1
|
{"args": {"mode": "bp", "dataset": "teacher", "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": 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_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": 2.660649061203003}, {"step": 200, "epoch": 0, "train_loss": 0.996557891368866}, {"epoch_end": 0, "step": 352, "eval_split": "validation", "eval_acc": 0.5742, "eval_loss": 1.1206134765625, "val_acc": 0.5742, "val_loss": 1.1206134765625}, {"step": 400, "epoch": 1, "train_loss": 0.8633809685707092}, {"step": 600, "epoch": 1, "train_loss": 0.853786826133728}, {"epoch_end": 1, "step": 704, "eval_split": "validation", "eval_acc": 0.5862, "eval_loss": 1.09737041015625, "val_acc": 0.5862, "val_loss": 1.09737041015625}, {"step": 800, "epoch": 2, "train_loss": 0.8752877712249756}, {"step": 1000, "epoch": 2, "train_loss": 0.8614870309829712}, {"epoch_end": 2, "step": 1056, "eval_split": "validation", "eval_acc": 0.5826, "eval_loss": 1.1256653564453125, "val_acc": 0.5826, "val_loss": 1.1256653564453125}, {"step": 1200, "epoch": 3, "train_loss": 0.9537748694419861}, {"step": 1400, "epoch": 3, "train_loss": 1.0066492557525635}, {"epoch_end": 3, "step": 1408, "eval_split": "validation", "eval_acc": 0.5828, "eval_loss": 1.1298255126953125, "val_acc": 0.5828, "val_loss": 1.1298255126953125}, {"step": 1600, "epoch": 4, "train_loss": 0.8712270259857178}, {"epoch_end": 4, "step": 1760, "eval_split": "validation", "eval_acc": 0.5826, "eval_loss": 1.1128714599609375, "val_acc": 0.5826, "val_loss": 1.1128714599609375}, {"step": 1800, "epoch": 5, "train_loss": 1.0536023378372192}, {"step": 2000, "epoch": 5, "train_loss": 0.8444521427154541}, {"epoch_end": 5, "step": 2112, "eval_split": "validation", "eval_acc": 0.5782, "eval_loss": 1.1357642822265626, "val_acc": 0.5782, "val_loss": 1.1357642822265626}, {"step": 2200, "epoch": 6, "train_loss": 1.0200448036193848}, {"step": 2400, "epoch": 6, "train_loss": 0.8217188715934753}, {"epoch_end": 6, "step": 2464, "eval_split": "validation", "eval_acc": 0.5936, "eval_loss": 1.09222138671875, "val_acc": 0.5936, "val_loss": 1.09222138671875}, {"step": 2600, "epoch": 7, "train_loss": 0.8908798694610596}, {"step": 2800, "epoch": 7, "train_loss": 1.0150463581085205}, {"epoch_end": 7, "step": 2816, "eval_split": "validation", "eval_acc": 0.579, "eval_loss": 1.136682080078125, "val_acc": 0.579, "val_loss": 1.136682080078125}, {"step": 3000, "epoch": 8, "train_loss": 0.9362335801124573}, {"epoch_end": 8, "step": 3168, "eval_split": "validation", "eval_acc": 0.5864, "eval_loss": 1.1225859130859375, "val_acc": 0.5864, "val_loss": 1.1225859130859375}, {"step": 3200, "epoch": 9, "train_loss": 0.8840282559394836}, {"step": 3400, "epoch": 9, "train_loss": 0.8676167726516724}, {"epoch_end": 9, "step": 3520, "eval_split": "validation", "eval_acc": 0.5856, "eval_loss": 1.109516064453125, "val_acc": 0.5856, "val_loss": 1.109516064453125}, {"step": 3600, "epoch": 10, "train_loss": 0.9329383969306946}, {"step": 3800, "epoch": 10, "train_loss": 0.9458080530166626}, {"epoch_end": 10, "step": 3872, "eval_split": "validation", "eval_acc": 0.5974, "eval_loss": 1.0999067749023437, "val_acc": 0.5974, "val_loss": 1.0999067749023437}, {"step": 4000, "epoch": 11, "train_loss": 0.9005988240242004}, {"step": 4200, "epoch": 11, "train_loss": 1.045232892036438}, {"epoch_end": 11, "step": 4224, "eval_split": "validation", "eval_acc": 0.5816, "eval_loss": 1.134672412109375, "val_acc": 0.5816, "val_loss": 1.134672412109375}, {"step": 4400, "epoch": 12, "train_loss": 0.9206005930900574}, {"epoch_end": 12, "step": 4576, "eval_split": "validation", "eval_acc": 0.5906, "eval_loss": 1.1100499755859374, "val_acc": 0.5906, "val_loss": 1.1100499755859374}, {"step": 4600, "epoch": 13, "train_loss": 1.116690754890442}, {"step": 4800, "epoch": 13, "train_loss": 0.8835233449935913}, {"epoch_end": 13, "step": 4928, "eval_split": "validation", "eval_acc": 0.586, "eval_loss": 1.13546455078125, "val_acc": 0.586, "val_loss": 1.13546455078125}, {"step": 5000, "epoch": 14, "train_loss": 1.0082675218582153}, {"step": 5200, "epoch": 14, "train_loss": 0.8944704532623291}, {"epoch_end": 14, "step": 5280, "eval_split": "validation", "eval_acc": 0.5874, "eval_loss": 1.1310967041015625, "val_acc": 0.5874, "val_loss": 1.1310967041015625}, {"step": 5400, "epoch": 15, "train_loss": 0.9900583028793335}, {"step": 5600, "epoch": 15, "train_loss": 1.0172821283340454}, {"epoch_end": 15, "step": 5632, "eval_split": "validation", "eval_acc": 0.5854, "eval_loss": 1.13110517578125, "val_acc": 0.5854, "val_loss": 1.13110517578125}, {"step": 5800, "epoch": 16, "train_loss": 0.9112899303436279}, {"epoch_end": 16, "step": 5984, "eval_split": "validation", "eval_acc": 0.6012, "eval_loss": 1.101458935546875, "val_acc": 0.6012, "val_loss": 1.101458935546875}, {"step": 6000, "epoch": 17, "train_loss": 0.8773520588874817}, {"step": 6200, "epoch": 17, "train_loss": 0.8771717548370361}, {"epoch_end": 17, "step": 6336, "eval_split": "validation", "eval_acc": 0.5892, "eval_loss": 1.1259810546875, "val_acc": 0.5892, "val_loss": 1.1259810546875}, {"step": 6400, "epoch": 18, "train_loss": 1.0280790328979492}, {"step": 6600, "epoch": 18, "train_loss": 0.8629776835441589}, {"epoch_end": 18, "step": 6688, "eval_split": "validation", "eval_acc": 0.5904, "eval_loss": 1.1227839599609375, "val_acc": 0.5904, "val_loss": 1.1227839599609375}, {"step": 6800, "epoch": 19, "train_loss": 1.1421175003051758}, {"step": 7000, "epoch": 19, "train_loss": 0.8250914812088013}, {"epoch_end": 19, "step": 7040, "eval_split": "validation", "eval_acc": 0.5898, "eval_loss": 1.116978271484375, "val_acc": 0.5898, "val_loss": 1.116978271484375}, {"step": 7200, "epoch": 20, "train_loss": 0.9006116986274719}, {"epoch_end": 20, "step": 7392, "eval_split": "validation", "eval_acc": 0.5898, "eval_loss": 1.1075101440429687, "val_acc": 0.5898, "val_loss": 1.1075101440429687}, {"step": 7400, "epoch": 21, "train_loss": 0.9425082802772522}, {"step": 7600, "epoch": 21, "train_loss": 1.0669807195663452}, {"epoch_end": 21, "step": 7744, "eval_split": "validation", "eval_acc": 0.5824, "eval_loss": 1.128408251953125, "val_acc": 0.5824, "val_loss": 1.128408251953125}, {"step": 7800, "epoch": 22, "train_loss": 0.9231097102165222}, {"step": 8000, "epoch": 22, "train_loss": 0.8883748054504395}, {"epoch_end": 22, "step": 8096, "eval_split": "validation", "eval_acc": 0.588, "eval_loss": 1.127149560546875, "val_acc": 0.588, "val_loss": 1.127149560546875}, {"step": 8200, "epoch": 23, "train_loss": 0.9372089505195618}, {"step": 8400, "epoch": 23, "train_loss": 1.0033732652664185}, {"epoch_end": 23, "step": 8448, "eval_split": "validation", "eval_acc": 0.5848, "eval_loss": 1.130660205078125, "val_acc": 0.5848, "val_loss": 1.130660205078125}, {"step": 8600, "epoch": 24, "train_loss": 0.7973468899726868}, {"epoch_end": 24, "step": 8800, "eval_split": "validation", "eval_acc": 0.5832, "eval_loss": 1.1286564453125, "val_acc": 0.5832, "val_loss": 1.1286564453125}], "final": {"eval_split": "validation", "eval_acc": 0.5832, "eval_loss": 1.1286564453125, "wall_s": 35.32112979888916, "val_acc": 0.5832, "val_loss": 1.1286564453125}}
|
