results: confirm standard-depth SDIL scaling

1 files changed, 2281 insertions, 0 deletions

diff --git a/results/oral_a_dynamic_scaling_v2/dfa_d56_s14.json b/results/oral_a_dynamic_scaling_v2/dfa_d56_s14.json
new file mode 100644
index 0000000..828f83e
--- /dev/null
+++ b/results/oral_a_dynamic_scaling_v2/dfa_d56_s14.json
@@ -0,0 +1,2281 @@
+{
+  "architecture": {
+    "adaptive_apical_parameters": 6389760,
+    "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": 532480,
+    "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": 1064960,
+    "residual_scale": 1.0,
+    "vectorizer_mode": "spatial_template",
+    "vectorizer_parameters": 5324800
+  },
+  "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": 56,
+    "device": "cuda",
+    "epochs": 200,
+    "eta_A": 0.01,
+    "eta_P": 0.01,
+    "eval_every": 0,
+    "eval_split": "test",
+    "learn_P": 0,
+    "loader_seed": 14,
+    "lr": 0.03,
+    "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_d56_s14.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": 14,
+    "split_seed": 2027,
+    "traffic_calibration_examples": 64,
+    "traffic_ratio": 4.0,
+    "traffic_rule": "innovation",
+    "traffic_seed": 4000,
+    "train_limit": 0,
+    "use_residual": 1,
+    "val_examples": 0,
+    "vectorizer_mode": "spatial_template",
+    "warmup_epochs": 0,
+    "weight_decay": 0.0001,
+    "weight_scale": 1.0,
+    "width": 16
+  },
+  "calibration_metric_space": "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": 0.0003577964930122511,
+    "innovation_negative_gradient_cosine": [
+      0.002081854734569788,
+      -6.581633351743221e-05,
+      0.00041557435179129243,
+      0.0008862123358994722,
+      -0.0005109517369419336,
+      0.00042458175448700786,
+      0.0027301099617034197,
+      0.001137684565037489,
+      -0.0005082294228486717,
+      -0.0018658648477867246,
+      0.0010939594358205795,
+      -0.0005554755916818976,
+      -0.0008758855401538312,
+      0.001304553123190999,
+      -0.0017698571318760514,
+      0.0007898603216744959,
+      0.0013096316251903772,
+      0.0004183952696621418,
+      -0.0008684020722284913,
+      -5.6430697441101074e-05,
+      -0.0011641886085271835,
+      -0.0017775511369109154,
+      0.0027877171523869038,
+      0.0006546652875840664,
+      0.0018238320481032133,
+      0.0007054642192088068,
+      -0.00016221619443967938,
+      0.003270443296059966,
+      0.0008029259042814374,
+      0.0014108323957771063,
+      0.002188350073993206,
+      0.0003249559667892754,
+      0.0020819089841097593,
+      0.0025527561083436012,
+      0.00035986059810966253,
+      0.0027287080883979797,
+      -0.0007787201902829111,
+      0.009754732251167297,
+      0.005875065922737122,
+      0.005507401190698147,
+      0.00709145050495863,
+      0.008006663993000984,
+      0.011602742597460747,
+      0.011238088831305504,
+      0.011543793603777885,
+      0.008512252941727638,
+      0.014168035238981247,
+      0.006046508438885212,
+      0.00905943289399147,
+      0.006326874252408743,
+      0.00905679538846016,
+      0.016640275716781616,
+      0.019679304212331772,
+      0.013295481912791729,
+      0.017547961324453354
+    ],
+    "normalization_state": "training_batch_stats_without_running_update",
+    "raw_negative_gradient_cosine": [
+      0.002081854734569788,
+      -6.581633351743221e-05,
+      0.00041557435179129243,
+      0.0008862123358994722,
+      -0.0005109517369419336,
+      0.00042458175448700786,
+      0.0027301099617034197,
+      0.001137684565037489,
+      -0.0005082294228486717,
+      -0.0018658648477867246,
+      0.0010939594358205795,
+      -0.0005554755916818976,
+      -0.0008758855401538312,
+      0.001304553123190999,
+      -0.0017698571318760514,
+      0.0007898603216744959,
+      0.0013096316251903772,
+      0.0004183952696621418,
+      -0.0008684020722284913,
+      -5.6430697441101074e-05,
+      -0.0011641886085271835,
+      -0.0017775511369109154,
+      0.0027877171523869038,
+      0.0006546652875840664,
+      0.0018238320481032133,
+      0.0007054642192088068,
+      -0.00016221619443967938,
+      0.003270443296059966,
+      0.0008029259042814374,
+      0.0014108323957771063,
+      0.002188350073993206,
+      0.0003249559667892754,
+      0.0020819089841097593,
+      0.0025527561083436012,
+      0.00035986059810966253,
+      0.0027287080883979797,
+      -0.0007787201902829111,
+      0.009754732251167297,
+      0.005875065922737122,
+      0.005507401190698147,
+      0.00709145050495863,
+      0.008006663993000984,
+      0.011602742597460747,
+      0.011238088831305504,
+      0.011543793603777885,
+      0.008512252941727638,
+      0.014168035238981247,
+      0.006046508438885212,
+      0.00905943289399147,
+      0.006326874252408743,
+      0.00905679538846016,
+      0.016640275716781616,
+      0.019679304212331772,
+      0.013295481912791729,
+      0.017547961324453354
+    ],
+    "teaching_negative_gradient_cosine": [
+      0.002081854734569788,
+      -6.581633351743221e-05,
+      0.00041557435179129243,
+      0.0008862123358994722,
+      -0.0005109517369419336,
+      0.00042458175448700786,
+      0.0027301099617034197,
+      0.001137684565037489,
+      -0.0005082294228486717,
+      -0.0018658648477867246,
+      0.0010939594358205795,
+      -0.0005554755916818976,
+      -0.0008758855401538312,
+      0.001304553123190999,
+      -0.0017698571318760514,
+      0.0007898603216744959,
+      0.0013096316251903772,
+      0.0004183952696621418,
+      -0.0008684020722284913,
+      -5.6430697441101074e-05,
+      -0.0011641886085271835,
+      -0.0017775511369109154,
+      0.0027877171523869038,
+      0.0006546652875840664,
+      0.0018238320481032133,
+      0.0007054642192088068,
+      -0.00016221619443967938,
+      0.003270443296059966,
+      0.0008029259042814374,
+      0.0014108323957771063,
+      0.002188350073993206,
+      0.0003249559667892754,
+      0.0020819089841097593,
+      0.0025527561083436012,
+      0.00035986059810966253,
+      0.0027287080883979797,
+      -0.0007787201902829111,
+      0.009754732251167297,
+      0.005875065922737122,
+      0.005507401190698147,
+      0.00709145050495863,
+      0.008006663993000984,
+      0.011602742597460747,
+      0.011238088831305504,
+      0.011543793603777885,
+      0.008512252941727638,
+      0.014168035238981247,
+      0.006046508438885212,
+      0.00905943289399147,
+      0.006326874252408743,
+      0.00905679538846016,
+      0.016640275716781616,
+      0.019679304212331772,
+      0.013295481912791729,
+      0.017547961324453354
+    ],
+    "wall_s": 0.2199254035949707
+  },
+  "epochs": [
+    {
+      "epoch": 1,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 391,
+      "train_examples": 50000,
+      "train_loss": 20.490372321777343
+    },
+    {
+      "epoch": 2,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 782,
+      "train_examples": 50000,
+      "train_loss": 16.882818517456055
+    },
+    {
+      "epoch": 3,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 1173,
+      "train_examples": 50000,
+      "train_loss": 22.28293617919922
+    },
+    {
+      "epoch": 4,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 1564,
+      "train_examples": 50000,
+      "train_loss": 32.22360906982422
+    },
+    {
+      "epoch": 5,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 1955,
+      "train_examples": 50000,
+      "train_loss": 46.52625292114258
+    },
+    {
+      "epoch": 6,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 2346,
+      "train_examples": 50000,
+      "train_loss": 61.080078745117184
+    },
+    {
+      "epoch": 7,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 2737,
+      "train_examples": 50000,
+      "train_loss": 69.91611538208008
+    },
+    {
+      "epoch": 8,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 3128,
+      "train_examples": 50000,
+      "train_loss": 92.61200556030273
+    },
+    {
+      "epoch": 9,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 3519,
+      "train_examples": 50000,
+      "train_loss": 110.42092873291016
+    },
+    {
+      "epoch": 10,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 3910,
+      "train_examples": 50000,
+      "train_loss": 139.7512891064453
+    },
+    {
+      "epoch": 11,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 4301,
+      "train_examples": 50000,
+      "train_loss": 169.34016850585937
+    },
+    {
+      "epoch": 12,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 4692,
+      "train_examples": 50000,
+      "train_loss": 227.0122556982422
+    },
+    {
+      "epoch": 13,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 5083,
+      "train_examples": 50000,
+      "train_loss": 292.92110482421873
+    },
+    {
+      "epoch": 14,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 5474,
+      "train_examples": 50000,
+      "train_loss": 361.45023318359375
+    },
+    {
+      "epoch": 15,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 5865,
+      "train_examples": 50000,
+      "train_loss": 426.76469272460935
+    },
+    {
+      "epoch": 16,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 6256,
+      "train_examples": 50000,
+      "train_loss": 500.7480604199219
+    },
+    {
+      "epoch": 17,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 6647,
+      "train_examples": 50000,
+      "train_loss": 546.9457539648438
+    },
+    {
+      "epoch": 18,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 7038,
+      "train_examples": 50000,
+      "train_loss": 724.0486225195312
+    },
+    {
+      "epoch": 19,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 7429,
+      "train_examples": 50000,
+      "train_loss": 833.3311554785156
+    },
+    {
+      "epoch": 20,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 7820,
+      "train_examples": 50000,
+      "train_loss": 1041.0364372070312
+    },
+    {
+      "epoch": 21,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 8211,
+      "train_examples": 50000,
+      "train_loss": 1248.4833929296874
+    },
+    {
+      "epoch": 22,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 8602,
+      "train_examples": 50000,
+      "train_loss": 1350.1294877734374
+    },
+    {
+      "epoch": 23,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 8993,
+      "train_examples": 50000,
+      "train_loss": 1710.98063296875
+    },
+    {
+      "epoch": 24,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 9384,
+      "train_examples": 50000,
+      "train_loss": 2209.1547815625
+    },
+    {
+      "epoch": 25,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 9775,
+      "train_examples": 50000,
+      "train_loss": 2441.934182890625
+    },
+    {
+      "epoch": 26,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 10166,
+      "train_examples": 50000,
+      "train_loss": 2529.25472796875
+    },
+    {
+      "epoch": 27,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 10557,
+      "train_examples": 50000,
+      "train_loss": 2887.71826140625
+    },
+    {
+      "epoch": 28,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 10948,
+      "train_examples": 50000,
+      "train_loss": 3126.71301359375
+    },
+    {
+      "epoch": 29,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 11339,
+      "train_examples": 50000,
+      "train_loss": 3453.0569534375
+    },
+    {
+      "epoch": 30,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 11730,
+      "train_examples": 50000,
+      "train_loss": 4313.30736875
+    },
+    {
+      "epoch": 31,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 12121,
+      "train_examples": 50000,
+      "train_loss": 5114.32865453125
+    },
+    {
+      "epoch": 32,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 12512,
+      "train_examples": 50000,
+      "train_loss": 5369.68033453125
+    },
+    {
+      "epoch": 33,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 12903,
+      "train_examples": 50000,
+      "train_loss": 5650.06040875
+    },
+    {
+      "epoch": 34,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 13294,
+      "train_examples": 50000,
+      "train_loss": 6635.95551671875
+    },
+    {
+      "epoch": 35,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 13685,
+      "train_examples": 50000,
+      "train_loss": 6293.6670246875
+    },
+    {
+      "epoch": 36,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 14076,
+      "train_examples": 50000,
+      "train_loss": 7230.133379375
+    },
+    {
+      "epoch": 37,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 14467,
+      "train_examples": 50000,
+      "train_loss": 8361.1046715625
+    },
+    {
+      "epoch": 38,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 14858,
+      "train_examples": 50000,
+      "train_loss": 8847.93686375
+    },
+    {
+      "epoch": 39,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 15249,
+      "train_examples": 50000,
+      "train_loss": 9081.9119225
+    },
+    {
+      "epoch": 40,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 15640,
+      "train_examples": 50000,
+      "train_loss": 10225.6603215625
+    },
+    {
+      "epoch": 41,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 16031,
+      "train_examples": 50000,
+      "train_loss": 10560.681274375
+    },
+    {
+      "epoch": 42,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 16422,
+      "train_examples": 50000,
+      "train_loss": 10014.091368125
+    },
+    {
+      "epoch": 43,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 16813,
+      "train_examples": 50000,
+      "train_loss": 12802.432016875
+    },
+    {
+      "epoch": 44,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 17204,
+      "train_examples": 50000,
+      "train_loss": 12514.8965871875
+    },
+    {
+      "epoch": 45,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 17595,
+      "train_examples": 50000,
+      "train_loss": 13962.953141875
+    },
+    {
+      "epoch": 46,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 17986,
+      "train_examples": 50000,
+      "train_loss": 15282.5079678125
+    },
+    {
+      "epoch": 47,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 18377,
+      "train_examples": 50000,
+      "train_loss": 14999.5047015625
+    },
+    {
+      "epoch": 48,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 18768,
+      "train_examples": 50000,
+      "train_loss": 15210.9302703125
+    },
+    {
+      "epoch": 49,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 19159,
+      "train_examples": 50000,
+      "train_loss": 17634.892895625
+    },
+    {
+      "epoch": 50,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 19550,
+      "train_examples": 50000,
+      "train_loss": 16799.24489625
+    },
+    {
+      "epoch": 51,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 19941,
+      "train_examples": 50000,
+      "train_loss": 19577.6702021875
+    },
+    {
+      "epoch": 52,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 20332,
+      "train_examples": 50000,
+      "train_loss": 18435.589159375
+    },
+    {
+      "epoch": 53,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 20723,
+      "train_examples": 50000,
+      "train_loss": 22150.93953
+    },
+    {
+      "epoch": 54,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 21114,
+      "train_examples": 50000,
+      "train_loss": 19658.92122875
+    },
+    {
+      "epoch": 55,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 21505,
+      "train_examples": 50000,
+      "train_loss": 21042.8679015625
+    },
+    {
+      "epoch": 56,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 21896,
+      "train_examples": 50000,
+      "train_loss": 21857.45334375
+    },
+    {
+      "epoch": 57,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 22287,
+      "train_examples": 50000,
+      "train_loss": 25517.614735625
+    },
+    {
+      "epoch": 58,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 22678,
+      "train_examples": 50000,
+      "train_loss": 25244.549959375
+    },
+    {
+      "epoch": 59,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 23069,
+      "train_examples": 50000,
+      "train_loss": 26335.128986875
+    },
+    {
+      "epoch": 60,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 23460,
+      "train_examples": 50000,
+      "train_loss": 27385.4901175
+    },
+    {
+      "epoch": 61,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 23851,
+      "train_examples": 50000,
+      "train_loss": 28137.03504
+    },
+    {
+      "epoch": 62,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 24242,
+      "train_examples": 50000,
+      "train_loss": 27979.586675
+    },
+    {
+      "epoch": 63,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 24633,
+      "train_examples": 50000,
+      "train_loss": 33889.63128375
+    },
+    {
+      "epoch": 64,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 25024,
+      "train_examples": 50000,
+      "train_loss": 32506.89203625
+    },
+    {
+      "epoch": 65,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 25415,
+      "train_examples": 50000,
+      "train_loss": 34838.36573
+    },
+    {
+      "epoch": 66,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 25806,
+      "train_examples": 50000,
+      "train_loss": 32596.39202875
+    },
+    {
+      "epoch": 67,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 26197,
+      "train_examples": 50000,
+      "train_loss": 39124.6518253125
+    },
+    {
+      "epoch": 68,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 26588,
+      "train_examples": 50000,
+      "train_loss": 36587.51864875
+    },
+    {
+      "epoch": 69,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 26979,
+      "train_examples": 50000,
+      "train_loss": 38581.61796375
+    },
+    {
+      "epoch": 70,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 27370,
+      "train_examples": 50000,
+      "train_loss": 39131.21147625
+    },
+    {
+      "epoch": 71,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 27761,
+      "train_examples": 50000,
+      "train_loss": 39419.66956375
+    },
+    {
+      "epoch": 72,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 28152,
+      "train_examples": 50000,
+      "train_loss": 45149.3276175
+    },
+    {
+      "epoch": 73,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 28543,
+      "train_examples": 50000,
+      "train_loss": 43675.2919675
+    },
+    {
+      "epoch": 74,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 28934,
+      "train_examples": 50000,
+      "train_loss": 43209.59469
+    },
+    {
+      "epoch": 75,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 29325,
+      "train_examples": 50000,
+      "train_loss": 47960.806325
+    },
+    {
+      "epoch": 76,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 29716,
+      "train_examples": 50000,
+      "train_loss": 45578.00503375
+    },
+    {
+      "epoch": 77,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 30107,
+      "train_examples": 50000,
+      "train_loss": 50038.49809
+    },
+    {
+      "epoch": 78,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 30498,
+      "train_examples": 50000,
+      "train_loss": 50708.95643125
+    },
+    {
+      "epoch": 79,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 30889,
+      "train_examples": 50000,
+      "train_loss": 49882.63458
+    },
+    {
+      "epoch": 80,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 31280,
+      "train_examples": 50000,
+      "train_loss": 52694.5556475
+    },
+    {
+      "epoch": 81,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 31671,
+      "train_examples": 50000,
+      "train_loss": 52183.03786375
+    },
+    {
+      "epoch": 82,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 32062,
+      "train_examples": 50000,
+      "train_loss": 57154.52413375
+    },
+    {
+      "epoch": 83,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 32453,
+      "train_examples": 50000,
+      "train_loss": 61268.2211825
+    },
+    {
+      "epoch": 84,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 32844,
+      "train_examples": 50000,
+      "train_loss": 61062.56609
+    },
+    {
+      "epoch": 85,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 33235,
+      "train_examples": 50000,
+      "train_loss": 63559.830705
+    },
+    {
+      "epoch": 86,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 33626,
+      "train_examples": 50000,
+      "train_loss": 67166.77463
+    },
+    {
+      "epoch": 87,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 34017,
+      "train_examples": 50000,
+      "train_loss": 73203.6220075
+    },
+    {
+      "epoch": 88,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 34408,
+      "train_examples": 50000,
+      "train_loss": 62199.4245425
+    },
+    {
+      "epoch": 89,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 34799,
+      "train_examples": 50000,
+      "train_loss": 77680.6113875
+    },
+    {
+      "epoch": 90,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 35190,
+      "train_examples": 50000,
+      "train_loss": 64505.8588025
+    },
+    {
+      "epoch": 91,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 35581,
+      "train_examples": 50000,
+      "train_loss": 82246.832105
+    },
+    {
+      "epoch": 92,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 35972,
+      "train_examples": 50000,
+      "train_loss": 72240.0251025
+    },
+    {
+      "epoch": 93,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 36363,
+      "train_examples": 50000,
+      "train_loss": 74938.227655
+    },
+    {
+      "epoch": 94,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 36754,
+      "train_examples": 50000,
+      "train_loss": 73391.8108975
+    },
+    {
+      "epoch": 95,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 37145,
+      "train_examples": 50000,
+      "train_loss": 80380.57861
+    },
+    {
+      "epoch": 96,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 37536,
+      "train_examples": 50000,
+      "train_loss": 86301.0710025
+    },
+    {
+      "epoch": 97,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 37927,
+      "train_examples": 50000,
+      "train_loss": 80236.23262
+    },
+    {
+      "epoch": 98,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 38318,
+      "train_examples": 50000,
+      "train_loss": 86720.108125
+    },
+    {
+      "epoch": 99,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 38709,
+      "train_examples": 50000,
+      "train_loss": 85931.05704125
+    },
+    {
+      "epoch": 100,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 39100,
+      "train_examples": 50000,
+      "train_loss": 87120.586675
+    },
+    {
+      "epoch": 101,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 39491,
+      "train_examples": 50000,
+      "train_loss": 16651.018333125
+    },
+    {
+      "epoch": 102,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 39882,
+      "train_examples": 50000,
+      "train_loss": 12207.6655284375
+    },
+    {
+      "epoch": 103,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 40273,
+      "train_examples": 50000,
+      "train_loss": 11093.006951875
+    },
+    {
+      "epoch": 104,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 40664,
+      "train_examples": 50000,
+      "train_loss": 10483.7009175
+    },
+    {
+      "epoch": 105,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 41055,
+      "train_examples": 50000,
+      "train_loss": 10113.0312990625
+    },
+    {
+      "epoch": 106,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 41446,
+      "train_examples": 50000,
+      "train_loss": 9756.068164375
+    },
+    {
+      "epoch": 107,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 41837,
+      "train_examples": 50000,
+      "train_loss": 9243.780345625
+    },
+    {
+      "epoch": 108,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 42228,
+      "train_examples": 50000,
+      "train_loss": 8713.2818921875
+    },
+    {
+      "epoch": 109,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 42619,
+      "train_examples": 50000,
+      "train_loss": 8324.015185
+    },
+    {
+      "epoch": 110,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 43010,
+      "train_examples": 50000,
+      "train_loss": 7787.7575653125
+    },
+    {
+      "epoch": 111,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 43401,
+      "train_examples": 50000,
+      "train_loss": 7442.3282865625
+    },
+    {
+      "epoch": 112,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 43792,
+      "train_examples": 50000,
+      "train_loss": 7383.6853378125
+    },
+    {
+      "epoch": 113,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 44183,
+      "train_examples": 50000,
+      "train_loss": 6795.67533390625
+    },
+    {
+      "epoch": 114,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 44574,
+      "train_examples": 50000,
+      "train_loss": 7012.378193125
+    },
+    {
+      "epoch": 115,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 44965,
+      "train_examples": 50000,
+      "train_loss": 6524.51093484375
+    },
+    {
+      "epoch": 116,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 45356,
+      "train_examples": 50000,
+      "train_loss": 6777.3454996875
+    },
+    {
+      "epoch": 117,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 45747,
+      "train_examples": 50000,
+      "train_loss": 6447.0069984375
+    },
+    {
+      "epoch": 118,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 46138,
+      "train_examples": 50000,
+      "train_loss": 5999.93104765625
+    },
+    {
+      "epoch": 119,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 46529,
+      "train_examples": 50000,
+      "train_loss": 6029.22547125
+    },
+    {
+      "epoch": 120,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 46920,
+      "train_examples": 50000,
+      "train_loss": 6563.4999215625
+    },
+    {
+      "epoch": 121,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 47311,
+      "train_examples": 50000,
+      "train_loss": 6040.0361803125
+    },
+    {
+      "epoch": 122,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 47702,
+      "train_examples": 50000,
+      "train_loss": 6728.580324375
+    },
+    {
+      "epoch": 123,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 48093,
+      "train_examples": 50000,
+      "train_loss": 6468.49931953125
+    },
+    {
+      "epoch": 124,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 48484,
+      "train_examples": 50000,
+      "train_loss": 6330.526211875
+    },
+    {
+      "epoch": 125,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 48875,
+      "train_examples": 50000,
+      "train_loss": 6213.6745078125
+    },
+    {
+      "epoch": 126,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 49266,
+      "train_examples": 50000,
+      "train_loss": 6347.12538546875
+    },
+    {
+      "epoch": 127,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 49657,
+      "train_examples": 50000,
+      "train_loss": 6139.44533984375
+    },
+    {
+      "epoch": 128,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 50048,
+      "train_examples": 50000,
+      "train_loss": 6085.03322859375
+    },
+    {
+      "epoch": 129,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 50439,
+      "train_examples": 50000,
+      "train_loss": 6121.663730625
+    },
+    {
+      "epoch": 130,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 50830,
+      "train_examples": 50000,
+      "train_loss": 6252.5325828125
+    },
+    {
+      "epoch": 131,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 51221,
+      "train_examples": 50000,
+      "train_loss": 6137.2040996875
+    },
+    {
+      "epoch": 132,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 51612,
+      "train_examples": 50000,
+      "train_loss": 5885.20040140625
+    },
+    {
+      "epoch": 133,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 52003,
+      "train_examples": 50000,
+      "train_loss": 6218.637215625
+    },
+    {
+      "epoch": 134,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 52394,
+      "train_examples": 50000,
+      "train_loss": 5887.972265
+    },
+    {
+      "epoch": 135,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 52785,
+      "train_examples": 50000,
+      "train_loss": 6469.8672965625
+    },
+    {
+      "epoch": 136,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 53176,
+      "train_examples": 50000,
+      "train_loss": 5943.174265625
+    },
+    {
+      "epoch": 137,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 53567,
+      "train_examples": 50000,
+      "train_loss": 6216.42217015625
+    },
+    {
+      "epoch": 138,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 53958,
+      "train_examples": 50000,
+      "train_loss": 6384.10118359375
+    },
+    {
+      "epoch": 139,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 54349,
+      "train_examples": 50000,
+      "train_loss": 6455.990405
+    },
+    {
+      "epoch": 140,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 54740,
+      "train_examples": 50000,
+      "train_loss": 6119.31162203125
+    },
+    {
+      "epoch": 141,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 55131,
+      "train_examples": 50000,
+      "train_loss": 6515.5113884375
+    },
+    {
+      "epoch": 142,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 55522,
+      "train_examples": 50000,
+      "train_loss": 6712.253388125
+    },
+    {
+      "epoch": 143,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 55913,
+      "train_examples": 50000,
+      "train_loss": 6020.62121953125
+    },
+    {
+      "epoch": 144,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 56304,
+      "train_examples": 50000,
+      "train_loss": 6674.37236703125
+    },
+    {
+      "epoch": 145,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 56695,
+      "train_examples": 50000,
+      "train_loss": 6140.965126953125
+    },
+    {
+      "epoch": 146,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 57086,
+      "train_examples": 50000,
+      "train_loss": 6570.85379765625
+    },
+    {
+      "epoch": 147,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 57477,
+      "train_examples": 50000,
+      "train_loss": 6375.9274646875
+    },
+    {
+      "epoch": 148,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 57868,
+      "train_examples": 50000,
+      "train_loss": 6504.94818796875
+    },
+    {
+      "epoch": 149,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 58259,
+      "train_examples": 50000,
+      "train_loss": 5907.9925953125
+    },
+    {
+      "epoch": 150,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 58650,
+      "train_examples": 50000,
+      "train_loss": 7468.96049375
+    },
+    {
+      "epoch": 151,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 59041,
+      "train_examples": 50000,
+      "train_loss": 2355.716592265625
+    },
+    {
+      "epoch": 152,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 59432,
+      "train_examples": 50000,
+      "train_loss": 1971.07486875
+    },
+    {
+      "epoch": 153,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 59823,
+      "train_examples": 50000,
+      "train_loss": 1849.134235625
+    },
+    {
+      "epoch": 154,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 60214,
+      "train_examples": 50000,
+      "train_loss": 1727.903703046875
+    },
+    {
+      "epoch": 155,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 60605,
+      "train_examples": 50000,
+      "train_loss": 1663.20844578125
+    },
+    {
+      "epoch": 156,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 60996,
+      "train_examples": 50000,
+      "train_loss": 1579.936839765625
+    },
+    {
+      "epoch": 157,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 61387,
+      "train_examples": 50000,
+      "train_loss": 1518.787508203125
+    },
+    {
+      "epoch": 158,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 61778,
+      "train_examples": 50000,
+      "train_loss": 1477.2234746484376
+    },
+    {
+      "epoch": 159,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 62169,
+      "train_examples": 50000,
+      "train_loss": 1400.1150776953125
+    },
+    {
+      "epoch": 160,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 62560,
+      "train_examples": 50000,
+      "train_loss": 1354.4668286328124
+    },
+    {
+      "epoch": 161,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 62951,
+      "train_examples": 50000,
+      "train_loss": 1321.101190546875
+    },
+    {
+      "epoch": 162,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 63342,
+      "train_examples": 50000,
+      "train_loss": 1299.769038359375
+    },
+    {
+      "epoch": 163,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 63733,
+      "train_examples": 50000,
+      "train_loss": 1246.1335008984374
+    },
+    {
+      "epoch": 164,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 64124,
+      "train_examples": 50000,
+      "train_loss": 1228.830918359375
+    },
+    {
+      "epoch": 165,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 64515,
+      "train_examples": 50000,
+      "train_loss": 1192.2152896679688
+    },
+    {
+      "epoch": 166,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 64906,
+      "train_examples": 50000,
+      "train_loss": 1191.7070167578124
+    },
+    {
+      "epoch": 167,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 65297,
+      "train_examples": 50000,
+      "train_loss": 1168.252881796875
+    },
+    {
+      "epoch": 168,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 65688,
+      "train_examples": 50000,
+      "train_loss": 1166.2085495507813
+    },
+    {
+      "epoch": 169,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 66079,
+      "train_examples": 50000,
+      "train_loss": 1154.250678828125
+    },
+    {
+      "epoch": 170,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 66470,
+      "train_examples": 50000,
+      "train_loss": 1107.4836101171875
+    },
+    {
+      "epoch": 171,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 66861,
+      "train_examples": 50000,
+      "train_loss": 1105.0455069921875
+    },
+    {
+      "epoch": 172,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 67252,
+      "train_examples": 50000,
+      "train_loss": 1100.120903515625
+    },
+    {
+      "epoch": 173,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 67643,
+      "train_examples": 50000,
+      "train_loss": 1066.6065143359374
+    },
+    {
+      "epoch": 174,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 68034,
+      "train_examples": 50000,
+      "train_loss": 1069.9054935546876
+    },
+    {
+      "epoch": 175,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 68425,
+      "train_examples": 50000,
+      "train_loss": 1059.8130787109376
+    },
+    {
+      "epoch": 176,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 68816,
+      "train_examples": 50000,
+      "train_loss": 1079.388514296875
+    },
+    {
+      "epoch": 177,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 69207,
+      "train_examples": 50000,
+      "train_loss": 1055.166579140625
+    },
+    {
+      "epoch": 178,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 69598,
+      "train_examples": 50000,
+      "train_loss": 1069.33503328125
+    },
+    {
+      "epoch": 179,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 69989,
+      "train_examples": 50000,
+      "train_loss": 1045.348050234375
+    },
+    {
+      "epoch": 180,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 70380,
+      "train_examples": 50000,
+      "train_loss": 1053.1074065429686
+    },
+    {
+      "epoch": 181,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 70771,
+      "train_examples": 50000,
+      "train_loss": 1058.7408469921875
+    },
+    {
+      "epoch": 182,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 71162,
+      "train_examples": 50000,
+      "train_loss": 1009.6501013671875
+    },
+    {
+      "epoch": 183,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 71553,
+      "train_examples": 50000,
+      "train_loss": 1018.4961114453125
+    },
+    {
+      "epoch": 184,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 71944,
+      "train_examples": 50000,
+      "train_loss": 1004.7097914453125
+    },
+    {
+      "epoch": 185,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 72335,
+      "train_examples": 50000,
+      "train_loss": 1027.2876579492188
+    },
+    {
+      "epoch": 186,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 72726,
+      "train_examples": 50000,
+      "train_loss": 1027.0266332617186
+    },
+    {
+      "epoch": 187,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 73117,
+      "train_examples": 50000,
+      "train_loss": 995.8798777734376
+    },
+    {
+      "epoch": 188,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 73508,
+      "train_examples": 50000,
+      "train_loss": 1015.6011597265625
+    },
+    {
+      "epoch": 189,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 73899,
+      "train_examples": 50000,
+      "train_loss": 972.1149674609375
+    },
+    {
+      "epoch": 190,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 74290,
+      "train_examples": 50000,
+      "train_loss": 977.882844765625
+    },
+    {
+      "epoch": 191,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 74681,
+      "train_examples": 50000,
+      "train_loss": 1008.7832816796875
+    },
+    {
+      "epoch": 192,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 75072,
+      "train_examples": 50000,
+      "train_loss": 974.944484921875
+    },
+    {
+      "epoch": 193,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 75463,
+      "train_examples": 50000,
+      "train_loss": 1007.7110688671875
+    },
+    {
+      "epoch": 194,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 75854,
+      "train_examples": 50000,
+      "train_loss": 969.6423557421875
+    },
+    {
+      "epoch": 195,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 76245,
+      "train_examples": 50000,
+      "train_loss": 980.5617981445313
+    },
+    {
+      "epoch": 196,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 76636,
+      "train_examples": 50000,
+      "train_loss": 966.952641484375
+    },
+    {
+      "epoch": 197,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 77027,
+      "train_examples": 50000,
+      "train_loss": 988.5392790625
+    },
+    {
+      "epoch": 198,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 77418,
+      "train_examples": 50000,
+      "train_loss": 1018.6668183007813
+    },
+    {
+      "epoch": 199,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 77809,
+      "train_examples": 50000,
+      "train_loss": 974.0764008007812
+    },
+    {
+      "epoch": 200,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 78200,
+      "train_examples": 50000,
+      "train_loss": 962.7301450390624
+    }
+  ],
+  "evaluation_protocol": {
+    "test_evaluations": 1,
+    "test_used_for_selection": false,
+    "validation_evaluations": 0
+  },
+  "final": {
+    "accuracy": 0.2603,
+    "epoch": 200,
+    "evaluation_split": "test",
+    "finite": true,
+    "loss": 1081.2588375,
+    "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": 3339442688,
+    "peak_memory_reserved_bytes": 4676648960,
+    "torch_version": "2.3.1+cu118"
+  },
+  "protocol_family": "oral_a_cifar_local_resnet_development",
+  "provenance": {
+    "git_commit": "8f9364e8ee2e2a19eb83b58452fa0e8cb155353c",
+    "git_tracked_dirty": false
+  },
+  "schema_version": 1,
+  "split": {
+    "cifar_source_files": [
+      {
+        "bytes": 31035704,
+        "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/data_batch_1",
+        "sha256": "54636561a3ce25bd3e19253c6b0d8538147b0ae398331ac4a2d86c6d987368cd"
+      },
+      {
+        "bytes": 31035320,
+        "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/data_batch_2",
+        "sha256": "766b2cef9fbc745cf056b3152224f7cf77163b330ea9a15f9392beb8b89bc5a8"
+      },
+      {
+        "bytes": 31035999,
+        "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/data_batch_3",
+        "sha256": "0f00d98ebfb30b3ec0ad19f9756dc2630b89003e10525f5e148445e82aa6a1f9"
+      },
+      {
+        "bytes": 31035696,
+        "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/data_batch_4",
+        "sha256": "3f7bb240661948b8f4d53e36ec720d8306f5668bd0071dcb4e6c947f78e9682b"
+      },
+      {
+        "bytes": 31035623,
+        "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/data_batch_5",
+        "sha256": "d91802434d8376bbaeeadf58a737e3a1b12ac839077e931237e0dcd43adcb154"
+      },
+      {
+        "bytes": 31035526,
+        "path": "/home/yurenh2/sdrn/data/cifar-10-batches-py/test_batch",
+        "sha256": "f53d8d457504f7cff4ea9e021afcf0e0ad8e24a91f3fc42091b8adef61157831"
+      }
+    ],
+    "dataset": "cifar10",
+    "evaluation_split": "test",
+    "input_layout": "NCHW",
+    "input_shape": [
+      3,
+      32,
+      32
+    ],
+    "loader_seed": 14,
+    "normalization_mean": [
+      0.49140000343322754,
+      0.4821999967098236,
+      0.4465000033378601
+    ],
+    "normalization_std": [
+      0.24699999392032623,
+      0.2434999942779541,
+      0.26159998774528503
+    ],
+    "split_from_training_only": true,
+    "split_seed": null,
+    "test_examples": 10000,
+    "train_examples": 50000,
+    "training_augmentation": "random_crop_32_padding_4_zero_then_horizontal_flip_p0.5",
+    "validation_class_counts": {},
+    "validation_examples": 0,
+    "validation_index_sha256": null
+  },
+  "timing": {
+    "apical_warmup_wall_s": 0.0,
+    "evaluation_wall_s": 2.035142660140991,
+    "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": 10382.633543491364,
+    "train_wall_s": 10380.57179403305
+  },
+  "work": {
+    "apical_macs_per_example": 5324800,
+    "causal_scalar_observations": 0,
+    "components": {
+      "apical_projection_macs": 53248000000000,
+      "apical_regression_macs": 0,
+      "bp_reverse_macs_estimate": 0,
+      "kp_reciprocal_correlation_macs": 0,
+      "local_weight_correlation_macs": 1254856960000000,
+      "mirror_feedback_prediction_macs": 0,
+      "mirror_local_correlation_macs": 0,
+      "mirror_response_macs": 0,
+      "ordinary_forward_macs": 1254856960000000,
+      "perturbation_forward_macs": 0,
+      "warmup_clean_forward_macs": 0
+    },
+    "definition": "multiply-accumulates in conv/linear maps; one local weight correlation equals one forward-weight MAC count; BP reverse is estimated as one weight-gradient plus one activation-gradient convolution per forward convolution; mixed-traffic/predictor elementwise arithmetic is reported as a conservative operation estimate separately and is not folded into MACs",
+    "elementwise_operations_estimate": 0,
+    "forward_macs_per_example": 125485696,
+    "logical_batch_loss_queries": 0,
+    "mirror_probe_examples": 0,
+    "mirror_readout_probe_examples": 0,
+    "neutral_projection_observations": 0,
+    "per_example_cross_entropy_terms": 0,
+    "total_clean_forward_examples": 10000000,
+    "total_forward_equivalent_examples": 10000000,
+    "total_macs_estimate": 2562961920000000
+  }
+}