summaryrefslogtreecommitdiff
path: root/results/oral_a_dynamic_scaling_v2/dfa_d56_s11.json
diff options
context:
space:
mode:
Diffstat (limited to 'results/oral_a_dynamic_scaling_v2/dfa_d56_s11.json')
-rw-r--r--results/oral_a_dynamic_scaling_v2/dfa_d56_s11.json2281
1 files changed, 2281 insertions, 0 deletions
diff --git a/results/oral_a_dynamic_scaling_v2/dfa_d56_s11.json b/results/oral_a_dynamic_scaling_v2/dfa_d56_s11.json
new file mode 100644
index 0000000..3dff1aa
--- /dev/null
+++ b/results/oral_a_dynamic_scaling_v2/dfa_d56_s11.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": 11,
+ "lr": 0.03,
+ "lr_gamma": 0.1,
+ "lr_milestones": "100,150",
+ "lr_schedule": "step",
+ "max_steps": 0,
+ "mirror_batch_size": 1,
+ "mirror_eta": 0.1,
+ "mirror_every": 16,
+ "mirror_noise_std": 1.0,
+ "mirror_seed": 3000,
+ "mirror_warmup_steps": 0,
+ "mode": "dfa",
+ "momentum": 0.9,
+ "neutral_projection": 0,
+ "normalization": "batchnorm",
+ "nuisance_scale": 0.0,
+ "out": "results/oral_a_dynamic_scaling_v2/dfa_d56_s11.json",
+ "output_lr": 0.1,
+ "pert_directions": 1,
+ "pert_every": 4,
+ "pert_sigma": 0.01,
+ "perturb_seed": 1000,
+ "predictor_every": 0,
+ "predictor_mode": "nlms",
+ "predictor_warmup_steps": 0,
+ "residual_scale": null,
+ "seed": 11,
+ "split_seed": 2027,
+ "traffic_calibration_examples": 64,
+ "traffic_ratio": 4.0,
+ "traffic_rule": "innovation",
+ "traffic_seed": 4000,
+ "train_limit": 0,
+ "use_residual": 1,
+ "val_examples": 0,
+ "vectorizer_mode": "spatial_template",
+ "warmup_epochs": 0,
+ "weight_decay": 0.0001,
+ "weight_scale": 1.0,
+ "width": 16
+ },
+ "calibration_metric_space": "full_hidden_field",
+ "counters": {
+ "apical_warmup_examples": 0,
+ "calibration_event_examples": 0,
+ "causal_scalar_observations": 0,
+ "logical_batch_loss_queries": 0,
+ "mirror_conv_examples": 0,
+ "mirror_events": 0,
+ "mirror_readout_examples": 0,
+ "neutral_projection_examples": 0,
+ "ordinary_examples": 10000000,
+ "per_example_loss_terms": 0,
+ "perturbation_events": 0,
+ "perturbation_forward_examples": 0,
+ "predictor_update_examples": 0,
+ "predictor_warmup_examples": 0,
+ "traffic_audit_examples": 0,
+ "traffic_calibration_examples": 0
+ },
+ "diagnostics": {
+ "early_third_mean": 0.0005879553298452973,
+ "innovation_negative_gradient_cosine": [
+ 0.0011305890511721373,
+ -8.942060230765492e-05,
+ -0.0006630357238464057,
+ -0.000659326498862356,
+ -0.0011131414212286472,
+ 0.0027757694479078054,
+ 0.002825309056788683,
+ 0.0013577521312981844,
+ 0.0004187228623777628,
+ 0.000324989203363657,
+ 0.0002259696484543383,
+ 0.002006225287914276,
+ -0.0006687602144666016,
+ -2.0834908355027437e-05,
+ 9.226246038451791e-05,
+ 0.000342094455845654,
+ 0.0011339447228237987,
+ 0.0011640869779512286,
+ 0.0015215585008263588,
+ 0.003560976590961218,
+ 0.0063767931424081326,
+ 5.662994226440787e-05,
+ 0.0038221392314881086,
+ 0.0007309076609089971,
+ -0.0009009235654957592,
+ 0.0029084018897265196,
+ 0.0029845749959349632,
+ 0.002053938340395689,
+ 0.001768512767739594,
+ 0.003022806253284216,
+ 0.0038056422490626574,
+ 0.002493410836905241,
+ 0.0030819051899015903,
+ 0.005456170998513699,
+ 0.0029835072346031666,
+ 0.004010680131614208,
+ 0.0027565720956772566,
+ 0.01105463970452547,
+ 0.01161639578640461,
+ 0.009767036885023117,
+ 0.006078685168176889,
+ 0.010227176360785961,
+ 0.007261885330080986,
+ 0.008559858426451683,
+ 0.015124209225177765,
+ 0.0069428132846951485,
+ 0.010973193682730198,
+ 0.004366468172520399,
+ 0.009382097981870174,
+ 0.008147720247507095,
+ 0.010429740883409977,
+ 0.005681689362972975,
+ 0.015391110442578793,
+ 0.014982243999838829,
+ 0.02180664986371994
+ ],
+ "normalization_state": "training_batch_stats_without_running_update",
+ "raw_negative_gradient_cosine": [
+ 0.0011305890511721373,
+ -8.942060230765492e-05,
+ -0.0006630357238464057,
+ -0.000659326498862356,
+ -0.0011131414212286472,
+ 0.0027757694479078054,
+ 0.002825309056788683,
+ 0.0013577521312981844,
+ 0.0004187228623777628,
+ 0.000324989203363657,
+ 0.0002259696484543383,
+ 0.002006225287914276,
+ -0.0006687602144666016,
+ -2.0834908355027437e-05,
+ 9.226246038451791e-05,
+ 0.000342094455845654,
+ 0.0011339447228237987,
+ 0.0011640869779512286,
+ 0.0015215585008263588,
+ 0.003560976590961218,
+ 0.0063767931424081326,
+ 5.662994226440787e-05,
+ 0.0038221392314881086,
+ 0.0007309076609089971,
+ -0.0009009235654957592,
+ 0.0029084018897265196,
+ 0.0029845749959349632,
+ 0.002053938340395689,
+ 0.001768512767739594,
+ 0.003022806253284216,
+ 0.0038056422490626574,
+ 0.002493410836905241,
+ 0.0030819051899015903,
+ 0.005456170998513699,
+ 0.0029835072346031666,
+ 0.004010680131614208,
+ 0.0027565720956772566,
+ 0.01105463970452547,
+ 0.01161639578640461,
+ 0.009767036885023117,
+ 0.006078685168176889,
+ 0.010227176360785961,
+ 0.007261885330080986,
+ 0.008559858426451683,
+ 0.015124209225177765,
+ 0.0069428132846951485,
+ 0.010973193682730198,
+ 0.004366468172520399,
+ 0.009382097981870174,
+ 0.008147720247507095,
+ 0.010429740883409977,
+ 0.005681689362972975,
+ 0.015391110442578793,
+ 0.014982243999838829,
+ 0.02180664986371994
+ ],
+ "teaching_negative_gradient_cosine": [
+ 0.0011305890511721373,
+ -8.942060230765492e-05,
+ -0.0006630357238464057,
+ -0.000659326498862356,
+ -0.0011131414212286472,
+ 0.0027757694479078054,
+ 0.002825309056788683,
+ 0.0013577521312981844,
+ 0.0004187228623777628,
+ 0.000324989203363657,
+ 0.0002259696484543383,
+ 0.002006225287914276,
+ -0.0006687602144666016,
+ -2.0834908355027437e-05,
+ 9.226246038451791e-05,
+ 0.000342094455845654,
+ 0.0011339447228237987,
+ 0.0011640869779512286,
+ 0.0015215585008263588,
+ 0.003560976590961218,
+ 0.0063767931424081326,
+ 5.662994226440787e-05,
+ 0.0038221392314881086,
+ 0.0007309076609089971,
+ -0.0009009235654957592,
+ 0.0029084018897265196,
+ 0.0029845749959349632,
+ 0.002053938340395689,
+ 0.001768512767739594,
+ 0.003022806253284216,
+ 0.0038056422490626574,
+ 0.002493410836905241,
+ 0.0030819051899015903,
+ 0.005456170998513699,
+ 0.0029835072346031666,
+ 0.004010680131614208,
+ 0.0027565720956772566,
+ 0.01105463970452547,
+ 0.01161639578640461,
+ 0.009767036885023117,
+ 0.006078685168176889,
+ 0.010227176360785961,
+ 0.007261885330080986,
+ 0.008559858426451683,
+ 0.015124209225177765,
+ 0.0069428132846951485,
+ 0.010973193682730198,
+ 0.004366468172520399,
+ 0.009382097981870174,
+ 0.008147720247507095,
+ 0.010429740883409977,
+ 0.005681689362972975,
+ 0.015391110442578793,
+ 0.014982243999838829,
+ 0.02180664986371994
+ ],
+ "wall_s": 0.21522068977355957
+ },
+ "epochs": [
+ {
+ "epoch": 1,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 391,
+ "train_examples": 50000,
+ "train_loss": 21.822016004943848
+ },
+ {
+ "epoch": 2,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 782,
+ "train_examples": 50000,
+ "train_loss": 17.824559099121092
+ },
+ {
+ "epoch": 3,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 1173,
+ "train_examples": 50000,
+ "train_loss": 21.810150268554686
+ },
+ {
+ "epoch": 4,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 1564,
+ "train_examples": 50000,
+ "train_loss": 30.373354115600584
+ },
+ {
+ "epoch": 5,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 1955,
+ "train_examples": 50000,
+ "train_loss": 37.876012268066404
+ },
+ {
+ "epoch": 6,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 2346,
+ "train_examples": 50000,
+ "train_loss": 53.92637309448242
+ },
+ {
+ "epoch": 7,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 2737,
+ "train_examples": 50000,
+ "train_loss": 67.34847674072266
+ },
+ {
+ "epoch": 8,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 3128,
+ "train_examples": 50000,
+ "train_loss": 100.33222059082031
+ },
+ {
+ "epoch": 9,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 3519,
+ "train_examples": 50000,
+ "train_loss": 133.74985802246093
+ },
+ {
+ "epoch": 10,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 3910,
+ "train_examples": 50000,
+ "train_loss": 173.3882081298828
+ },
+ {
+ "epoch": 11,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 4301,
+ "train_examples": 50000,
+ "train_loss": 226.11262048339844
+ },
+ {
+ "epoch": 12,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 4692,
+ "train_examples": 50000,
+ "train_loss": 290.80861913085937
+ },
+ {
+ "epoch": 13,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 5083,
+ "train_examples": 50000,
+ "train_loss": 415.3767459765625
+ },
+ {
+ "epoch": 14,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 5474,
+ "train_examples": 50000,
+ "train_loss": 466.00302681640625
+ },
+ {
+ "epoch": 15,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 5865,
+ "train_examples": 50000,
+ "train_loss": 575.5295325
+ },
+ {
+ "epoch": 16,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 6256,
+ "train_examples": 50000,
+ "train_loss": 694.2015186328125
+ },
+ {
+ "epoch": 17,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 6647,
+ "train_examples": 50000,
+ "train_loss": 922.064875
+ },
+ {
+ "epoch": 18,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 7038,
+ "train_examples": 50000,
+ "train_loss": 1099.0480919140625
+ },
+ {
+ "epoch": 19,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 7429,
+ "train_examples": 50000,
+ "train_loss": 1343.856710390625
+ },
+ {
+ "epoch": 20,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 7820,
+ "train_examples": 50000,
+ "train_loss": 1639.4419911328125
+ },
+ {
+ "epoch": 21,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 8211,
+ "train_examples": 50000,
+ "train_loss": 1726.9874137890624
+ },
+ {
+ "epoch": 22,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 8602,
+ "train_examples": 50000,
+ "train_loss": 1736.739409453125
+ },
+ {
+ "epoch": 23,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 8993,
+ "train_examples": 50000,
+ "train_loss": 2409.0237875
+ },
+ {
+ "epoch": 24,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 9384,
+ "train_examples": 50000,
+ "train_loss": 2551.986870078125
+ },
+ {
+ "epoch": 25,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 9775,
+ "train_examples": 50000,
+ "train_loss": 2991.640676875
+ },
+ {
+ "epoch": 26,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 10166,
+ "train_examples": 50000,
+ "train_loss": 3164.708963671875
+ },
+ {
+ "epoch": 27,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 10557,
+ "train_examples": 50000,
+ "train_loss": 3593.382035234375
+ },
+ {
+ "epoch": 28,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 10948,
+ "train_examples": 50000,
+ "train_loss": 3886.67451828125
+ },
+ {
+ "epoch": 29,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 11339,
+ "train_examples": 50000,
+ "train_loss": 4288.502298125
+ },
+ {
+ "epoch": 30,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 11730,
+ "train_examples": 50000,
+ "train_loss": 4229.5054853125
+ },
+ {
+ "epoch": 31,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 12121,
+ "train_examples": 50000,
+ "train_loss": 4708.4159971875
+ },
+ {
+ "epoch": 32,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 12512,
+ "train_examples": 50000,
+ "train_loss": 5456.21266015625
+ },
+ {
+ "epoch": 33,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 12903,
+ "train_examples": 50000,
+ "train_loss": 6402.48859484375
+ },
+ {
+ "epoch": 34,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 13294,
+ "train_examples": 50000,
+ "train_loss": 6137.558204375
+ },
+ {
+ "epoch": 35,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 13685,
+ "train_examples": 50000,
+ "train_loss": 6842.644431875
+ },
+ {
+ "epoch": 36,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 14076,
+ "train_examples": 50000,
+ "train_loss": 7860.1643009375
+ },
+ {
+ "epoch": 37,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 14467,
+ "train_examples": 50000,
+ "train_loss": 7504.49326
+ },
+ {
+ "epoch": 38,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 14858,
+ "train_examples": 50000,
+ "train_loss": 8271.88913
+ },
+ {
+ "epoch": 39,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 15249,
+ "train_examples": 50000,
+ "train_loss": 8202.785735625
+ },
+ {
+ "epoch": 40,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 15640,
+ "train_examples": 50000,
+ "train_loss": 9878.64996890625
+ },
+ {
+ "epoch": 41,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 16031,
+ "train_examples": 50000,
+ "train_loss": 10267.02852890625
+ },
+ {
+ "epoch": 42,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 16422,
+ "train_examples": 50000,
+ "train_loss": 11784.039109375
+ },
+ {
+ "epoch": 43,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 16813,
+ "train_examples": 50000,
+ "train_loss": 11671.2003621875
+ },
+ {
+ "epoch": 44,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 17204,
+ "train_examples": 50000,
+ "train_loss": 11453.28779875
+ },
+ {
+ "epoch": 45,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 17595,
+ "train_examples": 50000,
+ "train_loss": 13273.710763125
+ },
+ {
+ "epoch": 46,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 17986,
+ "train_examples": 50000,
+ "train_loss": 12479.77634875
+ },
+ {
+ "epoch": 47,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 18377,
+ "train_examples": 50000,
+ "train_loss": 13303.353846875
+ },
+ {
+ "epoch": 48,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 18768,
+ "train_examples": 50000,
+ "train_loss": 14392.4950259375
+ },
+ {
+ "epoch": 49,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 19159,
+ "train_examples": 50000,
+ "train_loss": 16243.511494375
+ },
+ {
+ "epoch": 50,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 19550,
+ "train_examples": 50000,
+ "train_loss": 16262.6872675
+ },
+ {
+ "epoch": 51,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 19941,
+ "train_examples": 50000,
+ "train_loss": 18072.855780625
+ },
+ {
+ "epoch": 52,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 20332,
+ "train_examples": 50000,
+ "train_loss": 17862.5171825
+ },
+ {
+ "epoch": 53,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 20723,
+ "train_examples": 50000,
+ "train_loss": 19925.392656875
+ },
+ {
+ "epoch": 54,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 21114,
+ "train_examples": 50000,
+ "train_loss": 20454.8466
+ },
+ {
+ "epoch": 55,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 21505,
+ "train_examples": 50000,
+ "train_loss": 21781.7770725
+ },
+ {
+ "epoch": 56,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 21896,
+ "train_examples": 50000,
+ "train_loss": 21613.798714375
+ },
+ {
+ "epoch": 57,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 22287,
+ "train_examples": 50000,
+ "train_loss": 22883.51164625
+ },
+ {
+ "epoch": 58,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 22678,
+ "train_examples": 50000,
+ "train_loss": 22273.373185
+ },
+ {
+ "epoch": 59,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 23069,
+ "train_examples": 50000,
+ "train_loss": 25506.16983875
+ },
+ {
+ "epoch": 60,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 23460,
+ "train_examples": 50000,
+ "train_loss": 25642.85067625
+ },
+ {
+ "epoch": 61,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 23851,
+ "train_examples": 50000,
+ "train_loss": 24691.8515728125
+ },
+ {
+ "epoch": 62,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 24242,
+ "train_examples": 50000,
+ "train_loss": 25866.07558625
+ },
+ {
+ "epoch": 63,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 24633,
+ "train_examples": 50000,
+ "train_loss": 29144.162578125
+ },
+ {
+ "epoch": 64,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 25024,
+ "train_examples": 50000,
+ "train_loss": 27740.450170625
+ },
+ {
+ "epoch": 65,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 25415,
+ "train_examples": 50000,
+ "train_loss": 29032.3199528125
+ },
+ {
+ "epoch": 66,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 25806,
+ "train_examples": 50000,
+ "train_loss": 28706.324155
+ },
+ {
+ "epoch": 67,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 26197,
+ "train_examples": 50000,
+ "train_loss": 29711.05745
+ },
+ {
+ "epoch": 68,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 26588,
+ "train_examples": 50000,
+ "train_loss": 31706.905618125
+ },
+ {
+ "epoch": 69,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 26979,
+ "train_examples": 50000,
+ "train_loss": 36933.063976875
+ },
+ {
+ "epoch": 70,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 27370,
+ "train_examples": 50000,
+ "train_loss": 37287.953715
+ },
+ {
+ "epoch": 71,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 27761,
+ "train_examples": 50000,
+ "train_loss": 37194.62840875
+ },
+ {
+ "epoch": 72,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 28152,
+ "train_examples": 50000,
+ "train_loss": 36731.81285375
+ },
+ {
+ "epoch": 73,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 28543,
+ "train_examples": 50000,
+ "train_loss": 41938.34304
+ },
+ {
+ "epoch": 74,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 28934,
+ "train_examples": 50000,
+ "train_loss": 42232.847375625
+ },
+ {
+ "epoch": 75,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 29325,
+ "train_examples": 50000,
+ "train_loss": 41347.69572
+ },
+ {
+ "epoch": 76,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 29716,
+ "train_examples": 50000,
+ "train_loss": 40861.88819875
+ },
+ {
+ "epoch": 77,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 30107,
+ "train_examples": 50000,
+ "train_loss": 44865.75046875
+ },
+ {
+ "epoch": 78,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 30498,
+ "train_examples": 50000,
+ "train_loss": 43689.4772625
+ },
+ {
+ "epoch": 79,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 30889,
+ "train_examples": 50000,
+ "train_loss": 46145.1057175
+ },
+ {
+ "epoch": 80,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 31280,
+ "train_examples": 50000,
+ "train_loss": 50945.485785
+ },
+ {
+ "epoch": 81,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 31671,
+ "train_examples": 50000,
+ "train_loss": 48024.606185
+ },
+ {
+ "epoch": 82,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 32062,
+ "train_examples": 50000,
+ "train_loss": 46213.86058
+ },
+ {
+ "epoch": 83,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 32453,
+ "train_examples": 50000,
+ "train_loss": 48255.666905
+ },
+ {
+ "epoch": 84,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 32844,
+ "train_examples": 50000,
+ "train_loss": 55334.47507875
+ },
+ {
+ "epoch": 85,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 33235,
+ "train_examples": 50000,
+ "train_loss": 51214.2301425
+ },
+ {
+ "epoch": 86,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 33626,
+ "train_examples": 50000,
+ "train_loss": 57531.0495775
+ },
+ {
+ "epoch": 87,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 34017,
+ "train_examples": 50000,
+ "train_loss": 59073.374505
+ },
+ {
+ "epoch": 88,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 34408,
+ "train_examples": 50000,
+ "train_loss": 59636.9727625
+ },
+ {
+ "epoch": 89,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 34799,
+ "train_examples": 50000,
+ "train_loss": 58170.847806875
+ },
+ {
+ "epoch": 90,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 35190,
+ "train_examples": 50000,
+ "train_loss": 64804.94207
+ },
+ {
+ "epoch": 91,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 35581,
+ "train_examples": 50000,
+ "train_loss": 65841.375075
+ },
+ {
+ "epoch": 92,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 35972,
+ "train_examples": 50000,
+ "train_loss": 67606.9351425
+ },
+ {
+ "epoch": 93,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 36363,
+ "train_examples": 50000,
+ "train_loss": 66196.475085
+ },
+ {
+ "epoch": 94,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 36754,
+ "train_examples": 50000,
+ "train_loss": 64087.29575625
+ },
+ {
+ "epoch": 95,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 37145,
+ "train_examples": 50000,
+ "train_loss": 70802.3089975
+ },
+ {
+ "epoch": 96,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 37536,
+ "train_examples": 50000,
+ "train_loss": 72189.4987
+ },
+ {
+ "epoch": 97,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 37927,
+ "train_examples": 50000,
+ "train_loss": 77082.48973
+ },
+ {
+ "epoch": 98,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 38318,
+ "train_examples": 50000,
+ "train_loss": 77423.9751925
+ },
+ {
+ "epoch": 99,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 38709,
+ "train_examples": 50000,
+ "train_loss": 75377.29657375
+ },
+ {
+ "epoch": 100,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 39100,
+ "train_examples": 50000,
+ "train_loss": 69896.49668125
+ },
+ {
+ "epoch": 101,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 39491,
+ "train_examples": 50000,
+ "train_loss": 14645.348516875
+ },
+ {
+ "epoch": 102,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 39882,
+ "train_examples": 50000,
+ "train_loss": 11066.969068125
+ },
+ {
+ "epoch": 103,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 40273,
+ "train_examples": 50000,
+ "train_loss": 9616.640616875
+ },
+ {
+ "epoch": 104,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 40664,
+ "train_examples": 50000,
+ "train_loss": 9555.9286115625
+ },
+ {
+ "epoch": 105,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 41055,
+ "train_examples": 50000,
+ "train_loss": 8889.608115625
+ },
+ {
+ "epoch": 106,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 41446,
+ "train_examples": 50000,
+ "train_loss": 8763.56310015625
+ },
+ {
+ "epoch": 107,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 41837,
+ "train_examples": 50000,
+ "train_loss": 8327.51983515625
+ },
+ {
+ "epoch": 108,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 42228,
+ "train_examples": 50000,
+ "train_loss": 8319.9284575
+ },
+ {
+ "epoch": 109,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 42619,
+ "train_examples": 50000,
+ "train_loss": 7766.7889059375
+ },
+ {
+ "epoch": 110,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 43010,
+ "train_examples": 50000,
+ "train_loss": 7218.4834346875
+ },
+ {
+ "epoch": 111,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 43401,
+ "train_examples": 50000,
+ "train_loss": 6762.2607203125
+ },
+ {
+ "epoch": 112,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 43792,
+ "train_examples": 50000,
+ "train_loss": 6680.30279734375
+ },
+ {
+ "epoch": 113,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 44183,
+ "train_examples": 50000,
+ "train_loss": 6376.58596
+ },
+ {
+ "epoch": 114,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 44574,
+ "train_examples": 50000,
+ "train_loss": 6162.12700375
+ },
+ {
+ "epoch": 115,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 44965,
+ "train_examples": 50000,
+ "train_loss": 6278.253233046875
+ },
+ {
+ "epoch": 116,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 45356,
+ "train_examples": 50000,
+ "train_loss": 6274.7127134375
+ },
+ {
+ "epoch": 117,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 45747,
+ "train_examples": 50000,
+ "train_loss": 5905.41486
+ },
+ {
+ "epoch": 118,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 46138,
+ "train_examples": 50000,
+ "train_loss": 5748.2783265625
+ },
+ {
+ "epoch": 119,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 46529,
+ "train_examples": 50000,
+ "train_loss": 6044.13596671875
+ },
+ {
+ "epoch": 120,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 46920,
+ "train_examples": 50000,
+ "train_loss": 5841.58058859375
+ },
+ {
+ "epoch": 121,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 47311,
+ "train_examples": 50000,
+ "train_loss": 6033.3200440625
+ },
+ {
+ "epoch": 122,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 47702,
+ "train_examples": 50000,
+ "train_loss": 5670.89361828125
+ },
+ {
+ "epoch": 123,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 48093,
+ "train_examples": 50000,
+ "train_loss": 6045.1858071875
+ },
+ {
+ "epoch": 124,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 48484,
+ "train_examples": 50000,
+ "train_loss": 5674.30254171875
+ },
+ {
+ "epoch": 125,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 48875,
+ "train_examples": 50000,
+ "train_loss": 6182.4114390625
+ },
+ {
+ "epoch": 126,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 49266,
+ "train_examples": 50000,
+ "train_loss": 5883.83195515625
+ },
+ {
+ "epoch": 127,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 49657,
+ "train_examples": 50000,
+ "train_loss": 6481.4023884375
+ },
+ {
+ "epoch": 128,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 50048,
+ "train_examples": 50000,
+ "train_loss": 5811.732153203125
+ },
+ {
+ "epoch": 129,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 50439,
+ "train_examples": 50000,
+ "train_loss": 5859.4066059375
+ },
+ {
+ "epoch": 130,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 50830,
+ "train_examples": 50000,
+ "train_loss": 5458.5892265625
+ },
+ {
+ "epoch": 131,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 51221,
+ "train_examples": 50000,
+ "train_loss": 5791.215790625
+ },
+ {
+ "epoch": 132,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 51612,
+ "train_examples": 50000,
+ "train_loss": 5406.5100315625
+ },
+ {
+ "epoch": 133,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 52003,
+ "train_examples": 50000,
+ "train_loss": 5945.68840109375
+ },
+ {
+ "epoch": 134,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 52394,
+ "train_examples": 50000,
+ "train_loss": 5955.9906925
+ },
+ {
+ "epoch": 135,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 52785,
+ "train_examples": 50000,
+ "train_loss": 6088.13583
+ },
+ {
+ "epoch": 136,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 53176,
+ "train_examples": 50000,
+ "train_loss": 5708.5804115625
+ },
+ {
+ "epoch": 137,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 53567,
+ "train_examples": 50000,
+ "train_loss": 5858.96247203125
+ },
+ {
+ "epoch": 138,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 53958,
+ "train_examples": 50000,
+ "train_loss": 5559.608916796875
+ },
+ {
+ "epoch": 139,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 54349,
+ "train_examples": 50000,
+ "train_loss": 6280.13497953125
+ },
+ {
+ "epoch": 140,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 54740,
+ "train_examples": 50000,
+ "train_loss": 6046.8483821875
+ },
+ {
+ "epoch": 141,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 55131,
+ "train_examples": 50000,
+ "train_loss": 6157.783931953125
+ },
+ {
+ "epoch": 142,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 55522,
+ "train_examples": 50000,
+ "train_loss": 5573.92391609375
+ },
+ {
+ "epoch": 143,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 55913,
+ "train_examples": 50000,
+ "train_loss": 6364.5391053125
+ },
+ {
+ "epoch": 144,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 56304,
+ "train_examples": 50000,
+ "train_loss": 5800.603029921875
+ },
+ {
+ "epoch": 145,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 56695,
+ "train_examples": 50000,
+ "train_loss": 5515.605156015625
+ },
+ {
+ "epoch": 146,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 57086,
+ "train_examples": 50000,
+ "train_loss": 6255.1965696875
+ },
+ {
+ "epoch": 147,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 57477,
+ "train_examples": 50000,
+ "train_loss": 5524.64677375
+ },
+ {
+ "epoch": 148,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 57868,
+ "train_examples": 50000,
+ "train_loss": 5814.7748246875
+ },
+ {
+ "epoch": 149,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 58259,
+ "train_examples": 50000,
+ "train_loss": 5768.510907890625
+ },
+ {
+ "epoch": 150,
+ "lr": 0.003,
+ "output_lr": 0.010000000000000002,
+ "step": 58650,
+ "train_examples": 50000,
+ "train_loss": 6645.0301459375
+ },
+ {
+ "epoch": 151,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 59041,
+ "train_examples": 50000,
+ "train_loss": 2053.8047731640627
+ },
+ {
+ "epoch": 152,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 59432,
+ "train_examples": 50000,
+ "train_loss": 1795.6883098046876
+ },
+ {
+ "epoch": 153,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 59823,
+ "train_examples": 50000,
+ "train_loss": 1641.6798268359375
+ },
+ {
+ "epoch": 154,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 60214,
+ "train_examples": 50000,
+ "train_loss": 1529.563673671875
+ },
+ {
+ "epoch": 155,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 60605,
+ "train_examples": 50000,
+ "train_loss": 1432.554256171875
+ },
+ {
+ "epoch": 156,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 60996,
+ "train_examples": 50000,
+ "train_loss": 1364.2565076953124
+ },
+ {
+ "epoch": 157,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 61387,
+ "train_examples": 50000,
+ "train_loss": 1310.4285279882813
+ },
+ {
+ "epoch": 158,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 61778,
+ "train_examples": 50000,
+ "train_loss": 1249.9394315820311
+ },
+ {
+ "epoch": 159,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 62169,
+ "train_examples": 50000,
+ "train_loss": 1218.193686953125
+ },
+ {
+ "epoch": 160,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 62560,
+ "train_examples": 50000,
+ "train_loss": 1164.8607373828124
+ },
+ {
+ "epoch": 161,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 62951,
+ "train_examples": 50000,
+ "train_loss": 1151.4370148046876
+ },
+ {
+ "epoch": 162,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 63342,
+ "train_examples": 50000,
+ "train_loss": 1109.2520802734375
+ },
+ {
+ "epoch": 163,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 63733,
+ "train_examples": 50000,
+ "train_loss": 1062.9285024609376
+ },
+ {
+ "epoch": 164,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 64124,
+ "train_examples": 50000,
+ "train_loss": 1052.6393740625
+ },
+ {
+ "epoch": 165,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 64515,
+ "train_examples": 50000,
+ "train_loss": 1033.6278116796875
+ },
+ {
+ "epoch": 166,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 64906,
+ "train_examples": 50000,
+ "train_loss": 1011.65394265625
+ },
+ {
+ "epoch": 167,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 65297,
+ "train_examples": 50000,
+ "train_loss": 992.3858441992187
+ },
+ {
+ "epoch": 168,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 65688,
+ "train_examples": 50000,
+ "train_loss": 960.9110959765625
+ },
+ {
+ "epoch": 169,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 66079,
+ "train_examples": 50000,
+ "train_loss": 948.3309200390624
+ },
+ {
+ "epoch": 170,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 66470,
+ "train_examples": 50000,
+ "train_loss": 927.1700156445313
+ },
+ {
+ "epoch": 171,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 66861,
+ "train_examples": 50000,
+ "train_loss": 909.3944135546875
+ },
+ {
+ "epoch": 172,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 67252,
+ "train_examples": 50000,
+ "train_loss": 897.6867053515625
+ },
+ {
+ "epoch": 173,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 67643,
+ "train_examples": 50000,
+ "train_loss": 910.5499123828125
+ },
+ {
+ "epoch": 174,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 68034,
+ "train_examples": 50000,
+ "train_loss": 894.3773679882812
+ },
+ {
+ "epoch": 175,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 68425,
+ "train_examples": 50000,
+ "train_loss": 889.3560428515625
+ },
+ {
+ "epoch": 176,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 68816,
+ "train_examples": 50000,
+ "train_loss": 884.3611259765624
+ },
+ {
+ "epoch": 177,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 69207,
+ "train_examples": 50000,
+ "train_loss": 867.3764177734375
+ },
+ {
+ "epoch": 178,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 69598,
+ "train_examples": 50000,
+ "train_loss": 836.3760258203125
+ },
+ {
+ "epoch": 179,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 69989,
+ "train_examples": 50000,
+ "train_loss": 853.8147291015625
+ },
+ {
+ "epoch": 180,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 70380,
+ "train_examples": 50000,
+ "train_loss": 850.666523984375
+ },
+ {
+ "epoch": 181,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 70771,
+ "train_examples": 50000,
+ "train_loss": 878.9937553515625
+ },
+ {
+ "epoch": 182,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 71162,
+ "train_examples": 50000,
+ "train_loss": 858.25947765625
+ },
+ {
+ "epoch": 183,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 71553,
+ "train_examples": 50000,
+ "train_loss": 835.6640744921875
+ },
+ {
+ "epoch": 184,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 71944,
+ "train_examples": 50000,
+ "train_loss": 867.6950784570313
+ },
+ {
+ "epoch": 185,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 72335,
+ "train_examples": 50000,
+ "train_loss": 887.0921052734375
+ },
+ {
+ "epoch": 186,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 72726,
+ "train_examples": 50000,
+ "train_loss": 853.1495698828124
+ },
+ {
+ "epoch": 187,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 73117,
+ "train_examples": 50000,
+ "train_loss": 848.339279765625
+ },
+ {
+ "epoch": 188,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 73508,
+ "train_examples": 50000,
+ "train_loss": 845.5097470703125
+ },
+ {
+ "epoch": 189,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 73899,
+ "train_examples": 50000,
+ "train_loss": 840.1051002929687
+ },
+ {
+ "epoch": 190,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 74290,
+ "train_examples": 50000,
+ "train_loss": 846.5633176953125
+ },
+ {
+ "epoch": 191,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 74681,
+ "train_examples": 50000,
+ "train_loss": 864.5601016015625
+ },
+ {
+ "epoch": 192,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 75072,
+ "train_examples": 50000,
+ "train_loss": 823.3140968164063
+ },
+ {
+ "epoch": 193,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 75463,
+ "train_examples": 50000,
+ "train_loss": 817.0898944335937
+ },
+ {
+ "epoch": 194,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 75854,
+ "train_examples": 50000,
+ "train_loss": 857.8201413671875
+ },
+ {
+ "epoch": 195,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 76245,
+ "train_examples": 50000,
+ "train_loss": 818.969621796875
+ },
+ {
+ "epoch": 196,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 76636,
+ "train_examples": 50000,
+ "train_loss": 808.2602341015624
+ },
+ {
+ "epoch": 197,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 77027,
+ "train_examples": 50000,
+ "train_loss": 840.3612466796875
+ },
+ {
+ "epoch": 198,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 77418,
+ "train_examples": 50000,
+ "train_loss": 820.9785159765624
+ },
+ {
+ "epoch": 199,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 77809,
+ "train_examples": 50000,
+ "train_loss": 807.97784828125
+ },
+ {
+ "epoch": 200,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.0010000000000000002,
+ "step": 78200,
+ "train_examples": 50000,
+ "train_loss": 882.8125062890625
+ }
+ ],
+ "evaluation_protocol": {
+ "test_evaluations": 1,
+ "test_used_for_selection": false,
+ "validation_evaluations": 0
+ },
+ "final": {
+ "accuracy": 0.2604,
+ "epoch": 200,
+ "evaluation_split": "test",
+ "finite": true,
+ "loss": 835.0514375,
+ "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": 11,
+ "normalization_mean": [
+ 0.49140000343322754,
+ 0.4821999967098236,
+ 0.4465000033378601
+ ],
+ "normalization_std": [
+ 0.24699999392032623,
+ 0.2434999942779541,
+ 0.26159998774528503
+ ],
+ "split_from_training_only": true,
+ "split_seed": null,
+ "test_examples": 10000,
+ "train_examples": 50000,
+ "training_augmentation": "random_crop_32_padding_4_zero_then_horizontal_flip_p0.5",
+ "validation_class_counts": {},
+ "validation_examples": 0,
+ "validation_index_sha256": null
+ },
+ "timing": {
+ "apical_warmup_wall_s": 0.0,
+ "evaluation_wall_s": 2.036393165588379,
+ "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": 10413.29232931137,
+ "train_wall_s": 10411.230472564697
+ },
+ "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
+ }
+}
generated by cgit v1.2.3 (git 2.25.1) at 2026-08-01 00:53:05 +0000