summaryrefslogtreecommitdiff
path: root/results
diff options
context:
space:
mode:
authorYurenHao0426 <Blackhao0426@gmail.com>2026-07-22 13:22:24 -0500
committerYurenHao0426 <Blackhao0426@gmail.com>2026-07-22 13:22:24 -0500
commitc24a6a575bce15e7d7da541ecb752c65b5e68e47 (patch)
tree8166e8c7eebf2785cd811e4fbbc1578b185262bc /results
parent225459f7ddfcd3c8d89f0515a0d766bf2a15c848 (diff)
results: close fixed hierarchical FA short gate
Diffstat (limited to 'results')
-rw-r--r--results/oral_a_hfa_short/hfa_lr0.01.json534
-rw-r--r--results/oral_a_hfa_short/hfa_lr0.03.json534
-rw-r--r--results/oral_a_hfa_short/hfa_lr0.1.json534
-rw-r--r--results/oral_a_hfa_short_selection.json69
4 files changed, 1671 insertions, 0 deletions
diff --git a/results/oral_a_hfa_short/hfa_lr0.01.json b/results/oral_a_hfa_short/hfa_lr0.01.json
new file mode 100644
index 0000000..32a1365
--- /dev/null
+++ b/results/oral_a_hfa_short/hfa_lr0.01.json
@@ -0,0 +1,534 @@
+{
+ "architecture": {
+ "adaptive_apical_parameters": 0,
+ "base_width": 16,
+ "blocks_per_stage": 3,
+ "bn_eps": 1e-05,
+ "bn_momentum": 0.1,
+ "depth": 20,
+ "family": "CIFAR 6n+2 ResNet, option-A shortcuts",
+ "fixed_feedback_parameters": 267904,
+ "fixed_traffic_coefficients": 0,
+ "forward_parameters": 269722,
+ "hidden_shapes": [
+ [
+ 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
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ]
+ ],
+ "normalization": "batchnorm",
+ "predictor_parameters": 0,
+ "residual_scale": 1.0,
+ "vectorizer_mode": null,
+ "vectorizer_parameters": 0
+ },
+ "args": {
+ "a_scale": 1.0,
+ "a_warmup_steps": 0,
+ "alignment_probe": 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": 20,
+ "device": "cuda",
+ "epochs": 20,
+ "eta_A": 0.01,
+ "eta_P": 0.01,
+ "eval_every": 0,
+ "eval_split": "validation",
+ "learn_P": 0,
+ "loader_seed": 0,
+ "lr": 0.01,
+ "lr_gamma": 0.1,
+ "lr_milestones": "100,150",
+ "lr_schedule": "cosine",
+ "max_steps": 0,
+ "mode": "hfa",
+ "momentum": 0.9,
+ "normalization": "batchnorm",
+ "nuisance_scale": 0.0,
+ "out": "results/oral_a_hfa_short/hfa_lr0.01.json",
+ "output_lr": 0.1,
+ "pert_directions": 1,
+ "pert_every": 4,
+ "pert_sigma": 0.01,
+ "perturb_seed": 1000,
+ "predictor_warmup_steps": 0,
+ "residual_scale": null,
+ "seed": 0,
+ "split_seed": 2027,
+ "train_limit": 10000,
+ "use_residual": 1,
+ "val_examples": 5000,
+ "vectorizer_mode": "spatial_template",
+ "warmup_epochs": 0,
+ "weight_decay": 0.0001,
+ "weight_scale": 1.0,
+ "width": 16
+ },
+ "calibration_metric_space": null,
+ "counters": {
+ "apical_warmup_examples": 0,
+ "calibration_event_examples": 0,
+ "causal_scalar_observations": 0,
+ "logical_batch_loss_queries": 0,
+ "ordinary_examples": 200000,
+ "per_example_loss_terms": 0,
+ "perturbation_events": 0,
+ "perturbation_forward_examples": 0,
+ "predictor_warmup_examples": 0
+ },
+ "diagnostics": {
+ "early_third_mean": 0.004572224900281678,
+ "innovation_negative_gradient_cosine": [
+ 0.0020579462870955467,
+ 0.0012284447439014912,
+ 0.007407092023640871,
+ 0.00022282346617430449,
+ 0.016841813921928406,
+ -0.0003247710410505533,
+ 0.030386799946427345,
+ 0.0017005586996674538,
+ 0.060705218464136124,
+ 0.015753068029880524,
+ 0.09917206317186356,
+ 0.0011747075477614999,
+ 0.15302309393882751,
+ 0.006798408459872007,
+ 0.21235091984272003,
+ 0.10412535071372986,
+ 0.2994855046272278,
+ 0.13240748643875122,
+ 0.3866526484489441
+ ],
+ "normalization_state": "training_batch_stats_without_running_update",
+ "raw_negative_gradient_cosine": [
+ 0.0020579462870955467,
+ 0.0012284447439014912,
+ 0.007407092023640871,
+ 0.00022282346617430449,
+ 0.016841813921928406,
+ -0.0003247710410505533,
+ 0.030386799946427345,
+ 0.0017005586996674538,
+ 0.060705218464136124,
+ 0.015753068029880524,
+ 0.09917206317186356,
+ 0.0011747075477614999,
+ 0.15302309393882751,
+ 0.006798408459872007,
+ 0.21235091984272003,
+ 0.10412535071372986,
+ 0.2994855046272278,
+ 0.13240748643875122,
+ 0.3866526484489441
+ ],
+ "teaching_negative_gradient_cosine": [
+ 0.0020579462870955467,
+ 0.0012284447439014912,
+ 0.007407092023640871,
+ 0.00022282346617430449,
+ 0.016841813921928406,
+ -0.0003247710410505533,
+ 0.030386799946427345,
+ 0.0017005586996674538,
+ 0.060705218464136124,
+ 0.015753068029880524,
+ 0.09917206317186356,
+ 0.0011747075477614999,
+ 0.15302309393882751,
+ 0.006798408459872007,
+ 0.21235091984272003,
+ 0.10412535071372986,
+ 0.2994855046272278,
+ 0.13240748643875122,
+ 0.3866526484489441
+ ],
+ "wall_s": 0.12525105476379395
+ },
+ "epochs": [
+ {
+ "epoch": 1,
+ "lr": 0.01,
+ "output_lr": 0.1,
+ "step": 79,
+ "train_examples": 10000,
+ "train_loss": 2.7393385593414306
+ },
+ {
+ "epoch": 2,
+ "lr": 0.009938441702975689,
+ "output_lr": 0.0993844170297569,
+ "step": 158,
+ "train_examples": 10000,
+ "train_loss": 2.3863267654418947
+ },
+ {
+ "epoch": 3,
+ "lr": 0.009755282581475769,
+ "output_lr": 0.09755282581475769,
+ "step": 237,
+ "train_examples": 10000,
+ "train_loss": 2.37835657081604
+ },
+ {
+ "epoch": 4,
+ "lr": 0.00945503262094184,
+ "output_lr": 0.09455032620941839,
+ "step": 316,
+ "train_examples": 10000,
+ "train_loss": 2.443051778793335
+ },
+ {
+ "epoch": 5,
+ "lr": 0.009045084971874737,
+ "output_lr": 0.09045084971874738,
+ "step": 395,
+ "train_examples": 10000,
+ "train_loss": 2.3301113273620606
+ },
+ {
+ "epoch": 6,
+ "lr": 0.008535533905932738,
+ "output_lr": 0.08535533905932738,
+ "step": 474,
+ "train_examples": 10000,
+ "train_loss": 2.2099775257110594
+ },
+ {
+ "epoch": 7,
+ "lr": 0.007938926261462366,
+ "output_lr": 0.07938926261462366,
+ "step": 553,
+ "train_examples": 10000,
+ "train_loss": 2.256035614776611
+ },
+ {
+ "epoch": 8,
+ "lr": 0.007269952498697734,
+ "output_lr": 0.07269952498697733,
+ "step": 632,
+ "train_examples": 10000,
+ "train_loss": 2.191825396156311
+ },
+ {
+ "epoch": 9,
+ "lr": 0.006545084971874737,
+ "output_lr": 0.06545084971874737,
+ "step": 711,
+ "train_examples": 10000,
+ "train_loss": 2.0189443355560304
+ },
+ {
+ "epoch": 10,
+ "lr": 0.0057821723252011546,
+ "output_lr": 0.05782172325201155,
+ "step": 790,
+ "train_examples": 10000,
+ "train_loss": 1.9209733907699584
+ },
+ {
+ "epoch": 11,
+ "lr": 0.005,
+ "output_lr": 0.05,
+ "step": 869,
+ "train_examples": 10000,
+ "train_loss": 1.9395836170196534
+ },
+ {
+ "epoch": 12,
+ "lr": 0.004217827674798845,
+ "output_lr": 0.04217827674798845,
+ "step": 948,
+ "train_examples": 10000,
+ "train_loss": 1.8563231918334961
+ },
+ {
+ "epoch": 13,
+ "lr": 0.003454915028125263,
+ "output_lr": 0.03454915028125263,
+ "step": 1027,
+ "train_examples": 10000,
+ "train_loss": 1.8415284217834473
+ },
+ {
+ "epoch": 14,
+ "lr": 0.0027300475013022664,
+ "output_lr": 0.027300475013022664,
+ "step": 1106,
+ "train_examples": 10000,
+ "train_loss": 1.816683451461792
+ },
+ {
+ "epoch": 15,
+ "lr": 0.0020610737385376348,
+ "output_lr": 0.02061073738537635,
+ "step": 1185,
+ "train_examples": 10000,
+ "train_loss": 1.7413049175262452
+ },
+ {
+ "epoch": 16,
+ "lr": 0.0014644660940672626,
+ "output_lr": 0.014644660940672627,
+ "step": 1264,
+ "train_examples": 10000,
+ "train_loss": 1.7405946004867554
+ },
+ {
+ "epoch": 17,
+ "lr": 0.0009549150281252633,
+ "output_lr": 0.009549150281252633,
+ "step": 1343,
+ "train_examples": 10000,
+ "train_loss": 1.72834784450531
+ },
+ {
+ "epoch": 18,
+ "lr": 0.0005449673790581611,
+ "output_lr": 0.005449673790581611,
+ "step": 1422,
+ "train_examples": 10000,
+ "train_loss": 1.7004222576141357
+ },
+ {
+ "epoch": 19,
+ "lr": 0.00024471741852423234,
+ "output_lr": 0.0024471741852423235,
+ "step": 1501,
+ "train_examples": 10000,
+ "train_loss": 1.686814351272583
+ },
+ {
+ "epoch": 20,
+ "lr": 6.15582970243117e-05,
+ "output_lr": 0.0006155829702431171,
+ "step": 1580,
+ "train_examples": 10000,
+ "train_loss": 1.6884625955581665
+ }
+ ],
+ "evaluation_protocol": {
+ "test_evaluations": 0,
+ "test_used_for_selection": false,
+ "validation_evaluations": 1
+ },
+ "final": {
+ "accuracy": 0.3964,
+ "epoch": 20,
+ "evaluation_split": "validation",
+ "finite": true,
+ "loss": 1.6798031494140624,
+ "step": 1580
+ },
+ "hardware": {
+ "cuda_device_name": "NVIDIA GeForce GTX 1080",
+ "cuda_visible_devices": "5",
+ "device": "cuda",
+ "device_total_memory_bytes": 8507949056,
+ "peak_memory_allocated_bytes": 1730831872,
+ "peak_memory_reserved_bytes": 2390753280,
+ "torch_version": "2.3.1+cu118"
+ },
+ "protocol_family": "oral_a_cifar_local_resnet_development",
+ "provenance": {
+ "git_commit": "f37644eb452f0eb3d7f368740f8b39493f11b2aa",
+ "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": "validation",
+ "input_layout": "NCHW",
+ "input_shape": [
+ 3,
+ 32,
+ 32
+ ],
+ "loader_seed": 0,
+ "normalization_mean": [
+ 0.49140000343322754,
+ 0.4821999967098236,
+ 0.4465000033378601
+ ],
+ "normalization_std": [
+ 0.24699999392032623,
+ 0.2434999942779541,
+ 0.26159998774528503
+ ],
+ "split_from_training_only": true,
+ "split_seed": 2027,
+ "test_examples": 10000,
+ "train_examples": 10000,
+ "training_augmentation": "random_crop_32_padding_4_zero_then_horizontal_flip_p0.5",
+ "validation_class_counts": {
+ "0": 500,
+ "1": 500,
+ "2": 500,
+ "3": 500,
+ "4": 500,
+ "5": 500,
+ "6": 500,
+ "7": 500,
+ "8": 500,
+ "9": 500
+ },
+ "validation_examples": 5000,
+ "validation_index_sha256": "8328b206a97c420e49e54e3eca4abe3274c4756b084355784ea3fb8059e4515b"
+ },
+ "timing": {
+ "apical_warmup_wall_s": 0.0,
+ "evaluation_wall_s": 0.36951661109924316,
+ "predictor_warmup_wall_s": 0.0,
+ "timing_excludes_data_loading_hashing_and_model_construction": true,
+ "total_timed_wall_s": 89.53109097480774,
+ "train_wall_s": 89.15893292427063
+ },
+ "work": {
+ "apical_macs_per_example": 40108672,
+ "causal_scalar_observations": 0,
+ "components": {
+ "apical_projection_macs": 8021734400000,
+ "apical_regression_macs": 0,
+ "bp_reverse_macs_estimate": 0,
+ "local_weight_correlation_macs": 8110208000000,
+ "ordinary_forward_macs": 8110208000000,
+ "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; elementwise nonlinearities and optimizer arithmetic excluded",
+ "forward_macs_per_example": 40551040,
+ "logical_batch_loss_queries": 0,
+ "per_example_cross_entropy_terms": 0,
+ "total_clean_forward_examples": 200000,
+ "total_forward_equivalent_examples": 200000,
+ "total_macs_estimate": 24242150400000
+ }
+}
diff --git a/results/oral_a_hfa_short/hfa_lr0.03.json b/results/oral_a_hfa_short/hfa_lr0.03.json
new file mode 100644
index 0000000..113ea64
--- /dev/null
+++ b/results/oral_a_hfa_short/hfa_lr0.03.json
@@ -0,0 +1,534 @@
+{
+ "architecture": {
+ "adaptive_apical_parameters": 0,
+ "base_width": 16,
+ "blocks_per_stage": 3,
+ "bn_eps": 1e-05,
+ "bn_momentum": 0.1,
+ "depth": 20,
+ "family": "CIFAR 6n+2 ResNet, option-A shortcuts",
+ "fixed_feedback_parameters": 267904,
+ "fixed_traffic_coefficients": 0,
+ "forward_parameters": 269722,
+ "hidden_shapes": [
+ [
+ 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
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ]
+ ],
+ "normalization": "batchnorm",
+ "predictor_parameters": 0,
+ "residual_scale": 1.0,
+ "vectorizer_mode": null,
+ "vectorizer_parameters": 0
+ },
+ "args": {
+ "a_scale": 1.0,
+ "a_warmup_steps": 0,
+ "alignment_probe": 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": 20,
+ "device": "cuda",
+ "epochs": 20,
+ "eta_A": 0.01,
+ "eta_P": 0.01,
+ "eval_every": 0,
+ "eval_split": "validation",
+ "learn_P": 0,
+ "loader_seed": 0,
+ "lr": 0.03,
+ "lr_gamma": 0.1,
+ "lr_milestones": "100,150",
+ "lr_schedule": "cosine",
+ "max_steps": 0,
+ "mode": "hfa",
+ "momentum": 0.9,
+ "normalization": "batchnorm",
+ "nuisance_scale": 0.0,
+ "out": "results/oral_a_hfa_short/hfa_lr0.03.json",
+ "output_lr": 0.1,
+ "pert_directions": 1,
+ "pert_every": 4,
+ "pert_sigma": 0.01,
+ "perturb_seed": 1000,
+ "predictor_warmup_steps": 0,
+ "residual_scale": null,
+ "seed": 0,
+ "split_seed": 2027,
+ "train_limit": 10000,
+ "use_residual": 1,
+ "val_examples": 5000,
+ "vectorizer_mode": "spatial_template",
+ "warmup_epochs": 0,
+ "weight_decay": 0.0001,
+ "weight_scale": 1.0,
+ "width": 16
+ },
+ "calibration_metric_space": null,
+ "counters": {
+ "apical_warmup_examples": 0,
+ "calibration_event_examples": 0,
+ "causal_scalar_observations": 0,
+ "logical_batch_loss_queries": 0,
+ "ordinary_examples": 200000,
+ "per_example_loss_terms": 0,
+ "perturbation_events": 0,
+ "perturbation_forward_examples": 0,
+ "predictor_warmup_examples": 0
+ },
+ "diagnostics": {
+ "early_third_mean": 0.01133825796811531,
+ "innovation_negative_gradient_cosine": [
+ 0.005747726187109947,
+ 0.003406030824407935,
+ 0.01511126384139061,
+ 0.0035506836138665676,
+ 0.038720425218343735,
+ 0.0014934181235730648,
+ 0.07037845999002457,
+ 0.010694023221731186,
+ 0.11927632242441177,
+ 0.029549002647399902,
+ 0.18039162456989288,
+ 0.023218173533678055,
+ 0.2376384288072586,
+ 0.039402417838573456,
+ 0.30495548248291016,
+ 0.2014918029308319,
+ 0.38542941212654114,
+ 0.2127426713705063,
+ 0.4092516303062439
+ ],
+ "normalization_state": "training_batch_stats_without_running_update",
+ "raw_negative_gradient_cosine": [
+ 0.005747726187109947,
+ 0.003406030824407935,
+ 0.01511126384139061,
+ 0.0035506836138665676,
+ 0.038720425218343735,
+ 0.0014934181235730648,
+ 0.07037845999002457,
+ 0.010694023221731186,
+ 0.11927632242441177,
+ 0.029549002647399902,
+ 0.18039162456989288,
+ 0.023218173533678055,
+ 0.2376384288072586,
+ 0.039402417838573456,
+ 0.30495548248291016,
+ 0.2014918029308319,
+ 0.38542941212654114,
+ 0.2127426713705063,
+ 0.4092516303062439
+ ],
+ "teaching_negative_gradient_cosine": [
+ 0.005747726187109947,
+ 0.003406030824407935,
+ 0.01511126384139061,
+ 0.0035506836138665676,
+ 0.038720425218343735,
+ 0.0014934181235730648,
+ 0.07037845999002457,
+ 0.010694023221731186,
+ 0.11927632242441177,
+ 0.029549002647399902,
+ 0.18039162456989288,
+ 0.023218173533678055,
+ 0.2376384288072586,
+ 0.039402417838573456,
+ 0.30495548248291016,
+ 0.2014918029308319,
+ 0.38542941212654114,
+ 0.2127426713705063,
+ 0.4092516303062439
+ ],
+ "wall_s": 0.13352656364440918
+ },
+ "epochs": [
+ {
+ "epoch": 1,
+ "lr": 0.03,
+ "output_lr": 0.1,
+ "step": 79,
+ "train_examples": 10000,
+ "train_loss": 2.7139521350860596
+ },
+ {
+ "epoch": 2,
+ "lr": 0.029815325108927065,
+ "output_lr": 0.0993844170297569,
+ "step": 158,
+ "train_examples": 10000,
+ "train_loss": 2.6086353378295897
+ },
+ {
+ "epoch": 3,
+ "lr": 0.029265847744427302,
+ "output_lr": 0.09755282581475769,
+ "step": 237,
+ "train_examples": 10000,
+ "train_loss": 2.4976746765136717
+ },
+ {
+ "epoch": 4,
+ "lr": 0.028365097862825515,
+ "output_lr": 0.09455032620941839,
+ "step": 316,
+ "train_examples": 10000,
+ "train_loss": 2.4073143131256103
+ },
+ {
+ "epoch": 5,
+ "lr": 0.02713525491562421,
+ "output_lr": 0.09045084971874738,
+ "step": 395,
+ "train_examples": 10000,
+ "train_loss": 2.360751438522339
+ },
+ {
+ "epoch": 6,
+ "lr": 0.02560660171779821,
+ "output_lr": 0.08535533905932738,
+ "step": 474,
+ "train_examples": 10000,
+ "train_loss": 2.2522952722549436
+ },
+ {
+ "epoch": 7,
+ "lr": 0.023816778784387097,
+ "output_lr": 0.07938926261462366,
+ "step": 553,
+ "train_examples": 10000,
+ "train_loss": 2.279150248336792
+ },
+ {
+ "epoch": 8,
+ "lr": 0.0218098574960932,
+ "output_lr": 0.07269952498697733,
+ "step": 632,
+ "train_examples": 10000,
+ "train_loss": 2.2757634071350097
+ },
+ {
+ "epoch": 9,
+ "lr": 0.01963525491562421,
+ "output_lr": 0.06545084971874737,
+ "step": 711,
+ "train_examples": 10000,
+ "train_loss": 2.0739892147064207
+ },
+ {
+ "epoch": 10,
+ "lr": 0.017346516975603462,
+ "output_lr": 0.05782172325201155,
+ "step": 790,
+ "train_examples": 10000,
+ "train_loss": 1.8923955001831054
+ },
+ {
+ "epoch": 11,
+ "lr": 0.015,
+ "output_lr": 0.05,
+ "step": 869,
+ "train_examples": 10000,
+ "train_loss": 1.9085709817886352
+ },
+ {
+ "epoch": 12,
+ "lr": 0.012653483024396533,
+ "output_lr": 0.04217827674798845,
+ "step": 948,
+ "train_examples": 10000,
+ "train_loss": 1.8094817079544068
+ },
+ {
+ "epoch": 13,
+ "lr": 0.01036474508437579,
+ "output_lr": 0.03454915028125263,
+ "step": 1027,
+ "train_examples": 10000,
+ "train_loss": 1.7567100723266602
+ },
+ {
+ "epoch": 14,
+ "lr": 0.0081901425039068,
+ "output_lr": 0.027300475013022664,
+ "step": 1106,
+ "train_examples": 10000,
+ "train_loss": 1.7179550045013428
+ },
+ {
+ "epoch": 15,
+ "lr": 0.006183221215612904,
+ "output_lr": 0.02061073738537635,
+ "step": 1185,
+ "train_examples": 10000,
+ "train_loss": 1.6542307502746583
+ },
+ {
+ "epoch": 16,
+ "lr": 0.004393398282201788,
+ "output_lr": 0.014644660940672627,
+ "step": 1264,
+ "train_examples": 10000,
+ "train_loss": 1.6502908784866333
+ },
+ {
+ "epoch": 17,
+ "lr": 0.00286474508437579,
+ "output_lr": 0.009549150281252633,
+ "step": 1343,
+ "train_examples": 10000,
+ "train_loss": 1.6148505542755127
+ },
+ {
+ "epoch": 18,
+ "lr": 0.001634902137174483,
+ "output_lr": 0.005449673790581611,
+ "step": 1422,
+ "train_examples": 10000,
+ "train_loss": 1.5908274923324586
+ },
+ {
+ "epoch": 19,
+ "lr": 0.000734152255572697,
+ "output_lr": 0.0024471741852423235,
+ "step": 1501,
+ "train_examples": 10000,
+ "train_loss": 1.570961207962036
+ },
+ {
+ "epoch": 20,
+ "lr": 0.0001846748910729351,
+ "output_lr": 0.0006155829702431171,
+ "step": 1580,
+ "train_examples": 10000,
+ "train_loss": 1.5774935535430907
+ }
+ ],
+ "evaluation_protocol": {
+ "test_evaluations": 0,
+ "test_used_for_selection": false,
+ "validation_evaluations": 1
+ },
+ "final": {
+ "accuracy": 0.4158,
+ "epoch": 20,
+ "evaluation_split": "validation",
+ "finite": true,
+ "loss": 1.6006802001953124,
+ "step": 1580
+ },
+ "hardware": {
+ "cuda_device_name": "NVIDIA GeForce GTX 1080",
+ "cuda_visible_devices": "7",
+ "device": "cuda",
+ "device_total_memory_bytes": 8507949056,
+ "peak_memory_allocated_bytes": 1730831872,
+ "peak_memory_reserved_bytes": 2390753280,
+ "torch_version": "2.3.1+cu118"
+ },
+ "protocol_family": "oral_a_cifar_local_resnet_development",
+ "provenance": {
+ "git_commit": "f37644eb452f0eb3d7f368740f8b39493f11b2aa",
+ "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": "validation",
+ "input_layout": "NCHW",
+ "input_shape": [
+ 3,
+ 32,
+ 32
+ ],
+ "loader_seed": 0,
+ "normalization_mean": [
+ 0.49140000343322754,
+ 0.4821999967098236,
+ 0.4465000033378601
+ ],
+ "normalization_std": [
+ 0.24699999392032623,
+ 0.2434999942779541,
+ 0.26159998774528503
+ ],
+ "split_from_training_only": true,
+ "split_seed": 2027,
+ "test_examples": 10000,
+ "train_examples": 10000,
+ "training_augmentation": "random_crop_32_padding_4_zero_then_horizontal_flip_p0.5",
+ "validation_class_counts": {
+ "0": 500,
+ "1": 500,
+ "2": 500,
+ "3": 500,
+ "4": 500,
+ "5": 500,
+ "6": 500,
+ "7": 500,
+ "8": 500,
+ "9": 500
+ },
+ "validation_examples": 5000,
+ "validation_index_sha256": "8328b206a97c420e49e54e3eca4abe3274c4756b084355784ea3fb8059e4515b"
+ },
+ "timing": {
+ "apical_warmup_wall_s": 0.0,
+ "evaluation_wall_s": 0.362001895904541,
+ "predictor_warmup_wall_s": 0.0,
+ "timing_excludes_data_loading_hashing_and_model_construction": true,
+ "total_timed_wall_s": 89.29601311683655,
+ "train_wall_s": 88.93181085586548
+ },
+ "work": {
+ "apical_macs_per_example": 40108672,
+ "causal_scalar_observations": 0,
+ "components": {
+ "apical_projection_macs": 8021734400000,
+ "apical_regression_macs": 0,
+ "bp_reverse_macs_estimate": 0,
+ "local_weight_correlation_macs": 8110208000000,
+ "ordinary_forward_macs": 8110208000000,
+ "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; elementwise nonlinearities and optimizer arithmetic excluded",
+ "forward_macs_per_example": 40551040,
+ "logical_batch_loss_queries": 0,
+ "per_example_cross_entropy_terms": 0,
+ "total_clean_forward_examples": 200000,
+ "total_forward_equivalent_examples": 200000,
+ "total_macs_estimate": 24242150400000
+ }
+}
diff --git a/results/oral_a_hfa_short/hfa_lr0.1.json b/results/oral_a_hfa_short/hfa_lr0.1.json
new file mode 100644
index 0000000..6bcb4ba
--- /dev/null
+++ b/results/oral_a_hfa_short/hfa_lr0.1.json
@@ -0,0 +1,534 @@
+{
+ "architecture": {
+ "adaptive_apical_parameters": 0,
+ "base_width": 16,
+ "blocks_per_stage": 3,
+ "bn_eps": 1e-05,
+ "bn_momentum": 0.1,
+ "depth": 20,
+ "family": "CIFAR 6n+2 ResNet, option-A shortcuts",
+ "fixed_feedback_parameters": 267904,
+ "fixed_traffic_coefficients": 0,
+ "forward_parameters": 269722,
+ "hidden_shapes": [
+ [
+ 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
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ],
+ [
+ 64,
+ 8,
+ 8
+ ]
+ ],
+ "normalization": "batchnorm",
+ "predictor_parameters": 0,
+ "residual_scale": 1.0,
+ "vectorizer_mode": null,
+ "vectorizer_parameters": 0
+ },
+ "args": {
+ "a_scale": 1.0,
+ "a_warmup_steps": 0,
+ "alignment_probe": 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": 20,
+ "device": "cuda",
+ "epochs": 20,
+ "eta_A": 0.01,
+ "eta_P": 0.01,
+ "eval_every": 0,
+ "eval_split": "validation",
+ "learn_P": 0,
+ "loader_seed": 0,
+ "lr": 0.1,
+ "lr_gamma": 0.1,
+ "lr_milestones": "100,150",
+ "lr_schedule": "cosine",
+ "max_steps": 0,
+ "mode": "hfa",
+ "momentum": 0.9,
+ "normalization": "batchnorm",
+ "nuisance_scale": 0.0,
+ "out": "results/oral_a_hfa_short/hfa_lr0.1.json",
+ "output_lr": 0.1,
+ "pert_directions": 1,
+ "pert_every": 4,
+ "pert_sigma": 0.01,
+ "perturb_seed": 1000,
+ "predictor_warmup_steps": 0,
+ "residual_scale": null,
+ "seed": 0,
+ "split_seed": 2027,
+ "train_limit": 10000,
+ "use_residual": 1,
+ "val_examples": 5000,
+ "vectorizer_mode": "spatial_template",
+ "warmup_epochs": 0,
+ "weight_decay": 0.0001,
+ "weight_scale": 1.0,
+ "width": 16
+ },
+ "calibration_metric_space": null,
+ "counters": {
+ "apical_warmup_examples": 0,
+ "calibration_event_examples": 0,
+ "causal_scalar_observations": 0,
+ "logical_batch_loss_queries": 0,
+ "ordinary_examples": 200000,
+ "per_example_loss_terms": 0,
+ "perturbation_events": 0,
+ "perturbation_forward_examples": 0,
+ "predictor_warmup_examples": 0
+ },
+ "diagnostics": {
+ "early_third_mean": 0.040405964168409504,
+ "innovation_negative_gradient_cosine": [
+ 0.023671559989452362,
+ 0.013195669278502464,
+ 0.06748953461647034,
+ 0.017365306615829468,
+ 0.11135254800319672,
+ 0.009361166507005692,
+ 0.13842418789863586,
+ 0.01869494654238224,
+ 0.19049662351608276,
+ 0.023486969992518425,
+ 0.22018298506736755,
+ 0.033551692962646484,
+ 0.23453857004642487,
+ 0.03822024166584015,
+ 0.2642383873462677,
+ 0.18111935257911682,
+ 0.26726263761520386,
+ 0.16317608952522278,
+ 0.2346772849559784
+ ],
+ "normalization_state": "training_batch_stats_without_running_update",
+ "raw_negative_gradient_cosine": [
+ 0.023671559989452362,
+ 0.013195669278502464,
+ 0.06748953461647034,
+ 0.017365306615829468,
+ 0.11135254800319672,
+ 0.009361166507005692,
+ 0.13842418789863586,
+ 0.01869494654238224,
+ 0.19049662351608276,
+ 0.023486969992518425,
+ 0.22018298506736755,
+ 0.033551692962646484,
+ 0.23453857004642487,
+ 0.03822024166584015,
+ 0.2642383873462677,
+ 0.18111935257911682,
+ 0.26726263761520386,
+ 0.16317608952522278,
+ 0.2346772849559784
+ ],
+ "teaching_negative_gradient_cosine": [
+ 0.023671559989452362,
+ 0.013195669278502464,
+ 0.06748953461647034,
+ 0.017365306615829468,
+ 0.11135254800319672,
+ 0.009361166507005692,
+ 0.13842418789863586,
+ 0.01869494654238224,
+ 0.19049662351608276,
+ 0.023486969992518425,
+ 0.22018298506736755,
+ 0.033551692962646484,
+ 0.23453857004642487,
+ 0.03822024166584015,
+ 0.2642383873462677,
+ 0.18111935257911682,
+ 0.26726263761520386,
+ 0.16317608952522278,
+ 0.2346772849559784
+ ],
+ "wall_s": 0.12181758880615234
+ },
+ "epochs": [
+ {
+ "epoch": 1,
+ "lr": 0.1,
+ "output_lr": 0.1,
+ "step": 79,
+ "train_examples": 10000,
+ "train_loss": 3.256435613632202
+ },
+ {
+ "epoch": 2,
+ "lr": 0.0993844170297569,
+ "output_lr": 0.0993844170297569,
+ "step": 158,
+ "train_examples": 10000,
+ "train_loss": 3.5897572273254394
+ },
+ {
+ "epoch": 3,
+ "lr": 0.09755282581475769,
+ "output_lr": 0.09755282581475769,
+ "step": 237,
+ "train_examples": 10000,
+ "train_loss": 3.483150173187256
+ },
+ {
+ "epoch": 4,
+ "lr": 0.09455032620941839,
+ "output_lr": 0.09455032620941839,
+ "step": 316,
+ "train_examples": 10000,
+ "train_loss": 3.499296664428711
+ },
+ {
+ "epoch": 5,
+ "lr": 0.09045084971874738,
+ "output_lr": 0.09045084971874738,
+ "step": 395,
+ "train_examples": 10000,
+ "train_loss": 3.5787659603118898
+ },
+ {
+ "epoch": 6,
+ "lr": 0.08535533905932738,
+ "output_lr": 0.08535533905932738,
+ "step": 474,
+ "train_examples": 10000,
+ "train_loss": 2.639942427062988
+ },
+ {
+ "epoch": 7,
+ "lr": 0.07938926261462366,
+ "output_lr": 0.07938926261462366,
+ "step": 553,
+ "train_examples": 10000,
+ "train_loss": 3.3110560848236084
+ },
+ {
+ "epoch": 8,
+ "lr": 0.07269952498697733,
+ "output_lr": 0.07269952498697733,
+ "step": 632,
+ "train_examples": 10000,
+ "train_loss": 2.8327583209991456
+ },
+ {
+ "epoch": 9,
+ "lr": 0.06545084971874737,
+ "output_lr": 0.06545084971874737,
+ "step": 711,
+ "train_examples": 10000,
+ "train_loss": 2.2873985042572023
+ },
+ {
+ "epoch": 10,
+ "lr": 0.05782172325201155,
+ "output_lr": 0.05782172325201155,
+ "step": 790,
+ "train_examples": 10000,
+ "train_loss": 2.1511593353271485
+ },
+ {
+ "epoch": 11,
+ "lr": 0.05,
+ "output_lr": 0.05,
+ "step": 869,
+ "train_examples": 10000,
+ "train_loss": 2.080998848724365
+ },
+ {
+ "epoch": 12,
+ "lr": 0.04217827674798845,
+ "output_lr": 0.04217827674798845,
+ "step": 948,
+ "train_examples": 10000,
+ "train_loss": 1.880720576095581
+ },
+ {
+ "epoch": 13,
+ "lr": 0.03454915028125263,
+ "output_lr": 0.03454915028125263,
+ "step": 1027,
+ "train_examples": 10000,
+ "train_loss": 1.80372526512146
+ },
+ {
+ "epoch": 14,
+ "lr": 0.027300475013022664,
+ "output_lr": 0.027300475013022664,
+ "step": 1106,
+ "train_examples": 10000,
+ "train_loss": 1.7575418134689331
+ },
+ {
+ "epoch": 15,
+ "lr": 0.02061073738537635,
+ "output_lr": 0.02061073738537635,
+ "step": 1185,
+ "train_examples": 10000,
+ "train_loss": 1.6703623008728028
+ },
+ {
+ "epoch": 16,
+ "lr": 0.014644660940672627,
+ "output_lr": 0.014644660940672627,
+ "step": 1264,
+ "train_examples": 10000,
+ "train_loss": 1.6365677841186523
+ },
+ {
+ "epoch": 17,
+ "lr": 0.009549150281252633,
+ "output_lr": 0.009549150281252633,
+ "step": 1343,
+ "train_examples": 10000,
+ "train_loss": 1.6003902708053588
+ },
+ {
+ "epoch": 18,
+ "lr": 0.005449673790581611,
+ "output_lr": 0.005449673790581611,
+ "step": 1422,
+ "train_examples": 10000,
+ "train_loss": 1.5677705976486207
+ },
+ {
+ "epoch": 19,
+ "lr": 0.0024471741852423235,
+ "output_lr": 0.0024471741852423235,
+ "step": 1501,
+ "train_examples": 10000,
+ "train_loss": 1.5354159210205078
+ },
+ {
+ "epoch": 20,
+ "lr": 0.0006155829702431171,
+ "output_lr": 0.0006155829702431171,
+ "step": 1580,
+ "train_examples": 10000,
+ "train_loss": 1.5316140928268434
+ }
+ ],
+ "evaluation_protocol": {
+ "test_evaluations": 0,
+ "test_used_for_selection": false,
+ "validation_evaluations": 1
+ },
+ "final": {
+ "accuracy": 0.4352,
+ "epoch": 20,
+ "evaluation_split": "validation",
+ "finite": true,
+ "loss": 1.5658135009765626,
+ "step": 1580
+ },
+ "hardware": {
+ "cuda_device_name": "NVIDIA GeForce GTX 1080",
+ "cuda_visible_devices": "5",
+ "device": "cuda",
+ "device_total_memory_bytes": 8507949056,
+ "peak_memory_allocated_bytes": 1730831872,
+ "peak_memory_reserved_bytes": 2390753280,
+ "torch_version": "2.3.1+cu118"
+ },
+ "protocol_family": "oral_a_cifar_local_resnet_development",
+ "provenance": {
+ "git_commit": "f37644eb452f0eb3d7f368740f8b39493f11b2aa",
+ "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": "validation",
+ "input_layout": "NCHW",
+ "input_shape": [
+ 3,
+ 32,
+ 32
+ ],
+ "loader_seed": 0,
+ "normalization_mean": [
+ 0.49140000343322754,
+ 0.4821999967098236,
+ 0.4465000033378601
+ ],
+ "normalization_std": [
+ 0.24699999392032623,
+ 0.2434999942779541,
+ 0.26159998774528503
+ ],
+ "split_from_training_only": true,
+ "split_seed": 2027,
+ "test_examples": 10000,
+ "train_examples": 10000,
+ "training_augmentation": "random_crop_32_padding_4_zero_then_horizontal_flip_p0.5",
+ "validation_class_counts": {
+ "0": 500,
+ "1": 500,
+ "2": 500,
+ "3": 500,
+ "4": 500,
+ "5": 500,
+ "6": 500,
+ "7": 500,
+ "8": 500,
+ "9": 500
+ },
+ "validation_examples": 5000,
+ "validation_index_sha256": "8328b206a97c420e49e54e3eca4abe3274c4756b084355784ea3fb8059e4515b"
+ },
+ "timing": {
+ "apical_warmup_wall_s": 0.0,
+ "evaluation_wall_s": 0.3655712604522705,
+ "predictor_warmup_wall_s": 0.0,
+ "timing_excludes_data_loading_hashing_and_model_construction": true,
+ "total_timed_wall_s": 89.62506937980652,
+ "train_wall_s": 89.25713443756104
+ },
+ "work": {
+ "apical_macs_per_example": 40108672,
+ "causal_scalar_observations": 0,
+ "components": {
+ "apical_projection_macs": 8021734400000,
+ "apical_regression_macs": 0,
+ "bp_reverse_macs_estimate": 0,
+ "local_weight_correlation_macs": 8110208000000,
+ "ordinary_forward_macs": 8110208000000,
+ "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; elementwise nonlinearities and optimizer arithmetic excluded",
+ "forward_macs_per_example": 40551040,
+ "logical_batch_loss_queries": 0,
+ "per_example_cross_entropy_terms": 0,
+ "total_clean_forward_examples": 200000,
+ "total_forward_equivalent_examples": 200000,
+ "total_macs_estimate": 24242150400000
+ }
+}
diff --git a/results/oral_a_hfa_short_selection.json b/results/oral_a_hfa_short_selection.json
new file mode 100644
index 0000000..394de01
--- /dev/null
+++ b/results/oral_a_hfa_short_selection.json
@@ -0,0 +1,69 @@
+{
+ "comparators": {
+ "matched_short_bp_accuracy": 0.7494,
+ "matched_short_dfa_accuracy": 0.3716,
+ "matched_short_failed_v1_sdil_accuracy": 0.4198
+ },
+ "confirmation_test_seeds_touched": false,
+ "full_validation_threshold": 0.5,
+ "protocol": "convolutional_hfa_S1_v1",
+ "rows": [
+ {
+ "accuracy": 0.3964,
+ "early_third_alignment": 0.004572224900281678,
+ "finite": true,
+ "fixed_feedback_parameters": 267904,
+ "logical_batch_loss_queries": 0,
+ "loss": 1.6798031494140624,
+ "lr": 0.01,
+ "path": "results/oral_a_hfa_short/hfa_lr0.01.json",
+ "peak_memory_allocated_bytes": 1730831872,
+ "source_commit": "f37644eb452f0eb3d7f368740f8b39493f11b2aa",
+ "total_macs": 24242150400000,
+ "wall_s": 89.53109097480774
+ },
+ {
+ "accuracy": 0.4158,
+ "early_third_alignment": 0.01133825796811531,
+ "finite": true,
+ "fixed_feedback_parameters": 267904,
+ "logical_batch_loss_queries": 0,
+ "loss": 1.6006802001953124,
+ "lr": 0.03,
+ "path": "results/oral_a_hfa_short/hfa_lr0.03.json",
+ "peak_memory_allocated_bytes": 1730831872,
+ "source_commit": "f37644eb452f0eb3d7f368740f8b39493f11b2aa",
+ "total_macs": 24242150400000,
+ "wall_s": 89.29601311683655
+ },
+ {
+ "accuracy": 0.4352,
+ "early_third_alignment": 0.040405964168409504,
+ "finite": true,
+ "fixed_feedback_parameters": 267904,
+ "logical_batch_loss_queries": 0,
+ "loss": 1.5658135009765626,
+ "lr": 0.1,
+ "path": "results/oral_a_hfa_short/hfa_lr0.1.json",
+ "peak_memory_allocated_bytes": 1730831872,
+ "source_commit": "f37644eb452f0eb3d7f368740f8b39493f11b2aa",
+ "total_macs": 24242150400000,
+ "wall_s": 89.62506937980652
+ }
+ ],
+ "selected": {
+ "accuracy": 0.4352,
+ "early_third_alignment": 0.040405964168409504,
+ "finite": true,
+ "fixed_feedback_parameters": 267904,
+ "logical_batch_loss_queries": 0,
+ "loss": 1.5658135009765626,
+ "lr": 0.1,
+ "path": "results/oral_a_hfa_short/hfa_lr0.1.json",
+ "peak_memory_allocated_bytes": 1730831872,
+ "source_commit": "f37644eb452f0eb3d7f368740f8b39493f11b2aa",
+ "total_macs": 24242150400000,
+ "wall_s": 89.62506937980652
+ },
+ "status": "selected_full_closed"
+}