1
|
{"args": {"mode": "bp", "dataset": "hierarchical", "depth": 4, "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_d4_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": 2.494375228881836}, {"step": 200, "epoch": 0, "train_loss": 1.7668466567993164}, {"epoch_end": 0, "step": 352, "eval_split": "validation", "eval_acc": 0.3264, "eval_loss": 1.9003558837890624, "val_acc": 0.3264, "val_loss": 1.9003558837890624}, {"step": 400, "epoch": 1, "train_loss": 1.739091396331787}, {"step": 600, "epoch": 1, "train_loss": 1.7146934270858765}, {"epoch_end": 1, "step": 704, "eval_split": "validation", "eval_acc": 0.3452, "eval_loss": 1.8879978759765625, "val_acc": 0.3452, "val_loss": 1.8879978759765625}, {"step": 800, "epoch": 2, "train_loss": 1.6795015335083008}, {"step": 1000, "epoch": 2, "train_loss": 1.7471888065338135}, {"epoch_end": 2, "step": 1056, "eval_split": "validation", "eval_acc": 0.3564, "eval_loss": 1.84619541015625, "val_acc": 0.3564, "val_loss": 1.84619541015625}, {"step": 1200, "epoch": 3, "train_loss": 1.5223157405853271}, {"step": 1400, "epoch": 3, "train_loss": 1.6022671461105347}, {"epoch_end": 3, "step": 1408, "eval_split": "validation", "eval_acc": 0.3592, "eval_loss": 1.8341489013671874, "val_acc": 0.3592, "val_loss": 1.8341489013671874}, {"step": 1600, "epoch": 4, "train_loss": 1.653402328491211}, {"epoch_end": 4, "step": 1760, "eval_split": "validation", "eval_acc": 0.3732, "eval_loss": 1.80735146484375, "val_acc": 0.3732, "val_loss": 1.80735146484375}, {"step": 1800, "epoch": 5, "train_loss": 1.8385359048843384}, {"step": 2000, "epoch": 5, "train_loss": 1.563118577003479}, {"epoch_end": 5, "step": 2112, "eval_split": "validation", "eval_acc": 0.3716, "eval_loss": 1.781091552734375, "val_acc": 0.3716, "val_loss": 1.781091552734375}, {"step": 2200, "epoch": 6, "train_loss": 1.5558639764785767}, {"step": 2400, "epoch": 6, "train_loss": 1.5983037948608398}, {"epoch_end": 6, "step": 2464, "eval_split": "validation", "eval_acc": 0.3794, "eval_loss": 1.7652639404296875, "val_acc": 0.3794, "val_loss": 1.7652639404296875}, {"step": 2600, "epoch": 7, "train_loss": 1.5639969110488892}, {"step": 2800, "epoch": 7, "train_loss": 1.6544610261917114}, {"epoch_end": 7, "step": 2816, "eval_split": "validation", "eval_acc": 0.3876, "eval_loss": 1.742245166015625, "val_acc": 0.3876, "val_loss": 1.742245166015625}, {"step": 3000, "epoch": 8, "train_loss": 1.5272120237350464}, {"epoch_end": 8, "step": 3168, "eval_split": "validation", "eval_acc": 0.3952, "eval_loss": 1.7395301513671875, "val_acc": 0.3952, "val_loss": 1.7395301513671875}, {"step": 3200, "epoch": 9, "train_loss": 1.5225242376327515}, {"step": 3400, "epoch": 9, "train_loss": 1.4774686098098755}, {"epoch_end": 9, "step": 3520, "eval_split": "validation", "eval_acc": 0.388, "eval_loss": 1.74919970703125, "val_acc": 0.388, "val_loss": 1.74919970703125}, {"step": 3600, "epoch": 10, "train_loss": 1.4476593732833862}, {"step": 3800, "epoch": 10, "train_loss": 1.3502999544143677}, {"epoch_end": 10, "step": 3872, "eval_split": "validation", "eval_acc": 0.3986, "eval_loss": 1.7331639892578126, "val_acc": 0.3986, "val_loss": 1.7331639892578126}, {"step": 4000, "epoch": 11, "train_loss": 1.52412748336792}, {"step": 4200, "epoch": 11, "train_loss": 1.6068642139434814}, {"epoch_end": 11, "step": 4224, "eval_split": "validation", "eval_acc": 0.3982, "eval_loss": 1.7193469970703126, "val_acc": 0.3982, "val_loss": 1.7193469970703126}, {"step": 4400, "epoch": 12, "train_loss": 1.6691029071807861}, {"epoch_end": 12, "step": 4576, "eval_split": "validation", "eval_acc": 0.401, "eval_loss": 1.7348052734375, "val_acc": 0.401, "val_loss": 1.7348052734375}, {"step": 4600, "epoch": 13, "train_loss": 1.2623645067214966}, {"step": 4800, "epoch": 13, "train_loss": 1.5370402336120605}, {"epoch_end": 13, "step": 4928, "eval_split": "validation", "eval_acc": 0.3964, "eval_loss": 1.7474617919921875, "val_acc": 0.3964, "val_loss": 1.7474617919921875}, {"step": 5000, "epoch": 14, "train_loss": 1.1603879928588867}, {"step": 5200, "epoch": 14, "train_loss": 1.4278842210769653}, {"epoch_end": 14, "step": 5280, "eval_split": "validation", "eval_acc": 0.4012, "eval_loss": 1.7398791015625, "val_acc": 0.4012, "val_loss": 1.7398791015625}, {"step": 5400, "epoch": 15, "train_loss": 1.6403396129608154}, {"step": 5600, "epoch": 15, "train_loss": 1.3499090671539307}, {"epoch_end": 15, "step": 5632, "eval_split": "validation", "eval_acc": 0.405, "eval_loss": 1.71754091796875, "val_acc": 0.405, "val_loss": 1.71754091796875}, {"step": 5800, "epoch": 16, "train_loss": 1.3887730836868286}, {"epoch_end": 16, "step": 5984, "eval_split": "validation", "eval_acc": 0.4018, "eval_loss": 1.7518520263671875, "val_acc": 0.4018, "val_loss": 1.7518520263671875}, {"step": 6000, "epoch": 17, "train_loss": 1.3639968633651733}, {"step": 6200, "epoch": 17, "train_loss": 1.3673912286758423}, {"epoch_end": 17, "step": 6336, "eval_split": "validation", "eval_acc": 0.4054, "eval_loss": 1.771031689453125, "val_acc": 0.4054, "val_loss": 1.771031689453125}, {"step": 6400, "epoch": 18, "train_loss": 1.2472014427185059}, {"step": 6600, "epoch": 18, "train_loss": 1.231408953666687}, {"epoch_end": 18, "step": 6688, "eval_split": "validation", "eval_acc": 0.4008, "eval_loss": 1.760891259765625, "val_acc": 0.4008, "val_loss": 1.760891259765625}, {"step": 6800, "epoch": 19, "train_loss": 1.3055731058120728}, {"step": 7000, "epoch": 19, "train_loss": 1.2214292287826538}, {"epoch_end": 19, "step": 7040, "eval_split": "validation", "eval_acc": 0.4112, "eval_loss": 1.7406487548828125, "val_acc": 0.4112, "val_loss": 1.7406487548828125}, {"step": 7200, "epoch": 20, "train_loss": 1.275927186012268}, {"epoch_end": 20, "step": 7392, "eval_split": "validation", "eval_acc": 0.4022, "eval_loss": 1.7747889404296875, "val_acc": 0.4022, "val_loss": 1.7747889404296875}, {"step": 7400, "epoch": 21, "train_loss": 1.202966570854187}, {"step": 7600, "epoch": 21, "train_loss": 1.2848490476608276}, {"epoch_end": 21, "step": 7744, "eval_split": "validation", "eval_acc": 0.3968, "eval_loss": 1.77179033203125, "val_acc": 0.3968, "val_loss": 1.77179033203125}, {"step": 7800, "epoch": 22, "train_loss": 1.3983107805252075}, {"step": 8000, "epoch": 22, "train_loss": 1.370850682258606}, {"epoch_end": 22, "step": 8096, "eval_split": "validation", "eval_acc": 0.4064, "eval_loss": 1.7622962890625, "val_acc": 0.4064, "val_loss": 1.7622962890625}, {"step": 8200, "epoch": 23, "train_loss": 1.1442334651947021}, {"step": 8400, "epoch": 23, "train_loss": 1.3749542236328125}, {"epoch_end": 23, "step": 8448, "eval_split": "validation", "eval_acc": 0.4012, "eval_loss": 1.7840415283203126, "val_acc": 0.4012, "val_loss": 1.7840415283203126}, {"step": 8600, "epoch": 24, "train_loss": 1.2262749671936035}, {"epoch_end": 24, "step": 8800, "eval_split": "validation", "eval_acc": 0.4006, "eval_loss": 1.775607666015625, "val_acc": 0.4006, "val_loss": 1.775607666015625}], "final": {"eval_split": "validation", "eval_acc": 0.4006, "eval_loss": 1.775607666015625, "wall_s": 25.604931116104126, "val_acc": 0.4006, "val_loss": 1.775607666015625}}
|
