results: confirm standard-depth SDIL scaling

1 files changed, 1986 insertions, 0 deletions

diff --git a/results/oral_a_dynamic_scaling_v2/bp_d32_s13.json b/results/oral_a_dynamic_scaling_v2/bp_d32_s13.json
new file mode 100644
index 0000000..44985ce
--- /dev/null
+++ b/results/oral_a_dynamic_scaling_v2/bp_d32_s13.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": 13,
+    "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_s13.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": 13,
+    "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": 1.8805035388946534
+    },
+    {
+      "epoch": 2,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 782,
+      "train_examples": 50000,
+      "train_loss": 1.2631674034881593
+    },
+    {
+      "epoch": 3,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 1173,
+      "train_examples": 50000,
+      "train_loss": 0.982023414478302
+    },
+    {
+      "epoch": 4,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 1564,
+      "train_examples": 50000,
+      "train_loss": 0.8269909022331238
+    },
+    {
+      "epoch": 5,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 1955,
+      "train_examples": 50000,
+      "train_loss": 0.7033132957839966
+    },
+    {
+      "epoch": 6,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 2346,
+      "train_examples": 50000,
+      "train_loss": 0.6245331028175354
+    },
+    {
+      "epoch": 7,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 2737,
+      "train_examples": 50000,
+      "train_loss": 0.5728815777492523
+    },
+    {
+      "epoch": 8,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 3128,
+      "train_examples": 50000,
+      "train_loss": 0.52921937292099
+    },
+    {
+      "epoch": 9,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 3519,
+      "train_examples": 50000,
+      "train_loss": 0.4934953989982605
+    },
+    {
+      "epoch": 10,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 3910,
+      "train_examples": 50000,
+      "train_loss": 0.46322856472969054
+    },
+    {
+      "epoch": 11,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 4301,
+      "train_examples": 50000,
+      "train_loss": 0.4377773000526428
+    },
+    {
+      "epoch": 12,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 4692,
+      "train_examples": 50000,
+      "train_loss": 0.42375280329704285
+    },
+    {
+      "epoch": 13,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 5083,
+      "train_examples": 50000,
+      "train_loss": 0.4023252283382416
+    },
+    {
+      "epoch": 14,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 5474,
+      "train_examples": 50000,
+      "train_loss": 0.39270214566230777
+    },
+    {
+      "epoch": 15,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 5865,
+      "train_examples": 50000,
+      "train_loss": 0.37170859550476076
+    },
+    {
+      "epoch": 16,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 6256,
+      "train_examples": 50000,
+      "train_loss": 0.35657511695861815
+    },
+    {
+      "epoch": 17,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 6647,
+      "train_examples": 50000,
+      "train_loss": 0.35467035373687744
+    },
+    {
+      "epoch": 18,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 7038,
+      "train_examples": 50000,
+      "train_loss": 0.33651030123710635
+    },
+    {
+      "epoch": 19,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 7429,
+      "train_examples": 50000,
+      "train_loss": 0.3320449847984314
+    },
+    {
+      "epoch": 20,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 7820,
+      "train_examples": 50000,
+      "train_loss": 0.31947522946357726
+    },
+    {
+      "epoch": 21,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 8211,
+      "train_examples": 50000,
+      "train_loss": 0.30916595887184145
+    },
+    {
+      "epoch": 22,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 8602,
+      "train_examples": 50000,
+      "train_loss": 0.30547217082977296
+    },
+    {
+      "epoch": 23,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 8993,
+      "train_examples": 50000,
+      "train_loss": 0.29219140836715696
+    },
+    {
+      "epoch": 24,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 9384,
+      "train_examples": 50000,
+      "train_loss": 0.29400309326171875
+    },
+    {
+      "epoch": 25,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 9775,
+      "train_examples": 50000,
+      "train_loss": 0.28870507167339327
+    },
+    {
+      "epoch": 26,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 10166,
+      "train_examples": 50000,
+      "train_loss": 0.2748667608737946
+    },
+    {
+      "epoch": 27,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 10557,
+      "train_examples": 50000,
+      "train_loss": 0.27103702613830566
+    },
+    {
+      "epoch": 28,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 10948,
+      "train_examples": 50000,
+      "train_loss": 0.26546217414855955
+    },
+    {
+      "epoch": 29,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 11339,
+      "train_examples": 50000,
+      "train_loss": 0.2625025454044342
+    },
+    {
+      "epoch": 30,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 11730,
+      "train_examples": 50000,
+      "train_loss": 0.25851132451057435
+    },
+    {
+      "epoch": 31,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 12121,
+      "train_examples": 50000,
+      "train_loss": 0.25504755184173583
+    },
+    {
+      "epoch": 32,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 12512,
+      "train_examples": 50000,
+      "train_loss": 0.24713118095397948
+    },
+    {
+      "epoch": 33,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 12903,
+      "train_examples": 50000,
+      "train_loss": 0.24437052943229676
+    },
+    {
+      "epoch": 34,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 13294,
+      "train_examples": 50000,
+      "train_loss": 0.23974795582294464
+    },
+    {
+      "epoch": 35,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 13685,
+      "train_examples": 50000,
+      "train_loss": 0.23825712741851807
+    },
+    {
+      "epoch": 36,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 14076,
+      "train_examples": 50000,
+      "train_loss": 0.24186695692062377
+    },
+    {
+      "epoch": 37,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 14467,
+      "train_examples": 50000,
+      "train_loss": 0.23216649408817291
+    },
+    {
+      "epoch": 38,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 14858,
+      "train_examples": 50000,
+      "train_loss": 0.22886585109233856
+    },
+    {
+      "epoch": 39,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 15249,
+      "train_examples": 50000,
+      "train_loss": 0.22818136295318603
+    },
+    {
+      "epoch": 40,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 15640,
+      "train_examples": 50000,
+      "train_loss": 0.2233695182609558
+    },
+    {
+      "epoch": 41,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 16031,
+      "train_examples": 50000,
+      "train_loss": 0.22007071732521058
+    },
+    {
+      "epoch": 42,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 16422,
+      "train_examples": 50000,
+      "train_loss": 0.21992354313850404
+    },
+    {
+      "epoch": 43,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 16813,
+      "train_examples": 50000,
+      "train_loss": 0.21873045570373534
+    },
+    {
+      "epoch": 44,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 17204,
+      "train_examples": 50000,
+      "train_loss": 0.21697771926403045
+    },
+    {
+      "epoch": 45,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 17595,
+      "train_examples": 50000,
+      "train_loss": 0.20397321831226348
+    },
+    {
+      "epoch": 46,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 17986,
+      "train_examples": 50000,
+      "train_loss": 0.2084566786146164
+    },
+    {
+      "epoch": 47,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 18377,
+      "train_examples": 50000,
+      "train_loss": 0.20745201335906982
+    },
+    {
+      "epoch": 48,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 18768,
+      "train_examples": 50000,
+      "train_loss": 0.20686608068466186
+    },
+    {
+      "epoch": 49,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 19159,
+      "train_examples": 50000,
+      "train_loss": 0.20360097942352295
+    },
+    {
+      "epoch": 50,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 19550,
+      "train_examples": 50000,
+      "train_loss": 0.2000230358028412
+    },
+    {
+      "epoch": 51,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 19941,
+      "train_examples": 50000,
+      "train_loss": 0.19716575583457946
+    },
+    {
+      "epoch": 52,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 20332,
+      "train_examples": 50000,
+      "train_loss": 0.20097089833259582
+    },
+    {
+      "epoch": 53,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 20723,
+      "train_examples": 50000,
+      "train_loss": 0.19471816690921784
+    },
+    {
+      "epoch": 54,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 21114,
+      "train_examples": 50000,
+      "train_loss": 0.197814011220932
+    },
+    {
+      "epoch": 55,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 21505,
+      "train_examples": 50000,
+      "train_loss": 0.1954924241065979
+    },
+    {
+      "epoch": 56,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 21896,
+      "train_examples": 50000,
+      "train_loss": 0.19353147352218628
+    },
+    {
+      "epoch": 57,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 22287,
+      "train_examples": 50000,
+      "train_loss": 0.18696056705474853
+    },
+    {
+      "epoch": 58,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 22678,
+      "train_examples": 50000,
+      "train_loss": 0.19248952531337737
+    },
+    {
+      "epoch": 59,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 23069,
+      "train_examples": 50000,
+      "train_loss": 0.19309106615543364
+    },
+    {
+      "epoch": 60,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 23460,
+      "train_examples": 50000,
+      "train_loss": 0.18839726268291473
+    },
+    {
+      "epoch": 61,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 23851,
+      "train_examples": 50000,
+      "train_loss": 0.1834150011730194
+    },
+    {
+      "epoch": 62,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 24242,
+      "train_examples": 50000,
+      "train_loss": 0.18410958391189575
+    },
+    {
+      "epoch": 63,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 24633,
+      "train_examples": 50000,
+      "train_loss": 0.18338790824890136
+    },
+    {
+      "epoch": 64,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 25024,
+      "train_examples": 50000,
+      "train_loss": 0.1839743905353546
+    },
+    {
+      "epoch": 65,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 25415,
+      "train_examples": 50000,
+      "train_loss": 0.1801528820514679
+    },
+    {
+      "epoch": 66,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 25806,
+      "train_examples": 50000,
+      "train_loss": 0.1803589987921715
+    },
+    {
+      "epoch": 67,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 26197,
+      "train_examples": 50000,
+      "train_loss": 0.1827104704475403
+    },
+    {
+      "epoch": 68,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 26588,
+      "train_examples": 50000,
+      "train_loss": 0.17917614500522613
+    },
+    {
+      "epoch": 69,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 26979,
+      "train_examples": 50000,
+      "train_loss": 0.17485497699737548
+    },
+    {
+      "epoch": 70,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 27370,
+      "train_examples": 50000,
+      "train_loss": 0.1796244272518158
+    },
+    {
+      "epoch": 71,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 27761,
+      "train_examples": 50000,
+      "train_loss": 0.1771907129573822
+    },
+    {
+      "epoch": 72,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 28152,
+      "train_examples": 50000,
+      "train_loss": 0.1721141401386261
+    },
+    {
+      "epoch": 73,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 28543,
+      "train_examples": 50000,
+      "train_loss": 0.17699346287965775
+    },
+    {
+      "epoch": 74,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 28934,
+      "train_examples": 50000,
+      "train_loss": 0.17763278216362
+    },
+    {
+      "epoch": 75,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 29325,
+      "train_examples": 50000,
+      "train_loss": 0.17064625200748443
+    },
+    {
+      "epoch": 76,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 29716,
+      "train_examples": 50000,
+      "train_loss": 0.16618630229473114
+    },
+    {
+      "epoch": 77,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 30107,
+      "train_examples": 50000,
+      "train_loss": 0.17021148312091827
+    },
+    {
+      "epoch": 78,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 30498,
+      "train_examples": 50000,
+      "train_loss": 0.1735873026251793
+    },
+    {
+      "epoch": 79,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 30889,
+      "train_examples": 50000,
+      "train_loss": 0.17325835156440736
+    },
+    {
+      "epoch": 80,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 31280,
+      "train_examples": 50000,
+      "train_loss": 0.17159906162261962
+    },
+    {
+      "epoch": 81,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 31671,
+      "train_examples": 50000,
+      "train_loss": 0.16759568387508392
+    },
+    {
+      "epoch": 82,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 32062,
+      "train_examples": 50000,
+      "train_loss": 0.16298270482778549
+    },
+    {
+      "epoch": 83,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 32453,
+      "train_examples": 50000,
+      "train_loss": 0.16351762491226196
+    },
+    {
+      "epoch": 84,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 32844,
+      "train_examples": 50000,
+      "train_loss": 0.166046575756073
+    },
+    {
+      "epoch": 85,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 33235,
+      "train_examples": 50000,
+      "train_loss": 0.16746359069824218
+    },
+    {
+      "epoch": 86,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 33626,
+      "train_examples": 50000,
+      "train_loss": 0.16691714178085326
+    },
+    {
+      "epoch": 87,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 34017,
+      "train_examples": 50000,
+      "train_loss": 0.16248040862560273
+    },
+    {
+      "epoch": 88,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 34408,
+      "train_examples": 50000,
+      "train_loss": 0.16665220249652862
+    },
+    {
+      "epoch": 89,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 34799,
+      "train_examples": 50000,
+      "train_loss": 0.16087845814704896
+    },
+    {
+      "epoch": 90,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 35190,
+      "train_examples": 50000,
+      "train_loss": 0.16221520748138427
+    },
+    {
+      "epoch": 91,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 35581,
+      "train_examples": 50000,
+      "train_loss": 0.1628304146051407
+    },
+    {
+      "epoch": 92,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 35972,
+      "train_examples": 50000,
+      "train_loss": 0.15647470529079438
+    },
+    {
+      "epoch": 93,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 36363,
+      "train_examples": 50000,
+      "train_loss": 0.16293619762420655
+    },
+    {
+      "epoch": 94,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 36754,
+      "train_examples": 50000,
+      "train_loss": 0.16091344168663024
+    },
+    {
+      "epoch": 95,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 37145,
+      "train_examples": 50000,
+      "train_loss": 0.15934230563163757
+    },
+    {
+      "epoch": 96,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 37536,
+      "train_examples": 50000,
+      "train_loss": 0.1541907616329193
+    },
+    {
+      "epoch": 97,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 37927,
+      "train_examples": 50000,
+      "train_loss": 0.15469486718654632
+    },
+    {
+      "epoch": 98,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 38318,
+      "train_examples": 50000,
+      "train_loss": 0.15900839382886886
+    },
+    {
+      "epoch": 99,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 38709,
+      "train_examples": 50000,
+      "train_loss": 0.16167540033102035
+    },
+    {
+      "epoch": 100,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 39100,
+      "train_examples": 50000,
+      "train_loss": 0.15679400232315063
+    },
+    {
+      "epoch": 101,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 39491,
+      "train_examples": 50000,
+      "train_loss": 0.08926333676099778
+    },
+    {
+      "epoch": 102,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 39882,
+      "train_examples": 50000,
+      "train_loss": 0.06358985314369202
+    },
+    {
+      "epoch": 103,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 40273,
+      "train_examples": 50000,
+      "train_loss": 0.05420353449106217
+    },
+    {
+      "epoch": 104,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 40664,
+      "train_examples": 50000,
+      "train_loss": 0.04906956180870533
+    },
+    {
+      "epoch": 105,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 41055,
+      "train_examples": 50000,
+      "train_loss": 0.045049789768457416
+    },
+    {
+      "epoch": 106,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 41446,
+      "train_examples": 50000,
+      "train_loss": 0.040599410170316694
+    },
+    {
+      "epoch": 107,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 41837,
+      "train_examples": 50000,
+      "train_loss": 0.03790038890838623
+    },
+    {
+      "epoch": 108,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 42228,
+      "train_examples": 50000,
+      "train_loss": 0.03717838701725006
+    },
+    {
+      "epoch": 109,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 42619,
+      "train_examples": 50000,
+      "train_loss": 0.03402421202838421
+    },
+    {
+      "epoch": 110,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 43010,
+      "train_examples": 50000,
+      "train_loss": 0.03068182231426239
+    },
+    {
+      "epoch": 111,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 43401,
+      "train_examples": 50000,
+      "train_loss": 0.02941229148209095
+    },
+    {
+      "epoch": 112,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 43792,
+      "train_examples": 50000,
+      "train_loss": 0.02811083570063114
+    },
+    {
+      "epoch": 113,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 44183,
+      "train_examples": 50000,
+      "train_loss": 0.02811160319328308
+    },
+    {
+      "epoch": 114,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 44574,
+      "train_examples": 50000,
+      "train_loss": 0.02578263300061226
+    },
+    {
+      "epoch": 115,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 44965,
+      "train_examples": 50000,
+      "train_loss": 0.024468044933080673
+    },
+    {
+      "epoch": 116,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 45356,
+      "train_examples": 50000,
+      "train_loss": 0.022521375725567342
+    },
+    {
+      "epoch": 117,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 45747,
+      "train_examples": 50000,
+      "train_loss": 0.0226251800596714
+    },
+    {
+      "epoch": 118,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 46138,
+      "train_examples": 50000,
+      "train_loss": 0.021669687507152556
+    },
+    {
+      "epoch": 119,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 46529,
+      "train_examples": 50000,
+      "train_loss": 0.020501725584864617
+    },
+    {
+      "epoch": 120,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 46920,
+      "train_examples": 50000,
+      "train_loss": 0.02019407486155629
+    },
+    {
+      "epoch": 121,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 47311,
+      "train_examples": 50000,
+      "train_loss": 0.01957855082988739
+    },
+    {
+      "epoch": 122,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 47702,
+      "train_examples": 50000,
+      "train_loss": 0.01867858926534653
+    },
+    {
+      "epoch": 123,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 48093,
+      "train_examples": 50000,
+      "train_loss": 0.017759973801076412
+    },
+    {
+      "epoch": 124,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 48484,
+      "train_examples": 50000,
+      "train_loss": 0.017850231526196005
+    },
+    {
+      "epoch": 125,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 48875,
+      "train_examples": 50000,
+      "train_loss": 0.016029231131076814
+    },
+    {
+      "epoch": 126,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 49266,
+      "train_examples": 50000,
+      "train_loss": 0.015987017872333528
+    },
+    {
+      "epoch": 127,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 49657,
+      "train_examples": 50000,
+      "train_loss": 0.016004691554009916
+    },
+    {
+      "epoch": 128,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 50048,
+      "train_examples": 50000,
+      "train_loss": 0.015084156606793403
+    },
+    {
+      "epoch": 129,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 50439,
+      "train_examples": 50000,
+      "train_loss": 0.015370317788422109
+    },
+    {
+      "epoch": 130,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 50830,
+      "train_examples": 50000,
+      "train_loss": 0.014356411390155553
+    },
+    {
+      "epoch": 131,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 51221,
+      "train_examples": 50000,
+      "train_loss": 0.013674242456257344
+    },
+    {
+      "epoch": 132,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 51612,
+      "train_examples": 50000,
+      "train_loss": 0.01331195755392313
+    },
+    {
+      "epoch": 133,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 52003,
+      "train_examples": 50000,
+      "train_loss": 0.013287956985011697
+    },
+    {
+      "epoch": 134,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 52394,
+      "train_examples": 50000,
+      "train_loss": 0.014606211445629597
+    },
+    {
+      "epoch": 135,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 52785,
+      "train_examples": 50000,
+      "train_loss": 0.013067925884127618
+    },
+    {
+      "epoch": 136,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 53176,
+      "train_examples": 50000,
+      "train_loss": 0.012662553620785475
+    },
+    {
+      "epoch": 137,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 53567,
+      "train_examples": 50000,
+      "train_loss": 0.012065714224725962
+    },
+    {
+      "epoch": 138,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 53958,
+      "train_examples": 50000,
+      "train_loss": 0.011524995899796485
+    },
+    {
+      "epoch": 139,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 54349,
+      "train_examples": 50000,
+      "train_loss": 0.011675733849853277
+    },
+    {
+      "epoch": 140,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 54740,
+      "train_examples": 50000,
+      "train_loss": 0.01266656897187233
+    },
+    {
+      "epoch": 141,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 55131,
+      "train_examples": 50000,
+      "train_loss": 0.010533184029757977
+    },
+    {
+      "epoch": 142,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 55522,
+      "train_examples": 50000,
+      "train_loss": 0.010978941943645477
+    },
+    {
+      "epoch": 143,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 55913,
+      "train_examples": 50000,
+      "train_loss": 0.010318347289264201
+    },
+    {
+      "epoch": 144,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 56304,
+      "train_examples": 50000,
+      "train_loss": 0.010467673452198505
+    },
+    {
+      "epoch": 145,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 56695,
+      "train_examples": 50000,
+      "train_loss": 0.010728884733617305
+    },
+    {
+      "epoch": 146,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 57086,
+      "train_examples": 50000,
+      "train_loss": 0.010494428095519542
+    },
+    {
+      "epoch": 147,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 57477,
+      "train_examples": 50000,
+      "train_loss": 0.010445828742682933
+    },
+    {
+      "epoch": 148,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 57868,
+      "train_examples": 50000,
+      "train_loss": 0.009801693223714828
+    },
+    {
+      "epoch": 149,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 58259,
+      "train_examples": 50000,
+      "train_loss": 0.010948296121805907
+    },
+    {
+      "epoch": 150,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 58650,
+      "train_examples": 50000,
+      "train_loss": 0.009674194239675998
+    },
+    {
+      "epoch": 151,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 59041,
+      "train_examples": 50000,
+      "train_loss": 0.008697000300362707
+    },
+    {
+      "epoch": 152,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 59432,
+      "train_examples": 50000,
+      "train_loss": 0.00895926752321422
+    },
+    {
+      "epoch": 153,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 59823,
+      "train_examples": 50000,
+      "train_loss": 0.008276405196115375
+    },
+    {
+      "epoch": 154,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 60214,
+      "train_examples": 50000,
+      "train_loss": 0.007976829496622085
+    },
+    {
+      "epoch": 155,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 60605,
+      "train_examples": 50000,
+      "train_loss": 0.0072691182130575184
+    },
+    {
+      "epoch": 156,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 60996,
+      "train_examples": 50000,
+      "train_loss": 0.007365617476552725
+    },
+    {
+      "epoch": 157,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 61387,
+      "train_examples": 50000,
+      "train_loss": 0.007068894384503365
+    },
+    {
+      "epoch": 158,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 61778,
+      "train_examples": 50000,
+      "train_loss": 0.006627400624677539
+    },
+    {
+      "epoch": 159,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 62169,
+      "train_examples": 50000,
+      "train_loss": 0.006712161379158497
+    },
+    {
+      "epoch": 160,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 62560,
+      "train_examples": 50000,
+      "train_loss": 0.006901890129819513
+    },
+    {
+      "epoch": 161,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 62951,
+      "train_examples": 50000,
+      "train_loss": 0.007002437942326069
+    },
+    {
+      "epoch": 162,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 63342,
+      "train_examples": 50000,
+      "train_loss": 0.007407158368527889
+    },
+    {
+      "epoch": 163,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 63733,
+      "train_examples": 50000,
+      "train_loss": 0.006923494810536504
+    },
+    {
+      "epoch": 164,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 64124,
+      "train_examples": 50000,
+      "train_loss": 0.006925872582197189
+    },
+    {
+      "epoch": 165,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 64515,
+      "train_examples": 50000,
+      "train_loss": 0.0067213862606883045
+    },
+    {
+      "epoch": 166,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 64906,
+      "train_examples": 50000,
+      "train_loss": 0.006998891127705574
+    },
+    {
+      "epoch": 167,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 65297,
+      "train_examples": 50000,
+      "train_loss": 0.006689337659217417
+    },
+    {
+      "epoch": 168,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 65688,
+      "train_examples": 50000,
+      "train_loss": 0.006929537190459668
+    },
+    {
+      "epoch": 169,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 66079,
+      "train_examples": 50000,
+      "train_loss": 0.006284183492511511
+    },
+    {
+      "epoch": 170,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 66470,
+      "train_examples": 50000,
+      "train_loss": 0.006663103199824691
+    },
+    {
+      "epoch": 171,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 66861,
+      "train_examples": 50000,
+      "train_loss": 0.006570300383940339
+    },
+    {
+      "epoch": 172,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 67252,
+      "train_examples": 50000,
+      "train_loss": 0.006503612499013543
+    },
+    {
+      "epoch": 173,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 67643,
+      "train_examples": 50000,
+      "train_loss": 0.006347955014780164
+    },
+    {
+      "epoch": 174,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 68034,
+      "train_examples": 50000,
+      "train_loss": 0.006226098643280565
+    },
+    {
+      "epoch": 175,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 68425,
+      "train_examples": 50000,
+      "train_loss": 0.0060284089615941045
+    },
+    {
+      "epoch": 176,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 68816,
+      "train_examples": 50000,
+      "train_loss": 0.006237343201041221
+    },
+    {
+      "epoch": 177,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 69207,
+      "train_examples": 50000,
+      "train_loss": 0.006058031805083156
+    },
+    {
+      "epoch": 178,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 69598,
+      "train_examples": 50000,
+      "train_loss": 0.0062861040268838405
+    },
+    {
+      "epoch": 179,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 69989,
+      "train_examples": 50000,
+      "train_loss": 0.005802006255611778
+    },
+    {
+      "epoch": 180,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 70380,
+      "train_examples": 50000,
+      "train_loss": 0.006082883627712726
+    },
+    {
+      "epoch": 181,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 70771,
+      "train_examples": 50000,
+      "train_loss": 0.006188844832703471
+    },
+    {
+      "epoch": 182,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 71162,
+      "train_examples": 50000,
+      "train_loss": 0.0057979475711286066
+    },
+    {
+      "epoch": 183,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 71553,
+      "train_examples": 50000,
+      "train_loss": 0.0063627321705967185
+    },
+    {
+      "epoch": 184,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 71944,
+      "train_examples": 50000,
+      "train_loss": 0.006196509729810059
+    },
+    {
+      "epoch": 185,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 72335,
+      "train_examples": 50000,
+      "train_loss": 0.00635628104865551
+    },
+    {
+      "epoch": 186,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 72726,
+      "train_examples": 50000,
+      "train_loss": 0.005718365462571383
+    },
+    {
+      "epoch": 187,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 73117,
+      "train_examples": 50000,
+      "train_loss": 0.006907654369175434
+    },
+    {
+      "epoch": 188,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 73508,
+      "train_examples": 50000,
+      "train_loss": 0.00584992028363049
+    },
+    {
+      "epoch": 189,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 73899,
+      "train_examples": 50000,
+      "train_loss": 0.005576714002341032
+    },
+    {
+      "epoch": 190,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 74290,
+      "train_examples": 50000,
+      "train_loss": 0.005562419883161783
+    },
+    {
+      "epoch": 191,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 74681,
+      "train_examples": 50000,
+      "train_loss": 0.006202604257687926
+    },
+    {
+      "epoch": 192,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 75072,
+      "train_examples": 50000,
+      "train_loss": 0.005533065688759088
+    },
+    {
+      "epoch": 193,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 75463,
+      "train_examples": 50000,
+      "train_loss": 0.005961296213939786
+    },
+    {
+      "epoch": 194,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 75854,
+      "train_examples": 50000,
+      "train_loss": 0.005992568579241633
+    },
+    {
+      "epoch": 195,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 76245,
+      "train_examples": 50000,
+      "train_loss": 0.006175138069316745
+    },
+    {
+      "epoch": 196,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 76636,
+      "train_examples": 50000,
+      "train_loss": 0.0058324388906359674
+    },
+    {
+      "epoch": 197,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 77027,
+      "train_examples": 50000,
+      "train_loss": 0.005814422636181116
+    },
+    {
+      "epoch": 198,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 77418,
+      "train_examples": 50000,
+      "train_loss": 0.006527215814217925
+    },
+    {
+      "epoch": 199,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 77809,
+      "train_examples": 50000,
+      "train_loss": 0.006064307056963444
+    },
+    {
+      "epoch": 200,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 78200,
+      "train_examples": 50000,
+      "train_loss": 0.005333981849998236
+    }
+  ],
+  "evaluation_protocol": {
+    "test_evaluations": 1,
+    "test_used_for_selection": false,
+    "validation_evaluations": 0
+  },
+  "final": {
+    "accuracy": 0.9262,
+    "epoch": 200,
+    "evaluation_split": "test",
+    "finite": true,
+    "loss": 0.3735280639648437,
+    "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": 13,
+    "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": 1.16585111618042,
+    "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": 5830.384622097015,
+    "train_wall_s": 5829.188536167145
+  },
+  "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
+  }
+}