results: confirm standard-depth SDIL scaling

1 files changed, 2106 insertions, 0 deletions

diff --git a/results/oral_a_dynamic_scaling_v2/bp_d56_s11.json b/results/oral_a_dynamic_scaling_v2/bp_d56_s11.json
new file mode 100644
index 0000000..e8e6c4a
--- /dev/null
+++ b/results/oral_a_dynamic_scaling_v2/bp_d56_s11.json
@@ -0,0 +1,2106 @@
+{
+  "architecture": {
+    "adaptive_apical_parameters": 0,
+    "adaptive_feedback_parameters": 0,
+    "base_width": 16,
+    "blocks_per_stage": 9,
+    "bn_eps": 1e-05,
+    "bn_momentum": 0.1,
+    "depth": 56,
+    "family": "CIFAR 6n+2 ResNet, option-A shortcuts",
+    "fixed_feedback_parameters": 0,
+    "fixed_traffic_coefficients": 0,
+    "forward_parameters": 853018,
+    "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
+      ],
+      [
+        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
+      ],
+      [
+        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
+      ],
+      [
+        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": 56,
+    "device": "cuda",
+    "epochs": 200,
+    "eta_A": 0.01,
+    "eta_P": 0.01,
+    "eval_every": 0,
+    "eval_split": "test",
+    "learn_P": 0,
+    "loader_seed": 11,
+    "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_d56_s11.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": 11,
+    "split_seed": 2027,
+    "traffic_calibration_examples": 64,
+    "traffic_ratio": 4.0,
+    "traffic_rule": "innovation",
+    "traffic_seed": 4000,
+    "train_limit": 0,
+    "use_residual": 1,
+    "val_examples": 0,
+    "vectorizer_mode": "spatial_template",
+    "warmup_epochs": 0,
+    "weight_decay": 0.0001,
+    "weight_scale": 1.0,
+    "width": 16
+  },
+  "calibration_metric_space": null,
+  "counters": {
+    "apical_warmup_examples": 0,
+    "calibration_event_examples": 0,
+    "causal_scalar_observations": 0,
+    "logical_batch_loss_queries": 0,
+    "mirror_conv_examples": 0,
+    "mirror_events": 0,
+    "mirror_readout_examples": 0,
+    "neutral_projection_examples": 0,
+    "ordinary_examples": 10000000,
+    "per_example_loss_terms": 0,
+    "perturbation_events": 0,
+    "perturbation_forward_examples": 0,
+    "predictor_update_examples": 0,
+    "predictor_warmup_examples": 0,
+    "traffic_audit_examples": 0,
+    "traffic_calibration_examples": 0
+  },
+  "diagnostics": null,
+  "epochs": [
+    {
+      "epoch": 1,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 391,
+      "train_examples": 50000,
+      "train_loss": 2.2441480759429933
+    },
+    {
+      "epoch": 2,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 782,
+      "train_examples": 50000,
+      "train_loss": 1.6520451573562622
+    },
+    {
+      "epoch": 3,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 1173,
+      "train_examples": 50000,
+      "train_loss": 1.2762053594589233
+    },
+    {
+      "epoch": 4,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 1564,
+      "train_examples": 50000,
+      "train_loss": 1.0366795889663696
+    },
+    {
+      "epoch": 5,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 1955,
+      "train_examples": 50000,
+      "train_loss": 0.8838742491340638
+    },
+    {
+      "epoch": 6,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 2346,
+      "train_examples": 50000,
+      "train_loss": 0.7775179610824585
+    },
+    {
+      "epoch": 7,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 2737,
+      "train_examples": 50000,
+      "train_loss": 0.7004733442497253
+    },
+    {
+      "epoch": 8,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 3128,
+      "train_examples": 50000,
+      "train_loss": 0.6399433501815796
+    },
+    {
+      "epoch": 9,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 3519,
+      "train_examples": 50000,
+      "train_loss": 0.5957588003158569
+    },
+    {
+      "epoch": 10,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 3910,
+      "train_examples": 50000,
+      "train_loss": 0.5524389961814881
+    },
+    {
+      "epoch": 11,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 4301,
+      "train_examples": 50000,
+      "train_loss": 0.5249286603546143
+    },
+    {
+      "epoch": 12,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 4692,
+      "train_examples": 50000,
+      "train_loss": 0.4941679028701782
+    },
+    {
+      "epoch": 13,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 5083,
+      "train_examples": 50000,
+      "train_loss": 0.4750888855743408
+    },
+    {
+      "epoch": 14,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 5474,
+      "train_examples": 50000,
+      "train_loss": 0.45399185165405276
+    },
+    {
+      "epoch": 15,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 5865,
+      "train_examples": 50000,
+      "train_loss": 0.43263239362716677
+    },
+    {
+      "epoch": 16,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 6256,
+      "train_examples": 50000,
+      "train_loss": 0.41556921830177307
+    },
+    {
+      "epoch": 17,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 6647,
+      "train_examples": 50000,
+      "train_loss": 0.3966397378349304
+    },
+    {
+      "epoch": 18,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 7038,
+      "train_examples": 50000,
+      "train_loss": 0.3875173846626282
+    },
+    {
+      "epoch": 19,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 7429,
+      "train_examples": 50000,
+      "train_loss": 0.3735868306541443
+    },
+    {
+      "epoch": 20,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 7820,
+      "train_examples": 50000,
+      "train_loss": 0.358007550201416
+    },
+    {
+      "epoch": 21,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 8211,
+      "train_examples": 50000,
+      "train_loss": 0.3514554388427734
+    },
+    {
+      "epoch": 22,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 8602,
+      "train_examples": 50000,
+      "train_loss": 0.3362429839515686
+    },
+    {
+      "epoch": 23,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 8993,
+      "train_examples": 50000,
+      "train_loss": 0.328383513584137
+    },
+    {
+      "epoch": 24,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 9384,
+      "train_examples": 50000,
+      "train_loss": 0.324324482755661
+    },
+    {
+      "epoch": 25,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 9775,
+      "train_examples": 50000,
+      "train_loss": 0.3128098294448853
+    },
+    {
+      "epoch": 26,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 10166,
+      "train_examples": 50000,
+      "train_loss": 0.30343968832015994
+    },
+    {
+      "epoch": 27,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 10557,
+      "train_examples": 50000,
+      "train_loss": 0.2968874151802063
+    },
+    {
+      "epoch": 28,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 10948,
+      "train_examples": 50000,
+      "train_loss": 0.29214453186035155
+    },
+    {
+      "epoch": 29,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 11339,
+      "train_examples": 50000,
+      "train_loss": 0.2794385500907898
+    },
+    {
+      "epoch": 30,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 11730,
+      "train_examples": 50000,
+      "train_loss": 0.2844532305288315
+    },
+    {
+      "epoch": 31,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 12121,
+      "train_examples": 50000,
+      "train_loss": 0.27004383907318114
+    },
+    {
+      "epoch": 32,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 12512,
+      "train_examples": 50000,
+      "train_loss": 0.2688684576606751
+    },
+    {
+      "epoch": 33,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 12903,
+      "train_examples": 50000,
+      "train_loss": 0.2632033786582947
+    },
+    {
+      "epoch": 34,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 13294,
+      "train_examples": 50000,
+      "train_loss": 0.25779811316013335
+    },
+    {
+      "epoch": 35,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 13685,
+      "train_examples": 50000,
+      "train_loss": 0.2536117401123047
+    },
+    {
+      "epoch": 36,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 14076,
+      "train_examples": 50000,
+      "train_loss": 0.24904733840465546
+    },
+    {
+      "epoch": 37,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 14467,
+      "train_examples": 50000,
+      "train_loss": 0.24340575067520143
+    },
+    {
+      "epoch": 38,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 14858,
+      "train_examples": 50000,
+      "train_loss": 0.24684725905418395
+    },
+    {
+      "epoch": 39,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 15249,
+      "train_examples": 50000,
+      "train_loss": 0.23614188342094422
+    },
+    {
+      "epoch": 40,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 15640,
+      "train_examples": 50000,
+      "train_loss": 0.23577110658168793
+    },
+    {
+      "epoch": 41,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 16031,
+      "train_examples": 50000,
+      "train_loss": 0.23184648150920867
+    },
+    {
+      "epoch": 42,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 16422,
+      "train_examples": 50000,
+      "train_loss": 0.22659937084197998
+    },
+    {
+      "epoch": 43,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 16813,
+      "train_examples": 50000,
+      "train_loss": 0.22489133757591248
+    },
+    {
+      "epoch": 44,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 17204,
+      "train_examples": 50000,
+      "train_loss": 0.22232432385444642
+    },
+    {
+      "epoch": 45,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 17595,
+      "train_examples": 50000,
+      "train_loss": 0.21497296612262726
+    },
+    {
+      "epoch": 46,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 17986,
+      "train_examples": 50000,
+      "train_loss": 0.213139944896698
+    },
+    {
+      "epoch": 47,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 18377,
+      "train_examples": 50000,
+      "train_loss": 0.21386680399894714
+    },
+    {
+      "epoch": 48,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 18768,
+      "train_examples": 50000,
+      "train_loss": 0.21408246718883514
+    },
+    {
+      "epoch": 49,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 19159,
+      "train_examples": 50000,
+      "train_loss": 0.21116155992031096
+    },
+    {
+      "epoch": 50,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 19550,
+      "train_examples": 50000,
+      "train_loss": 0.2054360934305191
+    },
+    {
+      "epoch": 51,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 19941,
+      "train_examples": 50000,
+      "train_loss": 0.20042337512493133
+    },
+    {
+      "epoch": 52,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 20332,
+      "train_examples": 50000,
+      "train_loss": 0.20099475722312926
+    },
+    {
+      "epoch": 53,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 20723,
+      "train_examples": 50000,
+      "train_loss": 0.20083012888431548
+    },
+    {
+      "epoch": 54,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 21114,
+      "train_examples": 50000,
+      "train_loss": 0.1945902212357521
+    },
+    {
+      "epoch": 55,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 21505,
+      "train_examples": 50000,
+      "train_loss": 0.19098259191513062
+    },
+    {
+      "epoch": 56,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 21896,
+      "train_examples": 50000,
+      "train_loss": 0.19665462752342225
+    },
+    {
+      "epoch": 57,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 22287,
+      "train_examples": 50000,
+      "train_loss": 0.1944172574710846
+    },
+    {
+      "epoch": 58,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 22678,
+      "train_examples": 50000,
+      "train_loss": 0.18747816185951233
+    },
+    {
+      "epoch": 59,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 23069,
+      "train_examples": 50000,
+      "train_loss": 0.19039372524738313
+    },
+    {
+      "epoch": 60,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 23460,
+      "train_examples": 50000,
+      "train_loss": 0.18545300397872924
+    },
+    {
+      "epoch": 61,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 23851,
+      "train_examples": 50000,
+      "train_loss": 0.1798163256597519
+    },
+    {
+      "epoch": 62,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 24242,
+      "train_examples": 50000,
+      "train_loss": 0.18622890206336976
+    },
+    {
+      "epoch": 63,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 24633,
+      "train_examples": 50000,
+      "train_loss": 0.1808734233045578
+    },
+    {
+      "epoch": 64,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 25024,
+      "train_examples": 50000,
+      "train_loss": 0.1849400773715973
+    },
+    {
+      "epoch": 65,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 25415,
+      "train_examples": 50000,
+      "train_loss": 0.17425392099857331
+    },
+    {
+      "epoch": 66,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 25806,
+      "train_examples": 50000,
+      "train_loss": 0.17621845196723937
+    },
+    {
+      "epoch": 67,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 26197,
+      "train_examples": 50000,
+      "train_loss": 0.17704346359729767
+    },
+    {
+      "epoch": 68,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 26588,
+      "train_examples": 50000,
+      "train_loss": 0.17074057980060578
+    },
+    {
+      "epoch": 69,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 26979,
+      "train_examples": 50000,
+      "train_loss": 0.1723531908416748
+    },
+    {
+      "epoch": 70,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 27370,
+      "train_examples": 50000,
+      "train_loss": 0.17592787893772124
+    },
+    {
+      "epoch": 71,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 27761,
+      "train_examples": 50000,
+      "train_loss": 0.16684336119174958
+    },
+    {
+      "epoch": 72,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 28152,
+      "train_examples": 50000,
+      "train_loss": 0.1709837682747841
+    },
+    {
+      "epoch": 73,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 28543,
+      "train_examples": 50000,
+      "train_loss": 0.1664174505519867
+    },
+    {
+      "epoch": 74,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 28934,
+      "train_examples": 50000,
+      "train_loss": 0.16713098608970642
+    },
+    {
+      "epoch": 75,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 29325,
+      "train_examples": 50000,
+      "train_loss": 0.1617482682275772
+    },
+    {
+      "epoch": 76,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 29716,
+      "train_examples": 50000,
+      "train_loss": 0.16305960576057435
+    },
+    {
+      "epoch": 77,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 30107,
+      "train_examples": 50000,
+      "train_loss": 0.1654076874256134
+    },
+    {
+      "epoch": 78,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 30498,
+      "train_examples": 50000,
+      "train_loss": 0.16277417621135712
+    },
+    {
+      "epoch": 79,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 30889,
+      "train_examples": 50000,
+      "train_loss": 0.16408164291858673
+    },
+    {
+      "epoch": 80,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 31280,
+      "train_examples": 50000,
+      "train_loss": 0.15663245958328248
+    },
+    {
+      "epoch": 81,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 31671,
+      "train_examples": 50000,
+      "train_loss": 0.16177598266601562
+    },
+    {
+      "epoch": 82,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 32062,
+      "train_examples": 50000,
+      "train_loss": 0.15847330315113067
+    },
+    {
+      "epoch": 83,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 32453,
+      "train_examples": 50000,
+      "train_loss": 0.15664955557346344
+    },
+    {
+      "epoch": 84,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 32844,
+      "train_examples": 50000,
+      "train_loss": 0.16187899160861968
+    },
+    {
+      "epoch": 85,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 33235,
+      "train_examples": 50000,
+      "train_loss": 0.15342998539447786
+    },
+    {
+      "epoch": 86,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 33626,
+      "train_examples": 50000,
+      "train_loss": 0.15301464160442352
+    },
+    {
+      "epoch": 87,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 34017,
+      "train_examples": 50000,
+      "train_loss": 0.15818682832717895
+    },
+    {
+      "epoch": 88,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 34408,
+      "train_examples": 50000,
+      "train_loss": 0.15367517629146576
+    },
+    {
+      "epoch": 89,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 34799,
+      "train_examples": 50000,
+      "train_loss": 0.15664457752227784
+    },
+    {
+      "epoch": 90,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 35190,
+      "train_examples": 50000,
+      "train_loss": 0.15240934684515
+    },
+    {
+      "epoch": 91,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 35581,
+      "train_examples": 50000,
+      "train_loss": 0.15182420168876648
+    },
+    {
+      "epoch": 92,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 35972,
+      "train_examples": 50000,
+      "train_loss": 0.15646429315567018
+    },
+    {
+      "epoch": 93,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 36363,
+      "train_examples": 50000,
+      "train_loss": 0.15114465425491333
+    },
+    {
+      "epoch": 94,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 36754,
+      "train_examples": 50000,
+      "train_loss": 0.14849061069011688
+    },
+    {
+      "epoch": 95,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 37145,
+      "train_examples": 50000,
+      "train_loss": 0.14975485221862794
+    },
+    {
+      "epoch": 96,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 37536,
+      "train_examples": 50000,
+      "train_loss": 0.1467229773426056
+    },
+    {
+      "epoch": 97,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 37927,
+      "train_examples": 50000,
+      "train_loss": 0.1438916012096405
+    },
+    {
+      "epoch": 98,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 38318,
+      "train_examples": 50000,
+      "train_loss": 0.15089732598304748
+    },
+    {
+      "epoch": 99,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 38709,
+      "train_examples": 50000,
+      "train_loss": 0.14869188279151915
+    },
+    {
+      "epoch": 100,
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 39100,
+      "train_examples": 50000,
+      "train_loss": 0.14303171578645707
+    },
+    {
+      "epoch": 101,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 39491,
+      "train_examples": 50000,
+      "train_loss": 0.07701349678874016
+    },
+    {
+      "epoch": 102,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 39882,
+      "train_examples": 50000,
+      "train_loss": 0.052145924034118656
+    },
+    {
+      "epoch": 103,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 40273,
+      "train_examples": 50000,
+      "train_loss": 0.04184103404998779
+    },
+    {
+      "epoch": 104,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 40664,
+      "train_examples": 50000,
+      "train_loss": 0.03869001036405564
+    },
+    {
+      "epoch": 105,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 41055,
+      "train_examples": 50000,
+      "train_loss": 0.034316612124443056
+    },
+    {
+      "epoch": 106,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 41446,
+      "train_examples": 50000,
+      "train_loss": 0.030132720137834548
+    },
+    {
+      "epoch": 107,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 41837,
+      "train_examples": 50000,
+      "train_loss": 0.027590993185043336
+    },
+    {
+      "epoch": 108,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 42228,
+      "train_examples": 50000,
+      "train_loss": 0.025615437606573106
+    },
+    {
+      "epoch": 109,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 42619,
+      "train_examples": 50000,
+      "train_loss": 0.023910948980748654
+    },
+    {
+      "epoch": 110,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 43010,
+      "train_examples": 50000,
+      "train_loss": 0.022141397398114205
+    },
+    {
+      "epoch": 111,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 43401,
+      "train_examples": 50000,
+      "train_loss": 0.020847919241786004
+    },
+    {
+      "epoch": 112,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 43792,
+      "train_examples": 50000,
+      "train_loss": 0.01950891440361738
+    },
+    {
+      "epoch": 113,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 44183,
+      "train_examples": 50000,
+      "train_loss": 0.01815674560159445
+    },
+    {
+      "epoch": 114,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 44574,
+      "train_examples": 50000,
+      "train_loss": 0.01754834236353636
+    },
+    {
+      "epoch": 115,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 44965,
+      "train_examples": 50000,
+      "train_loss": 0.017942451514899732
+    },
+    {
+      "epoch": 116,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 45356,
+      "train_examples": 50000,
+      "train_loss": 0.016193494759500028
+    },
+    {
+      "epoch": 117,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 45747,
+      "train_examples": 50000,
+      "train_loss": 0.01466585360556841
+    },
+    {
+      "epoch": 118,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 46138,
+      "train_examples": 50000,
+      "train_loss": 0.014408022256270052
+    },
+    {
+      "epoch": 119,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 46529,
+      "train_examples": 50000,
+      "train_loss": 0.014385275693535805
+    },
+    {
+      "epoch": 120,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 46920,
+      "train_examples": 50000,
+      "train_loss": 0.012758618231117725
+    },
+    {
+      "epoch": 121,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 47311,
+      "train_examples": 50000,
+      "train_loss": 0.012632293147146702
+    },
+    {
+      "epoch": 122,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 47702,
+      "train_examples": 50000,
+      "train_loss": 0.011592723081409932
+    },
+    {
+      "epoch": 123,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 48093,
+      "train_examples": 50000,
+      "train_loss": 0.010172025595158339
+    },
+    {
+      "epoch": 124,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 48484,
+      "train_examples": 50000,
+      "train_loss": 0.011078532488048076
+    },
+    {
+      "epoch": 125,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 48875,
+      "train_examples": 50000,
+      "train_loss": 0.011358492282927036
+    },
+    {
+      "epoch": 126,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 49266,
+      "train_examples": 50000,
+      "train_loss": 0.010191038188040256
+    },
+    {
+      "epoch": 127,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 49657,
+      "train_examples": 50000,
+      "train_loss": 0.00990184272825718
+    },
+    {
+      "epoch": 128,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 50048,
+      "train_examples": 50000,
+      "train_loss": 0.009352294146567583
+    },
+    {
+      "epoch": 129,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 50439,
+      "train_examples": 50000,
+      "train_loss": 0.010037282456979156
+    },
+    {
+      "epoch": 130,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 50830,
+      "train_examples": 50000,
+      "train_loss": 0.009517962433807552
+    },
+    {
+      "epoch": 131,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 51221,
+      "train_examples": 50000,
+      "train_loss": 0.009325559734627604
+    },
+    {
+      "epoch": 132,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 51612,
+      "train_examples": 50000,
+      "train_loss": 0.00858199987411499
+    },
+    {
+      "epoch": 133,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 52003,
+      "train_examples": 50000,
+      "train_loss": 0.009094842192158102
+    },
+    {
+      "epoch": 134,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 52394,
+      "train_examples": 50000,
+      "train_loss": 0.009128495811074971
+    },
+    {
+      "epoch": 135,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 52785,
+      "train_examples": 50000,
+      "train_loss": 0.008741873390451073
+    },
+    {
+      "epoch": 136,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 53176,
+      "train_examples": 50000,
+      "train_loss": 0.008326001418977975
+    },
+    {
+      "epoch": 137,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 53567,
+      "train_examples": 50000,
+      "train_loss": 0.0077708227922022345
+    },
+    {
+      "epoch": 138,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 53958,
+      "train_examples": 50000,
+      "train_loss": 0.008901972435489297
+    },
+    {
+      "epoch": 139,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 54349,
+      "train_examples": 50000,
+      "train_loss": 0.0086203098308295
+    },
+    {
+      "epoch": 140,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 54740,
+      "train_examples": 50000,
+      "train_loss": 0.007726845970079303
+    },
+    {
+      "epoch": 141,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 55131,
+      "train_examples": 50000,
+      "train_loss": 0.0070906995633989575
+    },
+    {
+      "epoch": 142,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 55522,
+      "train_examples": 50000,
+      "train_loss": 0.007010017248280346
+    },
+    {
+      "epoch": 143,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 55913,
+      "train_examples": 50000,
+      "train_loss": 0.007219308181889355
+    },
+    {
+      "epoch": 144,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 56304,
+      "train_examples": 50000,
+      "train_loss": 0.007353578768074512
+    },
+    {
+      "epoch": 145,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 56695,
+      "train_examples": 50000,
+      "train_loss": 0.007247117817439139
+    },
+    {
+      "epoch": 146,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 57086,
+      "train_examples": 50000,
+      "train_loss": 0.006711059629842639
+    },
+    {
+      "epoch": 147,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 57477,
+      "train_examples": 50000,
+      "train_loss": 0.006540186242870986
+    },
+    {
+      "epoch": 148,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 57868,
+      "train_examples": 50000,
+      "train_loss": 0.00603294903755188
+    },
+    {
+      "epoch": 149,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 58259,
+      "train_examples": 50000,
+      "train_loss": 0.006158079775236547
+    },
+    {
+      "epoch": 150,
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 58650,
+      "train_examples": 50000,
+      "train_loss": 0.0055025485432520505
+    },
+    {
+      "epoch": 151,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 59041,
+      "train_examples": 50000,
+      "train_loss": 0.005286827752552926
+    },
+    {
+      "epoch": 152,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 59432,
+      "train_examples": 50000,
+      "train_loss": 0.005041609901227057
+    },
+    {
+      "epoch": 153,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 59823,
+      "train_examples": 50000,
+      "train_loss": 0.0048378835561685265
+    },
+    {
+      "epoch": 154,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 60214,
+      "train_examples": 50000,
+      "train_loss": 0.004796428178362548
+    },
+    {
+      "epoch": 155,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 60605,
+      "train_examples": 50000,
+      "train_loss": 0.004337230419516563
+    },
+    {
+      "epoch": 156,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 60996,
+      "train_examples": 50000,
+      "train_loss": 0.004345615052171052
+    },
+    {
+      "epoch": 157,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 61387,
+      "train_examples": 50000,
+      "train_loss": 0.004179104437511415
+    },
+    {
+      "epoch": 158,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 61778,
+      "train_examples": 50000,
+      "train_loss": 0.004085922962166369
+    },
+    {
+      "epoch": 159,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 62169,
+      "train_examples": 50000,
+      "train_loss": 0.0038037374853342773
+    },
+    {
+      "epoch": 160,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 62560,
+      "train_examples": 50000,
+      "train_loss": 0.004268488837471232
+    },
+    {
+      "epoch": 161,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 62951,
+      "train_examples": 50000,
+      "train_loss": 0.004093922774810344
+    },
+    {
+      "epoch": 162,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 63342,
+      "train_examples": 50000,
+      "train_loss": 0.004019515627874061
+    },
+    {
+      "epoch": 163,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 63733,
+      "train_examples": 50000,
+      "train_loss": 0.004371743677668273
+    },
+    {
+      "epoch": 164,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 64124,
+      "train_examples": 50000,
+      "train_loss": 0.004517479064129293
+    },
+    {
+      "epoch": 165,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 64515,
+      "train_examples": 50000,
+      "train_loss": 0.004032488136291504
+    },
+    {
+      "epoch": 166,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 64906,
+      "train_examples": 50000,
+      "train_loss": 0.003401089366339147
+    },
+    {
+      "epoch": 167,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 65297,
+      "train_examples": 50000,
+      "train_loss": 0.003850532835051417
+    },
+    {
+      "epoch": 168,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 65688,
+      "train_examples": 50000,
+      "train_loss": 0.00388622621614486
+    },
+    {
+      "epoch": 169,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 66079,
+      "train_examples": 50000,
+      "train_loss": 0.003748435208313167
+    },
+    {
+      "epoch": 170,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 66470,
+      "train_examples": 50000,
+      "train_loss": 0.0034598920269683004
+    },
+    {
+      "epoch": 171,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 66861,
+      "train_examples": 50000,
+      "train_loss": 0.0035332952127605675
+    },
+    {
+      "epoch": 172,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 67252,
+      "train_examples": 50000,
+      "train_loss": 0.003315019896812737
+    },
+    {
+      "epoch": 173,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 67643,
+      "train_examples": 50000,
+      "train_loss": 0.003480002392269671
+    },
+    {
+      "epoch": 174,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 68034,
+      "train_examples": 50000,
+      "train_loss": 0.0035039778316766023
+    },
+    {
+      "epoch": 175,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 68425,
+      "train_examples": 50000,
+      "train_loss": 0.0036558511624112727
+    },
+    {
+      "epoch": 176,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 68816,
+      "train_examples": 50000,
+      "train_loss": 0.0035984811946749686
+    },
+    {
+      "epoch": 177,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 69207,
+      "train_examples": 50000,
+      "train_loss": 0.0035086333359684793
+    },
+    {
+      "epoch": 178,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 69598,
+      "train_examples": 50000,
+      "train_loss": 0.003361219122521579
+    },
+    {
+      "epoch": 179,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 69989,
+      "train_examples": 50000,
+      "train_loss": 0.0036342795138806106
+    },
+    {
+      "epoch": 180,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 70380,
+      "train_examples": 50000,
+      "train_loss": 0.0031355652825161813
+    },
+    {
+      "epoch": 181,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 70771,
+      "train_examples": 50000,
+      "train_loss": 0.0031834338897280393
+    },
+    {
+      "epoch": 182,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 71162,
+      "train_examples": 50000,
+      "train_loss": 0.003165430543273687
+    },
+    {
+      "epoch": 183,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 71553,
+      "train_examples": 50000,
+      "train_loss": 0.0033229424182325603
+    },
+    {
+      "epoch": 184,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 71944,
+      "train_examples": 50000,
+      "train_loss": 0.0033445316010527314
+    },
+    {
+      "epoch": 185,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 72335,
+      "train_examples": 50000,
+      "train_loss": 0.003054041137304157
+    },
+    {
+      "epoch": 186,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 72726,
+      "train_examples": 50000,
+      "train_loss": 0.003402996248789132
+    },
+    {
+      "epoch": 187,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 73117,
+      "train_examples": 50000,
+      "train_loss": 0.003288349488051608
+    },
+    {
+      "epoch": 188,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 73508,
+      "train_examples": 50000,
+      "train_loss": 0.003205289632026106
+    },
+    {
+      "epoch": 189,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 73899,
+      "train_examples": 50000,
+      "train_loss": 0.003335792209059
+    },
+    {
+      "epoch": 190,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 74290,
+      "train_examples": 50000,
+      "train_loss": 0.0031124157635495065
+    },
+    {
+      "epoch": 191,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 74681,
+      "train_examples": 50000,
+      "train_loss": 0.0034116411933861673
+    },
+    {
+      "epoch": 192,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 75072,
+      "train_examples": 50000,
+      "train_loss": 0.002717412325926125
+    },
+    {
+      "epoch": 193,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 75463,
+      "train_examples": 50000,
+      "train_loss": 0.0032805248819850383
+    },
+    {
+      "epoch": 194,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 75854,
+      "train_examples": 50000,
+      "train_loss": 0.0030963659314624963
+    },
+    {
+      "epoch": 195,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 76245,
+      "train_examples": 50000,
+      "train_loss": 0.0032434921885840596
+    },
+    {
+      "epoch": 196,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 76636,
+      "train_examples": 50000,
+      "train_loss": 0.003127295686490834
+    },
+    {
+      "epoch": 197,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 77027,
+      "train_examples": 50000,
+      "train_loss": 0.00306669891493395
+    },
+    {
+      "epoch": 198,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 77418,
+      "train_examples": 50000,
+      "train_loss": 0.0028870508106052877
+    },
+    {
+      "epoch": 199,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 77809,
+      "train_examples": 50000,
+      "train_loss": 0.0027870662285573783
+    },
+    {
+      "epoch": 200,
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 78200,
+      "train_examples": 50000,
+      "train_loss": 0.0027346119079738854
+    }
+  ],
+  "evaluation_protocol": {
+    "test_evaluations": 1,
+    "test_used_for_selection": false,
+    "validation_evaluations": 0
+  },
+  "final": {
+    "accuracy": 0.9275,
+    "epoch": 200,
+    "evaluation_split": "test",
+    "finite": true,
+    "loss": 0.4151858093261719,
+    "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": 3320109568,
+    "peak_memory_reserved_bytes": 4542431232,
+    "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": 11,
+    "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": 2.0695407390594482,
+    "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": 10217.597524166107,
+    "train_wall_s": 10215.493811130524
+  },
+  "work": {
+    "apical_macs_per_example": 0,
+    "causal_scalar_observations": 0,
+    "components": {
+      "apical_projection_macs": 0,
+      "apical_regression_macs": 0,
+      "bp_reverse_macs_estimate": 2509713920000000,
+      "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": 1254856960000000,
+      "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": 125485696,
+    "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": 3764570880000000
+  }
+}