results: confirm standard-depth SDIL scaling

1 files changed, 2089 insertions, 0 deletions

diff --git a/results/oral_a_dynamic_scaling_v2/dfa_d32_s11.json b/results/oral_a_dynamic_scaling_v2/dfa_d32_s11.json
new file mode 100644
index 0000000..d1663db
--- /dev/null
+++ b/results/oral_a_dynamic_scaling_v2/dfa_d32_s11.json
@@ -0,0 +1,2089 @@
+{
+  "architecture": {
+    "adaptive_apical_parameters": 3637248,
+    "adaptive_feedback_parameters": 0,
+    "base_width": 16,
+    "blocks_per_stage": 5,
+    "bn_eps": 1e-05,
+    "bn_momentum": 0.1,
+    "depth": 32,
+    "family": "CIFAR 6n+2 ResNet, option-A shortcuts",
+    "fixed_feedback_parameters": 0,
+    "fixed_traffic_coefficients": 303104,
+    "forward_parameters": 464154,
+    "hidden_shapes": [
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        16,
+        32,
+        32
+      ],
+      [
+        32,
+        16,
+        16
+      ],
+      [
+        32,
+        16,
+        16
+      ],
+      [
+        32,
+        16,
+        16
+      ],
+      [
+        32,
+        16,
+        16
+      ],
+      [
+        32,
+        16,
+        16
+      ],
+      [
+        32,
+        16,
+        16
+      ],
+      [
+        32,
+        16,
+        16
+      ],
+      [
+        32,
+        16,
+        16
+      ],
+      [
+        32,
+        16,
+        16
+      ],
+      [
+        32,
+        16,
+        16
+      ],
+      [
+        64,
+        8,
+        8
+      ],
+      [
+        64,
+        8,
+        8
+      ],
+      [
+        64,
+        8,
+        8
+      ],
+      [
+        64,
+        8,
+        8
+      ],
+      [
+        64,
+        8,
+        8
+      ],
+      [
+        64,
+        8,
+        8
+      ],
+      [
+        64,
+        8,
+        8
+      ],
+      [
+        64,
+        8,
+        8
+      ],
+      [
+        64,
+        8,
+        8
+      ],
+      [
+        64,
+        8,
+        8
+      ]
+    ],
+    "normalization": "batchnorm",
+    "predictor_parameters": 606208,
+    "residual_scale": 1.0,
+    "vectorizer_mode": "spatial_template",
+    "vectorizer_parameters": 3031040
+  },
+  "args": {
+    "a_scale": 1.0,
+    "a_warmup_steps": 0,
+    "alignment_probe": 32,
+    "apical_calibration_mode": "unit_targets",
+    "apical_seed": null,
+    "augment_train": 1,
+    "batch_size": 128,
+    "bn_eps": 1e-05,
+    "bn_momentum": 0.1,
+    "data_dir": "/home/yurenh2/sdrn/data",
+    "depth": 32,
+    "device": "cuda",
+    "epochs": 200,
+    "eta_A": 0.01,
+    "eta_P": 0.01,
+    "eval_every": 0,
+    "eval_split": "test",
+    "learn_P": 0,
+    "loader_seed": 11,
+    "lr": 0.03,
+    "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": "dfa",
+    "momentum": 0.9,
+    "neutral_projection": 0,
+    "normalization": "batchnorm",
+    "nuisance_scale": 0.0,
+    "out": "results/oral_a_dynamic_scaling_v2/dfa_d32_s11.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": 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": "full_hidden_field",
+  "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": {
+    "early_third_mean": 0.00026144811999984086,
+    "innovation_negative_gradient_cosine": [
+      -0.0004968902794644237,
+      0.0013761291047558188,
+      0.00010523246601223946,
+      2.608442446216941e-05,
+      0.0005782860680483282,
+      -0.0003741116961464286,
+      0.0006768519524484873,
+      0.0011184385512024164,
+      -0.0004898424376733601,
+      9.430304635316133e-05,
+      -0.0013928385451436043,
+      0.0001874021254479885,
+      -0.000301648979075253,
+      0.0015553411794826388,
+      0.0013286168687045574,
+      -0.00010815382120199502,
+      0.004846717230975628,
+      -0.0006915669655427337,
+      0.0048540569841861725,
+      -0.000358888937626034,
+      -0.0011774199083447456,
+      0.007732397876679897,
+      0.009962676092982292,
+      0.016084017232060432,
+      0.011626636609435081,
+      0.012131083756685257,
+      0.014776289463043213,
+      0.007380135823041201,
+      0.015810808166861534,
+      0.020616794005036354,
+      0.025164347141981125
+    ],
+    "normalization_state": "training_batch_stats_without_running_update",
+    "raw_negative_gradient_cosine": [
+      -0.0004968902794644237,
+      0.0013761291047558188,
+      0.00010523246601223946,
+      2.608442446216941e-05,
+      0.0005782860680483282,
+      -0.0003741116961464286,
+      0.0006768519524484873,
+      0.0011184385512024164,
+      -0.0004898424376733601,
+      9.430304635316133e-05,
+      -0.0013928385451436043,
+      0.0001874021254479885,
+      -0.000301648979075253,
+      0.0015553411794826388,
+      0.0013286168687045574,
+      -0.00010815382120199502,
+      0.004846717230975628,
+      -0.0006915669655427337,
+      0.0048540569841861725,
+      -0.000358888937626034,
+      -0.0011774199083447456,
+      0.007732397876679897,
+      0.009962676092982292,
+      0.016084017232060432,
+      0.011626636609435081,
+      0.012131083756685257,
+      0.014776289463043213,
+      0.007380135823041201,
+      0.015810808166861534,
+      0.020616794005036354,
+      0.025164347141981125
+    ],
+    "teaching_negative_gradient_cosine": [
+      -0.0004968902794644237,
+      0.0013761291047558188,
+      0.00010523246601223946,
+      2.608442446216941e-05,
+      0.0005782860680483282,
+      -0.0003741116961464286,
+      0.0006768519524484873,
+      0.0011184385512024164,
+      -0.0004898424376733601,
+      9.430304635316133e-05,
+      -0.0013928385451436043,
+      0.0001874021254479885,
+      -0.000301648979075253,
+      0.0015553411794826388,
+      0.0013286168687045574,
+      -0.00010815382120199502,
+      0.004846717230975628,
+      -0.0006915669655427337,
+      0.0048540569841861725,
+      -0.000358888937626034,
+      -0.0011774199083447456,
+      0.007732397876679897,
+      0.009962676092982292,
+      0.016084017232060432,
+      0.011626636609435081,
+      0.012131083756685257,
+      0.014776289463043213,
+      0.007380135823041201,
+      0.015810808166861534,
+      0.020616794005036354,
+      0.025164347141981125
+    ],
+    "wall_s": 0.12825512886047363
+  },
+  "epochs": [
+    {
+      "epoch": 1,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 391,
+      "train_examples": 50000,
+      "train_loss": 9.430248809814453
+    },
+    {
+      "epoch": 2,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 782,
+      "train_examples": 50000,
+      "train_loss": 9.950192311401366
+    },
+    {
+      "epoch": 3,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 1173,
+      "train_examples": 50000,
+      "train_loss": 12.234537961730958
+    },
+    {
+      "epoch": 4,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 1564,
+      "train_examples": 50000,
+      "train_loss": 16.58897698913574
+    },
+    {
+      "epoch": 5,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 1955,
+      "train_examples": 50000,
+      "train_loss": 22.226687786865234
+    },
+    {
+      "epoch": 6,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 2346,
+      "train_examples": 50000,
+      "train_loss": 30.449551181640626
+    },
+    {
+      "epoch": 7,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 2737,
+      "train_examples": 50000,
+      "train_loss": 36.34107797546387
+    },
+    {
+      "epoch": 8,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 3128,
+      "train_examples": 50000,
+      "train_loss": 46.41980688720703
+    },
+    {
+      "epoch": 9,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 3519,
+      "train_examples": 50000,
+      "train_loss": 64.11765653808594
+    },
+    {
+      "epoch": 10,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 3910,
+      "train_examples": 50000,
+      "train_loss": 82.97689662597656
+    },
+    {
+      "epoch": 11,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 4301,
+      "train_examples": 50000,
+      "train_loss": 102.98532927246094
+    },
+    {
+      "epoch": 12,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 4692,
+      "train_examples": 50000,
+      "train_loss": 120.77116177001953
+    },
+    {
+      "epoch": 13,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 5083,
+      "train_examples": 50000,
+      "train_loss": 140.28046770751953
+    },
+    {
+      "epoch": 14,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 5474,
+      "train_examples": 50000,
+      "train_loss": 157.43301540039062
+    },
+    {
+      "epoch": 15,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 5865,
+      "train_examples": 50000,
+      "train_loss": 205.60902872070312
+    },
+    {
+      "epoch": 16,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 6256,
+      "train_examples": 50000,
+      "train_loss": 230.8937860522461
+    },
+    {
+      "epoch": 17,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 6647,
+      "train_examples": 50000,
+      "train_loss": 253.97255029785157
+    },
+    {
+      "epoch": 18,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 7038,
+      "train_examples": 50000,
+      "train_loss": 317.77819536132813
+    },
+    {
+      "epoch": 19,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 7429,
+      "train_examples": 50000,
+      "train_loss": 343.6580914453125
+    },
+    {
+      "epoch": 20,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 7820,
+      "train_examples": 50000,
+      "train_loss": 407.3898486328125
+    },
+    {
+      "epoch": 21,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 8211,
+      "train_examples": 50000,
+      "train_loss": 477.45794138671874
+    },
+    {
+      "epoch": 22,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 8602,
+      "train_examples": 50000,
+      "train_loss": 524.5825004492187
+    },
+    {
+      "epoch": 23,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 8993,
+      "train_examples": 50000,
+      "train_loss": 557.1729856640625
+    },
+    {
+      "epoch": 24,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 9384,
+      "train_examples": 50000,
+      "train_loss": 662.8630315625
+    },
+    {
+      "epoch": 25,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 9775,
+      "train_examples": 50000,
+      "train_loss": 652.32735375
+    },
+    {
+      "epoch": 26,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 10166,
+      "train_examples": 50000,
+      "train_loss": 747.8931953320313
+    },
+    {
+      "epoch": 27,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 10557,
+      "train_examples": 50000,
+      "train_loss": 763.5256337109375
+    },
+    {
+      "epoch": 28,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 10948,
+      "train_examples": 50000,
+      "train_loss": 835.3264582617187
+    },
+    {
+      "epoch": 29,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 11339,
+      "train_examples": 50000,
+      "train_loss": 1058.0156810351561
+    },
+    {
+      "epoch": 30,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 11730,
+      "train_examples": 50000,
+      "train_loss": 905.8362717578125
+    },
+    {
+      "epoch": 31,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 12121,
+      "train_examples": 50000,
+      "train_loss": 1242.1385985546874
+    },
+    {
+      "epoch": 32,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 12512,
+      "train_examples": 50000,
+      "train_loss": 1134.6342613671875
+    },
+    {
+      "epoch": 33,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 12903,
+      "train_examples": 50000,
+      "train_loss": 1470.7842817578126
+    },
+    {
+      "epoch": 34,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 13294,
+      "train_examples": 50000,
+      "train_loss": 1540.7826381640625
+    },
+    {
+      "epoch": 35,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 13685,
+      "train_examples": 50000,
+      "train_loss": 1596.2331265625
+    },
+    {
+      "epoch": 36,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 14076,
+      "train_examples": 50000,
+      "train_loss": 1612.490711953125
+    },
+    {
+      "epoch": 37,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 14467,
+      "train_examples": 50000,
+      "train_loss": 1783.94997171875
+    },
+    {
+      "epoch": 38,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 14858,
+      "train_examples": 50000,
+      "train_loss": 1837.744021875
+    },
+    {
+      "epoch": 39,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 15249,
+      "train_examples": 50000,
+      "train_loss": 2026.28462453125
+    },
+    {
+      "epoch": 40,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 15640,
+      "train_examples": 50000,
+      "train_loss": 2168.968517890625
+    },
+    {
+      "epoch": 41,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 16031,
+      "train_examples": 50000,
+      "train_loss": 2237.83662625
+    },
+    {
+      "epoch": 42,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 16422,
+      "train_examples": 50000,
+      "train_loss": 2496.417368984375
+    },
+    {
+      "epoch": 43,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 16813,
+      "train_examples": 50000,
+      "train_loss": 2410.90841
+    },
+    {
+      "epoch": 44,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 17204,
+      "train_examples": 50000,
+      "train_loss": 2439.45786546875
+    },
+    {
+      "epoch": 45,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 17595,
+      "train_examples": 50000,
+      "train_loss": 2617.1114828125
+    },
+    {
+      "epoch": 46,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 17986,
+      "train_examples": 50000,
+      "train_loss": 2940.740397421875
+    },
+    {
+      "epoch": 47,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 18377,
+      "train_examples": 50000,
+      "train_loss": 2858.628394921875
+    },
+    {
+      "epoch": 48,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 18768,
+      "train_examples": 50000,
+      "train_loss": 3138.9177421875
+    },
+    {
+      "epoch": 49,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 19159,
+      "train_examples": 50000,
+      "train_loss": 3438.415418515625
+    },
+    {
+      "epoch": 50,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 19550,
+      "train_examples": 50000,
+      "train_loss": 3396.234941796875
+    },
+    {
+      "epoch": 51,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 19941,
+      "train_examples": 50000,
+      "train_loss": 3606.21058734375
+    },
+    {
+      "epoch": 52,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 20332,
+      "train_examples": 50000,
+      "train_loss": 4012.36167828125
+    },
+    {
+      "epoch": 53,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 20723,
+      "train_examples": 50000,
+      "train_loss": 4042.69706984375
+    },
+    {
+      "epoch": 54,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 21114,
+      "train_examples": 50000,
+      "train_loss": 4233.11868765625
+    },
+    {
+      "epoch": 55,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 21505,
+      "train_examples": 50000,
+      "train_loss": 4537.82246765625
+    },
+    {
+      "epoch": 56,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 21896,
+      "train_examples": 50000,
+      "train_loss": 4367.230910859375
+    },
+    {
+      "epoch": 57,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 22287,
+      "train_examples": 50000,
+      "train_loss": 4864.74235125
+    },
+    {
+      "epoch": 58,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 22678,
+      "train_examples": 50000,
+      "train_loss": 5054.6145271875
+    },
+    {
+      "epoch": 59,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 23069,
+      "train_examples": 50000,
+      "train_loss": 4842.692507109375
+    },
+    {
+      "epoch": 60,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 23460,
+      "train_examples": 50000,
+      "train_loss": 5127.00258875
+    },
+    {
+      "epoch": 61,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 23851,
+      "train_examples": 50000,
+      "train_loss": 4986.25878015625
+    },
+    {
+      "epoch": 62,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 24242,
+      "train_examples": 50000,
+      "train_loss": 5355.6228540625
+    },
+    {
+      "epoch": 63,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 24633,
+      "train_examples": 50000,
+      "train_loss": 5600.529760234375
+    },
+    {
+      "epoch": 64,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 25024,
+      "train_examples": 50000,
+      "train_loss": 5942.63540984375
+    },
+    {
+      "epoch": 65,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 25415,
+      "train_examples": 50000,
+      "train_loss": 6220.42150453125
+    },
+    {
+      "epoch": 66,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 25806,
+      "train_examples": 50000,
+      "train_loss": 6801.0142440625
+    },
+    {
+      "epoch": 67,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 26197,
+      "train_examples": 50000,
+      "train_loss": 6335.4780465625
+    },
+    {
+      "epoch": 68,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 26588,
+      "train_examples": 50000,
+      "train_loss": 6611.8889453125
+    },
+    {
+      "epoch": 69,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 26979,
+      "train_examples": 50000,
+      "train_loss": 7561.9144178125
+    },
+    {
+      "epoch": 70,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 27370,
+      "train_examples": 50000,
+      "train_loss": 7706.893026875
+    },
+    {
+      "epoch": 71,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 27761,
+      "train_examples": 50000,
+      "train_loss": 7683.9504371875
+    },
+    {
+      "epoch": 72,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 28152,
+      "train_examples": 50000,
+      "train_loss": 8865.1978615625
+    },
+    {
+      "epoch": 73,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 28543,
+      "train_examples": 50000,
+      "train_loss": 8177.27299953125
+    },
+    {
+      "epoch": 74,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 28934,
+      "train_examples": 50000,
+      "train_loss": 8050.2239734375
+    },
+    {
+      "epoch": 75,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 29325,
+      "train_examples": 50000,
+      "train_loss": 8500.7814559375
+    },
+    {
+      "epoch": 76,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 29716,
+      "train_examples": 50000,
+      "train_loss": 8926.3932909375
+    },
+    {
+      "epoch": 77,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 30107,
+      "train_examples": 50000,
+      "train_loss": 8798.44918875
+    },
+    {
+      "epoch": 78,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 30498,
+      "train_examples": 50000,
+      "train_loss": 9649.652635625
+    },
+    {
+      "epoch": 79,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 30889,
+      "train_examples": 50000,
+      "train_loss": 10129.6659884375
+    },
+    {
+      "epoch": 80,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 31280,
+      "train_examples": 50000,
+      "train_loss": 9658.81137625
+    },
+    {
+      "epoch": 81,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 31671,
+      "train_examples": 50000,
+      "train_loss": 10053.70207609375
+    },
+    {
+      "epoch": 82,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 32062,
+      "train_examples": 50000,
+      "train_loss": 10468.5874915625
+    },
+    {
+      "epoch": 83,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 32453,
+      "train_examples": 50000,
+      "train_loss": 10511.939255625
+    },
+    {
+      "epoch": 84,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 32844,
+      "train_examples": 50000,
+      "train_loss": 11405.4140125
+    },
+    {
+      "epoch": 85,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 33235,
+      "train_examples": 50000,
+      "train_loss": 13573.540273125
+    },
+    {
+      "epoch": 86,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 33626,
+      "train_examples": 50000,
+      "train_loss": 11169.5221915625
+    },
+    {
+      "epoch": 87,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 34017,
+      "train_examples": 50000,
+      "train_loss": 12833.8986875
+    },
+    {
+      "epoch": 88,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 34408,
+      "train_examples": 50000,
+      "train_loss": 12101.8815859375
+    },
+    {
+      "epoch": 89,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 34799,
+      "train_examples": 50000,
+      "train_loss": 12950.2093375
+    },
+    {
+      "epoch": 90,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 35190,
+      "train_examples": 50000,
+      "train_loss": 12925.1472246875
+    },
+    {
+      "epoch": 91,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 35581,
+      "train_examples": 50000,
+      "train_loss": 13137.740811875
+    },
+    {
+      "epoch": 92,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 35972,
+      "train_examples": 50000,
+      "train_loss": 13952.244633125
+    },
+    {
+      "epoch": 93,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 36363,
+      "train_examples": 50000,
+      "train_loss": 14272.286441875
+    },
+    {
+      "epoch": 94,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 36754,
+      "train_examples": 50000,
+      "train_loss": 14429.253629375
+    },
+    {
+      "epoch": 95,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 37145,
+      "train_examples": 50000,
+      "train_loss": 14849.0209059375
+    },
+    {
+      "epoch": 96,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 37536,
+      "train_examples": 50000,
+      "train_loss": 15120.0954565625
+    },
+    {
+      "epoch": 97,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 37927,
+      "train_examples": 50000,
+      "train_loss": 15741.8414025
+    },
+    {
+      "epoch": 98,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 38318,
+      "train_examples": 50000,
+      "train_loss": 16062.643003125
+    },
+    {
+      "epoch": 99,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 38709,
+      "train_examples": 50000,
+      "train_loss": 16957.89691375
+    },
+    {
+      "epoch": 100,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 39100,
+      "train_examples": 50000,
+      "train_loss": 16931.83966375
+    },
+    {
+      "epoch": 101,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 39491,
+      "train_examples": 50000,
+      "train_loss": 4130.02909078125
+    },
+    {
+      "epoch": 102,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 39882,
+      "train_examples": 50000,
+      "train_loss": 2515.256910625
+    },
+    {
+      "epoch": 103,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 40273,
+      "train_examples": 50000,
+      "train_loss": 2213.87051734375
+    },
+    {
+      "epoch": 104,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 40664,
+      "train_examples": 50000,
+      "train_loss": 2094.392287890625
+    },
+    {
+      "epoch": 105,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 41055,
+      "train_examples": 50000,
+      "train_loss": 1940.17430265625
+    },
+    {
+      "epoch": 106,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 41446,
+      "train_examples": 50000,
+      "train_loss": 1783.9225576953124
+    },
+    {
+      "epoch": 107,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 41837,
+      "train_examples": 50000,
+      "train_loss": 1632.972233046875
+    },
+    {
+      "epoch": 108,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 42228,
+      "train_examples": 50000,
+      "train_loss": 1545.16946125
+    },
+    {
+      "epoch": 109,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 42619,
+      "train_examples": 50000,
+      "train_loss": 1479.5323171875
+    },
+    {
+      "epoch": 110,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 43010,
+      "train_examples": 50000,
+      "train_loss": 1357.0434798046874
+    },
+    {
+      "epoch": 111,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 43401,
+      "train_examples": 50000,
+      "train_loss": 1282.2471298828125
+    },
+    {
+      "epoch": 112,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 43792,
+      "train_examples": 50000,
+      "train_loss": 1228.005009921875
+    },
+    {
+      "epoch": 113,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 44183,
+      "train_examples": 50000,
+      "train_loss": 1256.486622109375
+    },
+    {
+      "epoch": 114,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 44574,
+      "train_examples": 50000,
+      "train_loss": 1209.0455446875
+    },
+    {
+      "epoch": 115,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 44965,
+      "train_examples": 50000,
+      "train_loss": 1157.4591321484374
+    },
+    {
+      "epoch": 116,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 45356,
+      "train_examples": 50000,
+      "train_loss": 1197.2893698046876
+    },
+    {
+      "epoch": 117,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 45747,
+      "train_examples": 50000,
+      "train_loss": 1168.0829266796875
+    },
+    {
+      "epoch": 118,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 46138,
+      "train_examples": 50000,
+      "train_loss": 1162.90478453125
+    },
+    {
+      "epoch": 119,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 46529,
+      "train_examples": 50000,
+      "train_loss": 1268.49559921875
+    },
+    {
+      "epoch": 120,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 46920,
+      "train_examples": 50000,
+      "train_loss": 1171.266567890625
+    },
+    {
+      "epoch": 121,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 47311,
+      "train_examples": 50000,
+      "train_loss": 1259.3495566992187
+    },
+    {
+      "epoch": 122,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 47702,
+      "train_examples": 50000,
+      "train_loss": 1226.919266015625
+    },
+    {
+      "epoch": 123,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 48093,
+      "train_examples": 50000,
+      "train_loss": 1272.689976875
+    },
+    {
+      "epoch": 124,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 48484,
+      "train_examples": 50000,
+      "train_loss": 1207.8942051953125
+    },
+    {
+      "epoch": 125,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 48875,
+      "train_examples": 50000,
+      "train_loss": 1252.519639140625
+    },
+    {
+      "epoch": 126,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 49266,
+      "train_examples": 50000,
+      "train_loss": 1290.55786078125
+    },
+    {
+      "epoch": 127,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 49657,
+      "train_examples": 50000,
+      "train_loss": 1228.877055078125
+    },
+    {
+      "epoch": 128,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 50048,
+      "train_examples": 50000,
+      "train_loss": 1228.168096953125
+    },
+    {
+      "epoch": 129,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 50439,
+      "train_examples": 50000,
+      "train_loss": 1226.1055788867188
+    },
+    {
+      "epoch": 130,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 50830,
+      "train_examples": 50000,
+      "train_loss": 1368.034173828125
+    },
+    {
+      "epoch": 131,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 51221,
+      "train_examples": 50000,
+      "train_loss": 1175.8067020703124
+    },
+    {
+      "epoch": 132,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 51612,
+      "train_examples": 50000,
+      "train_loss": 1334.3213462109375
+    },
+    {
+      "epoch": 133,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 52003,
+      "train_examples": 50000,
+      "train_loss": 1247.6840731835937
+    },
+    {
+      "epoch": 134,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 52394,
+      "train_examples": 50000,
+      "train_loss": 1313.6224965625
+    },
+    {
+      "epoch": 135,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 52785,
+      "train_examples": 50000,
+      "train_loss": 1221.8752206640625
+    },
+    {
+      "epoch": 136,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 53176,
+      "train_examples": 50000,
+      "train_loss": 1328.462998671875
+    },
+    {
+      "epoch": 137,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 53567,
+      "train_examples": 50000,
+      "train_loss": 1275.940348515625
+    },
+    {
+      "epoch": 138,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 53958,
+      "train_examples": 50000,
+      "train_loss": 1252.2436762109376
+    },
+    {
+      "epoch": 139,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 54349,
+      "train_examples": 50000,
+      "train_loss": 1305.7217723828126
+    },
+    {
+      "epoch": 140,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 54740,
+      "train_examples": 50000,
+      "train_loss": 1340.4838646484375
+    },
+    {
+      "epoch": 141,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 55131,
+      "train_examples": 50000,
+      "train_loss": 1228.717749453125
+    },
+    {
+      "epoch": 142,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 55522,
+      "train_examples": 50000,
+      "train_loss": 1456.4811312109375
+    },
+    {
+      "epoch": 143,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 55913,
+      "train_examples": 50000,
+      "train_loss": 1288.603296171875
+    },
+    {
+      "epoch": 144,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 56304,
+      "train_examples": 50000,
+      "train_loss": 1299.126255078125
+    },
+    {
+      "epoch": 145,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 56695,
+      "train_examples": 50000,
+      "train_loss": 1265.7983320703124
+    },
+    {
+      "epoch": 146,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 57086,
+      "train_examples": 50000,
+      "train_loss": 1291.65321625
+    },
+    {
+      "epoch": 147,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 57477,
+      "train_examples": 50000,
+      "train_loss": 1210.3560226171876
+    },
+    {
+      "epoch": 148,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 57868,
+      "train_examples": 50000,
+      "train_loss": 1396.2364810546876
+    },
+    {
+      "epoch": 149,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 58259,
+      "train_examples": 50000,
+      "train_loss": 1282.69460765625
+    },
+    {
+      "epoch": 150,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 58650,
+      "train_examples": 50000,
+      "train_loss": 1252.808085390625
+    },
+    {
+      "epoch": 151,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 59041,
+      "train_examples": 50000,
+      "train_loss": 409.71976322265624
+    },
+    {
+      "epoch": 152,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 59432,
+      "train_examples": 50000,
+      "train_loss": 361.3236141210937
+    },
+    {
+      "epoch": 153,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 59823,
+      "train_examples": 50000,
+      "train_loss": 336.7964623730469
+    },
+    {
+      "epoch": 154,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 60214,
+      "train_examples": 50000,
+      "train_loss": 316.91355252929685
+    },
+    {
+      "epoch": 155,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 60605,
+      "train_examples": 50000,
+      "train_loss": 301.974838515625
+    },
+    {
+      "epoch": 156,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 60996,
+      "train_examples": 50000,
+      "train_loss": 290.2720501855469
+    },
+    {
+      "epoch": 157,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 61387,
+      "train_examples": 50000,
+      "train_loss": 280.62712240234373
+    },
+    {
+      "epoch": 158,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 61778,
+      "train_examples": 50000,
+      "train_loss": 272.9833394970703
+    },
+    {
+      "epoch": 159,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 62169,
+      "train_examples": 50000,
+      "train_loss": 259.4398801367187
+    },
+    {
+      "epoch": 160,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 62560,
+      "train_examples": 50000,
+      "train_loss": 249.6847511279297
+    },
+    {
+      "epoch": 161,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 62951,
+      "train_examples": 50000,
+      "train_loss": 246.0232903857422
+    },
+    {
+      "epoch": 162,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 63342,
+      "train_examples": 50000,
+      "train_loss": 230.45652196777343
+    },
+    {
+      "epoch": 163,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 63733,
+      "train_examples": 50000,
+      "train_loss": 225.3271287988281
+    },
+    {
+      "epoch": 164,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 64124,
+      "train_examples": 50000,
+      "train_loss": 218.68095298828126
+    },
+    {
+      "epoch": 165,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 64515,
+      "train_examples": 50000,
+      "train_loss": 222.911926796875
+    },
+    {
+      "epoch": 166,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 64906,
+      "train_examples": 50000,
+      "train_loss": 216.82613666992188
+    },
+    {
+      "epoch": 167,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 65297,
+      "train_examples": 50000,
+      "train_loss": 206.59793936523437
+    },
+    {
+      "epoch": 168,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 65688,
+      "train_examples": 50000,
+      "train_loss": 200.18790624023438
+    },
+    {
+      "epoch": 169,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 66079,
+      "train_examples": 50000,
+      "train_loss": 202.72470991210938
+    },
+    {
+      "epoch": 170,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 66470,
+      "train_examples": 50000,
+      "train_loss": 199.54272793945313
+    },
+    {
+      "epoch": 171,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 66861,
+      "train_examples": 50000,
+      "train_loss": 188.69470041992187
+    },
+    {
+      "epoch": 172,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 67252,
+      "train_examples": 50000,
+      "train_loss": 190.51382822753905
+    },
+    {
+      "epoch": 173,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 67643,
+      "train_examples": 50000,
+      "train_loss": 188.02758157714842
+    },
+    {
+      "epoch": 174,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 68034,
+      "train_examples": 50000,
+      "train_loss": 186.09762303710937
+    },
+    {
+      "epoch": 175,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 68425,
+      "train_examples": 50000,
+      "train_loss": 183.76290793945313
+    },
+    {
+      "epoch": 176,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 68816,
+      "train_examples": 50000,
+      "train_loss": 181.99154977539064
+    },
+    {
+      "epoch": 177,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 69207,
+      "train_examples": 50000,
+      "train_loss": 185.6296263623047
+    },
+    {
+      "epoch": 178,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 69598,
+      "train_examples": 50000,
+      "train_loss": 185.9326528515625
+    },
+    {
+      "epoch": 179,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 69989,
+      "train_examples": 50000,
+      "train_loss": 173.08992820800782
+    },
+    {
+      "epoch": 180,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 70380,
+      "train_examples": 50000,
+      "train_loss": 173.0970648486328
+    },
+    {
+      "epoch": 181,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 70771,
+      "train_examples": 50000,
+      "train_loss": 174.8073498828125
+    },
+    {
+      "epoch": 182,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 71162,
+      "train_examples": 50000,
+      "train_loss": 173.26389734375
+    },
+    {
+      "epoch": 183,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 71553,
+      "train_examples": 50000,
+      "train_loss": 178.76224380859375
+    },
+    {
+      "epoch": 184,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 71944,
+      "train_examples": 50000,
+      "train_loss": 177.17293513183594
+    },
+    {
+      "epoch": 185,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 72335,
+      "train_examples": 50000,
+      "train_loss": 166.40748666503907
+    },
+    {
+      "epoch": 186,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 72726,
+      "train_examples": 50000,
+      "train_loss": 173.59480986816405
+    },
+    {
+      "epoch": 187,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 73117,
+      "train_examples": 50000,
+      "train_loss": 179.32762776367187
+    },
+    {
+      "epoch": 188,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 73508,
+      "train_examples": 50000,
+      "train_loss": 171.93528721679687
+    },
+    {
+      "epoch": 189,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 73899,
+      "train_examples": 50000,
+      "train_loss": 169.06438075195314
+    },
+    {
+      "epoch": 190,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 74290,
+      "train_examples": 50000,
+      "train_loss": 165.37803157226563
+    },
+    {
+      "epoch": 191,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 74681,
+      "train_examples": 50000,
+      "train_loss": 174.83478853515626
+    },
+    {
+      "epoch": 192,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 75072,
+      "train_examples": 50000,
+      "train_loss": 170.7552182470703
+    },
+    {
+      "epoch": 193,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 75463,
+      "train_examples": 50000,
+      "train_loss": 165.30579611816407
+    },
+    {
+      "epoch": 194,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 75854,
+      "train_examples": 50000,
+      "train_loss": 171.17787514648438
+    },
+    {
+      "epoch": 195,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 76245,
+      "train_examples": 50000,
+      "train_loss": 164.37082291015625
+    },
+    {
+      "epoch": 196,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 76636,
+      "train_examples": 50000,
+      "train_loss": 184.00646500976563
+    },
+    {
+      "epoch": 197,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 77027,
+      "train_examples": 50000,
+      "train_loss": 164.81874224365234
+    },
+    {
+      "epoch": 198,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 77418,
+      "train_examples": 50000,
+      "train_loss": 165.01198385253906
+    },
+    {
+      "epoch": 199,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 77809,
+      "train_examples": 50000,
+      "train_loss": 168.083658046875
+    },
+    {
+      "epoch": 200,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 78200,
+      "train_examples": 50000,
+      "train_loss": 166.8696224951172
+    }
+  ],
+  "evaluation_protocol": {
+    "test_evaluations": 1,
+    "test_used_for_selection": false,
+    "validation_evaluations": 0
+  },
+  "final": {
+    "accuracy": 0.3264,
+    "epoch": 200,
+    "evaluation_split": "test",
+    "finite": true,
+    "loss": 131.45226796875,
+    "step": 78200
+  },
+  "hardware": {
+    "cuda_device_name": "NVIDIA GeForce GTX 1080",
+    "cuda_visible_devices": "7",
+    "device": "cuda",
+    "device_total_memory_bytes": 8507949056,
+    "peak_memory_allocated_bytes": 2537912832,
+    "peak_memory_reserved_bytes": 3323985920,
+    "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": 1.146092176437378,
+    "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": 5907.313760757446,
+    "train_wall_s": 5906.142598867416
+  },
+  "work": {
+    "apical_macs_per_example": 3031040,
+    "causal_scalar_observations": 0,
+    "components": {
+      "apical_projection_macs": 30310400000000,
+      "apical_regression_macs": 0,
+      "bp_reverse_macs_estimate": 0,
+      "kp_reciprocal_correlation_macs": 0,
+      "local_weight_correlation_macs": 688625920000000,
+      "mirror_feedback_prediction_macs": 0,
+      "mirror_local_correlation_macs": 0,
+      "mirror_response_macs": 0,
+      "ordinary_forward_macs": 688625920000000,
+      "perturbation_forward_macs": 0,
+      "warmup_clean_forward_macs": 0
+    },
+    "definition": "multiply-accumulates in conv/linear maps; one local weight correlation equals one forward-weight MAC count; BP reverse is estimated as one weight-gradient plus one activation-gradient convolution per forward convolution; mixed-traffic/predictor elementwise arithmetic is reported as a conservative operation estimate separately and is not folded into MACs",
+    "elementwise_operations_estimate": 0,
+    "forward_macs_per_example": 68862592,
+    "logical_batch_loss_queries": 0,
+    "mirror_probe_examples": 0,
+    "mirror_readout_probe_examples": 0,
+    "neutral_projection_observations": 0,
+    "per_example_cross_entropy_terms": 0,
+    "total_clean_forward_examples": 10000000,
+    "total_forward_equivalent_examples": 10000000,
+    "total_macs_estimate": 1407562240000000
+  }
+}