1
|
{"args": {"mode": "bp", "dataset": "teacher", "depth": 2, "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_d2_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": 2.3527565002441406}, {"step": 200, "epoch": 0, "train_loss": 0.9289120435714722}, {"epoch_end": 0, "step": 352, "eval_split": "validation", "eval_acc": 0.5702, "eval_loss": 1.142773486328125, "val_acc": 0.5702, "val_loss": 1.142773486328125}, {"step": 400, "epoch": 1, "train_loss": 0.8508673906326294}, {"step": 600, "epoch": 1, "train_loss": 0.8245355486869812}, {"epoch_end": 1, "step": 704, "eval_split": "validation", "eval_acc": 0.586, "eval_loss": 1.106931298828125, "val_acc": 0.586, "val_loss": 1.106931298828125}, {"step": 800, "epoch": 2, "train_loss": 0.8913167119026184}, {"step": 1000, "epoch": 2, "train_loss": 0.8256174325942993}, {"epoch_end": 2, "step": 1056, "eval_split": "validation", "eval_acc": 0.5864, "eval_loss": 1.122022705078125, "val_acc": 0.5864, "val_loss": 1.122022705078125}, {"step": 1200, "epoch": 3, "train_loss": 0.9590059518814087}, {"step": 1400, "epoch": 3, "train_loss": 1.1054637432098389}, {"epoch_end": 3, "step": 1408, "eval_split": "validation", "eval_acc": 0.5782, "eval_loss": 1.13520244140625, "val_acc": 0.5782, "val_loss": 1.13520244140625}, {"step": 1600, "epoch": 4, "train_loss": 0.9183905720710754}, {"epoch_end": 4, "step": 1760, "eval_split": "validation", "eval_acc": 0.5904, "eval_loss": 1.1116853759765626, "val_acc": 0.5904, "val_loss": 1.1116853759765626}, {"step": 1800, "epoch": 5, "train_loss": 0.9968322515487671}, {"step": 2000, "epoch": 5, "train_loss": 0.8338309526443481}, {"epoch_end": 5, "step": 2112, "eval_split": "validation", "eval_acc": 0.5792, "eval_loss": 1.1295319091796876, "val_acc": 0.5792, "val_loss": 1.1295319091796876}, {"step": 2200, "epoch": 6, "train_loss": 1.0942028760910034}, {"step": 2400, "epoch": 6, "train_loss": 0.8760504126548767}, {"epoch_end": 6, "step": 2464, "eval_split": "validation", "eval_acc": 0.599, "eval_loss": 1.0858934814453125, "val_acc": 0.599, "val_loss": 1.0858934814453125}, {"step": 2600, "epoch": 7, "train_loss": 0.962939441204071}, {"step": 2800, "epoch": 7, "train_loss": 0.9768973588943481}, {"epoch_end": 7, "step": 2816, "eval_split": "validation", "eval_acc": 0.583, "eval_loss": 1.13339775390625, "val_acc": 0.583, "val_loss": 1.13339775390625}, {"step": 3000, "epoch": 8, "train_loss": 0.9216310977935791}, {"epoch_end": 8, "step": 3168, "eval_split": "validation", "eval_acc": 0.5938, "eval_loss": 1.1107552490234376, "val_acc": 0.5938, "val_loss": 1.1107552490234376}, {"step": 3200, "epoch": 9, "train_loss": 0.9256529808044434}, {"step": 3400, "epoch": 9, "train_loss": 0.834572970867157}, {"epoch_end": 9, "step": 3520, "eval_split": "validation", "eval_acc": 0.5906, "eval_loss": 1.11136337890625, "val_acc": 0.5906, "val_loss": 1.11136337890625}, {"step": 3600, "epoch": 10, "train_loss": 0.9862117767333984}, {"step": 3800, "epoch": 10, "train_loss": 0.9779149293899536}, {"epoch_end": 10, "step": 3872, "eval_split": "validation", "eval_acc": 0.594, "eval_loss": 1.0945914794921876, "val_acc": 0.594, "val_loss": 1.0945914794921876}, {"step": 4000, "epoch": 11, "train_loss": 0.984391987323761}, {"step": 4200, "epoch": 11, "train_loss": 1.0463182926177979}, {"epoch_end": 11, "step": 4224, "eval_split": "validation", "eval_acc": 0.5858, "eval_loss": 1.11790849609375, "val_acc": 0.5858, "val_loss": 1.11790849609375}, {"step": 4400, "epoch": 12, "train_loss": 0.8927826881408691}, {"epoch_end": 12, "step": 4576, "eval_split": "validation", "eval_acc": 0.5882, "eval_loss": 1.111215966796875, "val_acc": 0.5882, "val_loss": 1.111215966796875}, {"step": 4600, "epoch": 13, "train_loss": 1.0913441181182861}, {"step": 4800, "epoch": 13, "train_loss": 0.8871700167655945}, {"epoch_end": 13, "step": 4928, "eval_split": "validation", "eval_acc": 0.5878, "eval_loss": 1.119411083984375, "val_acc": 0.5878, "val_loss": 1.119411083984375}, {"step": 5000, "epoch": 14, "train_loss": 1.0868006944656372}, {"step": 5200, "epoch": 14, "train_loss": 1.0118368864059448}, {"epoch_end": 14, "step": 5280, "eval_split": "validation", "eval_acc": 0.5918, "eval_loss": 1.09024091796875, "val_acc": 0.5918, "val_loss": 1.09024091796875}, {"step": 5400, "epoch": 15, "train_loss": 1.0145263671875}, {"step": 5600, "epoch": 15, "train_loss": 0.9607892632484436}, {"epoch_end": 15, "step": 5632, "eval_split": "validation", "eval_acc": 0.589, "eval_loss": 1.103532421875, "val_acc": 0.589, "val_loss": 1.103532421875}, {"step": 5800, "epoch": 16, "train_loss": 0.986981213092804}, {"epoch_end": 16, "step": 5984, "eval_split": "validation", "eval_acc": 0.6032, "eval_loss": 1.0763969482421876, "val_acc": 0.6032, "val_loss": 1.0763969482421876}, {"step": 6000, "epoch": 17, "train_loss": 1.0223749876022339}, {"step": 6200, "epoch": 17, "train_loss": 0.8838664293289185}, {"epoch_end": 17, "step": 6336, "eval_split": "validation", "eval_acc": 0.59, "eval_loss": 1.1092682861328125, "val_acc": 0.59, "val_loss": 1.1092682861328125}, {"step": 6400, "epoch": 18, "train_loss": 1.1250075101852417}, {"step": 6600, "epoch": 18, "train_loss": 0.897415041923523}, {"epoch_end": 18, "step": 6688, "eval_split": "validation", "eval_acc": 0.6006, "eval_loss": 1.099941552734375, "val_acc": 0.6006, "val_loss": 1.099941552734375}, {"step": 6800, "epoch": 19, "train_loss": 1.1121262311935425}, {"step": 7000, "epoch": 19, "train_loss": 0.7941070795059204}, {"epoch_end": 19, "step": 7040, "eval_split": "validation", "eval_acc": 0.6014, "eval_loss": 1.09330634765625, "val_acc": 0.6014, "val_loss": 1.09330634765625}, {"step": 7200, "epoch": 20, "train_loss": 0.835390567779541}, {"epoch_end": 20, "step": 7392, "eval_split": "validation", "eval_acc": 0.5932, "eval_loss": 1.1007385009765625, "val_acc": 0.5932, "val_loss": 1.1007385009765625}, {"step": 7400, "epoch": 21, "train_loss": 1.0114003419876099}, {"step": 7600, "epoch": 21, "train_loss": 0.957643985748291}, {"epoch_end": 21, "step": 7744, "eval_split": "validation", "eval_acc": 0.5958, "eval_loss": 1.09158408203125, "val_acc": 0.5958, "val_loss": 1.09158408203125}, {"step": 7800, "epoch": 22, "train_loss": 1.0202916860580444}, {"step": 8000, "epoch": 22, "train_loss": 0.8028672337532043}, {"epoch_end": 22, "step": 8096, "eval_split": "validation", "eval_acc": 0.5936, "eval_loss": 1.0905138671875, "val_acc": 0.5936, "val_loss": 1.0905138671875}, {"step": 8200, "epoch": 23, "train_loss": 0.8601924777030945}, {"step": 8400, "epoch": 23, "train_loss": 0.9645736813545227}, {"epoch_end": 23, "step": 8448, "eval_split": "validation", "eval_acc": 0.589, "eval_loss": 1.1072741943359374, "val_acc": 0.589, "val_loss": 1.1072741943359374}, {"step": 8600, "epoch": 24, "train_loss": 0.854410707950592}, {"epoch_end": 24, "step": 8800, "eval_split": "validation", "eval_acc": 0.5936, "eval_loss": 1.0944353759765626, "val_acc": 0.5936, "val_loss": 1.0944353759765626}], "final": {"eval_split": "validation", "eval_acc": 0.5936, "eval_loss": 1.0944353759765626, "wall_s": 18.031754732131958, "val_acc": 0.5936, "val_loss": 1.0944353759765626}}
|
