summaryrefslogtreecommitdiff
path: root/results/scale_v3_cifar10_dfa_w64_d5_s3.json
blob: 5da25bbeb3fff74193b3cc1d714b3018c1a20de5 (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": 3, "device": "cuda", "log_every": 500, "max_steps": 0, "probe_bs": 256, "outdir": "results", "tag": "scale_v3_cifar10_dfa_w64_d5_s3", "n_in": 3072, "n_out": 10}, "provenance": {"git_commit": "d4aef1a571c21e6e4e4125c90aa58da7f14e62e3", "git_dirty": false}, "steps": [{"step": 0, "epoch": 0, "train_loss": 2.5312678813934326, "cos_r_negg": [0.01098020002245903, 0.019480306655168533, 0.05898767337203026, 0.007838631048798561, 0.016951166093349457], "cos_innovation_negg": [0.01098020002245903, 0.019480306655168533, 0.05898767337203026, 0.007838631048798561, 0.016951166093349457], "cos_apical_negg": [0.01098020002245903, 0.019480306655168533, 0.05898767337203026, 0.007838631048798561, 0.016951166093349457], "cos_Ac_negg": [0.01098020002245903, 0.019480306655168533, 0.05898767337203026, 0.007838631048798561, 0.016951166093349457], "r_norm": [2.503025531768799, 2.3611538410186768, 2.4595303535461426, 2.4575695991516113, 2.559494972229004]}, {"epoch_end": 0, "step": 391, "test_acc": 0.3358, "test_loss": 1.9005482666015625}, {"step": 500, "epoch": 1, "train_loss": 1.7816979885101318, "cos_r_negg": [0.5328232049942017, 0.0057297623716294765, 0.017736613750457764, -0.02176375314593315, 0.026649020612239838], "cos_innovation_negg": [0.5328232049942017, 0.0057297623716294765, 0.017736613750457764, -0.02176375314593315, 0.026649020612239838], "cos_apical_negg": [0.5328232049942017, 0.0057297623716294765, 0.017736613750457764, -0.02176375314593315, 0.026649020612239838], "cos_Ac_negg": [0.5328232049942017, 0.0057297623716294765, 0.017736613750457764, -0.02176375314593315, 0.026649020612239838], "r_norm": [2.2604079246520996, 2.119000196456909, 2.196664571762085, 2.1956048011779785, 2.278520107269287]}, {"epoch_end": 1, "step": 782, "test_acc": 0.3323, "test_loss": 1.8884764404296874}, {"step": 1000, "epoch": 2, "train_loss": 1.808479905128479, "cos_r_negg": [0.5211073756217957, 0.07690542191267014, 0.12718281149864197, 0.008513258770108223, 0.06285284459590912], "cos_innovation_negg": [0.5211073756217957, 0.07690542191267014, 0.12718281149864197, 0.008513258770108223, 0.06285284459590912], "cos_apical_negg": [0.5211073756217957, 0.07690542191267014, 0.12718281149864197, 0.008513258770108223, 0.06285284459590912], "cos_Ac_negg": [0.5211073756217957, 0.07690542191267014, 0.12718281149864197, 0.008513258770108223, 0.06285284459590912], "r_norm": [2.214585304260254, 2.070432662963867, 2.1482949256896973, 2.143156051635742, 2.2267370223999023]}, {"epoch_end": 2, "step": 1173, "test_acc": 0.3461, "test_loss": 1.872576904296875}, {"step": 1500, "epoch": 3, "train_loss": 1.8156404495239258, "cos_r_negg": [0.48965734243392944, 0.1334024965763092, 0.17901238799095154, 0.024159131571650505, 0.14413520693778992], "cos_innovation_negg": [0.48965734243392944, 0.1334024965763092, 0.17901238799095154, 0.024159131571650505, 0.14413520693778992], "cos_apical_negg": [0.48965734243392944, 0.1334024965763092, 0.17901238799095154, 0.024159131571650505, 0.14413520693778992], "cos_Ac_negg": [0.48965734243392944, 0.1334024965763092, 0.17901238799095154, 0.024159131571650505, 0.14413520693778992], "r_norm": [2.230515241622925, 2.0890302658081055, 2.168339252471924, 2.157806396484375, 2.254725456237793]}, {"epoch_end": 3, "step": 1564, "test_acc": 0.3507, "test_loss": 1.8340133666992187}, {"epoch_end": 4, "step": 1955, "test_acc": 0.3761, "test_loss": 1.7844658813476562}], "final": {"test_acc": 0.3761, "test_loss": 1.7844658813476562, "wall_s": 4.991127252578735, "cos_r_negg": [0.48362597823143005, 0.20218965411186218, 0.21574810147285461, 0.07610452175140381, 0.20764309167861938], "cos_innovation_negg": [0.48362597823143005, 0.20218965411186218, 0.21574810147285461, 0.07610452175140381, 0.20764309167861938], "cos_apical_negg": [0.48362597823143005, 0.20218965411186218, 0.21574810147285461, 0.07610452175140381, 0.20764309167861938], "cos_Ac_negg": [0.48362597823143005, 0.20218965411186218, 0.21574810147285461, 0.07610452175140381, 0.20764309167861938]}}