summaryrefslogtreecommitdiff
path: root/results/scale_v3_cifar10_dfa_w64_d5_s0.json
blob: 62665deaa3a9dc97d95be77811f373a148a55f32 (plain)
1
{"args": {"mode": "dfa", "dataset": "cifar10", "depth": 5, "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_d5_s0", "n_in": 3072, "n_out": 10}, "provenance": {"git_commit": "d4aef1a571c21e6e4e4125c90aa58da7f14e62e3", "git_dirty": false}, "steps": [{"step": 0, "epoch": 0, "train_loss": 2.403229236602783, "cos_r_negg": [0.02148173376917839, -0.002783835167065263, -0.08420182019472122, -0.04247276857495308, 0.030795220285654068], "cos_innovation_negg": [0.02148173376917839, -0.002783835167065263, -0.08420182019472122, -0.04247276857495308, 0.030795220285654068], "cos_apical_negg": [0.02148173376917839, -0.002783835167065263, -0.08420182019472122, -0.04247276857495308, 0.030795220285654068], "cos_Ac_negg": [0.02148173376917839, -0.002783835167065263, -0.08420182019472122, -0.04247276857495308, 0.030795220285654068], "r_norm": [2.402596950531006, 2.5663161277770996, 2.507201910018921, 2.3726465702056885, 2.3823375701904297]}, {"epoch_end": 0, "step": 391, "test_acc": 0.3396, "test_loss": 1.8752569213867187}, {"step": 500, "epoch": 1, "train_loss": 1.7766005992889404, "cos_r_negg": [0.48811987042427063, 0.03559161350131035, -0.022757787257432938, 0.0048194159753620625, -0.01352208573371172], "cos_innovation_negg": [0.48811987042427063, 0.03559161350131035, -0.022757787257432938, 0.0048194159753620625, -0.01352208573371172], "cos_apical_negg": [0.48811987042427063, 0.03559161350131035, -0.022757787257432938, 0.0048194159753620625, -0.01352208573371172], "cos_Ac_negg": [0.48811987042427063, 0.03559161350131035, -0.022757787257432938, 0.0048194159753620625, -0.01352208573371172], "r_norm": [2.148845672607422, 2.289592742919922, 2.229010581970215, 2.0900979042053223, 2.1220974922180176]}, {"epoch_end": 1, "step": 782, "test_acc": 0.3627, "test_loss": 1.8362994140625}, {"step": 1000, "epoch": 2, "train_loss": 1.8073923587799072, "cos_r_negg": [0.49647217988967896, 0.06527790427207947, 0.11554482579231262, 0.07530932873487473, 0.04652947187423706], "cos_innovation_negg": [0.49647217988967896, 0.06527790427207947, 0.11554482579231262, 0.07530932873487473, 0.04652947187423706], "cos_apical_negg": [0.49647217988967896, 0.06527790427207947, 0.11554482579231262, 0.07530932873487473, 0.04652947187423706], "cos_Ac_negg": [0.49647217988967896, 0.06527790427207947, 0.11554482579231262, 0.07530932873487473, 0.04652947187423706], "r_norm": [2.1908528804779053, 2.3290905952453613, 2.2665624618530273, 2.126660108566284, 2.165654420852661]}, {"epoch_end": 2, "step": 1173, "test_acc": 0.3463, "test_loss": 1.8442423217773438}, {"step": 1500, "epoch": 3, "train_loss": 1.7471094131469727, "cos_r_negg": [0.45534342527389526, 0.1711357831954956, 0.0959051325917244, 0.08062054216861725, 0.06809711456298828], "cos_innovation_negg": [0.45534342527389526, 0.1711357831954956, 0.0959051325917244, 0.08062054216861725, 0.06809711456298828], "cos_apical_negg": [0.45534342527389526, 0.1711357831954956, 0.0959051325917244, 0.08062054216861725, 0.06809711456298828], "cos_Ac_negg": [0.45534342527389526, 0.1711357831954956, 0.0959051325917244, 0.08062054216861725, 0.06809711456298828], "r_norm": [2.155304193496704, 2.302260398864746, 2.227808952331543, 2.0961947441101074, 2.1185240745544434]}, {"epoch_end": 3, "step": 1564, "test_acc": 0.365, "test_loss": 1.8174841186523438}, {"epoch_end": 4, "step": 1955, "test_acc": 0.3844, "test_loss": 1.767465771484375}], "final": {"test_acc": 0.3844, "test_loss": 1.767465771484375, "wall_s": 4.812915563583374, "cos_r_negg": [0.5412516593933105, 0.23857612907886505, 0.14359042048454285, 0.16168369352817535, 0.10131145268678665], "cos_innovation_negg": [0.5412516593933105, 0.23857612907886505, 0.14359042048454285, 0.16168369352817535, 0.10131145268678665], "cos_apical_negg": [0.5412516593933105, 0.23857612907886505, 0.14359042048454285, 0.16168369352817535, 0.10131145268678665], "cos_Ac_negg": [0.5412516593933105, 0.23857612907886505, 0.14359042048454285, 0.16168369352817535, 0.10131145268678665]}}