [0.4909083843231201, 0.14414826035499573, 0.05920688062906265, 0.08365839719772339, 0.07772965729236603, 0.1297677755355835, 0.07531890273094177, 0.07135607302188873, 0.07624480128288269, 0.059515342116355896], "cos_Ac_negg": [0.4909083843231201, 0.14414826035499573, 0.05920688062906265, 0.08365839719772339, 0.07772965729236603, 0.1297677755355835, 0.07531890273094177, 0.07135607302188873, 0.07624480128288269, 0.059515342116355896]}}
|