results: confirm standard-depth SDIL scaling

1 files changed, 2281 insertions, 0 deletions

diff --git a/results/oral_a_dynamic_scaling_v2/dfa_d56_s10.json b/results/oral_a_dynamic_scaling_v2/dfa_d56_s10.json
new file mode 100644
index 0000000..e9f577f
--- /dev/null
+++ b/results/oral_a_dynamic_scaling_v2/dfa_d56_s10.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": 10,
+    "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_s10.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": 10,
+    "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": 0.00022253473434830084,
+    "innovation_negative_gradient_cosine": [
+      -0.0007358535658568144,
+      1.0973031749017537e-05,
+      0.00028803618624806404,
+      0.0018082298338413239,
+      0.0017930353060364723,
+      0.0026586810126900673,
+      -0.0005942065035924315,
+      -0.00015428336337208748,
+      -0.0017858229111880064,
+      -0.0009746977593749762,
+      0.0017651284579187632,
+      -0.0018736472120508552,
+      0.0012291024904698133,
+      -0.0015807488234713674,
+      0.0010712535586208105,
+      -0.0008963914588093758,
+      0.0010419270256534219,
+      0.0009349099127575755,
+      0.0007942283409647644,
+      0.0022612982429564,
+      0.0029903706163167953,
+      -0.0006342272972688079,
+      -0.0012854530941694975,
+      0.0017757457681000233,
+      0.0026695728302001953,
+      0.0005210742820054293,
+      0.0027406071312725544,
+      0.003057739231735468,
+      0.004302687011659145,
+      -0.0002525838790461421,
+      0.00313854543492198,
+      0.0034365016035735607,
+      0.0020225904881954193,
+      0.0012177969329059124,
+      0.0021197060123085976,
+      0.0009309356100857258,
+      0.0001798642915673554,
+      0.006009994074702263,
+      0.00971902348101139,
+      0.011476058512926102,
+      0.011707705445587635,
+      0.0097414031624794,
+      0.014721017330884933,
+      0.01197153888642788,
+      0.010873015969991684,
+      0.013532187789678574,
+      0.017107289284467697,
+      0.013830152340233326,
+      0.019055891782045364,
+      0.009968715719878674,
+      0.011132730171084404,
+      0.015165727585554123,
+      0.013714831322431564,
+      0.014137480407953262,
+      0.016913186758756638
+    ],
+    "normalization_state": "training_batch_stats_without_running_update",
+    "raw_negative_gradient_cosine": [
+      -0.0007358535658568144,
+      1.0973031749017537e-05,
+      0.00028803618624806404,
+      0.0018082298338413239,
+      0.0017930353060364723,
+      0.0026586810126900673,
+      -0.0005942065035924315,
+      -0.00015428336337208748,
+      -0.0017858229111880064,
+      -0.0009746977593749762,
+      0.0017651284579187632,
+      -0.0018736472120508552,
+      0.0012291024904698133,
+      -0.0015807488234713674,
+      0.0010712535586208105,
+      -0.0008963914588093758,
+      0.0010419270256534219,
+      0.0009349099127575755,
+      0.0007942283409647644,
+      0.0022612982429564,
+      0.0029903706163167953,
+      -0.0006342272972688079,
+      -0.0012854530941694975,
+      0.0017757457681000233,
+      0.0026695728302001953,
+      0.0005210742820054293,
+      0.0027406071312725544,
+      0.003057739231735468,
+      0.004302687011659145,
+      -0.0002525838790461421,
+      0.00313854543492198,
+      0.0034365016035735607,
+      0.0020225904881954193,
+      0.0012177969329059124,
+      0.0021197060123085976,
+      0.0009309356100857258,
+      0.0001798642915673554,
+      0.006009994074702263,
+      0.00971902348101139,
+      0.011476058512926102,
+      0.011707705445587635,
+      0.0097414031624794,
+      0.014721017330884933,
+      0.01197153888642788,
+      0.010873015969991684,
+      0.013532187789678574,
+      0.017107289284467697,
+      0.013830152340233326,
+      0.019055891782045364,
+      0.009968715719878674,
+      0.011132730171084404,
+      0.015165727585554123,
+      0.013714831322431564,
+      0.014137480407953262,
+      0.016913186758756638
+    ],
+    "teaching_negative_gradient_cosine": [
+      -0.0007358535658568144,
+      1.0973031749017537e-05,
+      0.00028803618624806404,
+      0.0018082298338413239,
+      0.0017930353060364723,
+      0.0026586810126900673,
+      -0.0005942065035924315,
+      -0.00015428336337208748,
+      -0.0017858229111880064,
+      -0.0009746977593749762,
+      0.0017651284579187632,
+      -0.0018736472120508552,
+      0.0012291024904698133,
+      -0.0015807488234713674,
+      0.0010712535586208105,
+      -0.0008963914588093758,
+      0.0010419270256534219,
+      0.0009349099127575755,
+      0.0007942283409647644,
+      0.0022612982429564,
+      0.0029903706163167953,
+      -0.0006342272972688079,
+      -0.0012854530941694975,
+      0.0017757457681000233,
+      0.0026695728302001953,
+      0.0005210742820054293,
+      0.0027406071312725544,
+      0.003057739231735468,
+      0.004302687011659145,
+      -0.0002525838790461421,
+      0.00313854543492198,
+      0.0034365016035735607,
+      0.0020225904881954193,
+      0.0012177969329059124,
+      0.0021197060123085976,
+      0.0009309356100857258,
+      0.0001798642915673554,
+      0.006009994074702263,
+      0.00971902348101139,
+      0.011476058512926102,
+      0.011707705445587635,
+      0.0097414031624794,
+      0.014721017330884933,
+      0.01197153888642788,
+      0.010873015969991684,
+      0.013532187789678574,
+      0.017107289284467697,
+      0.013830152340233326,
+      0.019055891782045364,
+      0.009968715719878674,
+      0.011132730171084404,
+      0.015165727585554123,
+      0.013714831322431564,
+      0.014137480407953262,
+      0.016913186758756638
+    ],
+    "wall_s": 0.20293188095092773
+  },
+  "epochs": [
+    {
+      "epoch": 1,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 391,
+      "train_examples": 50000,
+      "train_loss": 15.54466657989502
+    },
+    {
+      "epoch": 2,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 782,
+      "train_examples": 50000,
+      "train_loss": 14.547235075073242
+    },
+    {
+      "epoch": 3,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 1173,
+      "train_examples": 50000,
+      "train_loss": 17.414953399658202
+    },
+    {
+      "epoch": 4,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 1564,
+      "train_examples": 50000,
+      "train_loss": 25.90386300109863
+    },
+    {
+      "epoch": 5,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 1955,
+      "train_examples": 50000,
+      "train_loss": 39.68947647094726
+    },
+    {
+      "epoch": 6,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 2346,
+      "train_examples": 50000,
+      "train_loss": 54.04766418945312
+    },
+    {
+      "epoch": 7,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 2737,
+      "train_examples": 50000,
+      "train_loss": 72.88010504882813
+    },
+    {
+      "epoch": 8,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 3128,
+      "train_examples": 50000,
+      "train_loss": 108.43115664550781
+    },
+    {
+      "epoch": 9,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 3519,
+      "train_examples": 50000,
+      "train_loss": 154.87382209960938
+    },
+    {
+      "epoch": 10,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 3910,
+      "train_examples": 50000,
+      "train_loss": 168.45801689208983
+    },
+    {
+      "epoch": 11,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 4301,
+      "train_examples": 50000,
+      "train_loss": 252.97475659179688
+    },
+    {
+      "epoch": 12,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 4692,
+      "train_examples": 50000,
+      "train_loss": 292.0759301953125
+    },
+    {
+      "epoch": 13,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 5083,
+      "train_examples": 50000,
+      "train_loss": 332.7507882080078
+    },
+    {
+      "epoch": 14,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 5474,
+      "train_examples": 50000,
+      "train_loss": 442.96318801757815
+    },
+    {
+      "epoch": 15,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 5865,
+      "train_examples": 50000,
+      "train_loss": 564.6440700390625
+    },
+    {
+      "epoch": 16,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 6256,
+      "train_examples": 50000,
+      "train_loss": 572.4222674023438
+    },
+    {
+      "epoch": 17,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 6647,
+      "train_examples": 50000,
+      "train_loss": 840.05808921875
+    },
+    {
+      "epoch": 18,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 7038,
+      "train_examples": 50000,
+      "train_loss": 865.4898714257813
+    },
+    {
+      "epoch": 19,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 7429,
+      "train_examples": 50000,
+      "train_loss": 1098.5925456054688
+    },
+    {
+      "epoch": 20,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 7820,
+      "train_examples": 50000,
+      "train_loss": 1292.8258465625
+    },
+    {
+      "epoch": 21,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 8211,
+      "train_examples": 50000,
+      "train_loss": 1583.3709852734376
+    },
+    {
+      "epoch": 22,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 8602,
+      "train_examples": 50000,
+      "train_loss": 1701.3120577734376
+    },
+    {
+      "epoch": 23,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 8993,
+      "train_examples": 50000,
+      "train_loss": 2075.83830796875
+    },
+    {
+      "epoch": 24,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 9384,
+      "train_examples": 50000,
+      "train_loss": 2293.434503203125
+    },
+    {
+      "epoch": 25,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 9775,
+      "train_examples": 50000,
+      "train_loss": 2558.8233353125
+    },
+    {
+      "epoch": 26,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 10166,
+      "train_examples": 50000,
+      "train_loss": 3020.405894765625
+    },
+    {
+      "epoch": 27,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 10557,
+      "train_examples": 50000,
+      "train_loss": 3351.99866171875
+    },
+    {
+      "epoch": 28,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 10948,
+      "train_examples": 50000,
+      "train_loss": 3603.296686875
+    },
+    {
+      "epoch": 29,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 11339,
+      "train_examples": 50000,
+      "train_loss": 4202.71841546875
+    },
+    {
+      "epoch": 30,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 11730,
+      "train_examples": 50000,
+      "train_loss": 4266.887469375
+    },
+    {
+      "epoch": 31,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 12121,
+      "train_examples": 50000,
+      "train_loss": 4414.28546203125
+    },
+    {
+      "epoch": 32,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 12512,
+      "train_examples": 50000,
+      "train_loss": 4593.823394375
+    },
+    {
+      "epoch": 33,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 12903,
+      "train_examples": 50000,
+      "train_loss": 5191.17040109375
+    },
+    {
+      "epoch": 34,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 13294,
+      "train_examples": 50000,
+      "train_loss": 5839.81197625
+    },
+    {
+      "epoch": 35,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 13685,
+      "train_examples": 50000,
+      "train_loss": 6529.7030828125
+    },
+    {
+      "epoch": 36,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 14076,
+      "train_examples": 50000,
+      "train_loss": 6368.560588125
+    },
+    {
+      "epoch": 37,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 14467,
+      "train_examples": 50000,
+      "train_loss": 7541.834451015625
+    },
+    {
+      "epoch": 38,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 14858,
+      "train_examples": 50000,
+      "train_loss": 7486.9357040625
+    },
+    {
+      "epoch": 39,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 15249,
+      "train_examples": 50000,
+      "train_loss": 8868.19930328125
+    },
+    {
+      "epoch": 40,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 15640,
+      "train_examples": 50000,
+      "train_loss": 9579.1515990625
+    },
+    {
+      "epoch": 41,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 16031,
+      "train_examples": 50000,
+      "train_loss": 9371.672059375
+    },
+    {
+      "epoch": 42,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 16422,
+      "train_examples": 50000,
+      "train_loss": 10813.829093125
+    },
+    {
+      "epoch": 43,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 16813,
+      "train_examples": 50000,
+      "train_loss": 10746.49713625
+    },
+    {
+      "epoch": 44,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 17204,
+      "train_examples": 50000,
+      "train_loss": 12715.11721
+    },
+    {
+      "epoch": 45,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 17595,
+      "train_examples": 50000,
+      "train_loss": 12201.307511875
+    },
+    {
+      "epoch": 46,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 17986,
+      "train_examples": 50000,
+      "train_loss": 12907.2360246875
+    },
+    {
+      "epoch": 47,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 18377,
+      "train_examples": 50000,
+      "train_loss": 13705.78240875
+    },
+    {
+      "epoch": 48,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 18768,
+      "train_examples": 50000,
+      "train_loss": 13940.5551475
+    },
+    {
+      "epoch": 49,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 19159,
+      "train_examples": 50000,
+      "train_loss": 15038.75103625
+    },
+    {
+      "epoch": 50,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 19550,
+      "train_examples": 50000,
+      "train_loss": 15487.918846875
+    },
+    {
+      "epoch": 51,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 19941,
+      "train_examples": 50000,
+      "train_loss": 15641.1139175
+    },
+    {
+      "epoch": 52,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 20332,
+      "train_examples": 50000,
+      "train_loss": 17282.7993675
+    },
+    {
+      "epoch": 53,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 20723,
+      "train_examples": 50000,
+      "train_loss": 18200.64971375
+    },
+    {
+      "epoch": 54,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 21114,
+      "train_examples": 50000,
+      "train_loss": 18207.58564625
+    },
+    {
+      "epoch": 55,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 21505,
+      "train_examples": 50000,
+      "train_loss": 21211.5489225
+    },
+    {
+      "epoch": 56,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 21896,
+      "train_examples": 50000,
+      "train_loss": 21569.2429675
+    },
+    {
+      "epoch": 57,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 22287,
+      "train_examples": 50000,
+      "train_loss": 20786.201261875
+    },
+    {
+      "epoch": 58,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 22678,
+      "train_examples": 50000,
+      "train_loss": 22278.53631875
+    },
+    {
+      "epoch": 59,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 23069,
+      "train_examples": 50000,
+      "train_loss": 25304.68420375
+    },
+    {
+      "epoch": 60,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 23460,
+      "train_examples": 50000,
+      "train_loss": 22680.234508125
+    },
+    {
+      "epoch": 61,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 23851,
+      "train_examples": 50000,
+      "train_loss": 25914.68491625
+    },
+    {
+      "epoch": 62,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 24242,
+      "train_examples": 50000,
+      "train_loss": 23488.31530125
+    },
+    {
+      "epoch": 63,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 24633,
+      "train_examples": 50000,
+      "train_loss": 26916.453524375
+    },
+    {
+      "epoch": 64,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 25024,
+      "train_examples": 50000,
+      "train_loss": 27199.517476875
+    },
+    {
+      "epoch": 65,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 25415,
+      "train_examples": 50000,
+      "train_loss": 29014.880584375
+    },
+    {
+      "epoch": 66,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 25806,
+      "train_examples": 50000,
+      "train_loss": 31759.2104296875
+    },
+    {
+      "epoch": 67,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 26197,
+      "train_examples": 50000,
+      "train_loss": 32718.88177625
+    },
+    {
+      "epoch": 68,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 26588,
+      "train_examples": 50000,
+      "train_loss": 30877.23190875
+    },
+    {
+      "epoch": 69,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 26979,
+      "train_examples": 50000,
+      "train_loss": 33513.7631775
+    },
+    {
+      "epoch": 70,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 27370,
+      "train_examples": 50000,
+      "train_loss": 34085.73262625
+    },
+    {
+      "epoch": 71,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 27761,
+      "train_examples": 50000,
+      "train_loss": 34196.8135575
+    },
+    {
+      "epoch": 72,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 28152,
+      "train_examples": 50000,
+      "train_loss": 33159.024895
+    },
+    {
+      "epoch": 73,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 28543,
+      "train_examples": 50000,
+      "train_loss": 37637.4728825
+    },
+    {
+      "epoch": 74,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 28934,
+      "train_examples": 50000,
+      "train_loss": 36281.769501875
+    },
+    {
+      "epoch": 75,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 29325,
+      "train_examples": 50000,
+      "train_loss": 39038.17320375
+    },
+    {
+      "epoch": 76,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 29716,
+      "train_examples": 50000,
+      "train_loss": 37316.9266325
+    },
+    {
+      "epoch": 77,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 30107,
+      "train_examples": 50000,
+      "train_loss": 41881.991545
+    },
+    {
+      "epoch": 78,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 30498,
+      "train_examples": 50000,
+      "train_loss": 39957.9294975
+    },
+    {
+      "epoch": 79,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 30889,
+      "train_examples": 50000,
+      "train_loss": 42248.13161625
+    },
+    {
+      "epoch": 80,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 31280,
+      "train_examples": 50000,
+      "train_loss": 45955.48953875
+    },
+    {
+      "epoch": 81,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 31671,
+      "train_examples": 50000,
+      "train_loss": 42655.75375875
+    },
+    {
+      "epoch": 82,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 32062,
+      "train_examples": 50000,
+      "train_loss": 49212.42627
+    },
+    {
+      "epoch": 83,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 32453,
+      "train_examples": 50000,
+      "train_loss": 53806.118375625
+    },
+    {
+      "epoch": 84,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 32844,
+      "train_examples": 50000,
+      "train_loss": 53437.919545
+    },
+    {
+      "epoch": 85,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 33235,
+      "train_examples": 50000,
+      "train_loss": 56606.1347175
+    },
+    {
+      "epoch": 86,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 33626,
+      "train_examples": 50000,
+      "train_loss": 49374.35269125
+    },
+    {
+      "epoch": 87,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 34017,
+      "train_examples": 50000,
+      "train_loss": 51668.7037975
+    },
+    {
+      "epoch": 88,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 34408,
+      "train_examples": 50000,
+      "train_loss": 48634.9196875
+    },
+    {
+      "epoch": 89,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 34799,
+      "train_examples": 50000,
+      "train_loss": 58801.28231625
+    },
+    {
+      "epoch": 90,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 35190,
+      "train_examples": 50000,
+      "train_loss": 57543.91696875
+    },
+    {
+      "epoch": 91,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 35581,
+      "train_examples": 50000,
+      "train_loss": 57933.4511
+    },
+    {
+      "epoch": 92,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 35972,
+      "train_examples": 50000,
+      "train_loss": 55352.67518375
+    },
+    {
+      "epoch": 93,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 36363,
+      "train_examples": 50000,
+      "train_loss": 58914.55379
+    },
+    {
+      "epoch": 94,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 36754,
+      "train_examples": 50000,
+      "train_loss": 65179.508695
+    },
+    {
+      "epoch": 95,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 37145,
+      "train_examples": 50000,
+      "train_loss": 58191.2278625
+    },
+    {
+      "epoch": 96,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 37536,
+      "train_examples": 50000,
+      "train_loss": 66903.70380875
+    },
+    {
+      "epoch": 97,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 37927,
+      "train_examples": 50000,
+      "train_loss": 69500.40155
+    },
+    {
+      "epoch": 98,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 38318,
+      "train_examples": 50000,
+      "train_loss": 65830.5844625
+    },
+    {
+      "epoch": 99,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 38709,
+      "train_examples": 50000,
+      "train_loss": 70910.9999825
+    },
+    {
+      "epoch": 100,
+      "lr": 0.03,
+      "output_lr": 0.1,
+      "step": 39100,
+      "train_examples": 50000,
+      "train_loss": 78901.0362975
+    },
+    {
+      "epoch": 101,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 39491,
+      "train_examples": 50000,
+      "train_loss": 13774.071160625
+    },
+    {
+      "epoch": 102,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 39882,
+      "train_examples": 50000,
+      "train_loss": 10134.08951875
+    },
+    {
+      "epoch": 103,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 40273,
+      "train_examples": 50000,
+      "train_loss": 9220.26762609375
+    },
+    {
+      "epoch": 104,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 40664,
+      "train_examples": 50000,
+      "train_loss": 8444.07215015625
+    },
+    {
+      "epoch": 105,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 41055,
+      "train_examples": 50000,
+      "train_loss": 8040.99920953125
+    },
+    {
+      "epoch": 106,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 41446,
+      "train_examples": 50000,
+      "train_loss": 7679.8396934375
+    },
+    {
+      "epoch": 107,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 41837,
+      "train_examples": 50000,
+      "train_loss": 7400.2494
+    },
+    {
+      "epoch": 108,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 42228,
+      "train_examples": 50000,
+      "train_loss": 7037.24590375
+    },
+    {
+      "epoch": 109,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 42619,
+      "train_examples": 50000,
+      "train_loss": 6829.8353040625
+    },
+    {
+      "epoch": 110,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 43010,
+      "train_examples": 50000,
+      "train_loss": 6355.3798890625
+    },
+    {
+      "epoch": 111,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 43401,
+      "train_examples": 50000,
+      "train_loss": 5975.3924253125
+    },
+    {
+      "epoch": 112,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 43792,
+      "train_examples": 50000,
+      "train_loss": 5699.65335578125
+    },
+    {
+      "epoch": 113,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 44183,
+      "train_examples": 50000,
+      "train_loss": 5832.3879453125
+    },
+    {
+      "epoch": 114,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 44574,
+      "train_examples": 50000,
+      "train_loss": 5617.43853234375
+    },
+    {
+      "epoch": 115,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 44965,
+      "train_examples": 50000,
+      "train_loss": 5341.032689375
+    },
+    {
+      "epoch": 116,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 45356,
+      "train_examples": 50000,
+      "train_loss": 5310.4768853125
+    },
+    {
+      "epoch": 117,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 45747,
+      "train_examples": 50000,
+      "train_loss": 5184.67361703125
+    },
+    {
+      "epoch": 118,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 46138,
+      "train_examples": 50000,
+      "train_loss": 5230.7552275
+    },
+    {
+      "epoch": 119,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 46529,
+      "train_examples": 50000,
+      "train_loss": 5096.06707984375
+    },
+    {
+      "epoch": 120,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 46920,
+      "train_examples": 50000,
+      "train_loss": 5188.968658125
+    },
+    {
+      "epoch": 121,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 47311,
+      "train_examples": 50000,
+      "train_loss": 5028.6196609375
+    },
+    {
+      "epoch": 122,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 47702,
+      "train_examples": 50000,
+      "train_loss": 5261.3139678125
+    },
+    {
+      "epoch": 123,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 48093,
+      "train_examples": 50000,
+      "train_loss": 4988.89028703125
+    },
+    {
+      "epoch": 124,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 48484,
+      "train_examples": 50000,
+      "train_loss": 5692.444181875
+    },
+    {
+      "epoch": 125,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 48875,
+      "train_examples": 50000,
+      "train_loss": 5410.727960625
+    },
+    {
+      "epoch": 126,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 49266,
+      "train_examples": 50000,
+      "train_loss": 4953.4976396875
+    },
+    {
+      "epoch": 127,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 49657,
+      "train_examples": 50000,
+      "train_loss": 5121.6054525
+    },
+    {
+      "epoch": 128,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 50048,
+      "train_examples": 50000,
+      "train_loss": 5625.810119375
+    },
+    {
+      "epoch": 129,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 50439,
+      "train_examples": 50000,
+      "train_loss": 5545.74374234375
+    },
+    {
+      "epoch": 130,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 50830,
+      "train_examples": 50000,
+      "train_loss": 5527.4545828125
+    },
+    {
+      "epoch": 131,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 51221,
+      "train_examples": 50000,
+      "train_loss": 5033.5081665625
+    },
+    {
+      "epoch": 132,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 51612,
+      "train_examples": 50000,
+      "train_loss": 5074.50682484375
+    },
+    {
+      "epoch": 133,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 52003,
+      "train_examples": 50000,
+      "train_loss": 5297.9041115625
+    },
+    {
+      "epoch": 134,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 52394,
+      "train_examples": 50000,
+      "train_loss": 5031.92348703125
+    },
+    {
+      "epoch": 135,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 52785,
+      "train_examples": 50000,
+      "train_loss": 5217.82683203125
+    },
+    {
+      "epoch": 136,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 53176,
+      "train_examples": 50000,
+      "train_loss": 5220.15557703125
+    },
+    {
+      "epoch": 137,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 53567,
+      "train_examples": 50000,
+      "train_loss": 5376.6422384375
+    },
+    {
+      "epoch": 138,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 53958,
+      "train_examples": 50000,
+      "train_loss": 5205.66982625
+    },
+    {
+      "epoch": 139,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 54349,
+      "train_examples": 50000,
+      "train_loss": 5085.99358359375
+    },
+    {
+      "epoch": 140,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 54740,
+      "train_examples": 50000,
+      "train_loss": 5418.01852515625
+    },
+    {
+      "epoch": 141,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 55131,
+      "train_examples": 50000,
+      "train_loss": 5023.23353546875
+    },
+    {
+      "epoch": 142,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 55522,
+      "train_examples": 50000,
+      "train_loss": 5162.695340078125
+    },
+    {
+      "epoch": 143,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 55913,
+      "train_examples": 50000,
+      "train_loss": 5160.7980140625
+    },
+    {
+      "epoch": 144,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 56304,
+      "train_examples": 50000,
+      "train_loss": 5663.014644375
+    },
+    {
+      "epoch": 145,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 56695,
+      "train_examples": 50000,
+      "train_loss": 5296.452984296875
+    },
+    {
+      "epoch": 146,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 57086,
+      "train_examples": 50000,
+      "train_loss": 5229.27388703125
+    },
+    {
+      "epoch": 147,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 57477,
+      "train_examples": 50000,
+      "train_loss": 5626.622417578125
+    },
+    {
+      "epoch": 148,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 57868,
+      "train_examples": 50000,
+      "train_loss": 5444.78413109375
+    },
+    {
+      "epoch": 149,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 58259,
+      "train_examples": 50000,
+      "train_loss": 5456.180879921875
+    },
+    {
+      "epoch": 150,
+      "lr": 0.003,
+      "output_lr": 0.010000000000000002,
+      "step": 58650,
+      "train_examples": 50000,
+      "train_loss": 5576.560659453125
+    },
+    {
+      "epoch": 151,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 59041,
+      "train_examples": 50000,
+      "train_loss": 1782.5386803515626
+    },
+    {
+      "epoch": 152,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 59432,
+      "train_examples": 50000,
+      "train_loss": 1514.4121122265626
+    },
+    {
+      "epoch": 153,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 59823,
+      "train_examples": 50000,
+      "train_loss": 1410.9323049609375
+    },
+    {
+      "epoch": 154,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 60214,
+      "train_examples": 50000,
+      "train_loss": 1311.8951089453126
+    },
+    {
+      "epoch": 155,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 60605,
+      "train_examples": 50000,
+      "train_loss": 1214.2603724609376
+    },
+    {
+      "epoch": 156,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 60996,
+      "train_examples": 50000,
+      "train_loss": 1161.9654675
+    },
+    {
+      "epoch": 157,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 61387,
+      "train_examples": 50000,
+      "train_loss": 1120.774872265625
+    },
+    {
+      "epoch": 158,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 61778,
+      "train_examples": 50000,
+      "train_loss": 1080.1324510546874
+    },
+    {
+      "epoch": 159,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 62169,
+      "train_examples": 50000,
+      "train_loss": 1043.7089065625
+    },
+    {
+      "epoch": 160,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 62560,
+      "train_examples": 50000,
+      "train_loss": 1012.7494857421875
+    },
+    {
+      "epoch": 161,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 62951,
+      "train_examples": 50000,
+      "train_loss": 980.0301846875
+    },
+    {
+      "epoch": 162,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 63342,
+      "train_examples": 50000,
+      "train_loss": 946.735686796875
+    },
+    {
+      "epoch": 163,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 63733,
+      "train_examples": 50000,
+      "train_loss": 946.5751611132813
+    },
+    {
+      "epoch": 164,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 64124,
+      "train_examples": 50000,
+      "train_loss": 914.8629967578125
+    },
+    {
+      "epoch": 165,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 64515,
+      "train_examples": 50000,
+      "train_loss": 914.1067816796875
+    },
+    {
+      "epoch": 166,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 64906,
+      "train_examples": 50000,
+      "train_loss": 912.10888921875
+    },
+    {
+      "epoch": 167,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 65297,
+      "train_examples": 50000,
+      "train_loss": 884.2077573828125
+    },
+    {
+      "epoch": 168,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 65688,
+      "train_examples": 50000,
+      "train_loss": 902.1880673632812
+    },
+    {
+      "epoch": 169,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 66079,
+      "train_examples": 50000,
+      "train_loss": 878.38894671875
+    },
+    {
+      "epoch": 170,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 66470,
+      "train_examples": 50000,
+      "train_loss": 833.30157421875
+    },
+    {
+      "epoch": 171,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 66861,
+      "train_examples": 50000,
+      "train_loss": 866.5463003125
+    },
+    {
+      "epoch": 172,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 67252,
+      "train_examples": 50000,
+      "train_loss": 855.2307503515625
+    },
+    {
+      "epoch": 173,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 67643,
+      "train_examples": 50000,
+      "train_loss": 835.0665584375
+    },
+    {
+      "epoch": 174,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 68034,
+      "train_examples": 50000,
+      "train_loss": 841.1369849804687
+    },
+    {
+      "epoch": 175,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 68425,
+      "train_examples": 50000,
+      "train_loss": 821.0857187304688
+    },
+    {
+      "epoch": 176,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 68816,
+      "train_examples": 50000,
+      "train_loss": 815.5904630078124
+    },
+    {
+      "epoch": 177,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 69207,
+      "train_examples": 50000,
+      "train_loss": 829.5957598242187
+    },
+    {
+      "epoch": 178,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 69598,
+      "train_examples": 50000,
+      "train_loss": 815.3975285742188
+    },
+    {
+      "epoch": 179,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 69989,
+      "train_examples": 50000,
+      "train_loss": 828.9468991796875
+    },
+    {
+      "epoch": 180,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 70380,
+      "train_examples": 50000,
+      "train_loss": 827.04718890625
+    },
+    {
+      "epoch": 181,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 70771,
+      "train_examples": 50000,
+      "train_loss": 823.0365430078125
+    },
+    {
+      "epoch": 182,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 71162,
+      "train_examples": 50000,
+      "train_loss": 813.0535859179688
+    },
+    {
+      "epoch": 183,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 71553,
+      "train_examples": 50000,
+      "train_loss": 817.4472450195312
+    },
+    {
+      "epoch": 184,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 71944,
+      "train_examples": 50000,
+      "train_loss": 801.30095453125
+    },
+    {
+      "epoch": 185,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 72335,
+      "train_examples": 50000,
+      "train_loss": 796.7622555859375
+    },
+    {
+      "epoch": 186,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 72726,
+      "train_examples": 50000,
+      "train_loss": 789.8825111328125
+    },
+    {
+      "epoch": 187,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 73117,
+      "train_examples": 50000,
+      "train_loss": 793.7847402929688
+    },
+    {
+      "epoch": 188,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 73508,
+      "train_examples": 50000,
+      "train_loss": 779.5417561328125
+    },
+    {
+      "epoch": 189,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 73899,
+      "train_examples": 50000,
+      "train_loss": 793.7602258203125
+    },
+    {
+      "epoch": 190,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 74290,
+      "train_examples": 50000,
+      "train_loss": 806.3877822070313
+    },
+    {
+      "epoch": 191,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 74681,
+      "train_examples": 50000,
+      "train_loss": 797.7902808984375
+    },
+    {
+      "epoch": 192,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 75072,
+      "train_examples": 50000,
+      "train_loss": 755.6947116015625
+    },
+    {
+      "epoch": 193,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 75463,
+      "train_examples": 50000,
+      "train_loss": 779.7081112304687
+    },
+    {
+      "epoch": 194,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 75854,
+      "train_examples": 50000,
+      "train_loss": 803.3361581640625
+    },
+    {
+      "epoch": 195,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 76245,
+      "train_examples": 50000,
+      "train_loss": 772.281200703125
+    },
+    {
+      "epoch": 196,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 76636,
+      "train_examples": 50000,
+      "train_loss": 782.7138865039062
+    },
+    {
+      "epoch": 197,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 77027,
+      "train_examples": 50000,
+      "train_loss": 780.0067841015625
+    },
+    {
+      "epoch": 198,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 77418,
+      "train_examples": 50000,
+      "train_loss": 778.2932036523438
+    },
+    {
+      "epoch": 199,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 77809,
+      "train_examples": 50000,
+      "train_loss": 756.139004296875
+    },
+    {
+      "epoch": 200,
+      "lr": 0.00030000000000000003,
+      "output_lr": 0.0010000000000000002,
+      "step": 78200,
+      "train_examples": 50000,
+      "train_loss": 788.0550848046875
+    }
+  ],
+  "evaluation_protocol": {
+    "test_evaluations": 1,
+    "test_used_for_selection": false,
+    "validation_evaluations": 0
+  },
+  "final": {
+    "accuracy": 0.3348,
+    "epoch": 200,
+    "evaluation_split": "test",
+    "finite": true,
+    "loss": 831.0552375,
+    "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": 10,
+    "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": 2.034506320953369,
+    "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": 10377.251267910004,
+    "train_wall_s": 10375.189896583557
+  },
+  "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
+  }
+}