1
|
{"args": {"mode": "bp", "dataset": "teacher", "depth": 6, "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": 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_w64_d6_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": 2.6210927963256836}, {"step": 200, "epoch": 0, "train_loss": 0.8622533679008484}, {"epoch_end": 0, "step": 352, "eval_split": "validation", "eval_acc": 0.612, "eval_loss": 1.0252984497070312, "val_acc": 0.612, "val_loss": 1.0252984497070312}, {"step": 400, "epoch": 1, "train_loss": 0.77760910987854}, {"step": 600, "epoch": 1, "train_loss": 0.7434456944465637}, {"epoch_end": 1, "step": 704, "eval_split": "validation", "eval_acc": 0.6186, "eval_loss": 1.0156798828125, "val_acc": 0.6186, "val_loss": 1.0156798828125}, {"step": 800, "epoch": 2, "train_loss": 0.7321397662162781}, {"step": 1000, "epoch": 2, "train_loss": 0.7317296862602234}, {"epoch_end": 2, "step": 1056, "eval_split": "validation", "eval_acc": 0.6136, "eval_loss": 1.0229468505859376, "val_acc": 0.6136, "val_loss": 1.0229468505859376}, {"step": 1200, "epoch": 3, "train_loss": 0.8721165657043457}, {"step": 1400, "epoch": 3, "train_loss": 0.9030106067657471}, {"epoch_end": 3, "step": 1408, "eval_split": "validation", "eval_acc": 0.6102, "eval_loss": 1.0437629760742186, "val_acc": 0.6102, "val_loss": 1.0437629760742186}, {"step": 1600, "epoch": 4, "train_loss": 0.727023720741272}, {"epoch_end": 4, "step": 1760, "eval_split": "validation", "eval_acc": 0.6184, "eval_loss": 1.0310751220703125, "val_acc": 0.6184, "val_loss": 1.0310751220703125}, {"step": 1800, "epoch": 5, "train_loss": 0.797160267829895}, {"step": 2000, "epoch": 5, "train_loss": 0.6620596051216125}, {"epoch_end": 5, "step": 2112, "eval_split": "validation", "eval_acc": 0.6126, "eval_loss": 1.0606338989257813, "val_acc": 0.6126, "val_loss": 1.0606338989257813}, {"step": 2200, "epoch": 6, "train_loss": 0.8294033408164978}, {"step": 2400, "epoch": 6, "train_loss": 0.6756864786148071}, {"epoch_end": 6, "step": 2464, "eval_split": "validation", "eval_acc": 0.62, "eval_loss": 1.0471800903320312, "val_acc": 0.62, "val_loss": 1.0471800903320312}, {"step": 2600, "epoch": 7, "train_loss": 0.7365713119506836}, {"step": 2800, "epoch": 7, "train_loss": 0.8060657978057861}, {"epoch_end": 7, "step": 2816, "eval_split": "validation", "eval_acc": 0.604, "eval_loss": 1.102895166015625, "val_acc": 0.604, "val_loss": 1.102895166015625}, {"step": 3000, "epoch": 8, "train_loss": 0.6958932280540466}, {"epoch_end": 8, "step": 3168, "eval_split": "validation", "eval_acc": 0.6106, "eval_loss": 1.0889811767578126, "val_acc": 0.6106, "val_loss": 1.0889811767578126}, {"step": 3200, "epoch": 9, "train_loss": 0.6129227876663208}, {"step": 3400, "epoch": 9, "train_loss": 0.5703102946281433}, {"epoch_end": 9, "step": 3520, "eval_split": "validation", "eval_acc": 0.611, "eval_loss": 1.1095732177734374, "val_acc": 0.611, "val_loss": 1.1095732177734374}, {"step": 3600, "epoch": 10, "train_loss": 0.6207870841026306}, {"step": 3800, "epoch": 10, "train_loss": 0.692367672920227}, {"epoch_end": 10, "step": 3872, "eval_split": "validation", "eval_acc": 0.611, "eval_loss": 1.11912265625, "val_acc": 0.611, "val_loss": 1.11912265625}, {"step": 4000, "epoch": 11, "train_loss": 0.7206392288208008}, {"step": 4200, "epoch": 11, "train_loss": 0.6902603507041931}, {"epoch_end": 11, "step": 4224, "eval_split": "validation", "eval_acc": 0.6044, "eval_loss": 1.167743212890625, "val_acc": 0.6044, "val_loss": 1.167743212890625}, {"step": 4400, "epoch": 12, "train_loss": 0.5203877091407776}, {"epoch_end": 12, "step": 4576, "eval_split": "validation", "eval_acc": 0.6016, "eval_loss": 1.1713595458984376, "val_acc": 0.6016, "val_loss": 1.1713595458984376}, {"step": 4600, "epoch": 13, "train_loss": 0.650784969329834}, {"step": 4800, "epoch": 13, "train_loss": 0.560661792755127}, {"epoch_end": 13, "step": 4928, "eval_split": "validation", "eval_acc": 0.5876, "eval_loss": 1.2254881103515625, "val_acc": 0.5876, "val_loss": 1.2254881103515625}, {"step": 5000, "epoch": 14, "train_loss": 0.6800820231437683}, {"step": 5200, "epoch": 14, "train_loss": 0.5899974703788757}, {"epoch_end": 14, "step": 5280, "eval_split": "validation", "eval_acc": 0.5984, "eval_loss": 1.2158177978515625, "val_acc": 0.5984, "val_loss": 1.2158177978515625}, {"step": 5400, "epoch": 15, "train_loss": 0.6280472278594971}, {"step": 5600, "epoch": 15, "train_loss": 0.6516405344009399}, {"epoch_end": 15, "step": 5632, "eval_split": "validation", "eval_acc": 0.5912, "eval_loss": 1.249712353515625, "val_acc": 0.5912, "val_loss": 1.249712353515625}, {"step": 5800, "epoch": 16, "train_loss": 0.6372895836830139}, {"epoch_end": 16, "step": 5984, "eval_split": "validation", "eval_acc": 0.5978, "eval_loss": 1.2696236328125, "val_acc": 0.5978, "val_loss": 1.2696236328125}, {"step": 6000, "epoch": 17, "train_loss": 0.5179353952407837}, {"step": 6200, "epoch": 17, "train_loss": 0.5249927639961243}, {"epoch_end": 17, "step": 6336, "eval_split": "validation", "eval_acc": 0.5928, "eval_loss": 1.3020653076171875, "val_acc": 0.5928, "val_loss": 1.3020653076171875}, {"step": 6400, "epoch": 18, "train_loss": 0.6160680055618286}, {"step": 6600, "epoch": 18, "train_loss": 0.6579370498657227}, {"epoch_end": 18, "step": 6688, "eval_split": "validation", "eval_acc": 0.5894, "eval_loss": 1.3265169921875, "val_acc": 0.5894, "val_loss": 1.3265169921875}, {"step": 6800, "epoch": 19, "train_loss": 0.6454458236694336}, {"step": 7000, "epoch": 19, "train_loss": 0.39317917823791504}, {"epoch_end": 19, "step": 7040, "eval_split": "validation", "eval_acc": 0.5888, "eval_loss": 1.3616431640625, "val_acc": 0.5888, "val_loss": 1.3616431640625}, {"step": 7200, "epoch": 20, "train_loss": 0.5328649282455444}, {"epoch_end": 20, "step": 7392, "eval_split": "validation", "eval_acc": 0.5848, "eval_loss": 1.401586328125, "val_acc": 0.5848, "val_loss": 1.401586328125}, {"step": 7400, "epoch": 21, "train_loss": 0.5527584552764893}, {"step": 7600, "epoch": 21, "train_loss": 0.4882585108280182}, {"epoch_end": 21, "step": 7744, "eval_split": "validation", "eval_acc": 0.5864, "eval_loss": 1.4212303955078125, "val_acc": 0.5864, "val_loss": 1.4212303955078125}, {"step": 7800, "epoch": 22, "train_loss": 0.4863404631614685}, {"step": 8000, "epoch": 22, "train_loss": 0.4108026623725891}, {"epoch_end": 22, "step": 8096, "eval_split": "validation", "eval_acc": 0.5756, "eval_loss": 1.483331103515625, "val_acc": 0.5756, "val_loss": 1.483331103515625}, {"step": 8200, "epoch": 23, "train_loss": 0.392582505941391}, {"step": 8400, "epoch": 23, "train_loss": 0.616544783115387}, {"epoch_end": 23, "step": 8448, "eval_split": "validation", "eval_acc": 0.5834, "eval_loss": 1.49861015625, "val_acc": 0.5834, "val_loss": 1.49861015625}, {"step": 8600, "epoch": 24, "train_loss": 0.4344726800918579}, {"epoch_end": 24, "step": 8800, "eval_split": "validation", "eval_acc": 0.5714, "eval_loss": 1.5335162353515626, "val_acc": 0.5714, "val_loss": 1.5335162353515626}], "final": {"eval_split": "validation", "eval_acc": 0.5714, "eval_loss": 1.5335162353515626, "wall_s": 35.77094578742981, "val_acc": 0.5714, "val_loss": 1.5335162353515626}}
|
