experiment: add completed A6000 crossover resultsa6000-x1-results

1 files changed, 2717 insertions, 0 deletions

diff --git a/results/resnet_crossover/r2/resnet-r2-bp-d56-lr0p1.json b/results/resnet_crossover/r2/resnet-r2-bp-d56-lr0p1.json
new file mode 100644
index 0000000..933fa96
--- /dev/null
+++ b/results/resnet_crossover/r2/resnet-r2-bp-d56-lr0p1.json
@@ -0,0 +1,2717 @@
+{
+  "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": "/orion/u/oscarwan/scratch/sdil-data",
+    "depth": 56,
+    "device": "cuda",
+    "epochs": 200,
+    "eta_A": 0.01,
+    "eta_P": 0.01,
+    "eval_every": 1,
+    "eval_split": "validation",
+    "learn_P": 0,
+    "loader_seed": 0,
+    "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": "/orion/u/oscarwan/scratch/sdil-work/sdil-a6000-full-rerun/results/resnet_crossover/r2/resnet-r2-bp-d56-lr0p1.json",
+    "output_lr": 0.1,
+    "pert_directions": 1,
+    "pert_every": 4,
+    "pert_sigma": 0.01,
+    "perturb_seed": 1000,
+    "predictor_every": 0,
+    "predictor_mode": "nlms",
+    "predictor_warmup_steps": 0,
+    "residual_scale": null,
+    "seed": 0,
+    "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": 5000,
+    "vectorizer_mode": "spatial_template",
+    "warmup_epochs": 5,
+    "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": 9000000,
+    "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,
+      "eval_accuracy": 0.4062,
+      "eval_loss": 1.6383858154296875,
+      "eval_split": "validation",
+      "lr": 0.02,
+      "output_lr": 0.02,
+      "step": 352,
+      "train_examples": 45000,
+      "train_loss": 1.8120329790115357
+    },
+    {
+      "epoch": 2,
+      "eval_accuracy": 0.5926,
+      "eval_loss": 1.153287646484375,
+      "eval_split": "validation",
+      "lr": 0.04,
+      "output_lr": 0.04,
+      "step": 704,
+      "train_examples": 45000,
+      "train_loss": 1.3592044618182713
+    },
+    {
+      "epoch": 3,
+      "eval_accuracy": 0.5084,
+      "eval_loss": 1.90244970703125,
+      "eval_split": "validation",
+      "lr": 0.06000000000000001,
+      "output_lr": 0.06000000000000001,
+      "step": 1056,
+      "train_examples": 45000,
+      "train_loss": 1.0838589077843561
+    },
+    {
+      "epoch": 4,
+      "eval_accuracy": 0.693,
+      "eval_loss": 0.906397900390625,
+      "eval_split": "validation",
+      "lr": 0.08,
+      "output_lr": 0.08,
+      "step": 1408,
+      "train_examples": 45000,
+      "train_loss": 0.8900402983453539
+    },
+    {
+      "epoch": 5,
+      "eval_accuracy": 0.7638,
+      "eval_loss": 0.6916922607421875,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 1760,
+      "train_examples": 45000,
+      "train_loss": 0.7705084403991699
+    },
+    {
+      "epoch": 6,
+      "eval_accuracy": 0.7054,
+      "eval_loss": 0.94351083984375,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 2112,
+      "train_examples": 45000,
+      "train_loss": 0.6632061253759596
+    },
+    {
+      "epoch": 7,
+      "eval_accuracy": 0.7784,
+      "eval_loss": 0.6527540771484375,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 2464,
+      "train_examples": 45000,
+      "train_loss": 0.5979521300209893
+    },
+    {
+      "epoch": 8,
+      "eval_accuracy": 0.7632,
+      "eval_loss": 0.7341368408203125,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 2816,
+      "train_examples": 45000,
+      "train_loss": 0.55327621333864
+    },
+    {
+      "epoch": 9,
+      "eval_accuracy": 0.8116,
+      "eval_loss": 0.5726607421875,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 3168,
+      "train_examples": 45000,
+      "train_loss": 0.5106070032013788
+    },
+    {
+      "epoch": 10,
+      "eval_accuracy": 0.7938,
+      "eval_loss": 0.6018488891601562,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 3520,
+      "train_examples": 45000,
+      "train_loss": 0.47956937680774264
+    },
+    {
+      "epoch": 11,
+      "eval_accuracy": 0.8024,
+      "eval_loss": 0.61379228515625,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 3872,
+      "train_examples": 45000,
+      "train_loss": 0.45889593749046326
+    },
+    {
+      "epoch": 12,
+      "eval_accuracy": 0.8038,
+      "eval_loss": 0.6077332397460937,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 4224,
+      "train_examples": 45000,
+      "train_loss": 0.438798218621148
+    },
+    {
+      "epoch": 13,
+      "eval_accuracy": 0.7766,
+      "eval_loss": 0.7508157470703125,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 4576,
+      "train_examples": 45000,
+      "train_loss": 0.4087996634165446
+    },
+    {
+      "epoch": 14,
+      "eval_accuracy": 0.8144,
+      "eval_loss": 0.5816595092773438,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 4928,
+      "train_examples": 45000,
+      "train_loss": 0.39873176553514267
+    },
+    {
+      "epoch": 15,
+      "eval_accuracy": 0.84,
+      "eval_loss": 0.48526373901367187,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 5280,
+      "train_examples": 45000,
+      "train_loss": 0.3821409888108571
+    },
+    {
+      "epoch": 16,
+      "eval_accuracy": 0.832,
+      "eval_loss": 0.5142687622070312,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 5632,
+      "train_examples": 45000,
+      "train_loss": 0.36610251981947156
+    },
+    {
+      "epoch": 17,
+      "eval_accuracy": 0.8192,
+      "eval_loss": 0.5443266357421875,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 5984,
+      "train_examples": 45000,
+      "train_loss": 0.35214182429843477
+    },
+    {
+      "epoch": 18,
+      "eval_accuracy": 0.8386,
+      "eval_loss": 0.49019057006835937,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 6336,
+      "train_examples": 45000,
+      "train_loss": 0.341371633805169
+    },
+    {
+      "epoch": 19,
+      "eval_accuracy": 0.8502,
+      "eval_loss": 0.4660570251464844,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 6688,
+      "train_examples": 45000,
+      "train_loss": 0.333972642771403
+    },
+    {
+      "epoch": 20,
+      "eval_accuracy": 0.8532,
+      "eval_loss": 0.43096478271484373,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 7040,
+      "train_examples": 45000,
+      "train_loss": 0.31695578266514673
+    },
+    {
+      "epoch": 21,
+      "eval_accuracy": 0.8168,
+      "eval_loss": 0.60798935546875,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 7392,
+      "train_examples": 45000,
+      "train_loss": 0.30790028344260323
+    },
+    {
+      "epoch": 22,
+      "eval_accuracy": 0.8586,
+      "eval_loss": 0.4445894714355469,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 7744,
+      "train_examples": 45000,
+      "train_loss": 0.3030786698553297
+    },
+    {
+      "epoch": 23,
+      "eval_accuracy": 0.8612,
+      "eval_loss": 0.431785498046875,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 8096,
+      "train_examples": 45000,
+      "train_loss": 0.2980527818997701
+    },
+    {
+      "epoch": 24,
+      "eval_accuracy": 0.8438,
+      "eval_loss": 0.5205806335449219,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 8448,
+      "train_examples": 45000,
+      "train_loss": 0.2837291211393144
+    },
+    {
+      "epoch": 25,
+      "eval_accuracy": 0.863,
+      "eval_loss": 0.4326460021972656,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 8800,
+      "train_examples": 45000,
+      "train_loss": 0.27650729268391927
+    },
+    {
+      "epoch": 26,
+      "eval_accuracy": 0.8456,
+      "eval_loss": 0.5047505187988282,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 9152,
+      "train_examples": 45000,
+      "train_loss": 0.270019421143002
+    },
+    {
+      "epoch": 27,
+      "eval_accuracy": 0.8434,
+      "eval_loss": 0.5102612182617188,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 9504,
+      "train_examples": 45000,
+      "train_loss": 0.2668691424396303
+    },
+    {
+      "epoch": 28,
+      "eval_accuracy": 0.8578,
+      "eval_loss": 0.4414801574707031,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 9856,
+      "train_examples": 45000,
+      "train_loss": 0.25704341024292837
+    },
+    {
+      "epoch": 29,
+      "eval_accuracy": 0.8668,
+      "eval_loss": 0.4320094970703125,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 10208,
+      "train_examples": 45000,
+      "train_loss": 0.2572188554790285
+    },
+    {
+      "epoch": 30,
+      "eval_accuracy": 0.8714,
+      "eval_loss": 0.408448828125,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 10560,
+      "train_examples": 45000,
+      "train_loss": 0.250815947622723
+    },
+    {
+      "epoch": 31,
+      "eval_accuracy": 0.837,
+      "eval_loss": 0.5562254150390625,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 10912,
+      "train_examples": 45000,
+      "train_loss": 0.248208172082901
+    },
+    {
+      "epoch": 32,
+      "eval_accuracy": 0.8584,
+      "eval_loss": 0.46939553833007813,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 11264,
+      "train_examples": 45000,
+      "train_loss": 0.23669359466499754
+    },
+    {
+      "epoch": 33,
+      "eval_accuracy": 0.864,
+      "eval_loss": 0.44012159423828123,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 11616,
+      "train_examples": 45000,
+      "train_loss": 0.2358193315267563
+    },
+    {
+      "epoch": 34,
+      "eval_accuracy": 0.8726,
+      "eval_loss": 0.41900897827148437,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 11968,
+      "train_examples": 45000,
+      "train_loss": 0.23047334933280944
+    },
+    {
+      "epoch": 35,
+      "eval_accuracy": 0.8484,
+      "eval_loss": 0.5120837097167968,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 12320,
+      "train_examples": 45000,
+      "train_loss": 0.22745651706324682
+    },
+    {
+      "epoch": 36,
+      "eval_accuracy": 0.8572,
+      "eval_loss": 0.48343660888671874,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 12672,
+      "train_examples": 45000,
+      "train_loss": 0.2273322538084454
+    },
+    {
+      "epoch": 37,
+      "eval_accuracy": 0.8596,
+      "eval_loss": 0.4978451171875,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 13024,
+      "train_examples": 45000,
+      "train_loss": 0.22201022511985566
+    },
+    {
+      "epoch": 38,
+      "eval_accuracy": 0.8446,
+      "eval_loss": 0.5256515441894531,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 13376,
+      "train_examples": 45000,
+      "train_loss": 0.2145939707385169
+    },
+    {
+      "epoch": 39,
+      "eval_accuracy": 0.8696,
+      "eval_loss": 0.43493318481445314,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 13728,
+      "train_examples": 45000,
+      "train_loss": 0.20739925054709116
+    },
+    {
+      "epoch": 40,
+      "eval_accuracy": 0.8666,
+      "eval_loss": 0.4600316223144531,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 14080,
+      "train_examples": 45000,
+      "train_loss": 0.20801596263779534
+    },
+    {
+      "epoch": 41,
+      "eval_accuracy": 0.8632,
+      "eval_loss": 0.4669625122070313,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 14432,
+      "train_examples": 45000,
+      "train_loss": 0.20819860476387872
+    },
+    {
+      "epoch": 42,
+      "eval_accuracy": 0.8686,
+      "eval_loss": 0.4578409973144531,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 14784,
+      "train_examples": 45000,
+      "train_loss": 0.19786155010196899
+    },
+    {
+      "epoch": 43,
+      "eval_accuracy": 0.8662,
+      "eval_loss": 0.46103836059570313,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 15136,
+      "train_examples": 45000,
+      "train_loss": 0.20556741065979003
+    },
+    {
+      "epoch": 44,
+      "eval_accuracy": 0.8394,
+      "eval_loss": 0.5724481323242188,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 15488,
+      "train_examples": 45000,
+      "train_loss": 0.1993922583738963
+    },
+    {
+      "epoch": 45,
+      "eval_accuracy": 0.8724,
+      "eval_loss": 0.43330186157226563,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 15840,
+      "train_examples": 45000,
+      "train_loss": 0.19485141789913177
+    },
+    {
+      "epoch": 46,
+      "eval_accuracy": 0.8764,
+      "eval_loss": 0.42409732666015626,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 16192,
+      "train_examples": 45000,
+      "train_loss": 0.19267305657333797
+    },
+    {
+      "epoch": 47,
+      "eval_accuracy": 0.885,
+      "eval_loss": 0.39403966674804686,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 16544,
+      "train_examples": 45000,
+      "train_loss": 0.1875338579866621
+    },
+    {
+      "epoch": 48,
+      "eval_accuracy": 0.8796,
+      "eval_loss": 0.40434483642578123,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 16896,
+      "train_examples": 45000,
+      "train_loss": 0.18509351403978136
+    },
+    {
+      "epoch": 49,
+      "eval_accuracy": 0.88,
+      "eval_loss": 0.4242395751953125,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 17248,
+      "train_examples": 45000,
+      "train_loss": 0.18649417563014561
+    },
+    {
+      "epoch": 50,
+      "eval_accuracy": 0.8704,
+      "eval_loss": 0.4550154113769531,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 17600,
+      "train_examples": 45000,
+      "train_loss": 0.18717047963407304
+    },
+    {
+      "epoch": 51,
+      "eval_accuracy": 0.8744,
+      "eval_loss": 0.44483524169921873,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 17952,
+      "train_examples": 45000,
+      "train_loss": 0.1796641605509652
+    },
+    {
+      "epoch": 52,
+      "eval_accuracy": 0.8818,
+      "eval_loss": 0.40557909545898435,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 18304,
+      "train_examples": 45000,
+      "train_loss": 0.17563802019225228
+    },
+    {
+      "epoch": 53,
+      "eval_accuracy": 0.8742,
+      "eval_loss": 0.42595355224609377,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 18656,
+      "train_examples": 45000,
+      "train_loss": 0.17193800271352133
+    },
+    {
+      "epoch": 54,
+      "eval_accuracy": 0.8312,
+      "eval_loss": 0.6097184692382812,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 19008,
+      "train_examples": 45000,
+      "train_loss": 0.17408619323306612
+    },
+    {
+      "epoch": 55,
+      "eval_accuracy": 0.863,
+      "eval_loss": 0.47159163208007815,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 19360,
+      "train_examples": 45000,
+      "train_loss": 0.17443495961295233
+    },
+    {
+      "epoch": 56,
+      "eval_accuracy": 0.864,
+      "eval_loss": 0.47021598510742185,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 19712,
+      "train_examples": 45000,
+      "train_loss": 0.17730448887348174
+    },
+    {
+      "epoch": 57,
+      "eval_accuracy": 0.845,
+      "eval_loss": 0.5302630187988281,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 20064,
+      "train_examples": 45000,
+      "train_loss": 0.17513225899802315
+    },
+    {
+      "epoch": 58,
+      "eval_accuracy": 0.8764,
+      "eval_loss": 0.45107080078125,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 20416,
+      "train_examples": 45000,
+      "train_loss": 0.16688513896730212
+    },
+    {
+      "epoch": 59,
+      "eval_accuracy": 0.8506,
+      "eval_loss": 0.5074487365722656,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 20768,
+      "train_examples": 45000,
+      "train_loss": 0.17318025396664938
+    },
+    {
+      "epoch": 60,
+      "eval_accuracy": 0.872,
+      "eval_loss": 0.43073905639648435,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 21120,
+      "train_examples": 45000,
+      "train_loss": 0.16225387883716158
+    },
+    {
+      "epoch": 61,
+      "eval_accuracy": 0.8646,
+      "eval_loss": 0.4888678771972656,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 21472,
+      "train_examples": 45000,
+      "train_loss": 0.1628989502257771
+    },
+    {
+      "epoch": 62,
+      "eval_accuracy": 0.89,
+      "eval_loss": 0.3716917114257812,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 21824,
+      "train_examples": 45000,
+      "train_loss": 0.16541327353583443
+    },
+    {
+      "epoch": 63,
+      "eval_accuracy": 0.867,
+      "eval_loss": 0.48051559448242187,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 22176,
+      "train_examples": 45000,
+      "train_loss": 0.15507885179254743
+    },
+    {
+      "epoch": 64,
+      "eval_accuracy": 0.8856,
+      "eval_loss": 0.4047019226074219,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 22528,
+      "train_examples": 45000,
+      "train_loss": 0.15761819744375016
+    },
+    {
+      "epoch": 65,
+      "eval_accuracy": 0.8712,
+      "eval_loss": 0.4679113525390625,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 22880,
+      "train_examples": 45000,
+      "train_loss": 0.16318066529962752
+    },
+    {
+      "epoch": 66,
+      "eval_accuracy": 0.869,
+      "eval_loss": 0.46758770141601563,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 23232,
+      "train_examples": 45000,
+      "train_loss": 0.15842568448649513
+    },
+    {
+      "epoch": 67,
+      "eval_accuracy": 0.8764,
+      "eval_loss": 0.47932765502929686,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 23584,
+      "train_examples": 45000,
+      "train_loss": 0.15360924773481158
+    },
+    {
+      "epoch": 68,
+      "eval_accuracy": 0.8536,
+      "eval_loss": 0.52894326171875,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 23936,
+      "train_examples": 45000,
+      "train_loss": 0.1567973585764567
+    },
+    {
+      "epoch": 69,
+      "eval_accuracy": 0.8874,
+      "eval_loss": 0.3929457092285156,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 24288,
+      "train_examples": 45000,
+      "train_loss": 0.15907586552831862
+    },
+    {
+      "epoch": 70,
+      "eval_accuracy": 0.8816,
+      "eval_loss": 0.42247525634765626,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 24640,
+      "train_examples": 45000,
+      "train_loss": 0.15716197686725192
+    },
+    {
+      "epoch": 71,
+      "eval_accuracy": 0.8732,
+      "eval_loss": 0.4963585998535156,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 24992,
+      "train_examples": 45000,
+      "train_loss": 0.14983130187855825
+    },
+    {
+      "epoch": 72,
+      "eval_accuracy": 0.8732,
+      "eval_loss": 0.46758525390625,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 25344,
+      "train_examples": 45000,
+      "train_loss": 0.15397654530207316
+    },
+    {
+      "epoch": 73,
+      "eval_accuracy": 0.8784,
+      "eval_loss": 0.43674370727539064,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 25696,
+      "train_examples": 45000,
+      "train_loss": 0.14723493348757427
+    },
+    {
+      "epoch": 74,
+      "eval_accuracy": 0.8886,
+      "eval_loss": 0.39118277587890626,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 26048,
+      "train_examples": 45000,
+      "train_loss": 0.14652091862890457
+    },
+    {
+      "epoch": 75,
+      "eval_accuracy": 0.8752,
+      "eval_loss": 0.4687583251953125,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 26400,
+      "train_examples": 45000,
+      "train_loss": 0.1545022050380707
+    },
+    {
+      "epoch": 76,
+      "eval_accuracy": 0.876,
+      "eval_loss": 0.4439466125488281,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 26752,
+      "train_examples": 45000,
+      "train_loss": 0.14616850409242843
+    },
+    {
+      "epoch": 77,
+      "eval_accuracy": 0.8882,
+      "eval_loss": 0.39720289306640627,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 27104,
+      "train_examples": 45000,
+      "train_loss": 0.1465005384021335
+    },
+    {
+      "epoch": 78,
+      "eval_accuracy": 0.887,
+      "eval_loss": 0.4100860900878906,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 27456,
+      "train_examples": 45000,
+      "train_loss": 0.15173963310188718
+    },
+    {
+      "epoch": 79,
+      "eval_accuracy": 0.8672,
+      "eval_loss": 0.4720408081054687,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 27808,
+      "train_examples": 45000,
+      "train_loss": 0.1440488050142924
+    },
+    {
+      "epoch": 80,
+      "eval_accuracy": 0.8848,
+      "eval_loss": 0.4128118469238281,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 28160,
+      "train_examples": 45000,
+      "train_loss": 0.14068703662686877
+    },
+    {
+      "epoch": 81,
+      "eval_accuracy": 0.8622,
+      "eval_loss": 0.4955517456054688,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 28512,
+      "train_examples": 45000,
+      "train_loss": 0.14217469872898525
+    },
+    {
+      "epoch": 82,
+      "eval_accuracy": 0.8926,
+      "eval_loss": 0.3864740173339844,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 28864,
+      "train_examples": 45000,
+      "train_loss": 0.14880184517701467
+    },
+    {
+      "epoch": 83,
+      "eval_accuracy": 0.8698,
+      "eval_loss": 0.5053119140625,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 29216,
+      "train_examples": 45000,
+      "train_loss": 0.1299323884540134
+    },
+    {
+      "epoch": 84,
+      "eval_accuracy": 0.8806,
+      "eval_loss": 0.4496087341308594,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 29568,
+      "train_examples": 45000,
+      "train_loss": 0.146550003431903
+    },
+    {
+      "epoch": 85,
+      "eval_accuracy": 0.8928,
+      "eval_loss": 0.40095016479492185,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 29920,
+      "train_examples": 45000,
+      "train_loss": 0.14196164631048838
+    },
+    {
+      "epoch": 86,
+      "eval_accuracy": 0.8936,
+      "eval_loss": 0.38955796508789065,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 30272,
+      "train_examples": 45000,
+      "train_loss": 0.13579268198278216
+    },
+    {
+      "epoch": 87,
+      "eval_accuracy": 0.8726,
+      "eval_loss": 0.4710050598144531,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 30624,
+      "train_examples": 45000,
+      "train_loss": 0.14606884476078882
+    },
+    {
+      "epoch": 88,
+      "eval_accuracy": 0.8852,
+      "eval_loss": 0.40862398071289063,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 30976,
+      "train_examples": 45000,
+      "train_loss": 0.14596923381487528
+    },
+    {
+      "epoch": 89,
+      "eval_accuracy": 0.8938,
+      "eval_loss": 0.37538721313476564,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 31328,
+      "train_examples": 45000,
+      "train_loss": 0.13962470082706877
+    },
+    {
+      "epoch": 90,
+      "eval_accuracy": 0.8798,
+      "eval_loss": 0.45380982666015623,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 31680,
+      "train_examples": 45000,
+      "train_loss": 0.136911416721344
+    },
+    {
+      "epoch": 91,
+      "eval_accuracy": 0.8842,
+      "eval_loss": 0.4425386474609375,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 32032,
+      "train_examples": 45000,
+      "train_loss": 0.13629336178965037
+    },
+    {
+      "epoch": 92,
+      "eval_accuracy": 0.8908,
+      "eval_loss": 0.419219775390625,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 32384,
+      "train_examples": 45000,
+      "train_loss": 0.1369322536945343
+    },
+    {
+      "epoch": 93,
+      "eval_accuracy": 0.893,
+      "eval_loss": 0.3760003112792969,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 32736,
+      "train_examples": 45000,
+      "train_loss": 0.13725584357579548
+    },
+    {
+      "epoch": 94,
+      "eval_accuracy": 0.8704,
+      "eval_loss": 0.4654718566894531,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 33088,
+      "train_examples": 45000,
+      "train_loss": 0.13652064226733313
+    },
+    {
+      "epoch": 95,
+      "eval_accuracy": 0.8868,
+      "eval_loss": 0.44302418212890626,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 33440,
+      "train_examples": 45000,
+      "train_loss": 0.12910948644744025
+    },
+    {
+      "epoch": 96,
+      "eval_accuracy": 0.8864,
+      "eval_loss": 0.42267026977539063,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 33792,
+      "train_examples": 45000,
+      "train_loss": 0.13443772603140938
+    },
+    {
+      "epoch": 97,
+      "eval_accuracy": 0.8926,
+      "eval_loss": 0.3810096862792969,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 34144,
+      "train_examples": 45000,
+      "train_loss": 0.13513379742834303
+    },
+    {
+      "epoch": 98,
+      "eval_accuracy": 0.8934,
+      "eval_loss": 0.3915801513671875,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 34496,
+      "train_examples": 45000,
+      "train_loss": 0.12859395507971447
+    },
+    {
+      "epoch": 99,
+      "eval_accuracy": 0.8918,
+      "eval_loss": 0.4213328430175781,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 34848,
+      "train_examples": 45000,
+      "train_loss": 0.13369986132780712
+    },
+    {
+      "epoch": 100,
+      "eval_accuracy": 0.886,
+      "eval_loss": 0.45385445556640625,
+      "eval_split": "validation",
+      "lr": 0.1,
+      "output_lr": 0.1,
+      "step": 35200,
+      "train_examples": 45000,
+      "train_loss": 0.13087427303592364
+    },
+    {
+      "epoch": 101,
+      "eval_accuracy": 0.9204,
+      "eval_loss": 0.2958070129394531,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 35552,
+      "train_examples": 45000,
+      "train_loss": 0.07391229368348916
+    },
+    {
+      "epoch": 102,
+      "eval_accuracy": 0.9242,
+      "eval_loss": 0.28999569702148437,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 35904,
+      "train_examples": 45000,
+      "train_loss": 0.04571805165542497
+    },
+    {
+      "epoch": 103,
+      "eval_accuracy": 0.9242,
+      "eval_loss": 0.2971516357421875,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 36256,
+      "train_examples": 45000,
+      "train_loss": 0.036192750480605496
+    },
+    {
+      "epoch": 104,
+      "eval_accuracy": 0.9262,
+      "eval_loss": 0.3033703857421875,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 36608,
+      "train_examples": 45000,
+      "train_loss": 0.03154912928475274
+    },
+    {
+      "epoch": 105,
+      "eval_accuracy": 0.9256,
+      "eval_loss": 0.30697164306640623,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 36960,
+      "train_examples": 45000,
+      "train_loss": 0.02741032799548573
+    },
+    {
+      "epoch": 106,
+      "eval_accuracy": 0.9268,
+      "eval_loss": 0.30802741088867186,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 37312,
+      "train_examples": 45000,
+      "train_loss": 0.02428152202632692
+    },
+    {
+      "epoch": 107,
+      "eval_accuracy": 0.9262,
+      "eval_loss": 0.314338623046875,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 37664,
+      "train_examples": 45000,
+      "train_loss": 0.021879386997885174
+    },
+    {
+      "epoch": 108,
+      "eval_accuracy": 0.925,
+      "eval_loss": 0.3235135375976563,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 38016,
+      "train_examples": 45000,
+      "train_loss": 0.020878948243459067
+    },
+    {
+      "epoch": 109,
+      "eval_accuracy": 0.9252,
+      "eval_loss": 0.3302208251953125,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 38368,
+      "train_examples": 45000,
+      "train_loss": 0.01850255666706297
+    },
+    {
+      "epoch": 110,
+      "eval_accuracy": 0.9262,
+      "eval_loss": 0.32489039916992185,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 38720,
+      "train_examples": 45000,
+      "train_loss": 0.01726404068602456
+    },
+    {
+      "epoch": 111,
+      "eval_accuracy": 0.9268,
+      "eval_loss": 0.3281711608886719,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 39072,
+      "train_examples": 45000,
+      "train_loss": 0.016243692442443634
+    },
+    {
+      "epoch": 112,
+      "eval_accuracy": 0.9268,
+      "eval_loss": 0.3317916564941406,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 39424,
+      "train_examples": 45000,
+      "train_loss": 0.014255007772313224
+    },
+    {
+      "epoch": 113,
+      "eval_accuracy": 0.9268,
+      "eval_loss": 0.3346408752441406,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 39776,
+      "train_examples": 45000,
+      "train_loss": 0.014053582479059697
+    },
+    {
+      "epoch": 114,
+      "eval_accuracy": 0.9258,
+      "eval_loss": 0.3327147705078125,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 40128,
+      "train_examples": 45000,
+      "train_loss": 0.013204236159059737
+    },
+    {
+      "epoch": 115,
+      "eval_accuracy": 0.9264,
+      "eval_loss": 0.3379136962890625,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 40480,
+      "train_examples": 45000,
+      "train_loss": 0.012523622225059403
+    },
+    {
+      "epoch": 116,
+      "eval_accuracy": 0.925,
+      "eval_loss": 0.3409310119628906,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 40832,
+      "train_examples": 45000,
+      "train_loss": 0.011970065443052185
+    },
+    {
+      "epoch": 117,
+      "eval_accuracy": 0.9276,
+      "eval_loss": 0.341762646484375,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 41184,
+      "train_examples": 45000,
+      "train_loss": 0.010813051444954343
+    },
+    {
+      "epoch": 118,
+      "eval_accuracy": 0.9268,
+      "eval_loss": 0.338385888671875,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 41536,
+      "train_examples": 45000,
+      "train_loss": 0.011047906530731254
+    },
+    {
+      "epoch": 119,
+      "eval_accuracy": 0.924,
+      "eval_loss": 0.3458584655761719,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 41888,
+      "train_examples": 45000,
+      "train_loss": 0.01061292423490021
+    },
+    {
+      "epoch": 120,
+      "eval_accuracy": 0.9268,
+      "eval_loss": 0.34402819213867186,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 42240,
+      "train_examples": 45000,
+      "train_loss": 0.009872842174685663
+    },
+    {
+      "epoch": 121,
+      "eval_accuracy": 0.9262,
+      "eval_loss": 0.3493848388671875,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 42592,
+      "train_examples": 45000,
+      "train_loss": 0.008324120312101311
+    },
+    {
+      "epoch": 122,
+      "eval_accuracy": 0.9276,
+      "eval_loss": 0.3444608703613281,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 42944,
+      "train_examples": 45000,
+      "train_loss": 0.00849333224710491
+    },
+    {
+      "epoch": 123,
+      "eval_accuracy": 0.9264,
+      "eval_loss": 0.3520544494628906,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 43296,
+      "train_examples": 45000,
+      "train_loss": 0.008254868710206614
+    },
+    {
+      "epoch": 124,
+      "eval_accuracy": 0.9262,
+      "eval_loss": 0.3435565612792969,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 43648,
+      "train_examples": 45000,
+      "train_loss": 0.007909329082051084
+    },
+    {
+      "epoch": 125,
+      "eval_accuracy": 0.9258,
+      "eval_loss": 0.35690042114257814,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 44000,
+      "train_examples": 45000,
+      "train_loss": 0.008038395318388939
+    },
+    {
+      "epoch": 126,
+      "eval_accuracy": 0.9262,
+      "eval_loss": 0.35696865234375,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 44352,
+      "train_examples": 45000,
+      "train_loss": 0.007849043796294265
+    },
+    {
+      "epoch": 127,
+      "eval_accuracy": 0.9256,
+      "eval_loss": 0.369244384765625,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 44704,
+      "train_examples": 45000,
+      "train_loss": 0.008171430269297626
+    },
+    {
+      "epoch": 128,
+      "eval_accuracy": 0.924,
+      "eval_loss": 0.3637481201171875,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 45056,
+      "train_examples": 45000,
+      "train_loss": 0.007164580019977358
+    },
+    {
+      "epoch": 129,
+      "eval_accuracy": 0.9276,
+      "eval_loss": 0.35849393920898437,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 45408,
+      "train_examples": 45000,
+      "train_loss": 0.007030607246607542
+    },
+    {
+      "epoch": 130,
+      "eval_accuracy": 0.9268,
+      "eval_loss": 0.3662152770996094,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 45760,
+      "train_examples": 45000,
+      "train_loss": 0.006088745883686675
+    },
+    {
+      "epoch": 131,
+      "eval_accuracy": 0.9274,
+      "eval_loss": 0.355767724609375,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 46112,
+      "train_examples": 45000,
+      "train_loss": 0.006438479425261418
+    },
+    {
+      "epoch": 132,
+      "eval_accuracy": 0.9272,
+      "eval_loss": 0.3704664978027344,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 46464,
+      "train_examples": 45000,
+      "train_loss": 0.006546230017228259
+    },
+    {
+      "epoch": 133,
+      "eval_accuracy": 0.9276,
+      "eval_loss": 0.37374725952148435,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 46816,
+      "train_examples": 45000,
+      "train_loss": 0.005257100144318409
+    },
+    {
+      "epoch": 134,
+      "eval_accuracy": 0.9288,
+      "eval_loss": 0.3644460998535156,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 47168,
+      "train_examples": 45000,
+      "train_loss": 0.005803062219565941
+    },
+    {
+      "epoch": 135,
+      "eval_accuracy": 0.9268,
+      "eval_loss": 0.3769530700683594,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 47520,
+      "train_examples": 45000,
+      "train_loss": 0.0059276600245075925
+    },
+    {
+      "epoch": 136,
+      "eval_accuracy": 0.9252,
+      "eval_loss": 0.3732922607421875,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 47872,
+      "train_examples": 45000,
+      "train_loss": 0.005344935539116462
+    },
+    {
+      "epoch": 137,
+      "eval_accuracy": 0.9256,
+      "eval_loss": 0.3735669189453125,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 48224,
+      "train_examples": 45000,
+      "train_loss": 0.006024956390758355
+    },
+    {
+      "epoch": 138,
+      "eval_accuracy": 0.9274,
+      "eval_loss": 0.37350067749023436,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 48576,
+      "train_examples": 45000,
+      "train_loss": 0.00587833823416796
+    },
+    {
+      "epoch": 139,
+      "eval_accuracy": 0.93,
+      "eval_loss": 0.3658798034667969,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 48928,
+      "train_examples": 45000,
+      "train_loss": 0.005202372343217333
+    },
+    {
+      "epoch": 140,
+      "eval_accuracy": 0.928,
+      "eval_loss": 0.37819749145507814,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 49280,
+      "train_examples": 45000,
+      "train_loss": 0.00522608685038156
+    },
+    {
+      "epoch": 141,
+      "eval_accuracy": 0.9308,
+      "eval_loss": 0.3692561706542969,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 49632,
+      "train_examples": 45000,
+      "train_loss": 0.005239222901273105
+    },
+    {
+      "epoch": 142,
+      "eval_accuracy": 0.9276,
+      "eval_loss": 0.37637430419921875,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 49984,
+      "train_examples": 45000,
+      "train_loss": 0.004970019086450338
+    },
+    {
+      "epoch": 143,
+      "eval_accuracy": 0.9266,
+      "eval_loss": 0.3803944091796875,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 50336,
+      "train_examples": 45000,
+      "train_loss": 0.004805412318226364
+    },
+    {
+      "epoch": 144,
+      "eval_accuracy": 0.9282,
+      "eval_loss": 0.3740898742675781,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 50688,
+      "train_examples": 45000,
+      "train_loss": 0.004170282630084289
+    },
+    {
+      "epoch": 145,
+      "eval_accuracy": 0.9276,
+      "eval_loss": 0.38094301147460935,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 51040,
+      "train_examples": 45000,
+      "train_loss": 0.004932723968476057
+    },
+    {
+      "epoch": 146,
+      "eval_accuracy": 0.9262,
+      "eval_loss": 0.37821016845703126,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 51392,
+      "train_examples": 45000,
+      "train_loss": 0.00465814622434684
+    },
+    {
+      "epoch": 147,
+      "eval_accuracy": 0.9282,
+      "eval_loss": 0.374492822265625,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 51744,
+      "train_examples": 45000,
+      "train_loss": 0.004465697120130062
+    },
+    {
+      "epoch": 148,
+      "eval_accuracy": 0.9272,
+      "eval_loss": 0.37554210205078126,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 52096,
+      "train_examples": 45000,
+      "train_loss": 0.0049913340316878425
+    },
+    {
+      "epoch": 149,
+      "eval_accuracy": 0.9274,
+      "eval_loss": 0.3821321533203125,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 52448,
+      "train_examples": 45000,
+      "train_loss": 0.004511870892842611
+    },
+    {
+      "epoch": 150,
+      "eval_accuracy": 0.9278,
+      "eval_loss": 0.383819873046875,
+      "eval_split": "validation",
+      "lr": 0.010000000000000002,
+      "output_lr": 0.010000000000000002,
+      "step": 52800,
+      "train_examples": 45000,
+      "train_loss": 0.0036155341097464165
+    },
+    {
+      "epoch": 151,
+      "eval_accuracy": 0.9266,
+      "eval_loss": 0.3865296875,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 53152,
+      "train_examples": 45000,
+      "train_loss": 0.0034015125179870263
+    },
+    {
+      "epoch": 152,
+      "eval_accuracy": 0.93,
+      "eval_loss": 0.38060599365234377,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 53504,
+      "train_examples": 45000,
+      "train_loss": 0.0038181717752996417
+    },
+    {
+      "epoch": 153,
+      "eval_accuracy": 0.928,
+      "eval_loss": 0.38813572998046875,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 53856,
+      "train_examples": 45000,
+      "train_loss": 0.003342558716175457
+    },
+    {
+      "epoch": 154,
+      "eval_accuracy": 0.9288,
+      "eval_loss": 0.37930936889648437,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 54208,
+      "train_examples": 45000,
+      "train_loss": 0.0037702425530387294
+    },
+    {
+      "epoch": 155,
+      "eval_accuracy": 0.9294,
+      "eval_loss": 0.37951343994140624,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 54560,
+      "train_examples": 45000,
+      "train_loss": 0.0035697890337970523
+    },
+    {
+      "epoch": 156,
+      "eval_accuracy": 0.929,
+      "eval_loss": 0.37994603881835937,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 54912,
+      "train_examples": 45000,
+      "train_loss": 0.0036909519653353425
+    },
+    {
+      "epoch": 157,
+      "eval_accuracy": 0.9288,
+      "eval_loss": 0.38521229858398437,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 55264,
+      "train_examples": 45000,
+      "train_loss": 0.0035891185532013576
+    },
+    {
+      "epoch": 158,
+      "eval_accuracy": 0.9296,
+      "eval_loss": 0.38098255004882814,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 55616,
+      "train_examples": 45000,
+      "train_loss": 0.0032051278701103808
+    },
+    {
+      "epoch": 159,
+      "eval_accuracy": 0.93,
+      "eval_loss": 0.3787800048828125,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 55968,
+      "train_examples": 45000,
+      "train_loss": 0.003374717134113113
+    },
+    {
+      "epoch": 160,
+      "eval_accuracy": 0.9296,
+      "eval_loss": 0.38265086059570314,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 56320,
+      "train_examples": 45000,
+      "train_loss": 0.002994217939178149
+    },
+    {
+      "epoch": 161,
+      "eval_accuracy": 0.9286,
+      "eval_loss": 0.3790346862792969,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 56672,
+      "train_examples": 45000,
+      "train_loss": 0.003431437090711875
+    },
+    {
+      "epoch": 162,
+      "eval_accuracy": 0.929,
+      "eval_loss": 0.3793521911621094,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 57024,
+      "train_examples": 45000,
+      "train_loss": 0.0029321252649649977
+    },
+    {
+      "epoch": 163,
+      "eval_accuracy": 0.9296,
+      "eval_loss": 0.3795800048828125,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 57376,
+      "train_examples": 45000,
+      "train_loss": 0.0036018194581071534
+    },
+    {
+      "epoch": 164,
+      "eval_accuracy": 0.929,
+      "eval_loss": 0.3825399169921875,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 57728,
+      "train_examples": 45000,
+      "train_loss": 0.0027656607893606026
+    },
+    {
+      "epoch": 165,
+      "eval_accuracy": 0.928,
+      "eval_loss": 0.37968671264648435,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 58080,
+      "train_examples": 45000,
+      "train_loss": 0.0026229754688011277
+    },
+    {
+      "epoch": 166,
+      "eval_accuracy": 0.929,
+      "eval_loss": 0.38276746826171876,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 58432,
+      "train_examples": 45000,
+      "train_loss": 0.0029054207376307912
+    },
+    {
+      "epoch": 167,
+      "eval_accuracy": 0.9292,
+      "eval_loss": 0.3794607421875,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 58784,
+      "train_examples": 45000,
+      "train_loss": 0.003060741257626149
+    },
+    {
+      "epoch": 168,
+      "eval_accuracy": 0.929,
+      "eval_loss": 0.38132125244140624,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 59136,
+      "train_examples": 45000,
+      "train_loss": 0.002886012737949689
+    },
+    {
+      "epoch": 169,
+      "eval_accuracy": 0.929,
+      "eval_loss": 0.3822179992675781,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 59488,
+      "train_examples": 45000,
+      "train_loss": 0.003108068531089359
+    },
+    {
+      "epoch": 170,
+      "eval_accuracy": 0.9292,
+      "eval_loss": 0.38042718505859374,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 59840,
+      "train_examples": 45000,
+      "train_loss": 0.0029554093821181193
+    },
+    {
+      "epoch": 171,
+      "eval_accuracy": 0.929,
+      "eval_loss": 0.383385693359375,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 60192,
+      "train_examples": 45000,
+      "train_loss": 0.0030305214785453344
+    },
+    {
+      "epoch": 172,
+      "eval_accuracy": 0.9304,
+      "eval_loss": 0.38013319091796877,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 60544,
+      "train_examples": 45000,
+      "train_loss": 0.0031653479740954937
+    },
+    {
+      "epoch": 173,
+      "eval_accuracy": 0.9288,
+      "eval_loss": 0.38019481811523437,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 60896,
+      "train_examples": 45000,
+      "train_loss": 0.0032616001981827947
+    },
+    {
+      "epoch": 174,
+      "eval_accuracy": 0.9294,
+      "eval_loss": 0.3823200317382813,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 61248,
+      "train_examples": 45000,
+      "train_loss": 0.0026846566768984
+    },
+    {
+      "epoch": 175,
+      "eval_accuracy": 0.9294,
+      "eval_loss": 0.38054329833984374,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 61600,
+      "train_examples": 45000,
+      "train_loss": 0.0028761062923197945
+    },
+    {
+      "epoch": 176,
+      "eval_accuracy": 0.9294,
+      "eval_loss": 0.37557706298828125,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 61952,
+      "train_examples": 45000,
+      "train_loss": 0.002718526501969124
+    },
+    {
+      "epoch": 177,
+      "eval_accuracy": 0.9296,
+      "eval_loss": 0.38403538208007815,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 62304,
+      "train_examples": 45000,
+      "train_loss": 0.002765856211798059
+    },
+    {
+      "epoch": 178,
+      "eval_accuracy": 0.9288,
+      "eval_loss": 0.3827006591796875,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 62656,
+      "train_examples": 45000,
+      "train_loss": 0.002846009177507626
+    },
+    {
+      "epoch": 179,
+      "eval_accuracy": 0.9304,
+      "eval_loss": 0.37569268798828126,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 63008,
+      "train_examples": 45000,
+      "train_loss": 0.0028106547514183656
+    },
+    {
+      "epoch": 180,
+      "eval_accuracy": 0.9308,
+      "eval_loss": 0.37752535400390624,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 63360,
+      "train_examples": 45000,
+      "train_loss": 0.002670673717599776
+    },
+    {
+      "epoch": 181,
+      "eval_accuracy": 0.9294,
+      "eval_loss": 0.37972734375,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 63712,
+      "train_examples": 45000,
+      "train_loss": 0.0029900958141932887
+    },
+    {
+      "epoch": 182,
+      "eval_accuracy": 0.93,
+      "eval_loss": 0.3836199890136719,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 64064,
+      "train_examples": 45000,
+      "train_loss": 0.0028344500539824367
+    },
+    {
+      "epoch": 183,
+      "eval_accuracy": 0.9298,
+      "eval_loss": 0.382769482421875,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 64416,
+      "train_examples": 45000,
+      "train_loss": 0.002919924992824801
+    },
+    {
+      "epoch": 184,
+      "eval_accuracy": 0.93,
+      "eval_loss": 0.38172957153320314,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 64768,
+      "train_examples": 45000,
+      "train_loss": 0.0029668405563053157
+    },
+    {
+      "epoch": 185,
+      "eval_accuracy": 0.9286,
+      "eval_loss": 0.37836466674804686,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 65120,
+      "train_examples": 45000,
+      "train_loss": 0.002790116118432747
+    },
+    {
+      "epoch": 186,
+      "eval_accuracy": 0.9302,
+      "eval_loss": 0.3797643493652344,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 65472,
+      "train_examples": 45000,
+      "train_loss": 0.002748481237143278
+    },
+    {
+      "epoch": 187,
+      "eval_accuracy": 0.9288,
+      "eval_loss": 0.3819689208984375,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 65824,
+      "train_examples": 45000,
+      "train_loss": 0.002598481794156962
+    },
+    {
+      "epoch": 188,
+      "eval_accuracy": 0.929,
+      "eval_loss": 0.38522958984375,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 66176,
+      "train_examples": 45000,
+      "train_loss": 0.0023670312017202376
+    },
+    {
+      "epoch": 189,
+      "eval_accuracy": 0.9284,
+      "eval_loss": 0.38395161743164063,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 66528,
+      "train_examples": 45000,
+      "train_loss": 0.0026725557989130417
+    },
+    {
+      "epoch": 190,
+      "eval_accuracy": 0.9278,
+      "eval_loss": 0.3851235168457031,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 66880,
+      "train_examples": 45000,
+      "train_loss": 0.002891599158321818
+    },
+    {
+      "epoch": 191,
+      "eval_accuracy": 0.9292,
+      "eval_loss": 0.38445775756835937,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 67232,
+      "train_examples": 45000,
+      "train_loss": 0.0026930961543590658
+    },
+    {
+      "epoch": 192,
+      "eval_accuracy": 0.9286,
+      "eval_loss": 0.38256485595703127,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 67584,
+      "train_examples": 45000,
+      "train_loss": 0.002824814874927203
+    },
+    {
+      "epoch": 193,
+      "eval_accuracy": 0.93,
+      "eval_loss": 0.38405804443359376,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 67936,
+      "train_examples": 45000,
+      "train_loss": 0.002977210904844105
+    },
+    {
+      "epoch": 194,
+      "eval_accuracy": 0.93,
+      "eval_loss": 0.3784375061035156,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 68288,
+      "train_examples": 45000,
+      "train_loss": 0.0028413962472023237
+    },
+    {
+      "epoch": 195,
+      "eval_accuracy": 0.93,
+      "eval_loss": 0.3832097351074219,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 68640,
+      "train_examples": 45000,
+      "train_loss": 0.0030580059261837355
+    },
+    {
+      "epoch": 196,
+      "eval_accuracy": 0.9294,
+      "eval_loss": 0.37996414184570315,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 68992,
+      "train_examples": 45000,
+      "train_loss": 0.0026476701118155485
+    },
+    {
+      "epoch": 197,
+      "eval_accuracy": 0.9292,
+      "eval_loss": 0.38070816650390626,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 69344,
+      "train_examples": 45000,
+      "train_loss": 0.0029351543212205997
+    },
+    {
+      "epoch": 198,
+      "eval_accuracy": 0.9298,
+      "eval_loss": 0.3816475341796875,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 69696,
+      "train_examples": 45000,
+      "train_loss": 0.0026628320027556685
+    },
+    {
+      "epoch": 199,
+      "eval_accuracy": 0.9306,
+      "eval_loss": 0.3811450012207031,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 70048,
+      "train_examples": 45000,
+      "train_loss": 0.002589297404181626
+    },
+    {
+      "epoch": 200,
+      "eval_accuracy": 0.93,
+      "eval_loss": 0.3817935302734375,
+      "eval_split": "validation",
+      "lr": 0.0010000000000000002,
+      "output_lr": 0.0010000000000000002,
+      "step": 70400,
+      "train_examples": 45000,
+      "train_loss": 0.002858151467392842
+    }
+  ],
+  "evaluation_protocol": {
+    "test_evaluations": 0,
+    "test_used_for_selection": false,
+    "validation_evaluations": 201
+  },
+  "final": {
+    "accuracy": 0.93,
+    "epoch": 200,
+    "evaluation_split": "validation",
+    "finite": true,
+    "loss": 0.3817935302734375,
+    "step": 70400
+  },
+  "hardware": {
+    "cuda_device_name": "NVIDIA RTX A6000",
+    "cuda_visible_devices": "0",
+    "device": "cuda",
+    "device_total_memory_bytes": 50896568320,
+    "peak_memory_allocated_bytes": 3257588736,
+    "peak_memory_reserved_bytes": 4322230272,
+    "torch_version": "2.3.1+cu118"
+  },
+  "protocol_family": "oral_a_cifar_local_resnet_development",
+  "provenance": {
+    "git_commit": "f55872beaee658ff76ac15fd421b73017820c320",
+    "git_tracked_dirty": false
+  },
+  "schema_version": 1,
+  "split": {
+    "cifar_source_files": [
+      {
+        "bytes": 31035704,
+        "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/data_batch_1",
+        "sha256": "54636561a3ce25bd3e19253c6b0d8538147b0ae398331ac4a2d86c6d987368cd"
+      },
+      {
+        "bytes": 31035320,
+        "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/data_batch_2",
+        "sha256": "766b2cef9fbc745cf056b3152224f7cf77163b330ea9a15f9392beb8b89bc5a8"
+      },
+      {
+        "bytes": 31035999,
+        "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/data_batch_3",
+        "sha256": "0f00d98ebfb30b3ec0ad19f9756dc2630b89003e10525f5e148445e82aa6a1f9"
+      },
+      {
+        "bytes": 31035696,
+        "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/data_batch_4",
+        "sha256": "3f7bb240661948b8f4d53e36ec720d8306f5668bd0071dcb4e6c947f78e9682b"
+      },
+      {
+        "bytes": 31035623,
+        "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/data_batch_5",
+        "sha256": "d91802434d8376bbaeeadf58a737e3a1b12ac839077e931237e0dcd43adcb154"
+      },
+      {
+        "bytes": 31035526,
+        "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/test_batch",
+        "sha256": "f53d8d457504f7cff4ea9e021afcf0e0ad8e24a91f3fc42091b8adef61157831"
+      }
+    ],
+    "dataset": "cifar10",
+    "evaluation_split": "validation",
+    "input_layout": "NCHW",
+    "input_shape": [
+      3,
+      32,
+      32
+    ],
+    "loader_seed": 0,
+    "normalization_mean": [
+      0.49140000343322754,
+      0.4821999967098236,
+      0.4465000033378601
+    ],
+    "normalization_std": [
+      0.24699999392032623,
+      0.2434999942779541,
+      0.26159998774528503
+    ],
+    "split_from_training_only": true,
+    "split_seed": 2027,
+    "test_examples": 10000,
+    "train_examples": 45000,
+    "training_augmentation": "random_crop_32_padding_4_zero_then_horizontal_flip_p0.5",
+    "validation_class_counts": {
+      "0": 500,
+      "1": 500,
+      "2": 500,
+      "3": 500,
+      "4": 500,
+      "5": 500,
+      "6": 500,
+      "7": 500,
+      "8": 500,
+      "9": 500
+    },
+    "validation_examples": 5000,
+    "validation_index_sha256": "8328b206a97c420e49e54e3eca4abe3274c4756b084355784ea3fb8059e4515b"
+  },
+  "timing": {
+    "apical_warmup_wall_s": 0.0,
+    "evaluation_wall_s": 60.09430003166199,
+    "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": 3323.5966560840607,
+    "train_wall_s": 3263.4807090759277
+  },
+  "work": {
+    "apical_macs_per_example": 0,
+    "causal_scalar_observations": 0,
+    "components": {
+      "apical_projection_macs": 0,
+      "apical_regression_macs": 0,
+      "bp_reverse_macs_estimate": 2258742528000000,
+      "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": 1129371264000000,
+      "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": 9000000,
+    "total_forward_equivalent_examples": 9000000,
+    "total_macs_estimate": 3388113792000000
+  }
+}