1
|
{"args": {"mode": "bp", "dataset": "hierarchical", "depth": 6, "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": 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_w16_d6_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": 2.6297314167022705}, {"step": 200, "epoch": 0, "train_loss": 1.7909108400344849}, {"epoch_end": 0, "step": 352, "eval_split": "validation", "eval_acc": 0.3262, "eval_loss": 1.9040460205078125, "val_acc": 0.3262, "val_loss": 1.9040460205078125}, {"step": 400, "epoch": 1, "train_loss": 1.7910490036010742}, {"step": 600, "epoch": 1, "train_loss": 1.7194494009017944}, {"epoch_end": 1, "step": 704, "eval_split": "validation", "eval_acc": 0.3518, "eval_loss": 1.86377822265625, "val_acc": 0.3518, "val_loss": 1.86377822265625}, {"step": 800, "epoch": 2, "train_loss": 1.70653235912323}, {"step": 1000, "epoch": 2, "train_loss": 1.7814356088638306}, {"epoch_end": 2, "step": 1056, "eval_split": "validation", "eval_acc": 0.358, "eval_loss": 1.834432470703125, "val_acc": 0.358, "val_loss": 1.834432470703125}, {"step": 1200, "epoch": 3, "train_loss": 1.5530292987823486}, {"step": 1400, "epoch": 3, "train_loss": 1.6456412076950073}, {"epoch_end": 3, "step": 1408, "eval_split": "validation", "eval_acc": 0.3614, "eval_loss": 1.835983447265625, "val_acc": 0.3614, "val_loss": 1.835983447265625}, {"step": 1600, "epoch": 4, "train_loss": 1.676372766494751}, {"epoch_end": 4, "step": 1760, "eval_split": "validation", "eval_acc": 0.3686, "eval_loss": 1.8233727294921875, "val_acc": 0.3686, "val_loss": 1.8233727294921875}, {"step": 1800, "epoch": 5, "train_loss": 1.8964179754257202}, {"step": 2000, "epoch": 5, "train_loss": 1.6459848880767822}, {"epoch_end": 5, "step": 2112, "eval_split": "validation", "eval_acc": 0.3628, "eval_loss": 1.80917802734375, "val_acc": 0.3628, "val_loss": 1.80917802734375}, {"step": 2200, "epoch": 6, "train_loss": 1.6466037034988403}, {"step": 2400, "epoch": 6, "train_loss": 1.653733253479004}, {"epoch_end": 6, "step": 2464, "eval_split": "validation", "eval_acc": 0.3654, "eval_loss": 1.818825537109375, "val_acc": 0.3654, "val_loss": 1.818825537109375}, {"step": 2600, "epoch": 7, "train_loss": 1.6940566301345825}, {"step": 2800, "epoch": 7, "train_loss": 1.778384804725647}, {"epoch_end": 7, "step": 2816, "eval_split": "validation", "eval_acc": 0.3732, "eval_loss": 1.796735595703125, "val_acc": 0.3732, "val_loss": 1.796735595703125}, {"step": 3000, "epoch": 8, "train_loss": 1.6371135711669922}, {"epoch_end": 8, "step": 3168, "eval_split": "validation", "eval_acc": 0.3714, "eval_loss": 1.80244658203125, "val_acc": 0.3714, "val_loss": 1.80244658203125}, {"step": 3200, "epoch": 9, "train_loss": 1.6923823356628418}, {"step": 3400, "epoch": 9, "train_loss": 1.5949510335922241}, {"epoch_end": 9, "step": 3520, "eval_split": "validation", "eval_acc": 0.376, "eval_loss": 1.791606640625, "val_acc": 0.376, "val_loss": 1.791606640625}, {"step": 3600, "epoch": 10, "train_loss": 1.5921745300292969}, {"step": 3800, "epoch": 10, "train_loss": 1.5253223180770874}, {"epoch_end": 10, "step": 3872, "eval_split": "validation", "eval_acc": 0.3764, "eval_loss": 1.7941333740234375, "val_acc": 0.3764, "val_loss": 1.7941333740234375}, {"step": 4000, "epoch": 11, "train_loss": 1.6548943519592285}, {"step": 4200, "epoch": 11, "train_loss": 1.6630606651306152}, {"epoch_end": 11, "step": 4224, "eval_split": "validation", "eval_acc": 0.3788, "eval_loss": 1.7755791748046874, "val_acc": 0.3788, "val_loss": 1.7755791748046874}, {"step": 4400, "epoch": 12, "train_loss": 1.7819236516952515}, {"epoch_end": 12, "step": 4576, "eval_split": "validation", "eval_acc": 0.3718, "eval_loss": 1.8013405029296874, "val_acc": 0.3718, "val_loss": 1.8013405029296874}, {"step": 4600, "epoch": 13, "train_loss": 1.4850767850875854}, {"step": 4800, "epoch": 13, "train_loss": 1.6239793300628662}, {"epoch_end": 13, "step": 4928, "eval_split": "validation", "eval_acc": 0.377, "eval_loss": 1.7875544677734374, "val_acc": 0.377, "val_loss": 1.7875544677734374}, {"step": 5000, "epoch": 14, "train_loss": 1.3573062419891357}, {"step": 5200, "epoch": 14, "train_loss": 1.6061058044433594}, {"epoch_end": 14, "step": 5280, "eval_split": "validation", "eval_acc": 0.3732, "eval_loss": 1.795987060546875, "val_acc": 0.3732, "val_loss": 1.795987060546875}, {"step": 5400, "epoch": 15, "train_loss": 1.7622802257537842}, {"step": 5600, "epoch": 15, "train_loss": 1.4920891523361206}, {"epoch_end": 15, "step": 5632, "eval_split": "validation", "eval_acc": 0.3732, "eval_loss": 1.7809289306640625, "val_acc": 0.3732, "val_loss": 1.7809289306640625}, {"step": 5800, "epoch": 16, "train_loss": 1.5436149835586548}, {"epoch_end": 16, "step": 5984, "eval_split": "validation", "eval_acc": 0.3808, "eval_loss": 1.790147998046875, "val_acc": 0.3808, "val_loss": 1.790147998046875}, {"step": 6000, "epoch": 17, "train_loss": 1.525758981704712}, {"step": 6200, "epoch": 17, "train_loss": 1.4599571228027344}, {"epoch_end": 17, "step": 6336, "eval_split": "validation", "eval_acc": 0.378, "eval_loss": 1.7829792236328126, "val_acc": 0.378, "val_loss": 1.7829792236328126}, {"step": 6400, "epoch": 18, "train_loss": 1.442229986190796}, {"step": 6600, "epoch": 18, "train_loss": 1.5601450204849243}, {"epoch_end": 18, "step": 6688, "eval_split": "validation", "eval_acc": 0.3798, "eval_loss": 1.79341337890625, "val_acc": 0.3798, "val_loss": 1.79341337890625}, {"step": 6800, "epoch": 19, "train_loss": 1.5725939273834229}, {"step": 7000, "epoch": 19, "train_loss": 1.4079320430755615}, {"epoch_end": 19, "step": 7040, "eval_split": "validation", "eval_acc": 0.3802, "eval_loss": 1.76940849609375, "val_acc": 0.3802, "val_loss": 1.76940849609375}, {"step": 7200, "epoch": 20, "train_loss": 1.653389573097229}, {"epoch_end": 20, "step": 7392, "eval_split": "validation", "eval_acc": 0.3762, "eval_loss": 1.78190068359375, "val_acc": 0.3762, "val_loss": 1.78190068359375}, {"step": 7400, "epoch": 21, "train_loss": 1.5843520164489746}, {"step": 7600, "epoch": 21, "train_loss": 1.599625825881958}, {"epoch_end": 21, "step": 7744, "eval_split": "validation", "eval_acc": 0.3796, "eval_loss": 1.7869532958984375, "val_acc": 0.3796, "val_loss": 1.7869532958984375}, {"step": 7800, "epoch": 22, "train_loss": 1.5850491523742676}, {"step": 8000, "epoch": 22, "train_loss": 1.6007497310638428}, {"epoch_end": 22, "step": 8096, "eval_split": "validation", "eval_acc": 0.3816, "eval_loss": 1.779801416015625, "val_acc": 0.3816, "val_loss": 1.779801416015625}, {"step": 8200, "epoch": 23, "train_loss": 1.5141990184783936}, {"step": 8400, "epoch": 23, "train_loss": 1.683617115020752}, {"epoch_end": 23, "step": 8448, "eval_split": "validation", "eval_acc": 0.3802, "eval_loss": 1.773928466796875, "val_acc": 0.3802, "val_loss": 1.773928466796875}, {"step": 8600, "epoch": 24, "train_loss": 1.4998162984848022}, {"epoch_end": 24, "step": 8800, "eval_split": "validation", "eval_acc": 0.38, "eval_loss": 1.7654266357421875, "val_acc": 0.38, "val_loss": 1.7654266357421875}], "final": {"eval_split": "validation", "eval_acc": 0.38, "eval_loss": 1.7654266357421875, "wall_s": 33.10647225379944, "val_acc": 0.38, "val_loss": 1.7654266357421875}}
|
