{"args": {"mode": "bp", "dataset": "hierarchical", "depth": 8, "width": 64, "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": 6, "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_hierarchical_bp_w64_d8_t0_s0", "n_in": 64, "n_out": 10}, "split": {"dataset": "hierarchical", "task_seed": 0, "train_examples": 45000, "test_examples": 10000, "evaluation_split": "validation", "synthetic_generator": true, "split_seed": 2027, "validation_examples": 5000, "validation_index_sha256": "f5282ed201086357dbe24c23e70cc9eb86deaaf70d99c41553ffc78626ee8d6c", "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": 2.600351333618164}, {"step": 200, "epoch": 0, "train_loss": 1.7608060836791992}, {"epoch_end": 0, "step": 352, "eval_split": "validation", "eval_acc": 0.331, "eval_loss": 1.89740234375, "val_acc": 0.331, "val_loss": 1.89740234375}, {"step": 400, "epoch": 1, "train_loss": 1.7457090616226196}, {"step": 600, "epoch": 1, "train_loss": 1.7250763177871704}, {"epoch_end": 1, "step": 704, "eval_split": "validation", "eval_acc": 0.3472, "eval_loss": 1.87708076171875, "val_acc": 0.3472, "val_loss": 1.87708076171875}, {"step": 800, "epoch": 2, "train_loss": 1.6624048948287964}, {"step": 1000, "epoch": 2, "train_loss": 1.7809312343597412}, {"epoch_end": 2, "step": 1056, "eval_split": "validation", "eval_acc": 0.3576, "eval_loss": 1.8494482177734375, "val_acc": 0.3576, "val_loss": 1.8494482177734375}, {"step": 1200, "epoch": 3, "train_loss": 1.5753974914550781}, {"step": 1400, "epoch": 3, "train_loss": 1.5667771100997925}, {"epoch_end": 3, "step": 1408, "eval_split": "validation", "eval_acc": 0.3568, "eval_loss": 1.8406751220703126, "val_acc": 0.3568, "val_loss": 1.8406751220703126}, {"step": 1600, "epoch": 4, "train_loss": 1.6137046813964844}, {"epoch_end": 4, "step": 1760, "eval_split": "validation", "eval_acc": 0.372, "eval_loss": 1.8091213623046876, "val_acc": 0.372, "val_loss": 1.8091213623046876}, {"step": 1800, "epoch": 5, "train_loss": 1.7945579290390015}, {"step": 2000, "epoch": 5, "train_loss": 1.566240906715393}, {"epoch_end": 5, "step": 2112, "eval_split": "validation", "eval_acc": 0.3716, "eval_loss": 1.7866246337890626, "val_acc": 0.3716, "val_loss": 1.7866246337890626}, {"step": 2200, "epoch": 6, "train_loss": 1.534356951713562}, {"step": 2400, "epoch": 6, "train_loss": 1.512060523033142}, {"epoch_end": 6, "step": 2464, "eval_split": "validation", "eval_acc": 0.3768, "eval_loss": 1.784936767578125, "val_acc": 0.3768, "val_loss": 1.784936767578125}, {"step": 2600, "epoch": 7, "train_loss": 1.5933821201324463}, {"step": 2800, "epoch": 7, "train_loss": 1.6559200286865234}, {"epoch_end": 7, "step": 2816, "eval_split": "validation", "eval_acc": 0.3792, "eval_loss": 1.7693947998046875, "val_acc": 0.3792, "val_loss": 1.7693947998046875}, {"step": 3000, "epoch": 8, "train_loss": 1.5288678407669067}, {"epoch_end": 8, "step": 3168, "eval_split": "validation", "eval_acc": 0.3858, "eval_loss": 1.7760865234375, "val_acc": 0.3858, "val_loss": 1.7760865234375}, {"step": 3200, "epoch": 9, "train_loss": 1.5052134990692139}, {"step": 3400, "epoch": 9, "train_loss": 1.4993115663528442}, {"epoch_end": 9, "step": 3520, "eval_split": "validation", "eval_acc": 0.3836, "eval_loss": 1.7761372314453125, "val_acc": 0.3836, "val_loss": 1.7761372314453125}, {"step": 3600, "epoch": 10, "train_loss": 1.3986375331878662}, {"step": 3800, "epoch": 10, "train_loss": 1.369816780090332}, {"epoch_end": 10, "step": 3872, "eval_split": "validation", "eval_acc": 0.3814, "eval_loss": 1.7722346923828125, "val_acc": 0.3814, "val_loss": 1.7722346923828125}, {"step": 4000, "epoch": 11, "train_loss": 1.483936071395874}, {"step": 4200, "epoch": 11, "train_loss": 1.575093150138855}, {"epoch_end": 11, "step": 4224, "eval_split": "validation", "eval_acc": 0.3954, "eval_loss": 1.7554138671875, "val_acc": 0.3954, "val_loss": 1.7554138671875}, {"step": 4400, "epoch": 12, "train_loss": 1.622323989868164}, {"epoch_end": 12, "step": 4576, "eval_split": "validation", "eval_acc": 0.378, "eval_loss": 1.7892219970703125, "val_acc": 0.378, "val_loss": 1.7892219970703125}, {"step": 4600, "epoch": 13, "train_loss": 1.2166309356689453}, {"step": 4800, "epoch": 13, "train_loss": 1.4594851732254028}, {"epoch_end": 13, "step": 4928, "eval_split": "validation", "eval_acc": 0.3846, "eval_loss": 1.7835304443359374, "val_acc": 0.3846, "val_loss": 1.7835304443359374}, {"step": 5000, "epoch": 14, "train_loss": 1.175580382347107}, {"step": 5200, "epoch": 14, "train_loss": 1.457786202430725}, {"epoch_end": 14, "step": 5280, "eval_split": "validation", "eval_acc": 0.3844, "eval_loss": 1.78108515625, "val_acc": 0.3844, "val_loss": 1.78108515625}, {"step": 5400, "epoch": 15, "train_loss": 1.6417378187179565}, {"step": 5600, "epoch": 15, "train_loss": 1.3701038360595703}, {"epoch_end": 15, "step": 5632, "eval_split": "validation", "eval_acc": 0.3908, "eval_loss": 1.76001337890625, "val_acc": 0.3908, "val_loss": 1.76001337890625}, {"step": 5800, "epoch": 16, "train_loss": 1.2742304801940918}, {"epoch_end": 16, "step": 5984, "eval_split": "validation", "eval_acc": 0.3888, "eval_loss": 1.7802591064453126, "val_acc": 0.3888, "val_loss": 1.7802591064453126}, {"step": 6000, "epoch": 17, "train_loss": 1.2584271430969238}, {"step": 6200, "epoch": 17, "train_loss": 1.3066821098327637}, {"epoch_end": 17, "step": 6336, "eval_split": "validation", "eval_acc": 0.3816, "eval_loss": 1.814546533203125, "val_acc": 0.3816, "val_loss": 1.814546533203125}, {"step": 6400, "epoch": 18, "train_loss": 1.1564587354660034}, {"step": 6600, "epoch": 18, "train_loss": 1.29982590675354}, {"epoch_end": 18, "step": 6688, "eval_split": "validation", "eval_acc": 0.3882, "eval_loss": 1.8060728515625, "val_acc": 0.3882, "val_loss": 1.8060728515625}, {"step": 6800, "epoch": 19, "train_loss": 1.2510889768600464}, {"step": 7000, "epoch": 19, "train_loss": 1.2118326425552368}, {"epoch_end": 19, "step": 7040, "eval_split": "validation", "eval_acc": 0.3902, "eval_loss": 1.8025027587890625, "val_acc": 0.3902, "val_loss": 1.8025027587890625}, {"step": 7200, "epoch": 20, "train_loss": 1.2556395530700684}, {"epoch_end": 20, "step": 7392, "eval_split": "validation", "eval_acc": 0.3806, "eval_loss": 1.83795029296875, "val_acc": 0.3806, "val_loss": 1.83795029296875}, {"step": 7400, "epoch": 21, "train_loss": 1.0644207000732422}, {"step": 7600, "epoch": 21, "train_loss": 1.1828950643539429}, {"epoch_end": 21, "step": 7744, "eval_split": "validation", "eval_acc": 0.3828, "eval_loss": 1.8515244873046874, "val_acc": 0.3828, "val_loss": 1.8515244873046874}, {"step": 7800, "epoch": 22, "train_loss": 1.3489990234375}, {"step": 8000, "epoch": 22, "train_loss": 1.2202485799789429}, {"epoch_end": 22, "step": 8096, "eval_split": "validation", "eval_acc": 0.3836, "eval_loss": 1.8432701171875, "val_acc": 0.3836, "val_loss": 1.8432701171875}, {"step": 8200, "epoch": 23, "train_loss": 1.0749735832214355}, {"step": 8400, "epoch": 23, "train_loss": 1.3119491338729858}, {"epoch_end": 23, "step": 8448, "eval_split": "validation", "eval_acc": 0.3826, "eval_loss": 1.8644279052734376, "val_acc": 0.3826, "val_loss": 1.8644279052734376}, {"step": 8600, "epoch": 24, "train_loss": 1.2155643701553345}, {"epoch_end": 24, "step": 8800, "eval_split": "validation", "eval_acc": 0.3792, "eval_loss": 1.8842864990234376, "val_acc": 0.3792, "val_loss": 1.8842864990234376}], "final": {"eval_split": "validation", "eval_acc": 0.3792, "eval_loss": 1.8842864990234376, "wall_s": 42.9035120010376, "val_acc": 0.3792, "val_loss": 1.8842864990234376}}