results: confirm standard-depth SDIL scaling

1 files changed, 2089 insertions, 0 deletions

diff --git a/results/oral_a_dynamic_scaling_v2/dfa_d32_s12.json b/results/oral_a_dynamic_scaling_v2/dfa_d32_s12.json
new file mode 100644
index 0000000..782982f
--- /dev/null
+++ b/results/oral_a_dynamic_scaling_v2/dfa_d32_s12.json
@@ -0,0 +1,2089 @@
+{
+  "architecture": {
+    "adaptive_apical_parameters": 3637248,
+    "adaptive_feedback_parameters": 0,
+    "base_width": 16,
+    "blocks_per_stage": 5,
+    "bn_eps": 1e-05,
+    "bn_momentum": 0.1,
+    "depth": 32,
+    "family": "CIFAR 6n+2 ResNet, option-A shortcuts",
+    "fixed_feedback_parameters": 0,
+    "fixed_traffic_coefficients": 303104,
+    "forward_parameters": 464154,
+    "hidden_shapes": [
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        32,
+        16,
+        16
+      ],
+      [
+        32,
+        16,
+        16
+      ],
+      [
+        32,
+        16,
+        16
+      ],
+      [
+        32,
+        16,
+        16
+      ],
+      [
+        32,
+        16,
+        16
+      ],
+      [
+        32,
+        16,
+        16
+      ],
+      [
+        32,
+        16,
+        16
+      ],
+      [
+        32,
+        16,
+        16
+      ],
+      [
+        32,
+        16,
+        16
+      ],
+      [
+        32,
+        16,
+        16
+      ],
+      [
+        64,
+        8,
+        8
+      ],
+      [
+        64,
+        8,
+        8
+      ],
+      [
+        64,
+        8,
+        8
+      ],
+      [
+        64,
+        8,
+        8
+      ],
+      [
+        64,
+        8,
+        8
+      ],
+      [
+        64,
+        8,
+        8
+      ],
+      [
+        64,
+        8,
+        8
+      ],
+      [
+        64,
+        8,
+        8
+      ],
+      [
+        64,
+        8,
+        8
+      ],
+      [
+        64,
+        8,
+        8
+      ]
+    ],
+    "normalization": "batchnorm",
+    "predictor_parameters": 606208,
+    "residual_scale": 1.0,
+    "vectorizer_mode": "spatial_template",
+    "vectorizer_parameters": 3031040
+  },
+  "args": {
+    "a_scale": 1.0,
+    "a_warmup_steps": 0,
+    "alignment_probe": 32,
+    "apical_calibration_mode": "unit_targets",
+    "apical_seed": null,
+    "augment_train": 1,
+    "batch_size": 128,
+    "bn_eps": 1e-05,
+    "bn_momentum": 0.1,
+    "data_dir": "/home/yurenh2/sdrn/data",
+    "depth": 32,
+    "device": "cuda",
+    "epochs": 200,
+    "eta_A": 0.01,
+    "eta_P": 0.01,
+    "eval_every": 0,
+    "eval_split": "test",
+    "learn_P": 0,
+    "loader_seed": 12,
+    "lr": 0.03,
+    "lr_gamma": 0.1,
+    "lr_milestones": "100,150",
+    "lr_schedule": "step",
+    "max_steps": 0,
+    "mirror_batch_size": 1,
+    "mirror_eta": 0.1,
+    "mirror_every": 16,
+    "mirror_noise_std": 1.0,
+    "mirror_seed": 3000,
+    "mirror_warmup_steps": 0,
+    "mode": "dfa",
+    "momentum": 0.9,
+    "neutral_projection": 0,
+    "normalization": "batchnorm",
+    "nuisance_scale": 0.0,
+    "out": "results/oral_a_dynamic_scaling_v2/dfa_d32_s12.json",
+    "output_lr": 0.1,
+    "pert_directions": 1,
+    "pert_every": 4,
+    "pert_sigma": 0.01,
+    "perturb_seed": 1000,
+    "predictor_every": 0,
+    "predictor_mode": "nlms",
+    "predictor_warmup_steps": 0,
+    "residual_scale": null,
+    "seed": 12,
+    "split_seed": 2027,
+    "traffic_calibration_examples": 64,
+    "traffic_ratio": 4.0,
+    "traffic_rule": "innovation",
+    "traffic_seed": 4000,
+    "train_limit": 0,
+    "use_residual": 1,
+    "val_examples": 0,
+    "vectorizer_mode": "spatial_template",
+    "warmup_epochs": 0,
+    "weight_decay": 0.0001,
+    "weight_scale": 1.0,
+    "width": 16
+  },
+  "calibration_metric_space": "full_hidden_field",
+  "counters": {
+    "apical_warmup_examples": 0,
+    "calibration_event_examples": 0,
+    "causal_scalar_observations": 0,
+    "logical_batch_loss_queries": 0,
+    "mirror_conv_examples": 0,
+    "mirror_events": 0,
+    "mirror_readout_examples": 0,
+    "neutral_projection_examples": 0,
+    "ordinary_examples": 10000000,
+    "per_example_loss_terms": 0,
+    "perturbation_events": 0,
+    "perturbation_forward_examples": 0,
+    "predictor_update_examples": 0,
+    "predictor_warmup_examples": 0,
+    "traffic_audit_examples": 0,
+    "traffic_calibration_examples": 0
+  },
+  "diagnostics": {
+    "early_third_mean": 0.0006295824307017029,
+    "innovation_negative_gradient_cosine": [
+      -0.0017798964399844408,
+      -0.0002305394154973328,
+      -0.0003138069296255708,
+      0.0005825411644764245,
+      0.002634990494698286,
+      -0.0007164775161072612,
+      0.002833605743944645,
+      -0.0009813582291826606,
+      0.0018706656992435455,
+      0.0023960997350513935,
+      -0.001758429454639554,
+      0.004448906052857637,
+      0.005853395909070969,
+      0.004932241979986429,
+      0.0018295261543244123,
+      0.003428603522479534,
+      0.005708863027393818,
+      0.0003740524407476187,
+      0.00368067086674273,
+      0.0057233176194131374,
+      0.0038504861295223236,
+      0.014013165608048439,
+      0.013460400514304638,
+      0.010330073535442352,
+      0.013087764382362366,
+      0.00992564670741558,
+      0.013553736731410027,
+      0.01602642983198166,
+      0.017832670360803604,
+      0.01908412203192711,
+      0.03177158161997795
+    ],
+    "normalization_state": "training_batch_stats_without_running_update",
+    "raw_negative_gradient_cosine": [
+      -0.0017798964399844408,
+      -0.0002305394154973328,
+      -0.0003138069296255708,
+      0.0005825411644764245,
+      0.002634990494698286,
+      -0.0007164775161072612,
+      0.002833605743944645,
+      -0.0009813582291826606,
+      0.0018706656992435455,
+      0.0023960997350513935,
+      -0.001758429454639554,
+      0.004448906052857637,
+      0.005853395909070969,
+      0.004932241979986429,
+      0.0018295261543244123,
+      0.003428603522479534,
+      0.005708863027393818,
+      0.0003740524407476187,
+      0.00368067086674273,
+      0.0057233176194131374,
+      0.0038504861295223236,
+      0.014013165608048439,
+      0.013460400514304638,
+      0.010330073535442352,
+      0.013087764382362366,
+      0.00992564670741558,
+      0.013553736731410027,
+      0.01602642983198166,
+      0.017832670360803604,
+      0.01908412203192711,
+      0.03177158161997795
+    ],
+    "teaching_negative_gradient_cosine": [
+      -0.0017798964399844408,
+      -0.0002305394154973328,
+      -0.0003138069296255708,
+      0.0005825411644764245,
+      0.002634990494698286,
+      -0.0007164775161072612,
+      0.002833605743944645,
+      -0.0009813582291826606,
+      0.0018706656992435455,
+      0.0023960997350513935,
+      -0.001758429454639554,
+      0.004448906052857637,
+      0.005853395909070969,
+      0.004932241979986429,
+      0.0018295261543244123,
+      0.003428603522479534,
+      0.005708863027393818,
+      0.0003740524407476187,
+      0.00368067086674273,
+      0.0057233176194131374,
+      0.0038504861295223236,
+      0.014013165608048439,
+      0.013460400514304638,
+      0.010330073535442352,
+      0.013087764382362366,
+      0.00992564670741558,
+      0.013553736731410027,
+      0.01602642983198166,
+      0.017832670360803604,
+      0.01908412203192711,
+      0.03177158161997795
+    ],
+    "wall_s": 0.133530855178833
+  },
+  "epochs": [
+    {
+      "epoch": 1,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 391,
+      "train_examples": 50000,
+      "train_loss": 10.945983673706055
+    },
+    {
+      "epoch": 2,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 782,
+      "train_examples": 50000,
+      "train_loss": 9.268972373046875
+    },
+    {
+      "epoch": 3,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 1173,
+      "train_examples": 50000,
+      "train_loss": 11.631135036621094
+    },
+    {
+      "epoch": 4,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 1564,
+      "train_examples": 50000,
+      "train_loss": 15.33683402130127
+    },
+    {
+      "epoch": 5,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 1955,
+      "train_examples": 50000,
+      "train_loss": 23.313101820068358
+    },
+    {
+      "epoch": 6,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 2346,
+      "train_examples": 50000,
+      "train_loss": 25.09919560180664
+    },
+    {
+      "epoch": 7,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 2737,
+      "train_examples": 50000,
+      "train_loss": 32.680282828369144
+    },
+    {
+      "epoch": 8,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 3128,
+      "train_examples": 50000,
+      "train_loss": 45.92108037231446
+    },
+    {
+      "epoch": 9,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 3519,
+      "train_examples": 50000,
+      "train_loss": 59.816421459960935
+    },
+    {
+      "epoch": 10,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 3910,
+      "train_examples": 50000,
+      "train_loss": 72.5789655078125
+    },
+    {
+      "epoch": 11,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 4301,
+      "train_examples": 50000,
+      "train_loss": 96.03700706054687
+    },
+    {
+      "epoch": 12,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 4692,
+      "train_examples": 50000,
+      "train_loss": 119.57904437988282
+    },
+    {
+      "epoch": 13,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 5083,
+      "train_examples": 50000,
+      "train_loss": 137.12588658203126
+    },
+    {
+      "epoch": 14,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 5474,
+      "train_examples": 50000,
+      "train_loss": 161.45996876953126
+    },
+    {
+      "epoch": 15,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 5865,
+      "train_examples": 50000,
+      "train_loss": 206.0152785546875
+    },
+    {
+      "epoch": 16,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 6256,
+      "train_examples": 50000,
+      "train_loss": 242.59920668945313
+    },
+    {
+      "epoch": 17,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 6647,
+      "train_examples": 50000,
+      "train_loss": 274.3646112109375
+    },
+    {
+      "epoch": 18,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 7038,
+      "train_examples": 50000,
+      "train_loss": 322.8843250341797
+    },
+    {
+      "epoch": 19,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 7429,
+      "train_examples": 50000,
+      "train_loss": 353.91449166992186
+    },
+    {
+      "epoch": 20,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 7820,
+      "train_examples": 50000,
+      "train_loss": 425.56319271972654
+    },
+    {
+      "epoch": 21,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 8211,
+      "train_examples": 50000,
+      "train_loss": 457.8159681835937
+    },
+    {
+      "epoch": 22,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 8602,
+      "train_examples": 50000,
+      "train_loss": 519.7735980859375
+    },
+    {
+      "epoch": 23,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 8993,
+      "train_examples": 50000,
+      "train_loss": 576.164127421875
+    },
+    {
+      "epoch": 24,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 9384,
+      "train_examples": 50000,
+      "train_loss": 697.5201409960938
+    },
+    {
+      "epoch": 25,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 9775,
+      "train_examples": 50000,
+      "train_loss": 769.9046768164062
+    },
+    {
+      "epoch": 26,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 10166,
+      "train_examples": 50000,
+      "train_loss": 832.7179850195313
+    },
+    {
+      "epoch": 27,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 10557,
+      "train_examples": 50000,
+      "train_loss": 859.3094276660156
+    },
+    {
+      "epoch": 28,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 10948,
+      "train_examples": 50000,
+      "train_loss": 995.9832930078124
+    },
+    {
+      "epoch": 29,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 11339,
+      "train_examples": 50000,
+      "train_loss": 1094.9401513671876
+    },
+    {
+      "epoch": 30,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 11730,
+      "train_examples": 50000,
+      "train_loss": 1203.578955625
+    },
+    {
+      "epoch": 31,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 12121,
+      "train_examples": 50000,
+      "train_loss": 1192.4968762890626
+    },
+    {
+      "epoch": 32,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 12512,
+      "train_examples": 50000,
+      "train_loss": 1330.959121953125
+    },
+    {
+      "epoch": 33,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 12903,
+      "train_examples": 50000,
+      "train_loss": 1535.65795578125
+    },
+    {
+      "epoch": 34,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 13294,
+      "train_examples": 50000,
+      "train_loss": 1571.660817265625
+    },
+    {
+      "epoch": 35,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 13685,
+      "train_examples": 50000,
+      "train_loss": 1668.7655644140625
+    },
+    {
+      "epoch": 36,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 14076,
+      "train_examples": 50000,
+      "train_loss": 1800.386894921875
+    },
+    {
+      "epoch": 37,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 14467,
+      "train_examples": 50000,
+      "train_loss": 1840.12381953125
+    },
+    {
+      "epoch": 38,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 14858,
+      "train_examples": 50000,
+      "train_loss": 2142.220775390625
+    },
+    {
+      "epoch": 39,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 15249,
+      "train_examples": 50000,
+      "train_loss": 2094.68745
+    },
+    {
+      "epoch": 40,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 15640,
+      "train_examples": 50000,
+      "train_loss": 2175.59486328125
+    },
+    {
+      "epoch": 41,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 16031,
+      "train_examples": 50000,
+      "train_loss": 2477.9617656640626
+    },
+    {
+      "epoch": 42,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 16422,
+      "train_examples": 50000,
+      "train_loss": 2538.88201515625
+    },
+    {
+      "epoch": 43,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 16813,
+      "train_examples": 50000,
+      "train_loss": 2733.2605650390624
+    },
+    {
+      "epoch": 44,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 17204,
+      "train_examples": 50000,
+      "train_loss": 2910.646749453125
+    },
+    {
+      "epoch": 45,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 17595,
+      "train_examples": 50000,
+      "train_loss": 3267.03188890625
+    },
+    {
+      "epoch": 46,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 17986,
+      "train_examples": 50000,
+      "train_loss": 3110.582402890625
+    },
+    {
+      "epoch": 47,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 18377,
+      "train_examples": 50000,
+      "train_loss": 3444.714680859375
+    },
+    {
+      "epoch": 48,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 18768,
+      "train_examples": 50000,
+      "train_loss": 3504.92890515625
+    },
+    {
+      "epoch": 49,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 19159,
+      "train_examples": 50000,
+      "train_loss": 3936.67706125
+    },
+    {
+      "epoch": 50,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 19550,
+      "train_examples": 50000,
+      "train_loss": 4224.252715625
+    },
+    {
+      "epoch": 51,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 19941,
+      "train_examples": 50000,
+      "train_loss": 3919.4437746875
+    },
+    {
+      "epoch": 52,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 20332,
+      "train_examples": 50000,
+      "train_loss": 4145.83366421875
+    },
+    {
+      "epoch": 53,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 20723,
+      "train_examples": 50000,
+      "train_loss": 4586.4608271875
+    },
+    {
+      "epoch": 54,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 21114,
+      "train_examples": 50000,
+      "train_loss": 5002.82196125
+    },
+    {
+      "epoch": 55,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 21505,
+      "train_examples": 50000,
+      "train_loss": 5003.635096484375
+    },
+    {
+      "epoch": 56,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 21896,
+      "train_examples": 50000,
+      "train_loss": 5529.62706953125
+    },
+    {
+      "epoch": 57,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 22287,
+      "train_examples": 50000,
+      "train_loss": 5188.77828671875
+    },
+    {
+      "epoch": 58,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 22678,
+      "train_examples": 50000,
+      "train_loss": 5482.045545625
+    },
+    {
+      "epoch": 59,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 23069,
+      "train_examples": 50000,
+      "train_loss": 5559.4013565625
+    },
+    {
+      "epoch": 60,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 23460,
+      "train_examples": 50000,
+      "train_loss": 5908.454430625
+    },
+    {
+      "epoch": 61,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 23851,
+      "train_examples": 50000,
+      "train_loss": 5980.538468203125
+    },
+    {
+      "epoch": 62,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 24242,
+      "train_examples": 50000,
+      "train_loss": 6472.2701846875
+    },
+    {
+      "epoch": 63,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 24633,
+      "train_examples": 50000,
+      "train_loss": 7121.7553678125
+    },
+    {
+      "epoch": 64,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 25024,
+      "train_examples": 50000,
+      "train_loss": 6582.68953703125
+    },
+    {
+      "epoch": 65,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 25415,
+      "train_examples": 50000,
+      "train_loss": 7461.9865875
+    },
+    {
+      "epoch": 66,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 25806,
+      "train_examples": 50000,
+      "train_loss": 7860.72135984375
+    },
+    {
+      "epoch": 67,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 26197,
+      "train_examples": 50000,
+      "train_loss": 7357.63736375
+    },
+    {
+      "epoch": 68,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 26588,
+      "train_examples": 50000,
+      "train_loss": 8490.441175
+    },
+    {
+      "epoch": 69,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 26979,
+      "train_examples": 50000,
+      "train_loss": 7804.14765578125
+    },
+    {
+      "epoch": 70,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 27370,
+      "train_examples": 50000,
+      "train_loss": 9004.9752734375
+    },
+    {
+      "epoch": 71,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 27761,
+      "train_examples": 50000,
+      "train_loss": 8701.8584590625
+    },
+    {
+      "epoch": 72,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 28152,
+      "train_examples": 50000,
+      "train_loss": 9813.075638125
+    },
+    {
+      "epoch": 73,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 28543,
+      "train_examples": 50000,
+      "train_loss": 10199.4175478125
+    },
+    {
+      "epoch": 74,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 28934,
+      "train_examples": 50000,
+      "train_loss": 9049.182855
+    },
+    {
+      "epoch": 75,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 29325,
+      "train_examples": 50000,
+      "train_loss": 10973.45527125
+    },
+    {
+      "epoch": 76,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 29716,
+      "train_examples": 50000,
+      "train_loss": 9971.30596546875
+    },
+    {
+      "epoch": 77,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 30107,
+      "train_examples": 50000,
+      "train_loss": 10112.501990625
+    },
+    {
+      "epoch": 78,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 30498,
+      "train_examples": 50000,
+      "train_loss": 9919.8541425
+    },
+    {
+      "epoch": 79,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 30889,
+      "train_examples": 50000,
+      "train_loss": 10888.5804590625
+    },
+    {
+      "epoch": 80,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 31280,
+      "train_examples": 50000,
+      "train_loss": 11364.8928075
+    },
+    {
+      "epoch": 81,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 31671,
+      "train_examples": 50000,
+      "train_loss": 12158.661168125
+    },
+    {
+      "epoch": 82,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 32062,
+      "train_examples": 50000,
+      "train_loss": 13187.3304353125
+    },
+    {
+      "epoch": 83,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 32453,
+      "train_examples": 50000,
+      "train_loss": 12198.78574125
+    },
+    {
+      "epoch": 84,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 32844,
+      "train_examples": 50000,
+      "train_loss": 13902.523351875
+    },
+    {
+      "epoch": 85,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 33235,
+      "train_examples": 50000,
+      "train_loss": 13500.95309015625
+    },
+    {
+      "epoch": 86,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 33626,
+      "train_examples": 50000,
+      "train_loss": 12426.27800390625
+    },
+    {
+      "epoch": 87,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 34017,
+      "train_examples": 50000,
+      "train_loss": 13973.0024403125
+    },
+    {
+      "epoch": 88,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 34408,
+      "train_examples": 50000,
+      "train_loss": 15644.7464046875
+    },
+    {
+      "epoch": 89,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 34799,
+      "train_examples": 50000,
+      "train_loss": 14839.85543
+    },
+    {
+      "epoch": 90,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 35190,
+      "train_examples": 50000,
+      "train_loss": 14755.5351403125
+    },
+    {
+      "epoch": 91,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 35581,
+      "train_examples": 50000,
+      "train_loss": 14689.6583528125
+    },
+    {
+      "epoch": 92,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 35972,
+      "train_examples": 50000,
+      "train_loss": 14835.501405625
+    },
+    {
+      "epoch": 93,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 36363,
+      "train_examples": 50000,
+      "train_loss": 17823.8316465625
+    },
+    {
+      "epoch": 94,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 36754,
+      "train_examples": 50000,
+      "train_loss": 17181.902283125
+    },
+    {
+      "epoch": 95,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 37145,
+      "train_examples": 50000,
+      "train_loss": 17498.508451875
+    },
+    {
+      "epoch": 96,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 37536,
+      "train_examples": 50000,
+      "train_loss": 17212.2398271875
+    },
+    {
+      "epoch": 97,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 37927,
+      "train_examples": 50000,
+      "train_loss": 17416.0030225
+    },
+    {
+      "epoch": 98,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 38318,
+      "train_examples": 50000,
+      "train_loss": 18377.93776125
+    },
+    {
+      "epoch": 99,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 38709,
+      "train_examples": 50000,
+      "train_loss": 18028.026354375
+    },
+    {
+      "epoch": 100,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 39100,
+      "train_examples": 50000,
+      "train_loss": 17851.23628375
+    },
+    {
+      "epoch": 101,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 39491,
+      "train_examples": 50000,
+      "train_loss": 4473.729016640625
+    },
+    {
+      "epoch": 102,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 39882,
+      "train_examples": 50000,
+      "train_loss": 2481.37507703125
+    },
+    {
+      "epoch": 103,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 40273,
+      "train_examples": 50000,
+      "train_loss": 2250.18129578125
+    },
+    {
+      "epoch": 104,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 40664,
+      "train_examples": 50000,
+      "train_loss": 2149.58493734375
+    },
+    {
+      "epoch": 105,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 41055,
+      "train_examples": 50000,
+      "train_loss": 1967.60817125
+    },
+    {
+      "epoch": 106,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 41446,
+      "train_examples": 50000,
+      "train_loss": 1852.376851640625
+    },
+    {
+      "epoch": 107,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 41837,
+      "train_examples": 50000,
+      "train_loss": 1777.812954453125
+    },
+    {
+      "epoch": 108,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 42228,
+      "train_examples": 50000,
+      "train_loss": 1674.043941796875
+    },
+    {
+      "epoch": 109,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 42619,
+      "train_examples": 50000,
+      "train_loss": 1628.418797734375
+    },
+    {
+      "epoch": 110,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 43010,
+      "train_examples": 50000,
+      "train_loss": 1604.0854262109376
+    },
+    {
+      "epoch": 111,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 43401,
+      "train_examples": 50000,
+      "train_loss": 1504.91542609375
+    },
+    {
+      "epoch": 112,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 43792,
+      "train_examples": 50000,
+      "train_loss": 1479.6587116015626
+    },
+    {
+      "epoch": 113,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 44183,
+      "train_examples": 50000,
+      "train_loss": 1531.1797834375
+    },
+    {
+      "epoch": 114,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 44574,
+      "train_examples": 50000,
+      "train_loss": 1457.0685587109374
+    },
+    {
+      "epoch": 115,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 44965,
+      "train_examples": 50000,
+      "train_loss": 1466.0198483984375
+    },
+    {
+      "epoch": 116,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 45356,
+      "train_examples": 50000,
+      "train_loss": 1427.258835234375
+    },
+    {
+      "epoch": 117,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 45747,
+      "train_examples": 50000,
+      "train_loss": 1466.809107421875
+    },
+    {
+      "epoch": 118,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 46138,
+      "train_examples": 50000,
+      "train_loss": 1469.6581989453125
+    },
+    {
+      "epoch": 119,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 46529,
+      "train_examples": 50000,
+      "train_loss": 1487.5747997265626
+    },
+    {
+      "epoch": 120,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 46920,
+      "train_examples": 50000,
+      "train_loss": 1354.257206796875
+    },
+    {
+      "epoch": 121,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 47311,
+      "train_examples": 50000,
+      "train_loss": 1377.6707595703126
+    },
+    {
+      "epoch": 122,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 47702,
+      "train_examples": 50000,
+      "train_loss": 1522.67504984375
+    },
+    {
+      "epoch": 123,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 48093,
+      "train_examples": 50000,
+      "train_loss": 1490.97368515625
+    },
+    {
+      "epoch": 124,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 48484,
+      "train_examples": 50000,
+      "train_loss": 1427.966161015625
+    },
+    {
+      "epoch": 125,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 48875,
+      "train_examples": 50000,
+      "train_loss": 1320.568961640625
+    },
+    {
+      "epoch": 126,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 49266,
+      "train_examples": 50000,
+      "train_loss": 1398.082406484375
+    },
+    {
+      "epoch": 127,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 49657,
+      "train_examples": 50000,
+      "train_loss": 1448.611054453125
+    },
+    {
+      "epoch": 128,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 50048,
+      "train_examples": 50000,
+      "train_loss": 1373.0002104296875
+    },
+    {
+      "epoch": 129,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 50439,
+      "train_examples": 50000,
+      "train_loss": 1389.1033404296875
+    },
+    {
+      "epoch": 130,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 50830,
+      "train_examples": 50000,
+      "train_loss": 1495.4376337890626
+    },
+    {
+      "epoch": 131,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 51221,
+      "train_examples": 50000,
+      "train_loss": 1485.04863515625
+    },
+    {
+      "epoch": 132,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 51612,
+      "train_examples": 50000,
+      "train_loss": 1501.30015125
+    },
+    {
+      "epoch": 133,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 52003,
+      "train_examples": 50000,
+      "train_loss": 1416.8424656640625
+    },
+    {
+      "epoch": 134,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 52394,
+      "train_examples": 50000,
+      "train_loss": 1354.535560703125
+    },
+    {
+      "epoch": 135,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 52785,
+      "train_examples": 50000,
+      "train_loss": 1257.2732968359376
+    },
+    {
+      "epoch": 136,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 53176,
+      "train_examples": 50000,
+      "train_loss": 1552.74855265625
+    },
+    {
+      "epoch": 137,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 53567,
+      "train_examples": 50000,
+      "train_loss": 1409.460373125
+    },
+    {
+      "epoch": 138,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 53958,
+      "train_examples": 50000,
+      "train_loss": 1559.767197578125
+    },
+    {
+      "epoch": 139,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 54349,
+      "train_examples": 50000,
+      "train_loss": 1487.652125703125
+    },
+    {
+      "epoch": 140,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 54740,
+      "train_examples": 50000,
+      "train_loss": 1543.125969609375
+    },
+    {
+      "epoch": 141,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 55131,
+      "train_examples": 50000,
+      "train_loss": 1467.4839623046876
+    },
+    {
+      "epoch": 142,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 55522,
+      "train_examples": 50000,
+      "train_loss": 1560.67693390625
+    },
+    {
+      "epoch": 143,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 55913,
+      "train_examples": 50000,
+      "train_loss": 1279.8525775
+    },
+    {
+      "epoch": 144,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 56304,
+      "train_examples": 50000,
+      "train_loss": 1493.00962359375
+    },
+    {
+      "epoch": 145,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 56695,
+      "train_examples": 50000,
+      "train_loss": 1334.0112649609375
+    },
+    {
+      "epoch": 146,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 57086,
+      "train_examples": 50000,
+      "train_loss": 1468.115055546875
+    },
+    {
+      "epoch": 147,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 57477,
+      "train_examples": 50000,
+      "train_loss": 1388.5694150195313
+    },
+    {
+      "epoch": 148,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 57868,
+      "train_examples": 50000,
+      "train_loss": 1477.0068703125
+    },
+    {
+      "epoch": 149,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 58259,
+      "train_examples": 50000,
+      "train_loss": 1480.3632398046875
+    },
+    {
+      "epoch": 150,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 58650,
+      "train_examples": 50000,
+      "train_loss": 1345.044163203125
+    },
+    {
+      "epoch": 151,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 59041,
+      "train_examples": 50000,
+      "train_loss": 497.901114921875
+    },
+    {
+      "epoch": 152,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 59432,
+      "train_examples": 50000,
+      "train_loss": 410.2715630078125
+    },
+    {
+      "epoch": 153,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 59823,
+      "train_examples": 50000,
+      "train_loss": 382.8667728125
+    },
+    {
+      "epoch": 154,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 60214,
+      "train_examples": 50000,
+      "train_loss": 356.7439740234375
+    },
+    {
+      "epoch": 155,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 60605,
+      "train_examples": 50000,
+      "train_loss": 338.90097631835937
+    },
+    {
+      "epoch": 156,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 60996,
+      "train_examples": 50000,
+      "train_loss": 330.863975546875
+    },
+    {
+      "epoch": 157,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 61387,
+      "train_examples": 50000,
+      "train_loss": 315.0818691552734
+    },
+    {
+      "epoch": 158,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 61778,
+      "train_examples": 50000,
+      "train_loss": 302.8783972753906
+    },
+    {
+      "epoch": 159,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 62169,
+      "train_examples": 50000,
+      "train_loss": 291.8472606933594
+    },
+    {
+      "epoch": 160,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 62560,
+      "train_examples": 50000,
+      "train_loss": 287.8466359472656
+    },
+    {
+      "epoch": 161,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 62951,
+      "train_examples": 50000,
+      "train_loss": 278.3380192578125
+    },
+    {
+      "epoch": 162,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 63342,
+      "train_examples": 50000,
+      "train_loss": 269.89768504394533
+    },
+    {
+      "epoch": 163,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 63733,
+      "train_examples": 50000,
+      "train_loss": 270.5264351953125
+    },
+    {
+      "epoch": 164,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 64124,
+      "train_examples": 50000,
+      "train_loss": 262.18027779296875
+    },
+    {
+      "epoch": 165,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 64515,
+      "train_examples": 50000,
+      "train_loss": 248.15325690429688
+    },
+    {
+      "epoch": 166,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 64906,
+      "train_examples": 50000,
+      "train_loss": 248.11462529296875
+    },
+    {
+      "epoch": 167,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 65297,
+      "train_examples": 50000,
+      "train_loss": 242.40116031738282
+    },
+    {
+      "epoch": 168,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 65688,
+      "train_examples": 50000,
+      "train_loss": 240.77935544921874
+    },
+    {
+      "epoch": 169,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 66079,
+      "train_examples": 50000,
+      "train_loss": 235.85267098632812
+    },
+    {
+      "epoch": 170,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 66470,
+      "train_examples": 50000,
+      "train_loss": 229.2045337060547
+    },
+    {
+      "epoch": 171,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 66861,
+      "train_examples": 50000,
+      "train_loss": 228.73335375976563
+    },
+    {
+      "epoch": 172,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 67252,
+      "train_examples": 50000,
+      "train_loss": 222.9525912207031
+    },
+    {
+      "epoch": 173,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 67643,
+      "train_examples": 50000,
+      "train_loss": 226.910256953125
+    },
+    {
+      "epoch": 174,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 68034,
+      "train_examples": 50000,
+      "train_loss": 211.91460568359375
+    },
+    {
+      "epoch": 175,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 68425,
+      "train_examples": 50000,
+      "train_loss": 220.34502372070313
+    },
+    {
+      "epoch": 176,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 68816,
+      "train_examples": 50000,
+      "train_loss": 217.32782608398438
+    },
+    {
+      "epoch": 177,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 69207,
+      "train_examples": 50000,
+      "train_loss": 217.8464500390625
+    },
+    {
+      "epoch": 178,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 69598,
+      "train_examples": 50000,
+      "train_loss": 213.92846335449218
+    },
+    {
+      "epoch": 179,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 69989,
+      "train_examples": 50000,
+      "train_loss": 209.2568585546875
+    },
+    {
+      "epoch": 180,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 70380,
+      "train_examples": 50000,
+      "train_loss": 215.55283462402343
+    },
+    {
+      "epoch": 181,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 70771,
+      "train_examples": 50000,
+      "train_loss": 215.38840415527343
+    },
+    {
+      "epoch": 182,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 71162,
+      "train_examples": 50000,
+      "train_loss": 205.83364685058595
+    },
+    {
+      "epoch": 183,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 71553,
+      "train_examples": 50000,
+      "train_loss": 215.3645288623047
+    },
+    {
+      "epoch": 184,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 71944,
+      "train_examples": 50000,
+      "train_loss": 207.5909344824219
+    },
+    {
+      "epoch": 185,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 72335,
+      "train_examples": 50000,
+      "train_loss": 203.62799319335937
+    },
+    {
+      "epoch": 186,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 72726,
+      "train_examples": 50000,
+      "train_loss": 213.9936000732422
+    },
+    {
+      "epoch": 187,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 73117,
+      "train_examples": 50000,
+      "train_loss": 202.21468946777344
+    },
+    {
+      "epoch": 188,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 73508,
+      "train_examples": 50000,
+      "train_loss": 201.94267767089843
+    },
+    {
+      "epoch": 189,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 73899,
+      "train_examples": 50000,
+      "train_loss": 201.10531641601563
+    },
+    {
+      "epoch": 190,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 74290,
+      "train_examples": 50000,
+      "train_loss": 198.05497042480468
+    },
+    {
+      "epoch": 191,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 74681,
+      "train_examples": 50000,
+      "train_loss": 203.1494203613281
+    },
+    {
+      "epoch": 192,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 75072,
+      "train_examples": 50000,
+      "train_loss": 203.26228122070313
+    },
+    {
+      "epoch": 193,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 75463,
+      "train_examples": 50000,
+      "train_loss": 206.91613942871095
+    },
+    {
+      "epoch": 194,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 75854,
+      "train_examples": 50000,
+      "train_loss": 191.9774615234375
+    },
+    {
+      "epoch": 195,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 76245,
+      "train_examples": 50000,
+      "train_loss": 195.60639663085936
+    },
+    {
+      "epoch": 196,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 76636,
+      "train_examples": 50000,
+      "train_loss": 203.9393654003906
+    },
+    {
+      "epoch": 197,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 77027,
+      "train_examples": 50000,
+      "train_loss": 196.9610649609375
+    },
+    {
+      "epoch": 198,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 77418,
+      "train_examples": 50000,
+      "train_loss": 196.74086037597655
+    },
+    {
+      "epoch": 199,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 77809,
+      "train_examples": 50000,
+      "train_loss": 198.9787810498047
+    },
+    {
+      "epoch": 200,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 78200,
+      "train_examples": 50000,
+      "train_loss": 194.3087822363281
+    }
+  ],
+  "evaluation_protocol": {
+    "test_evaluations": 1,
+    "test_used_for_selection": false,
+    "validation_evaluations": 0
+  },
+  "final": {
+    "accuracy": 0.3404,
+    "epoch": 200,
+    "evaluation_split": "test",
+    "finite": true,
+    "loss": 160.818975,
+    "step": 78200
+  },
+  "hardware": {
+    "cuda_device_name": "NVIDIA GeForce GTX 1080",
+    "cuda_visible_devices": "7",
+    "device": "cuda",
+    "device_total_memory_bytes": 8507949056,
+    "peak_memory_allocated_bytes": 2537912832,
+    "peak_memory_reserved_bytes": 3323985920,
+    "torch_version": "2.3.1+cu118"
+  },
+  "protocol_family": "oral_a_cifar_local_resnet_development",
+  "provenance": {
+    "git_commit": "8f9364e8ee2e2a19eb83b58452fa0e8cb155353c",
+    "git_tracked_dirty": false
+  },
+  "schema_version": 1,
+  "split": {
+    "cifar_source_files": [
+      {
+        "bytes": 31035704,
+        "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/data_batch_1",
+        "sha256": "54636561a3ce25bd3e19253c6b0d8538147b0ae398331ac4a2d86c6d987368cd"
+      },
+      {
+        "bytes": 31035320,
+        "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/data_batch_2",
+        "sha256": "766b2cef9fbc745cf056b3152224f7cf77163b330ea9a15f9392beb8b89bc5a8"
+      },
+      {
+        "bytes": 31035999,
+        "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/data_batch_3",
+        "sha256": "0f00d98ebfb30b3ec0ad19f9756dc2630b89003e10525f5e148445e82aa6a1f9"
+      },
+      {
+        "bytes": 31035696,
+        "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/data_batch_4",
+        "sha256": "3f7bb240661948b8f4d53e36ec720d8306f5668bd0071dcb4e6c947f78e9682b"
+      },
+      {
+        "bytes": 31035623,
+        "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/data_batch_5",
+        "sha256": "d91802434d8376bbaeeadf58a737e3a1b12ac839077e931237e0dcd43adcb154"
+      },
+      {
+        "bytes": 31035526,
+        "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/test_batch",
+        "sha256": "f53d8d457504f7cff4ea9e021afcf0e0ad8e24a91f3fc42091b8adef61157831"
+      }
+    ],
+    "dataset": "cifar10",
+    "evaluation_split": "test",
+    "input_layout": "NCHW",
+    "input_shape": [
+      3,
+      32,
+      32
+    ],
+    "loader_seed": 12,
+    "normalization_mean": [
+      0.49140000343322754,
+      0.4821999967098236,
+      0.4465000033378601
+    ],
+    "normalization_std": [
+      0.24699999392032623,
+      0.2434999942779541,
+      0.26159998774528503
+    ],
+    "split_from_training_only": true,
+    "split_seed": null,
+    "test_examples": 10000,
+    "train_examples": 50000,
+    "training_augmentation": "random_crop_32_padding_4_zero_then_horizontal_flip_p0.5",
+    "validation_class_counts": {},
+    "validation_examples": 0,
+    "validation_index_sha256": null
+  },
+  "timing": {
+    "apical_warmup_wall_s": 0.0,
+    "evaluation_wall_s": 1.1457910537719727,
+    "mirror_warmup_wall_s": 0.0,
+    "predictor_warmup_wall_s": 0.0,
+    "timing_excludes_data_loading_hashing_and_model_construction": true,
+    "total_timed_wall_s": 5887.7735023498535,
+    "train_wall_s": 5886.60419011116
+  },
+  "work": {
+    "apical_macs_per_example": 3031040,
+    "causal_scalar_observations": 0,
+    "components": {
+      "apical_projection_macs": 30310400000000,
+      "apical_regression_macs": 0,
+      "bp_reverse_macs_estimate": 0,
+      "kp_reciprocal_correlation_macs": 0,
+      "local_weight_correlation_macs": 688625920000000,
+      "mirror_feedback_prediction_macs": 0,
+      "mirror_local_correlation_macs": 0,
+      "mirror_response_macs": 0,
+      "ordinary_forward_macs": 688625920000000,
+      "perturbation_forward_macs": 0,
+      "warmup_clean_forward_macs": 0
+    },
+    "definition": "multiply-accumulates in conv/linear maps; one local weight correlation equals one forward-weight MAC count; BP reverse is estimated as one weight-gradient plus one activation-gradient convolution per forward convolution; mixed-traffic/predictor elementwise arithmetic is reported as a conservative operation estimate separately and is not folded into MACs",
+    "elementwise_operations_estimate": 0,
+    "forward_macs_per_example": 68862592,
+    "logical_batch_loss_queries": 0,
+    "mirror_probe_examples": 0,
+    "mirror_readout_probe_examples": 0,
+    "neutral_projection_observations": 0,
+    "per_example_cross_entropy_terms": 0,
+    "total_clean_forward_examples": 10000000,
+    "total_forward_equivalent_examples": 10000000,
+    "total_macs_estimate": 1407562240000000
+  }
+}