results: confirm standard-depth SDIL scaling

1 files changed, 1986 insertions, 0 deletions

diff --git a/results/oral_a_dynamic_scaling_v2/bp_d32_s14.json b/results/oral_a_dynamic_scaling_v2/bp_d32_s14.json
new file mode 100644
index 0000000..ead9775
--- /dev/null
+++ b/results/oral_a_dynamic_scaling_v2/bp_d32_s14.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": 14,
+    "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_s14.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": 14,
+    "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": 1.9225823739242554
+    },
+    {
+      "epoch": 2,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 782,
+      "train_examples": 50000,
+      "train_loss": 1.4164960892868041
+    },
+    {
+      "epoch": 3,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 1173,
+      "train_examples": 50000,
+      "train_loss": 1.120760012626648
+    },
+    {
+      "epoch": 4,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 1564,
+      "train_examples": 50000,
+      "train_loss": 0.9091575496864319
+    },
+    {
+      "epoch": 5,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 1955,
+      "train_examples": 50000,
+      "train_loss": 0.7714499942970275
+    },
+    {
+      "epoch": 6,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 2346,
+      "train_examples": 50000,
+      "train_loss": 0.6778129623222351
+    },
+    {
+      "epoch": 7,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 2737,
+      "train_examples": 50000,
+      "train_loss": 0.6140075178146363
+    },
+    {
+      "epoch": 8,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 3128,
+      "train_examples": 50000,
+      "train_loss": 0.5595854544448853
+    },
+    {
+      "epoch": 9,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 3519,
+      "train_examples": 50000,
+      "train_loss": 0.5196639413070678
+    },
+    {
+      "epoch": 10,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 3910,
+      "train_examples": 50000,
+      "train_loss": 0.4925120655441284
+    },
+    {
+      "epoch": 11,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 4301,
+      "train_examples": 50000,
+      "train_loss": 0.46359059255599977
+    },
+    {
+      "epoch": 12,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 4692,
+      "train_examples": 50000,
+      "train_loss": 0.4498627822494507
+    },
+    {
+      "epoch": 13,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 5083,
+      "train_examples": 50000,
+      "train_loss": 0.4269271025943756
+    },
+    {
+      "epoch": 14,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 5474,
+      "train_examples": 50000,
+      "train_loss": 0.40440647114753725
+    },
+    {
+      "epoch": 15,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 5865,
+      "train_examples": 50000,
+      "train_loss": 0.38960297353744505
+    },
+    {
+      "epoch": 16,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 6256,
+      "train_examples": 50000,
+      "train_loss": 0.3763126535797119
+    },
+    {
+      "epoch": 17,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 6647,
+      "train_examples": 50000,
+      "train_loss": 0.366042610244751
+    },
+    {
+      "epoch": 18,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 7038,
+      "train_examples": 50000,
+      "train_loss": 0.3508842104625702
+    },
+    {
+      "epoch": 19,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 7429,
+      "train_examples": 50000,
+      "train_loss": 0.34554700902938845
+    },
+    {
+      "epoch": 20,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 7820,
+      "train_examples": 50000,
+      "train_loss": 0.3303641291618347
+    },
+    {
+      "epoch": 21,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 8211,
+      "train_examples": 50000,
+      "train_loss": 0.3270281967258453
+    },
+    {
+      "epoch": 22,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 8602,
+      "train_examples": 50000,
+      "train_loss": 0.31568570635795595
+    },
+    {
+      "epoch": 23,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 8993,
+      "train_examples": 50000,
+      "train_loss": 0.3150420280361175
+    },
+    {
+      "epoch": 24,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 9384,
+      "train_examples": 50000,
+      "train_loss": 0.30107805528640746
+    },
+    {
+      "epoch": 25,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 9775,
+      "train_examples": 50000,
+      "train_loss": 0.2947032307434082
+    },
+    {
+      "epoch": 26,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 10166,
+      "train_examples": 50000,
+      "train_loss": 0.294678018283844
+    },
+    {
+      "epoch": 27,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 10557,
+      "train_examples": 50000,
+      "train_loss": 0.2859807000780106
+    },
+    {
+      "epoch": 28,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 10948,
+      "train_examples": 50000,
+      "train_loss": 0.2766690892982483
+    },
+    {
+      "epoch": 29,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 11339,
+      "train_examples": 50000,
+      "train_loss": 0.27374145025253294
+    },
+    {
+      "epoch": 30,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 11730,
+      "train_examples": 50000,
+      "train_loss": 0.2638592168903351
+    },
+    {
+      "epoch": 31,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 12121,
+      "train_examples": 50000,
+      "train_loss": 0.2615679057645798
+    },
+    {
+      "epoch": 32,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 12512,
+      "train_examples": 50000,
+      "train_loss": 0.25975301442146304
+    },
+    {
+      "epoch": 33,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 12903,
+      "train_examples": 50000,
+      "train_loss": 0.2502034167432785
+    },
+    {
+      "epoch": 34,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 13294,
+      "train_examples": 50000,
+      "train_loss": 0.2520859577322006
+    },
+    {
+      "epoch": 35,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 13685,
+      "train_examples": 50000,
+      "train_loss": 0.24862939637184142
+    },
+    {
+      "epoch": 36,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 14076,
+      "train_examples": 50000,
+      "train_loss": 0.244263690366745
+    },
+    {
+      "epoch": 37,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 14467,
+      "train_examples": 50000,
+      "train_loss": 0.24465613820075988
+    },
+    {
+      "epoch": 38,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 14858,
+      "train_examples": 50000,
+      "train_loss": 0.24179527607917786
+    },
+    {
+      "epoch": 39,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 15249,
+      "train_examples": 50000,
+      "train_loss": 0.23272236390113832
+    },
+    {
+      "epoch": 40,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 15640,
+      "train_examples": 50000,
+      "train_loss": 0.22882037929534912
+    },
+    {
+      "epoch": 41,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 16031,
+      "train_examples": 50000,
+      "train_loss": 0.22537988799095154
+    },
+    {
+      "epoch": 42,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 16422,
+      "train_examples": 50000,
+      "train_loss": 0.22588210765838623
+    },
+    {
+      "epoch": 43,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 16813,
+      "train_examples": 50000,
+      "train_loss": 0.22358049662590027
+    },
+    {
+      "epoch": 44,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 17204,
+      "train_examples": 50000,
+      "train_loss": 0.22093470664978027
+    },
+    {
+      "epoch": 45,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 17595,
+      "train_examples": 50000,
+      "train_loss": 0.21793900159358978
+    },
+    {
+      "epoch": 46,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 17986,
+      "train_examples": 50000,
+      "train_loss": 0.21753683715820313
+    },
+    {
+      "epoch": 47,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 18377,
+      "train_examples": 50000,
+      "train_loss": 0.21588620354175567
+    },
+    {
+      "epoch": 48,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 18768,
+      "train_examples": 50000,
+      "train_loss": 0.21618832901000976
+    },
+    {
+      "epoch": 49,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 19159,
+      "train_examples": 50000,
+      "train_loss": 0.21097279445171357
+    },
+    {
+      "epoch": 50,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 19550,
+      "train_examples": 50000,
+      "train_loss": 0.20663198817253112
+    },
+    {
+      "epoch": 51,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 19941,
+      "train_examples": 50000,
+      "train_loss": 0.20597183012962342
+    },
+    {
+      "epoch": 52,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 20332,
+      "train_examples": 50000,
+      "train_loss": 0.19994518122673036
+    },
+    {
+      "epoch": 53,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 20723,
+      "train_examples": 50000,
+      "train_loss": 0.20113924821853638
+    },
+    {
+      "epoch": 54,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 21114,
+      "train_examples": 50000,
+      "train_loss": 0.19977485011100768
+    },
+    {
+      "epoch": 55,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 21505,
+      "train_examples": 50000,
+      "train_loss": 0.1999194315648079
+    },
+    {
+      "epoch": 56,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 21896,
+      "train_examples": 50000,
+      "train_loss": 0.19770239804267883
+    },
+    {
+      "epoch": 57,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 22287,
+      "train_examples": 50000,
+      "train_loss": 0.20051282331466674
+    },
+    {
+      "epoch": 58,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 22678,
+      "train_examples": 50000,
+      "train_loss": 0.19370327817440033
+    },
+    {
+      "epoch": 59,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 23069,
+      "train_examples": 50000,
+      "train_loss": 0.19730971799850464
+    },
+    {
+      "epoch": 60,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 23460,
+      "train_examples": 50000,
+      "train_loss": 0.19116241208076476
+    },
+    {
+      "epoch": 61,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 23851,
+      "train_examples": 50000,
+      "train_loss": 0.19146727406024933
+    },
+    {
+      "epoch": 62,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 24242,
+      "train_examples": 50000,
+      "train_loss": 0.1893305285835266
+    },
+    {
+      "epoch": 63,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 24633,
+      "train_examples": 50000,
+      "train_loss": 0.18181477528095244
+    },
+    {
+      "epoch": 64,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 25024,
+      "train_examples": 50000,
+      "train_loss": 0.1908838845062256
+    },
+    {
+      "epoch": 65,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 25415,
+      "train_examples": 50000,
+      "train_loss": 0.18901293563365937
+    },
+    {
+      "epoch": 66,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 25806,
+      "train_examples": 50000,
+      "train_loss": 0.1834484317302704
+    },
+    {
+      "epoch": 67,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 26197,
+      "train_examples": 50000,
+      "train_loss": 0.1857504557991028
+    },
+    {
+      "epoch": 68,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 26588,
+      "train_examples": 50000,
+      "train_loss": 0.18426053916454316
+    },
+    {
+      "epoch": 69,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 26979,
+      "train_examples": 50000,
+      "train_loss": 0.1842867259001732
+    },
+    {
+      "epoch": 70,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 27370,
+      "train_examples": 50000,
+      "train_loss": 0.17477602120399474
+    },
+    {
+      "epoch": 71,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 27761,
+      "train_examples": 50000,
+      "train_loss": 0.17921741899013519
+    },
+    {
+      "epoch": 72,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 28152,
+      "train_examples": 50000,
+      "train_loss": 0.17559958313941956
+    },
+    {
+      "epoch": 73,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 28543,
+      "train_examples": 50000,
+      "train_loss": 0.1802306512594223
+    },
+    {
+      "epoch": 74,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 28934,
+      "train_examples": 50000,
+      "train_loss": 0.17615691054821014
+    },
+    {
+      "epoch": 75,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 29325,
+      "train_examples": 50000,
+      "train_loss": 0.17121027718544007
+    },
+    {
+      "epoch": 76,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 29716,
+      "train_examples": 50000,
+      "train_loss": 0.17714182065963746
+    },
+    {
+      "epoch": 77,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 30107,
+      "train_examples": 50000,
+      "train_loss": 0.1763121262741089
+    },
+    {
+      "epoch": 78,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 30498,
+      "train_examples": 50000,
+      "train_loss": 0.17495869878768922
+    },
+    {
+      "epoch": 79,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 30889,
+      "train_examples": 50000,
+      "train_loss": 0.17348072474956514
+    },
+    {
+      "epoch": 80,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 31280,
+      "train_examples": 50000,
+      "train_loss": 0.17383595414400102
+    },
+    {
+      "epoch": 81,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 31671,
+      "train_examples": 50000,
+      "train_loss": 0.1727399858903885
+    },
+    {
+      "epoch": 82,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 32062,
+      "train_examples": 50000,
+      "train_loss": 0.1734715978717804
+    },
+    {
+      "epoch": 83,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 32453,
+      "train_examples": 50000,
+      "train_loss": 0.1673349199104309
+    },
+    {
+      "epoch": 84,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 32844,
+      "train_examples": 50000,
+      "train_loss": 0.16863740804672242
+    },
+    {
+      "epoch": 85,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 33235,
+      "train_examples": 50000,
+      "train_loss": 0.16489555994987487
+    },
+    {
+      "epoch": 86,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 33626,
+      "train_examples": 50000,
+      "train_loss": 0.1702830473613739
+    },
+    {
+      "epoch": 87,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 34017,
+      "train_examples": 50000,
+      "train_loss": 0.16995898442745208
+    },
+    {
+      "epoch": 88,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 34408,
+      "train_examples": 50000,
+      "train_loss": 0.1719565958404541
+    },
+    {
+      "epoch": 89,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 34799,
+      "train_examples": 50000,
+      "train_loss": 0.16844428910255432
+    },
+    {
+      "epoch": 90,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 35190,
+      "train_examples": 50000,
+      "train_loss": 0.16289416341781615
+    },
+    {
+      "epoch": 91,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 35581,
+      "train_examples": 50000,
+      "train_loss": 0.1642383770608902
+    },
+    {
+      "epoch": 92,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 35972,
+      "train_examples": 50000,
+      "train_loss": 0.16417108777999878
+    },
+    {
+      "epoch": 93,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 36363,
+      "train_examples": 50000,
+      "train_loss": 0.1624667575740814
+    },
+    {
+      "epoch": 94,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 36754,
+      "train_examples": 50000,
+      "train_loss": 0.16282382113456725
+    },
+    {
+      "epoch": 95,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 37145,
+      "train_examples": 50000,
+      "train_loss": 0.16381212590694427
+    },
+    {
+      "epoch": 96,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 37536,
+      "train_examples": 50000,
+      "train_loss": 0.15981725409269332
+    },
+    {
+      "epoch": 97,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 37927,
+      "train_examples": 50000,
+      "train_loss": 0.15901209310531617
+    },
+    {
+      "epoch": 98,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 38318,
+      "train_examples": 50000,
+      "train_loss": 0.16209107344150545
+    },
+    {
+      "epoch": 99,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 38709,
+      "train_examples": 50000,
+      "train_loss": 0.16062442711353303
+    },
+    {
+      "epoch": 100,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 39100,
+      "train_examples": 50000,
+      "train_loss": 0.16109401269435883
+    },
+    {
+      "epoch": 101,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 39491,
+      "train_examples": 50000,
+      "train_loss": 0.09330496985912323
+    },
+    {
+      "epoch": 102,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 39882,
+      "train_examples": 50000,
+      "train_loss": 0.06679853103995323
+    },
+    {
+      "epoch": 103,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 40273,
+      "train_examples": 50000,
+      "train_loss": 0.05833959693670273
+    },
+    {
+      "epoch": 104,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 40664,
+      "train_examples": 50000,
+      "train_loss": 0.05145497631549835
+    },
+    {
+      "epoch": 105,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 41055,
+      "train_examples": 50000,
+      "train_loss": 0.04714090671777725
+    },
+    {
+      "epoch": 106,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 41446,
+      "train_examples": 50000,
+      "train_loss": 0.04397652909398079
+    },
+    {
+      "epoch": 107,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 41837,
+      "train_examples": 50000,
+      "train_loss": 0.03884652088701725
+    },
+    {
+      "epoch": 108,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 42228,
+      "train_examples": 50000,
+      "train_loss": 0.03708200654268265
+    },
+    {
+      "epoch": 109,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 42619,
+      "train_examples": 50000,
+      "train_loss": 0.03475045755326748
+    },
+    {
+      "epoch": 110,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 43010,
+      "train_examples": 50000,
+      "train_loss": 0.03192536900877953
+    },
+    {
+      "epoch": 111,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 43401,
+      "train_examples": 50000,
+      "train_loss": 0.03199194766640663
+    },
+    {
+      "epoch": 112,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 43792,
+      "train_examples": 50000,
+      "train_loss": 0.0298711553555727
+    },
+    {
+      "epoch": 113,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 44183,
+      "train_examples": 50000,
+      "train_loss": 0.0275455548286438
+    },
+    {
+      "epoch": 114,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 44574,
+      "train_examples": 50000,
+      "train_loss": 0.027080157968699933
+    },
+    {
+      "epoch": 115,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 44965,
+      "train_examples": 50000,
+      "train_loss": 0.025380560950636864
+    },
+    {
+      "epoch": 116,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 45356,
+      "train_examples": 50000,
+      "train_loss": 0.02474314520716667
+    },
+    {
+      "epoch": 117,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 45747,
+      "train_examples": 50000,
+      "train_loss": 0.023645476237237454
+    },
+    {
+      "epoch": 118,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 46138,
+      "train_examples": 50000,
+      "train_loss": 0.02262053836196661
+    },
+    {
+      "epoch": 119,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 46529,
+      "train_examples": 50000,
+      "train_loss": 0.021375253983438015
+    },
+    {
+      "epoch": 120,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 46920,
+      "train_examples": 50000,
+      "train_loss": 0.02164223212003708
+    },
+    {
+      "epoch": 121,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 47311,
+      "train_examples": 50000,
+      "train_loss": 0.021175485536307097
+    },
+    {
+      "epoch": 122,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 47702,
+      "train_examples": 50000,
+      "train_loss": 0.020002913225889207
+    },
+    {
+      "epoch": 123,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 48093,
+      "train_examples": 50000,
+      "train_loss": 0.018995866283774376
+    },
+    {
+      "epoch": 124,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 48484,
+      "train_examples": 50000,
+      "train_loss": 0.01745662123799324
+    },
+    {
+      "epoch": 125,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 48875,
+      "train_examples": 50000,
+      "train_loss": 0.018009458032250403
+    },
+    {
+      "epoch": 126,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 49266,
+      "train_examples": 50000,
+      "train_loss": 0.017360723948776722
+    },
+    {
+      "epoch": 127,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 49657,
+      "train_examples": 50000,
+      "train_loss": 0.01653353522092104
+    },
+    {
+      "epoch": 128,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 50048,
+      "train_examples": 50000,
+      "train_loss": 0.016291869946420193
+    },
+    {
+      "epoch": 129,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 50439,
+      "train_examples": 50000,
+      "train_loss": 0.015165688739567995
+    },
+    {
+      "epoch": 130,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 50830,
+      "train_examples": 50000,
+      "train_loss": 0.014390518234670161
+    },
+    {
+      "epoch": 131,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 51221,
+      "train_examples": 50000,
+      "train_loss": 0.014876080607920885
+    },
+    {
+      "epoch": 132,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 51612,
+      "train_examples": 50000,
+      "train_loss": 0.01354152322858572
+    },
+    {
+      "epoch": 133,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 52003,
+      "train_examples": 50000,
+      "train_loss": 0.013892504596710206
+    },
+    {
+      "epoch": 134,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 52394,
+      "train_examples": 50000,
+      "train_loss": 0.01305307987049222
+    },
+    {
+      "epoch": 135,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 52785,
+      "train_examples": 50000,
+      "train_loss": 0.013101011054515839
+    },
+    {
+      "epoch": 136,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 53176,
+      "train_examples": 50000,
+      "train_loss": 0.011781157428771257
+    },
+    {
+      "epoch": 137,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 53567,
+      "train_examples": 50000,
+      "train_loss": 0.013396826519072055
+    },
+    {
+      "epoch": 138,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 53958,
+      "train_examples": 50000,
+      "train_loss": 0.012641742394864559
+    },
+    {
+      "epoch": 139,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 54349,
+      "train_examples": 50000,
+      "train_loss": 0.012545095380246639
+    },
+    {
+      "epoch": 140,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 54740,
+      "train_examples": 50000,
+      "train_loss": 0.01211236124843359
+    },
+    {
+      "epoch": 141,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 55131,
+      "train_examples": 50000,
+      "train_loss": 0.012181379710882902
+    },
+    {
+      "epoch": 142,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 55522,
+      "train_examples": 50000,
+      "train_loss": 0.011682115127891302
+    },
+    {
+      "epoch": 143,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 55913,
+      "train_examples": 50000,
+      "train_loss": 0.012197788713127374
+    },
+    {
+      "epoch": 144,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 56304,
+      "train_examples": 50000,
+      "train_loss": 0.011751104683354496
+    },
+    {
+      "epoch": 145,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 56695,
+      "train_examples": 50000,
+      "train_loss": 0.010575061148554086
+    },
+    {
+      "epoch": 146,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 57086,
+      "train_examples": 50000,
+      "train_loss": 0.010539636700376867
+    },
+    {
+      "epoch": 147,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 57477,
+      "train_examples": 50000,
+      "train_loss": 0.010397719995826483
+    },
+    {
+      "epoch": 148,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 57868,
+      "train_examples": 50000,
+      "train_loss": 0.009707565153241157
+    },
+    {
+      "epoch": 149,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 58259,
+      "train_examples": 50000,
+      "train_loss": 0.010903430490493774
+    },
+    {
+      "epoch": 150,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 58650,
+      "train_examples": 50000,
+      "train_loss": 0.010372428714781999
+    },
+    {
+      "epoch": 151,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 59041,
+      "train_examples": 50000,
+      "train_loss": 0.008965822239741683
+    },
+    {
+      "epoch": 152,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 59432,
+      "train_examples": 50000,
+      "train_loss": 0.008958354243934154
+    },
+    {
+      "epoch": 153,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 59823,
+      "train_examples": 50000,
+      "train_loss": 0.008743972501903773
+    },
+    {
+      "epoch": 154,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 60214,
+      "train_examples": 50000,
+      "train_loss": 0.008012467796206475
+    },
+    {
+      "epoch": 155,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 60605,
+      "train_examples": 50000,
+      "train_loss": 0.007939136294722557
+    },
+    {
+      "epoch": 156,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 60996,
+      "train_examples": 50000,
+      "train_loss": 0.00772694343701005
+    },
+    {
+      "epoch": 157,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 61387,
+      "train_examples": 50000,
+      "train_loss": 0.007006239338219165
+    },
+    {
+      "epoch": 158,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 61778,
+      "train_examples": 50000,
+      "train_loss": 0.008014692026376724
+    },
+    {
+      "epoch": 159,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 62169,
+      "train_examples": 50000,
+      "train_loss": 0.008064465202838182
+    },
+    {
+      "epoch": 160,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 62560,
+      "train_examples": 50000,
+      "train_loss": 0.0075252782972157
+    },
+    {
+      "epoch": 161,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 62951,
+      "train_examples": 50000,
+      "train_loss": 0.00778471342086792
+    },
+    {
+      "epoch": 162,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 63342,
+      "train_examples": 50000,
+      "train_loss": 0.006732449814379215
+    },
+    {
+      "epoch": 163,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 63733,
+      "train_examples": 50000,
+      "train_loss": 0.007805665935575962
+    },
+    {
+      "epoch": 164,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 64124,
+      "train_examples": 50000,
+      "train_loss": 0.006874608674496413
+    },
+    {
+      "epoch": 165,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 64515,
+      "train_examples": 50000,
+      "train_loss": 0.007625447829514742
+    },
+    {
+      "epoch": 166,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 64906,
+      "train_examples": 50000,
+      "train_loss": 0.007315959031283855
+    },
+    {
+      "epoch": 167,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 65297,
+      "train_examples": 50000,
+      "train_loss": 0.007561641323417425
+    },
+    {
+      "epoch": 168,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 65688,
+      "train_examples": 50000,
+      "train_loss": 0.007466862306669354
+    },
+    {
+      "epoch": 169,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 66079,
+      "train_examples": 50000,
+      "train_loss": 0.007192313821911812
+    },
+    {
+      "epoch": 170,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 66470,
+      "train_examples": 50000,
+      "train_loss": 0.00749102226883173
+    },
+    {
+      "epoch": 171,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 66861,
+      "train_examples": 50000,
+      "train_loss": 0.006988092204034328
+    },
+    {
+      "epoch": 172,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 67252,
+      "train_examples": 50000,
+      "train_loss": 0.0068182036101073025
+    },
+    {
+      "epoch": 173,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 67643,
+      "train_examples": 50000,
+      "train_loss": 0.005724681941866875
+    },
+    {
+      "epoch": 174,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 68034,
+      "train_examples": 50000,
+      "train_loss": 0.007182316865585745
+    },
+    {
+      "epoch": 175,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 68425,
+      "train_examples": 50000,
+      "train_loss": 0.006710217799684033
+    },
+    {
+      "epoch": 176,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 68816,
+      "train_examples": 50000,
+      "train_loss": 0.006433702199906111
+    },
+    {
+      "epoch": 177,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 69207,
+      "train_examples": 50000,
+      "train_loss": 0.006969278431907296
+    },
+    {
+      "epoch": 178,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 69598,
+      "train_examples": 50000,
+      "train_loss": 0.0060771925245970485
+    },
+    {
+      "epoch": 179,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 69989,
+      "train_examples": 50000,
+      "train_loss": 0.0069215068834275004
+    },
+    {
+      "epoch": 180,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 70380,
+      "train_examples": 50000,
+      "train_loss": 0.006596071783900261
+    },
+    {
+      "epoch": 181,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 70771,
+      "train_examples": 50000,
+      "train_loss": 0.006939981128051877
+    },
+    {
+      "epoch": 182,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 71162,
+      "train_examples": 50000,
+      "train_loss": 0.006849895872250199
+    },
+    {
+      "epoch": 183,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 71553,
+      "train_examples": 50000,
+      "train_loss": 0.006535437549203635
+    },
+    {
+      "epoch": 184,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 71944,
+      "train_examples": 50000,
+      "train_loss": 0.006412262945920229
+    },
+    {
+      "epoch": 185,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 72335,
+      "train_examples": 50000,
+      "train_loss": 0.006494907861351967
+    },
+    {
+      "epoch": 186,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 72726,
+      "train_examples": 50000,
+      "train_loss": 0.006900249844565987
+    },
+    {
+      "epoch": 187,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 73117,
+      "train_examples": 50000,
+      "train_loss": 0.0065344038428366185
+    },
+    {
+      "epoch": 188,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 73508,
+      "train_examples": 50000,
+      "train_loss": 0.005925394353866577
+    },
+    {
+      "epoch": 189,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 73899,
+      "train_examples": 50000,
+      "train_loss": 0.0064781298804283145
+    },
+    {
+      "epoch": 190,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 74290,
+      "train_examples": 50000,
+      "train_loss": 0.006599408498555422
+    },
+    {
+      "epoch": 191,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 74681,
+      "train_examples": 50000,
+      "train_loss": 0.006264180855304002
+    },
+    {
+      "epoch": 192,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 75072,
+      "train_examples": 50000,
+      "train_loss": 0.005788764647431672
+    },
+    {
+      "epoch": 193,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 75463,
+      "train_examples": 50000,
+      "train_loss": 0.006316005318537355
+    },
+    {
+      "epoch": 194,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 75854,
+      "train_examples": 50000,
+      "train_loss": 0.005849055158644915
+    },
+    {
+      "epoch": 195,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 76245,
+      "train_examples": 50000,
+      "train_loss": 0.006100576285738498
+    },
+    {
+      "epoch": 196,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 76636,
+      "train_examples": 50000,
+      "train_loss": 0.006544416576102376
+    },
+    {
+      "epoch": 197,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 77027,
+      "train_examples": 50000,
+      "train_loss": 0.006031918609589338
+    },
+    {
+      "epoch": 198,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 77418,
+      "train_examples": 50000,
+      "train_loss": 0.006268274684920907
+    },
+    {
+      "epoch": 199,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 77809,
+      "train_examples": 50000,
+      "train_loss": 0.005552465254440904
+    },
+    {
+      "epoch": 200,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 78200,
+      "train_examples": 50000,
+      "train_loss": 0.005661729123815894
+    }
+  ],
+  "evaluation_protocol": {
+    "test_evaluations": 1,
+    "test_used_for_selection": false,
+    "validation_evaluations": 0
+  },
+  "final": {
+    "accuracy": 0.9216,
+    "epoch": 200,
+    "evaluation_split": "test",
+    "finite": true,
+    "loss": 0.3806199676513672,
+    "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": 14,
+    "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": 1.1587669849395752,
+    "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": 5810.336062431335,
+    "train_wall_s": 5809.14767408371
+  },
+  "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
+  }
+}