{"args": {"mode": "sdil", "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": 0.02, "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_sdil_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": [0.00890412274748087, -0.010813121683895588, -0.07350920140743256, 0.013458739966154099, 0.03097783587872982, -0.03341919183731079, -0.05201931670308113, 0.020390721037983894, 0.0702129602432251, 0.023283839225769043], "cos_innovation_negg": [0.00890412274748087, -0.010813121683895588, -0.07350920140743256, 0.013458739966154099, 0.03097783587872982, -0.03341919183731079, -0.05201931670308113, 0.020390721037983894, 0.0702129602432251, 0.023283839225769043], "cos_apical_negg": [0.00890412274748087, -0.010813121683895588, -0.07350920140743256, 0.013458739966154099, 0.03097783587872982, -0.03341919183731079, -0.05201931670308113, 0.020390721037983894, 0.0702129602432251, 0.023283839225769043], "cos_Ac_negg": [0.00890412274748087, -0.010813121683895588, -0.07350920140743256, 0.013458739966154099, 0.03097783587872982, -0.03341919183731079, -0.05201931670308113, 0.020390721037983894, 0.0702129602432251, 0.023283839225769043], "r_norm": [2.454575538635254, 2.4658546447753906, 2.4461216926574707, 2.3404316902160645, 2.4380602836608887, 2.5399534702301025, 2.412501811981201, 2.3276968002319336, 2.3590633869171143, 2.4193224906921387], "ldr": {"dL_sdil": 0.0296630859375, "dL_bp": -0.25075697898864746, "ratio": -0.11829415897869362}}, {"epoch_end": 0, "step": 391, "test_acc": 0.3678, "test_loss": 1.8021219360351564}, {"step": 500, "epoch": 1, "train_loss": 1.643762469291687, "cos_r_negg": [0.668576180934906, 0.08138389885425568, -0.010578181594610214, 0.13176356256008148, 0.006847953423857689, 0.057518117129802704, 0.06511583924293518, 0.04449224844574928, 0.10441965609788895, 0.11995890736579895], "cos_innovation_negg": [0.668576180934906, 0.08138389885425568, -0.010578181594610214, 0.13176356256008148, 0.006847953423857689, 0.057518117129802704, 0.06511583924293518, 0.04449224844574928, 0.10441965609788895, 0.11995890736579895], "cos_apical_negg": [0.668576180934906, 0.08138389885425568, -0.010578181594610214, 0.13176356256008148, 0.006847953423857689, 0.057518117129802704, 0.06511583924293518, 0.04449224844574928, 0.10441965609788895, 0.11995890736579895], "cos_Ac_negg": [0.668576180934906, 0.08138389885425568, -0.010578181594610214, 0.13176356256008148, 0.006847953423857689, 0.057518117129802704, 0.06511583924293518, 0.04449224844574928, 0.10441965609788895, 0.11995890736579895], "r_norm": [1.803136944770813, 1.753688097000122, 1.7280343770980835, 1.6760941743850708, 1.7355520725250244, 1.8076618909835815, 1.7228456735610962, 1.6739610433578491, 1.6996949911117554, 1.741556167602539]}, {"epoch_end": 1, "step": 782, "test_acc": 0.3965, "test_loss": 1.743835400390625}, {"step": 1000, "epoch": 2, "train_loss": 1.7011975049972534, "cos_r_negg": [0.7419799566268921, 0.11188904941082001, 0.07698537409305573, 0.2065775990486145, 0.05961015447974205, 0.13213741779327393, 0.15764692425727844, 0.13687898218631744, 0.1703268587589264, 0.21145741641521454], "cos_innovation_negg": [0.7419799566268921, 0.11188904941082001, 0.07698537409305573, 0.2065775990486145, 0.05961015447974205, 0.13213741779327393, 0.15764692425727844, 0.13687898218631744, 0.1703268587589264, 0.21145741641521454], "cos_apical_negg": [0.7419799566268921, 0.11188904941082001, 0.07698537409305573, 0.2065775990486145, 0.05961015447974205, 0.13213741779327393, 0.15764692425727844, 0.13687898218631744, 0.1703268587589264, 0.21145741641521454], "cos_Ac_negg": [0.7419799566268921, 0.11188904941082001, 0.07698537409305573, 0.2065775990486145, 0.05961015447974205, 0.13213741779327393, 0.15764692425727844, 0.13687898218631744, 0.1703268587589264, 0.21145741641521454], "r_norm": [1.5534785985946655, 1.4312855005264282, 1.4071943759918213, 1.3801522254943848, 1.403579831123352, 1.4719865322113037, 1.4034450054168701, 1.3655714988708496, 1.3894765377044678, 1.4327127933502197]}, {"epoch_end": 2, "step": 1173, "test_acc": 0.4008, "test_loss": 1.7155303466796874}, {"step": 1500, "epoch": 3, "train_loss": 1.653611421585083, "cos_r_negg": [0.7580274343490601, 0.18283510208129883, 0.18835005164146423, 0.29169800877571106, 0.14787468314170837, 0.22053425014019012, 0.25823792815208435, 0.2636600136756897, 0.2753768265247345, 0.30040299892425537], "cos_innovation_negg": [0.7580274343490601, 0.18283510208129883, 0.18835005164146423, 0.29169800877571106, 0.14787468314170837, 0.22053425014019012, 0.25823792815208435, 0.2636600136756897, 0.2753768265247345, 0.30040299892425537], "cos_apical_negg": [0.7580274343490601, 0.18283510208129883, 0.18835005164146423, 0.29169800877571106, 0.14787468314170837, 0.22053425014019012, 0.25823792815208435, 0.2636600136756897, 0.2753768265247345, 0.30040299892425537], "cos_Ac_negg": [0.7580274343490601, 0.18283510208129883, 0.18835005164146423, 0.29169800877571106, 0.14787468314170837, 0.22053425014019012, 0.25823792815208435, 0.2636600136756897, 0.2753768265247345, 0.30040299892425537], "r_norm": [1.338926911354065, 1.165548324584961, 1.1345579624176025, 1.1281288862228394, 1.1352111101150513, 1.1927907466888428, 1.140704870223999, 1.114128589630127, 1.1318466663360596, 1.168143630027771]}, {"epoch_end": 3, "step": 1564, "test_acc": 0.4007, "test_loss": 1.7157528198242187}, {"epoch_end": 4, "step": 1955, "test_acc": 0.4135, "test_loss": 1.7036121704101563}], "final": {"test_acc": 0.4135, "test_loss": 1.7036121704101563, "wall_s": 16.892754316329956, "cos_r_negg": [0.7832787036895752, 0.3105674982070923, 0.27877748012542725, 0.40869736671447754, 0.25105783343315125, 0.3396332859992981, 0.35860276222229004, 0.39316368103027344, 0.38421788811683655, 0.42819368839263916], "cos_innovation_negg": [0.7832787036895752, 0.3105674982070923, 0.27877748012542725, 0.40869736671447754, 0.25105783343315125, 0.3396332859992981, 0.35860276222229004, 0.39316368103027344, 0.38421788811683655, 0.42819368839263916], "cos_apical_negg": [0.7832787036895752, 0.3105674982070923, 0.27877748012542725, 0.40869736671447754, 0.25105783343315125, 0.3396332859992981, 0.35860276222229004, 0.39316368103027344, 0.38421788811683655, 0.42819368839263916], "cos_Ac_negg": [0.7832787036895752, 0.3105674982070923, 0.27877748012542725, 0.40869736671447754, 0.25105783343315125, 0.3396332859992981, 0.35860276222229004, 0.39316368103027344, 0.38421788811683655, 0.42819368839263916]}}