1
|
{"args": {"mode": "bp", "dataset": "teacher", "depth": 12, "width": 16, "act": "tanh", "residual": 1, "epochs": 25, "batch_size": 128, "train_examples": 0, "val_examples": 5000, "split_seed": 2027, "eval_split": "validation", "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": 0.03, "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, "traffic_mode": "soma", "predictor_mode": "diagonal", "normalize_delta": 0, "settle_steps": 0, "kappa": 0.0, "feedback": "error", "seed": 0, "device": "cuda", "log_every": 200, "max_steps": 0, "probe_bs": 512, "outdir": "results", "tag": "depthtask_dev_v1_teacher_bp_w16_d12_t0_s0", "n_in": 128, "n_out": 10}, "split": {"dataset": "teacher", "task_seed": 0, "train_examples": 45000, "test_examples": 10000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 5000, "validation_index_sha256": "0ec30707a69784626a2d17128e5408bb8a8a8964e2ea753ef8e54b35ffb36355", "validation_class_counts": {"0": 500, "1": 500, "2": 500, "3": 500, "4": 500, "5": 500, "6": 500, "7": 500, "8": 500, "9": 500}, "split_from_training_only": true}, "provenance": {"git_commit": "a0696a3d72cc5bf69876e1716b74f7e1ac525dd0", "git_dirty": false}, "steps": [{"step": 0, "epoch": 0, "train_loss": 2.517413854598999}, {"step": 200, "epoch": 0, "train_loss": 0.9769093990325928}, {"epoch_end": 0, "step": 352, "eval_split": "validation", "eval_acc": 0.5692, "eval_loss": 1.1378795654296876, "val_acc": 0.5692, "val_loss": 1.1378795654296876}, {"step": 400, "epoch": 1, "train_loss": 0.9296771883964539}, {"step": 600, "epoch": 1, "train_loss": 0.8773318529129028}, {"epoch_end": 1, "step": 704, "eval_split": "validation", "eval_acc": 0.581, "eval_loss": 1.1087588134765625, "val_acc": 0.581, "val_loss": 1.1087588134765625}, {"step": 800, "epoch": 2, "train_loss": 0.8861347436904907}, {"step": 1000, "epoch": 2, "train_loss": 0.9042978882789612}, {"epoch_end": 2, "step": 1056, "eval_split": "validation", "eval_acc": 0.582, "eval_loss": 1.11873134765625, "val_acc": 0.582, "val_loss": 1.11873134765625}, {"step": 1200, "epoch": 3, "train_loss": 0.9459412693977356}, {"step": 1400, "epoch": 3, "train_loss": 1.064085841178894}, {"epoch_end": 3, "step": 1408, "eval_split": "validation", "eval_acc": 0.5866, "eval_loss": 1.120823095703125, "val_acc": 0.5866, "val_loss": 1.120823095703125}, {"step": 1600, "epoch": 4, "train_loss": 0.9249476790428162}, {"epoch_end": 4, "step": 1760, "eval_split": "validation", "eval_acc": 0.5834, "eval_loss": 1.11244638671875, "val_acc": 0.5834, "val_loss": 1.11244638671875}, {"step": 1800, "epoch": 5, "train_loss": 1.039158821105957}, {"step": 2000, "epoch": 5, "train_loss": 0.8265764713287354}, {"epoch_end": 5, "step": 2112, "eval_split": "validation", "eval_acc": 0.5856, "eval_loss": 1.1243849853515624, "val_acc": 0.5856, "val_loss": 1.1243849853515624}, {"step": 2200, "epoch": 6, "train_loss": 1.0884428024291992}, {"step": 2400, "epoch": 6, "train_loss": 0.8787632584571838}, {"epoch_end": 6, "step": 2464, "eval_split": "validation", "eval_acc": 0.5984, "eval_loss": 1.0845456909179687, "val_acc": 0.5984, "val_loss": 1.0845456909179687}, {"step": 2600, "epoch": 7, "train_loss": 0.9358597993850708}, {"step": 2800, "epoch": 7, "train_loss": 1.064014196395874}, {"epoch_end": 7, "step": 2816, "eval_split": "validation", "eval_acc": 0.5812, "eval_loss": 1.1327805419921875, "val_acc": 0.5812, "val_loss": 1.1327805419921875}, {"step": 3000, "epoch": 8, "train_loss": 0.9705671072006226}, {"epoch_end": 8, "step": 3168, "eval_split": "validation", "eval_acc": 0.594, "eval_loss": 1.0957290405273437, "val_acc": 0.594, "val_loss": 1.0957290405273437}, {"step": 3200, "epoch": 9, "train_loss": 0.8985673785209656}, {"step": 3400, "epoch": 9, "train_loss": 0.8335405588150024}, {"epoch_end": 9, "step": 3520, "eval_split": "validation", "eval_acc": 0.5878, "eval_loss": 1.1013695068359375, "val_acc": 0.5878, "val_loss": 1.1013695068359375}, {"step": 3600, "epoch": 10, "train_loss": 0.977665901184082}, {"step": 3800, "epoch": 10, "train_loss": 0.8490504622459412}, {"epoch_end": 10, "step": 3872, "eval_split": "validation", "eval_acc": 0.599, "eval_loss": 1.080225537109375, "val_acc": 0.599, "val_loss": 1.080225537109375}, {"step": 4000, "epoch": 11, "train_loss": 0.9312368035316467}, {"step": 4200, "epoch": 11, "train_loss": 1.0891715288162231}, {"epoch_end": 11, "step": 4224, "eval_split": "validation", "eval_acc": 0.5842, "eval_loss": 1.1082308471679687, "val_acc": 0.5842, "val_loss": 1.1082308471679687}, {"step": 4400, "epoch": 12, "train_loss": 0.9681256413459778}, {"epoch_end": 12, "step": 4576, "eval_split": "validation", "eval_acc": 0.5916, "eval_loss": 1.100140380859375, "val_acc": 0.5916, "val_loss": 1.100140380859375}, {"step": 4600, "epoch": 13, "train_loss": 1.093500018119812}, {"step": 4800, "epoch": 13, "train_loss": 0.9157226085662842}, {"epoch_end": 13, "step": 4928, "eval_split": "validation", "eval_acc": 0.579, "eval_loss": 1.1225397705078124, "val_acc": 0.579, "val_loss": 1.1225397705078124}, {"step": 5000, "epoch": 14, "train_loss": 0.9541612267494202}, {"step": 5200, "epoch": 14, "train_loss": 0.9279407262802124}, {"epoch_end": 14, "step": 5280, "eval_split": "validation", "eval_acc": 0.5928, "eval_loss": 1.1008900390625, "val_acc": 0.5928, "val_loss": 1.1008900390625}, {"step": 5400, "epoch": 15, "train_loss": 0.9405715465545654}, {"step": 5600, "epoch": 15, "train_loss": 0.9940947890281677}, {"epoch_end": 15, "step": 5632, "eval_split": "validation", "eval_acc": 0.5884, "eval_loss": 1.1134339721679687, "val_acc": 0.5884, "val_loss": 1.1134339721679687}, {"step": 5800, "epoch": 16, "train_loss": 0.8996315598487854}, {"epoch_end": 16, "step": 5984, "eval_split": "validation", "eval_acc": 0.5968, "eval_loss": 1.0805838745117187, "val_acc": 0.5968, "val_loss": 1.0805838745117187}, {"step": 6000, "epoch": 17, "train_loss": 0.8689146637916565}, {"step": 6200, "epoch": 17, "train_loss": 0.8290905356407166}, {"epoch_end": 17, "step": 6336, "eval_split": "validation", "eval_acc": 0.5896, "eval_loss": 1.1009724853515626, "val_acc": 0.5896, "val_loss": 1.1009724853515626}, {"step": 6400, "epoch": 18, "train_loss": 1.0169360637664795}, {"step": 6600, "epoch": 18, "train_loss": 0.9826826453208923}, {"epoch_end": 18, "step": 6688, "eval_split": "validation", "eval_acc": 0.5866, "eval_loss": 1.1058596435546875, "val_acc": 0.5866, "val_loss": 1.1058596435546875}, {"step": 6800, "epoch": 19, "train_loss": 1.1044045686721802}, {"step": 7000, "epoch": 19, "train_loss": 0.7680770754814148}, {"epoch_end": 19, "step": 7040, "eval_split": "validation", "eval_acc": 0.5974, "eval_loss": 1.1014206298828124, "val_acc": 0.5974, "val_loss": 1.1014206298828124}, {"step": 7200, "epoch": 20, "train_loss": 0.9752224087715149}, {"epoch_end": 20, "step": 7392, "eval_split": "validation", "eval_acc": 0.5942, "eval_loss": 1.0939255859375, "val_acc": 0.5942, "val_loss": 1.0939255859375}, {"step": 7400, "epoch": 21, "train_loss": 1.0379060506820679}, {"step": 7600, "epoch": 21, "train_loss": 0.9451590180397034}, {"epoch_end": 21, "step": 7744, "eval_split": "validation", "eval_acc": 0.5904, "eval_loss": 1.0944151123046875, "val_acc": 0.5904, "val_loss": 1.0944151123046875}, {"step": 7800, "epoch": 22, "train_loss": 0.9887235760688782}, {"step": 8000, "epoch": 22, "train_loss": 0.893457293510437}, {"epoch_end": 22, "step": 8096, "eval_split": "validation", "eval_acc": 0.5916, "eval_loss": 1.1057144287109375, "val_acc": 0.5916, "val_loss": 1.1057144287109375}, {"step": 8200, "epoch": 23, "train_loss": 0.9537519216537476}, {"step": 8400, "epoch": 23, "train_loss": 0.9157257676124573}, {"epoch_end": 23, "step": 8448, "eval_split": "validation", "eval_acc": 0.5894, "eval_loss": 1.1086910278320312, "val_acc": 0.5894, "val_loss": 1.1086910278320312}, {"step": 8600, "epoch": 24, "train_loss": 0.8472723364830017}, {"epoch_end": 24, "step": 8800, "eval_split": "validation", "eval_acc": 0.5884, "eval_loss": 1.107358544921875, "val_acc": 0.5884, "val_loss": 1.107358544921875}], "final": {"eval_split": "validation", "eval_acc": 0.5884, "eval_loss": 1.107358544921875, "wall_s": 58.53278994560242, "val_acc": 0.5884, "val_loss": 1.107358544921875}}
|
