results: confirm standard-depth SDIL scaling

1 files changed, 1986 insertions, 0 deletions

diff --git a/results/oral_a_dynamic_scaling_v2/bp_d32_s12.json b/results/oral_a_dynamic_scaling_v2/bp_d32_s12.json
new file mode 100644
index 0000000..8b71383
--- /dev/null
+++ b/results/oral_a_dynamic_scaling_v2/bp_d32_s12.json
@@ -0,0 +1,1986 @@
+{
+  "architecture": {
+    "adaptive_apical_parameters": 0,
+    "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": 0,
+    "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": 0,
+    "residual_scale": 1.0,
+    "vectorizer_mode": null,
+    "vectorizer_parameters": 0
+  },
+  "args": {
+    "a_scale": 1.0,
+    "a_warmup_steps": 0,
+    "alignment_probe": 0,
+    "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": 0.1,
+    "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": "bp",
+    "momentum": 0.9,
+    "neutral_projection": 0,
+    "normalization": "batchnorm",
+    "nuisance_scale": 0.0,
+    "out": "results/oral_a_dynamic_scaling_v2/bp_d32_s12.json",
+    "output_lr": null,
+    "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": null,
+  "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": null,
+  "epochs": [
+    {
+      "epoch": 1,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 391,
+      "train_examples": 50000,
+      "train_loss": 2.0109724760055543
+    },
+    {
+      "epoch": 2,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 782,
+      "train_examples": 50000,
+      "train_loss": 1.4362567902374268
+    },
+    {
+      "epoch": 3,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 1173,
+      "train_examples": 50000,
+      "train_loss": 1.0867284999084472
+    },
+    {
+      "epoch": 4,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 1564,
+      "train_examples": 50000,
+      "train_loss": 0.8709917178153992
+    },
+    {
+      "epoch": 5,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 1955,
+      "train_examples": 50000,
+      "train_loss": 0.7273440295028687
+    },
+    {
+      "epoch": 6,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 2346,
+      "train_examples": 50000,
+      "train_loss": 0.6438600259399414
+    },
+    {
+      "epoch": 7,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 2737,
+      "train_examples": 50000,
+      "train_loss": 0.5770112455368042
+    },
+    {
+      "epoch": 8,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 3128,
+      "train_examples": 50000,
+      "train_loss": 0.5361482039642333
+    },
+    {
+      "epoch": 9,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 3519,
+      "train_examples": 50000,
+      "train_loss": 0.5016758289909363
+    },
+    {
+      "epoch": 10,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 3910,
+      "train_examples": 50000,
+      "train_loss": 0.468095518579483
+    },
+    {
+      "epoch": 11,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 4301,
+      "train_examples": 50000,
+      "train_loss": 0.44619220394134523
+    },
+    {
+      "epoch": 12,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 4692,
+      "train_examples": 50000,
+      "train_loss": 0.42657073191642764
+    },
+    {
+      "epoch": 13,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 5083,
+      "train_examples": 50000,
+      "train_loss": 0.40628708760261534
+    },
+    {
+      "epoch": 14,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 5474,
+      "train_examples": 50000,
+      "train_loss": 0.38555097502708435
+    },
+    {
+      "epoch": 15,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 5865,
+      "train_examples": 50000,
+      "train_loss": 0.379112883605957
+    },
+    {
+      "epoch": 16,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 6256,
+      "train_examples": 50000,
+      "train_loss": 0.36289750562667844
+    },
+    {
+      "epoch": 17,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 6647,
+      "train_examples": 50000,
+      "train_loss": 0.35174879128456116
+    },
+    {
+      "epoch": 18,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 7038,
+      "train_examples": 50000,
+      "train_loss": 0.33422498554229735
+    },
+    {
+      "epoch": 19,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 7429,
+      "train_examples": 50000,
+      "train_loss": 0.33106016448020936
+    },
+    {
+      "epoch": 20,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 7820,
+      "train_examples": 50000,
+      "train_loss": 0.31727935015678405
+    },
+    {
+      "epoch": 21,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 8211,
+      "train_examples": 50000,
+      "train_loss": 0.3118976783657074
+    },
+    {
+      "epoch": 22,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 8602,
+      "train_examples": 50000,
+      "train_loss": 0.3070366426849365
+    },
+    {
+      "epoch": 23,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 8993,
+      "train_examples": 50000,
+      "train_loss": 0.29446419066429136
+    },
+    {
+      "epoch": 24,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 9384,
+      "train_examples": 50000,
+      "train_loss": 0.28701503961563113
+    },
+    {
+      "epoch": 25,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 9775,
+      "train_examples": 50000,
+      "train_loss": 0.2842537051582336
+    },
+    {
+      "epoch": 26,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 10166,
+      "train_examples": 50000,
+      "train_loss": 0.27581866315841674
+    },
+    {
+      "epoch": 27,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 10557,
+      "train_examples": 50000,
+      "train_loss": 0.2763401738643646
+    },
+    {
+      "epoch": 28,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 10948,
+      "train_examples": 50000,
+      "train_loss": 0.25907476749420166
+    },
+    {
+      "epoch": 29,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 11339,
+      "train_examples": 50000,
+      "train_loss": 0.2554603359699249
+    },
+    {
+      "epoch": 30,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 11730,
+      "train_examples": 50000,
+      "train_loss": 0.25803864082336425
+    },
+    {
+      "epoch": 31,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 12121,
+      "train_examples": 50000,
+      "train_loss": 0.2508873863697052
+    },
+    {
+      "epoch": 32,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 12512,
+      "train_examples": 50000,
+      "train_loss": 0.2454772821187973
+    },
+    {
+      "epoch": 33,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 12903,
+      "train_examples": 50000,
+      "train_loss": 0.2503407073020935
+    },
+    {
+      "epoch": 34,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 13294,
+      "train_examples": 50000,
+      "train_loss": 0.23940209719657898
+    },
+    {
+      "epoch": 35,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 13685,
+      "train_examples": 50000,
+      "train_loss": 0.22898770686149597
+    },
+    {
+      "epoch": 36,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 14076,
+      "train_examples": 50000,
+      "train_loss": 0.23579959624290467
+    },
+    {
+      "epoch": 37,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 14467,
+      "train_examples": 50000,
+      "train_loss": 0.23449487831115723
+    },
+    {
+      "epoch": 38,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 14858,
+      "train_examples": 50000,
+      "train_loss": 0.22400421281814575
+    },
+    {
+      "epoch": 39,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 15249,
+      "train_examples": 50000,
+      "train_loss": 0.22653502743721007
+    },
+    {
+      "epoch": 40,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 15640,
+      "train_examples": 50000,
+      "train_loss": 0.21716231523036958
+    },
+    {
+      "epoch": 41,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 16031,
+      "train_examples": 50000,
+      "train_loss": 0.22319153874397277
+    },
+    {
+      "epoch": 42,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 16422,
+      "train_examples": 50000,
+      "train_loss": 0.22315684301137925
+    },
+    {
+      "epoch": 43,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 16813,
+      "train_examples": 50000,
+      "train_loss": 0.21558603727817535
+    },
+    {
+      "epoch": 44,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 17204,
+      "train_examples": 50000,
+      "train_loss": 0.20974070155858993
+    },
+    {
+      "epoch": 45,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 17595,
+      "train_examples": 50000,
+      "train_loss": 0.21091635036945344
+    },
+    {
+      "epoch": 46,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 17986,
+      "train_examples": 50000,
+      "train_loss": 0.2082318727207184
+    },
+    {
+      "epoch": 47,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 18377,
+      "train_examples": 50000,
+      "train_loss": 0.20161246847629546
+    },
+    {
+      "epoch": 48,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 18768,
+      "train_examples": 50000,
+      "train_loss": 0.20585539951324464
+    },
+    {
+      "epoch": 49,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 19159,
+      "train_examples": 50000,
+      "train_loss": 0.2038133184146881
+    },
+    {
+      "epoch": 50,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 19550,
+      "train_examples": 50000,
+      "train_loss": 0.19946762784004213
+    },
+    {
+      "epoch": 51,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 19941,
+      "train_examples": 50000,
+      "train_loss": 0.19972815490245818
+    },
+    {
+      "epoch": 52,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 20332,
+      "train_examples": 50000,
+      "train_loss": 0.20028639941215515
+    },
+    {
+      "epoch": 53,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 20723,
+      "train_examples": 50000,
+      "train_loss": 0.1937701982116699
+    },
+    {
+      "epoch": 54,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 21114,
+      "train_examples": 50000,
+      "train_loss": 0.19515884229660035
+    },
+    {
+      "epoch": 55,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 21505,
+      "train_examples": 50000,
+      "train_loss": 0.19537211338996888
+    },
+    {
+      "epoch": 56,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 21896,
+      "train_examples": 50000,
+      "train_loss": 0.1926883613014221
+    },
+    {
+      "epoch": 57,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 22287,
+      "train_examples": 50000,
+      "train_loss": 0.19347582818984985
+    },
+    {
+      "epoch": 58,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 22678,
+      "train_examples": 50000,
+      "train_loss": 0.19204679753303527
+    },
+    {
+      "epoch": 59,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 23069,
+      "train_examples": 50000,
+      "train_loss": 0.18423490328788758
+    },
+    {
+      "epoch": 60,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 23460,
+      "train_examples": 50000,
+      "train_loss": 0.18769514951705932
+    },
+    {
+      "epoch": 61,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 23851,
+      "train_examples": 50000,
+      "train_loss": 0.18194034811496734
+    },
+    {
+      "epoch": 62,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 24242,
+      "train_examples": 50000,
+      "train_loss": 0.18289040670871734
+    },
+    {
+      "epoch": 63,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 24633,
+      "train_examples": 50000,
+      "train_loss": 0.1808374851989746
+    },
+    {
+      "epoch": 64,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 25024,
+      "train_examples": 50000,
+      "train_loss": 0.18032792008399964
+    },
+    {
+      "epoch": 65,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 25415,
+      "train_examples": 50000,
+      "train_loss": 0.1827612723827362
+    },
+    {
+      "epoch": 66,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 25806,
+      "train_examples": 50000,
+      "train_loss": 0.18270389512062074
+    },
+    {
+      "epoch": 67,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 26197,
+      "train_examples": 50000,
+      "train_loss": 0.18044097856521607
+    },
+    {
+      "epoch": 68,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 26588,
+      "train_examples": 50000,
+      "train_loss": 0.1740411812543869
+    },
+    {
+      "epoch": 69,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 26979,
+      "train_examples": 50000,
+      "train_loss": 0.1738271105015278
+    },
+    {
+      "epoch": 70,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 27370,
+      "train_examples": 50000,
+      "train_loss": 0.1737141721534729
+    },
+    {
+      "epoch": 71,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 27761,
+      "train_examples": 50000,
+      "train_loss": 0.17230419850826265
+    },
+    {
+      "epoch": 72,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 28152,
+      "train_examples": 50000,
+      "train_loss": 0.17472062274932862
+    },
+    {
+      "epoch": 73,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 28543,
+      "train_examples": 50000,
+      "train_loss": 0.17696084894180297
+    },
+    {
+      "epoch": 74,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 28934,
+      "train_examples": 50000,
+      "train_loss": 0.16926053122997284
+    },
+    {
+      "epoch": 75,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 29325,
+      "train_examples": 50000,
+      "train_loss": 0.17055362528800963
+    },
+    {
+      "epoch": 76,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 29716,
+      "train_examples": 50000,
+      "train_loss": 0.1706799511241913
+    },
+    {
+      "epoch": 77,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 30107,
+      "train_examples": 50000,
+      "train_loss": 0.16936428791046143
+    },
+    {
+      "epoch": 78,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 30498,
+      "train_examples": 50000,
+      "train_loss": 0.16855762898445129
+    },
+    {
+      "epoch": 79,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 30889,
+      "train_examples": 50000,
+      "train_loss": 0.16561252969503404
+    },
+    {
+      "epoch": 80,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 31280,
+      "train_examples": 50000,
+      "train_loss": 0.17293027634620667
+    },
+    {
+      "epoch": 81,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 31671,
+      "train_examples": 50000,
+      "train_loss": 0.17181514074802398
+    },
+    {
+      "epoch": 82,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 32062,
+      "train_examples": 50000,
+      "train_loss": 0.16485951325416565
+    },
+    {
+      "epoch": 83,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 32453,
+      "train_examples": 50000,
+      "train_loss": 0.16485164187431336
+    },
+    {
+      "epoch": 84,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 32844,
+      "train_examples": 50000,
+      "train_loss": 0.16326760728597642
+    },
+    {
+      "epoch": 85,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 33235,
+      "train_examples": 50000,
+      "train_loss": 0.1644478901576996
+    },
+    {
+      "epoch": 86,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 33626,
+      "train_examples": 50000,
+      "train_loss": 0.165506713347435
+    },
+    {
+      "epoch": 87,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 34017,
+      "train_examples": 50000,
+      "train_loss": 0.164615642824173
+    },
+    {
+      "epoch": 88,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 34408,
+      "train_examples": 50000,
+      "train_loss": 0.1611957668232918
+    },
+    {
+      "epoch": 89,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 34799,
+      "train_examples": 50000,
+      "train_loss": 0.16008895670890808
+    },
+    {
+      "epoch": 90,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 35190,
+      "train_examples": 50000,
+      "train_loss": 0.16092798224925994
+    },
+    {
+      "epoch": 91,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 35581,
+      "train_examples": 50000,
+      "train_loss": 0.1590071769285202
+    },
+    {
+      "epoch": 92,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 35972,
+      "train_examples": 50000,
+      "train_loss": 0.16325321798801423
+    },
+    {
+      "epoch": 93,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 36363,
+      "train_examples": 50000,
+      "train_loss": 0.15636748574256898
+    },
+    {
+      "epoch": 94,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 36754,
+      "train_examples": 50000,
+      "train_loss": 0.16012700378894806
+    },
+    {
+      "epoch": 95,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 37145,
+      "train_examples": 50000,
+      "train_loss": 0.15904670073986055
+    },
+    {
+      "epoch": 96,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 37536,
+      "train_examples": 50000,
+      "train_loss": 0.1563712153697014
+    },
+    {
+      "epoch": 97,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 37927,
+      "train_examples": 50000,
+      "train_loss": 0.15662149175167084
+    },
+    {
+      "epoch": 98,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 38318,
+      "train_examples": 50000,
+      "train_loss": 0.15754105580329894
+    },
+    {
+      "epoch": 99,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 38709,
+      "train_examples": 50000,
+      "train_loss": 0.1567109692144394
+    },
+    {
+      "epoch": 100,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 39100,
+      "train_examples": 50000,
+      "train_loss": 0.15609369305610657
+    },
+    {
+      "epoch": 101,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 39491,
+      "train_examples": 50000,
+      "train_loss": 0.08568393441200256
+    },
+    {
+      "epoch": 102,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 39882,
+      "train_examples": 50000,
+      "train_loss": 0.06317522408127785
+    },
+    {
+      "epoch": 103,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 40273,
+      "train_examples": 50000,
+      "train_loss": 0.05369775810599327
+    },
+    {
+      "epoch": 104,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 40664,
+      "train_examples": 50000,
+      "train_loss": 0.0476024976670742
+    },
+    {
+      "epoch": 105,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 41055,
+      "train_examples": 50000,
+      "train_loss": 0.042950796638131145
+    },
+    {
+      "epoch": 106,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 41446,
+      "train_examples": 50000,
+      "train_loss": 0.038947348670363424
+    },
+    {
+      "epoch": 107,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 41837,
+      "train_examples": 50000,
+      "train_loss": 0.037823320852518084
+    },
+    {
+      "epoch": 108,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 42228,
+      "train_examples": 50000,
+      "train_loss": 0.034411116131544114
+    },
+    {
+      "epoch": 109,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 42619,
+      "train_examples": 50000,
+      "train_loss": 0.03271289059400558
+    },
+    {
+      "epoch": 110,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 43010,
+      "train_examples": 50000,
+      "train_loss": 0.0314477956533432
+    },
+    {
+      "epoch": 111,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 43401,
+      "train_examples": 50000,
+      "train_loss": 0.029343869721889494
+    },
+    {
+      "epoch": 112,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 43792,
+      "train_examples": 50000,
+      "train_loss": 0.028077500077784062
+    },
+    {
+      "epoch": 113,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 44183,
+      "train_examples": 50000,
+      "train_loss": 0.026478302937150002
+    },
+    {
+      "epoch": 114,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 44574,
+      "train_examples": 50000,
+      "train_loss": 0.026096549355834723
+    },
+    {
+      "epoch": 115,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 44965,
+      "train_examples": 50000,
+      "train_loss": 0.02390244416564703
+    },
+    {
+      "epoch": 116,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 45356,
+      "train_examples": 50000,
+      "train_loss": 0.02377692777991295
+    },
+    {
+      "epoch": 117,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 45747,
+      "train_examples": 50000,
+      "train_loss": 0.021567717416882515
+    },
+    {
+      "epoch": 118,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 46138,
+      "train_examples": 50000,
+      "train_loss": 0.021640529872179032
+    },
+    {
+      "epoch": 119,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 46529,
+      "train_examples": 50000,
+      "train_loss": 0.020908026728034018
+    },
+    {
+      "epoch": 120,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 46920,
+      "train_examples": 50000,
+      "train_loss": 0.020051894177794457
+    },
+    {
+      "epoch": 121,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 47311,
+      "train_examples": 50000,
+      "train_loss": 0.019267218337357044
+    },
+    {
+      "epoch": 122,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 47702,
+      "train_examples": 50000,
+      "train_loss": 0.018688082472383977
+    },
+    {
+      "epoch": 123,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 48093,
+      "train_examples": 50000,
+      "train_loss": 0.018083301803544165
+    },
+    {
+      "epoch": 124,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 48484,
+      "train_examples": 50000,
+      "train_loss": 0.016567399821281432
+    },
+    {
+      "epoch": 125,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 48875,
+      "train_examples": 50000,
+      "train_loss": 0.015733772432953118
+    },
+    {
+      "epoch": 126,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 49266,
+      "train_examples": 50000,
+      "train_loss": 0.014920587013065815
+    },
+    {
+      "epoch": 127,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 49657,
+      "train_examples": 50000,
+      "train_loss": 0.014819765307158232
+    },
+    {
+      "epoch": 128,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 50048,
+      "train_examples": 50000,
+      "train_loss": 0.014428511834591626
+    },
+    {
+      "epoch": 129,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 50439,
+      "train_examples": 50000,
+      "train_loss": 0.014853895919919013
+    },
+    {
+      "epoch": 130,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 50830,
+      "train_examples": 50000,
+      "train_loss": 0.014573111248910427
+    },
+    {
+      "epoch": 131,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 51221,
+      "train_examples": 50000,
+      "train_loss": 0.013724009919166565
+    },
+    {
+      "epoch": 132,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 51612,
+      "train_examples": 50000,
+      "train_loss": 0.012956059173047542
+    },
+    {
+      "epoch": 133,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 52003,
+      "train_examples": 50000,
+      "train_loss": 0.011421179175302386
+    },
+    {
+      "epoch": 134,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 52394,
+      "train_examples": 50000,
+      "train_loss": 0.012258051640093326
+    },
+    {
+      "epoch": 135,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 52785,
+      "train_examples": 50000,
+      "train_loss": 0.011669609291255474
+    },
+    {
+      "epoch": 136,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 53176,
+      "train_examples": 50000,
+      "train_loss": 0.0116468328332901
+    },
+    {
+      "epoch": 137,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 53567,
+      "train_examples": 50000,
+      "train_loss": 0.012746921005249023
+    },
+    {
+      "epoch": 138,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 53958,
+      "train_examples": 50000,
+      "train_loss": 0.011482126246765256
+    },
+    {
+      "epoch": 139,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 54349,
+      "train_examples": 50000,
+      "train_loss": 0.011237891421169042
+    },
+    {
+      "epoch": 140,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 54740,
+      "train_examples": 50000,
+      "train_loss": 0.011759749162048102
+    },
+    {
+      "epoch": 141,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 55131,
+      "train_examples": 50000,
+      "train_loss": 0.010505393952429295
+    },
+    {
+      "epoch": 142,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 55522,
+      "train_examples": 50000,
+      "train_loss": 0.011218925432339311
+    },
+    {
+      "epoch": 143,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 55913,
+      "train_examples": 50000,
+      "train_loss": 0.01023633032079786
+    },
+    {
+      "epoch": 144,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 56304,
+      "train_examples": 50000,
+      "train_loss": 0.010636819964051247
+    },
+    {
+      "epoch": 145,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 56695,
+      "train_examples": 50000,
+      "train_loss": 0.010898309437930584
+    },
+    {
+      "epoch": 146,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 57086,
+      "train_examples": 50000,
+      "train_loss": 0.010507725623846054
+    },
+    {
+      "epoch": 147,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 57477,
+      "train_examples": 50000,
+      "train_loss": 0.010324308806657792
+    },
+    {
+      "epoch": 148,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 57868,
+      "train_examples": 50000,
+      "train_loss": 0.009738635150194169
+    },
+    {
+      "epoch": 149,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 58259,
+      "train_examples": 50000,
+      "train_loss": 0.008979119207412005
+    },
+    {
+      "epoch": 150,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 58650,
+      "train_examples": 50000,
+      "train_loss": 0.009272252855747938
+    },
+    {
+      "epoch": 151,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 59041,
+      "train_examples": 50000,
+      "train_loss": 0.007726711673140526
+    },
+    {
+      "epoch": 152,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 59432,
+      "train_examples": 50000,
+      "train_loss": 0.007999423414170741
+    },
+    {
+      "epoch": 153,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 59823,
+      "train_examples": 50000,
+      "train_loss": 0.007616992153301835
+    },
+    {
+      "epoch": 154,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 60214,
+      "train_examples": 50000,
+      "train_loss": 0.007475075915306806
+    },
+    {
+      "epoch": 155,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 60605,
+      "train_examples": 50000,
+      "train_loss": 0.006915905899107456
+    },
+    {
+      "epoch": 156,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 60996,
+      "train_examples": 50000,
+      "train_loss": 0.006768089570701122
+    },
+    {
+      "epoch": 157,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 61387,
+      "train_examples": 50000,
+      "train_loss": 0.0068142542649805545
+    },
+    {
+      "epoch": 158,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 61778,
+      "train_examples": 50000,
+      "train_loss": 0.0067272376950085165
+    },
+    {
+      "epoch": 159,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 62169,
+      "train_examples": 50000,
+      "train_loss": 0.006835442341864109
+    },
+    {
+      "epoch": 160,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 62560,
+      "train_examples": 50000,
+      "train_loss": 0.0066535683606565
+    },
+    {
+      "epoch": 161,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 62951,
+      "train_examples": 50000,
+      "train_loss": 0.0068024345889687535
+    },
+    {
+      "epoch": 162,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 63342,
+      "train_examples": 50000,
+      "train_loss": 0.006965562088042498
+    },
+    {
+      "epoch": 163,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 63733,
+      "train_examples": 50000,
+      "train_loss": 0.006457564204707742
+    },
+    {
+      "epoch": 164,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 64124,
+      "train_examples": 50000,
+      "train_loss": 0.006979998501986265
+    },
+    {
+      "epoch": 165,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 64515,
+      "train_examples": 50000,
+      "train_loss": 0.006785780550315976
+    },
+    {
+      "epoch": 166,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 64906,
+      "train_examples": 50000,
+      "train_loss": 0.0065035085116326805
+    },
+    {
+      "epoch": 167,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 65297,
+      "train_examples": 50000,
+      "train_loss": 0.006269088081642986
+    },
+    {
+      "epoch": 168,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 65688,
+      "train_examples": 50000,
+      "train_loss": 0.006208638264853507
+    },
+    {
+      "epoch": 169,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 66079,
+      "train_examples": 50000,
+      "train_loss": 0.0069993398681282995
+    },
+    {
+      "epoch": 170,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 66470,
+      "train_examples": 50000,
+      "train_loss": 0.006230590722858906
+    },
+    {
+      "epoch": 171,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 66861,
+      "train_examples": 50000,
+      "train_loss": 0.006384732846021653
+    },
+    {
+      "epoch": 172,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 67252,
+      "train_examples": 50000,
+      "train_loss": 0.006704631698206067
+    },
+    {
+      "epoch": 173,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 67643,
+      "train_examples": 50000,
+      "train_loss": 0.006025525711774826
+    },
+    {
+      "epoch": 174,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 68034,
+      "train_examples": 50000,
+      "train_loss": 0.005968988159000874
+    },
+    {
+      "epoch": 175,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 68425,
+      "train_examples": 50000,
+      "train_loss": 0.006251406453922391
+    },
+    {
+      "epoch": 176,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 68816,
+      "train_examples": 50000,
+      "train_loss": 0.006875649279206991
+    },
+    {
+      "epoch": 177,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 69207,
+      "train_examples": 50000,
+      "train_loss": 0.006274009262062609
+    },
+    {
+      "epoch": 178,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 69598,
+      "train_examples": 50000,
+      "train_loss": 0.005923532918095589
+    },
+    {
+      "epoch": 179,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 69989,
+      "train_examples": 50000,
+      "train_loss": 0.006215575973615051
+    },
+    {
+      "epoch": 180,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 70380,
+      "train_examples": 50000,
+      "train_loss": 0.005989656295031309
+    },
+    {
+      "epoch": 181,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 70771,
+      "train_examples": 50000,
+      "train_loss": 0.006137610604166985
+    },
+    {
+      "epoch": 182,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 71162,
+      "train_examples": 50000,
+      "train_loss": 0.005827424137964845
+    },
+    {
+      "epoch": 183,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 71553,
+      "train_examples": 50000,
+      "train_loss": 0.005676852166801691
+    },
+    {
+      "epoch": 184,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 71944,
+      "train_examples": 50000,
+      "train_loss": 0.005693613475114107
+    },
+    {
+      "epoch": 185,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 72335,
+      "train_examples": 50000,
+      "train_loss": 0.005433482741564512
+    },
+    {
+      "epoch": 186,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 72726,
+      "train_examples": 50000,
+      "train_loss": 0.006112713556215167
+    },
+    {
+      "epoch": 187,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 73117,
+      "train_examples": 50000,
+      "train_loss": 0.005977615090310574
+    },
+    {
+      "epoch": 188,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 73508,
+      "train_examples": 50000,
+      "train_loss": 0.005879531293958426
+    },
+    {
+      "epoch": 189,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 73899,
+      "train_examples": 50000,
+      "train_loss": 0.005727535793632269
+    },
+    {
+      "epoch": 190,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 74290,
+      "train_examples": 50000,
+      "train_loss": 0.005559209338799119
+    },
+    {
+      "epoch": 191,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 74681,
+      "train_examples": 50000,
+      "train_loss": 0.006220875388532877
+    },
+    {
+      "epoch": 192,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 75072,
+      "train_examples": 50000,
+      "train_loss": 0.005635415128320455
+    },
+    {
+      "epoch": 193,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 75463,
+      "train_examples": 50000,
+      "train_loss": 0.006039364632219076
+    },
+    {
+      "epoch": 194,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 75854,
+      "train_examples": 50000,
+      "train_loss": 0.005313597747534513
+    },
+    {
+      "epoch": 195,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 76245,
+      "train_examples": 50000,
+      "train_loss": 0.006019642222523689
+    },
+    {
+      "epoch": 196,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 76636,
+      "train_examples": 50000,
+      "train_loss": 0.005320607864558697
+    },
+    {
+      "epoch": 197,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 77027,
+      "train_examples": 50000,
+      "train_loss": 0.0062282088583707805
+    },
+    {
+      "epoch": 198,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 77418,
+      "train_examples": 50000,
+      "train_loss": 0.005710777210593224
+    },
+    {
+      "epoch": 199,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 77809,
+      "train_examples": 50000,
+      "train_loss": 0.005980389836505055
+    },
+    {
+      "epoch": 200,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 78200,
+      "train_examples": 50000,
+      "train_loss": 0.005370809080637992
+    }
+  ],
+  "evaluation_protocol": {
+    "test_evaluations": 1,
+    "test_used_for_selection": false,
+    "validation_evaluations": 0
+  },
+  "final": {
+    "accuracy": 0.9228,
+    "epoch": 200,
+    "evaluation_split": "test",
+    "finite": true,
+    "loss": 0.3814170867919922,
+    "step": 78200
+  },
+  "hardware": {
+    "cuda_device_name": "NVIDIA GeForce GTX 1080",
+    "cuda_visible_devices": "5",
+    "device": "cuda",
+    "device_total_memory_bytes": 8507949056,
+    "peak_memory_allocated_bytes": 2530507264,
+    "peak_memory_reserved_bytes": 3254779904,
+    "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": 1.1620919704437256,
+    "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": 5822.728997468948,
+    "train_wall_s": 5821.533981084824
+  },
+  "work": {
+    "apical_macs_per_example": 0,
+    "causal_scalar_observations": 0,
+    "components": {
+      "apical_projection_macs": 0,
+      "apical_regression_macs": 0,
+      "bp_reverse_macs_estimate": 1377251840000000,
+      "kp_reciprocal_correlation_macs": 0,
+      "local_weight_correlation_macs": 0,
+      "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": 2065877760000000
+  }
+}