results: pass Oral-A short accuracy screen

1 files changed, 639 insertions, 0 deletions

diff --git a/results/oral_a_short/sdil_channel_gated_lr0.01.json b/results/oral_a_short/sdil_channel_gated_lr0.01.json
new file mode 100644
index 0000000..d630417
--- /dev/null
+++ b/results/oral_a_short/sdil_channel_gated_lr0.01.json
@@ -0,0 +1,639 @@
+{
+  "apical_warmup": {
+    "last": {
+      "calibration_mse": 518.2959050717561,
+      "prediction_target_cosine": 0.00016553175127441687,
+      "target_power": 518.2958984375
+    },
+    "steps": 100
+  },
+  "architecture": {
+    "adaptive_apical_parameters": 390592,
+    "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_traffic_coefficients": 188416,
+    "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": 376832,
+    "residual_scale": 1.0,
+    "vectorizer_mode": "channel_gated",
+    "vectorizer_parameters": 13760
+  },
+  "args": {
+    "a_scale": 1.0,
+    "a_warmup_steps": 100,
+    "alignment_probe": 32,
+    "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:0",
+    "epochs": 20,
+    "eta_A": 0.001,
+    "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": "sdil",
+    "momentum": 0.9,
+    "normalization": "batchnorm",
+    "nuisance_scale": 0.0,
+    "out": "results/oral_a_short/sdil_channel_gated_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": "channel_gated",
+    "warmup_epochs": 0,
+    "weight_decay": 0.0001,
+    "weight_scale": 1.0,
+    "width": 16
+  },
+  "counters": {
+    "apical_warmup_examples": 12688,
+    "calibration_event_examples": 62688,
+    "causal_scalar_observations": 990,
+    "logical_batch_loss_queries": 990,
+    "ordinary_examples": 200000,
+    "per_example_loss_terms": 125376,
+    "perturbation_events": 495,
+    "perturbation_forward_examples": 125376,
+    "predictor_warmup_examples": 0
+  },
+  "diagnostics": {
+    "early_third_mean": 0.018784026537711423,
+    "innovation_negative_gradient_cosine": [
+      0.0064247846603393555,
+      0.04731021076440811,
+      0.016317252069711685,
+      0.0033069862984120846,
+      0.021622221916913986,
+      0.017722703516483307,
+      0.027213476598262787,
+      0.013780711218714714,
+      0.07152771204710007,
+      0.06467936187982559,
+      0.05392073467373848,
+      0.04071490839123726,
+      0.02924099937081337,
+      0.08977454900741577,
+      0.16485495865345,
+      0.19109880924224854,
+      0.19917020201683044,
+      0.2036983072757721,
+      0.12616632878780365
+    ],
+    "normalization_state": "training_batch_stats_without_running_update",
+    "raw_negative_gradient_cosine": [
+      0.0064247846603393555,
+      0.04731021076440811,
+      0.016317252069711685,
+      0.0033069862984120846,
+      0.021622221916913986,
+      0.017722703516483307,
+      0.027213476598262787,
+      0.013780711218714714,
+      0.07152771204710007,
+      0.06467936187982559,
+      0.05392073467373848,
+      0.04071490839123726,
+      0.02924099937081337,
+      0.08977454900741577,
+      0.16485495865345,
+      0.19109880924224854,
+      0.19917020201683044,
+      0.2036983072757721,
+      0.12616632878780365
+    ],
+    "teaching_negative_gradient_cosine": [
+      0.0064247846603393555,
+      0.04731021076440811,
+      0.016317252069711685,
+      0.0033069862984120846,
+      0.021622221916913986,
+      0.017722703516483307,
+      0.027213476598262787,
+      0.013780711218714714,
+      0.07152771204710007,
+      0.06467936187982559,
+      0.05392073467373848,
+      0.04071490839123726,
+      0.02924099937081337,
+      0.08977454900741577,
+      0.16485495865345,
+      0.19109880924224854,
+      0.19917020201683044,
+      0.2036983072757721,
+      0.12616632878780365
+    ],
+    "wall_s": 0.1185142993927002
+  },
+  "epochs": [
+    {
+      "calibration": {
+        "calibration_mse": 303.60023584691606,
+        "prediction_target_cosine": -7.053524919744465e-05,
+        "target_power": 303.60024019551014
+      },
+      "epoch": 1,
+      "lr": 0.01,
+      "output_lr": 0.1,
+      "step": 79,
+      "train_examples": 10000,
+      "train_loss": 3.0403768768310546
+    },
+    {
+      "calibration": {
+        "calibration_mse": 785.7472175463388,
+        "prediction_target_cosine": 2.4688466785111376e-05,
+        "target_power": 785.7471995187842
+      },
+      "epoch": 2,
+      "lr": 0.009938441702975689,
+      "output_lr": 0.0993844170297569,
+      "step": 158,
+      "train_examples": 10000,
+      "train_loss": 2.47521697807312
+    },
+    {
+      "calibration": {
+        "calibration_mse": 926.6549590877864,
+        "prediction_target_cosine": -8.481026661355656e-05,
+        "target_power": 926.6549154965772
+      },
+      "epoch": 3,
+      "lr": 0.009755282581475769,
+      "output_lr": 0.09755282581475769,
+      "step": 237,
+      "train_examples": 10000,
+      "train_loss": 2.327832473754883
+    },
+    {
+      "calibration": {
+        "calibration_mse": 524.3671950681531,
+        "prediction_target_cosine": -9.424448535305888e-06,
+        "target_power": 524.3671919719851
+      },
+      "epoch": 4,
+      "lr": 0.00945503262094184,
+      "output_lr": 0.09455032620941839,
+      "step": 316,
+      "train_examples": 10000,
+      "train_loss": 2.2477327514648437
+    },
+    {
+      "calibration": {
+        "calibration_mse": 906.0465460537567,
+        "prediction_target_cosine": 5.689712272068958e-05,
+        "target_power": 906.0465624182121
+      },
+      "epoch": 5,
+      "lr": 0.009045084971874737,
+      "output_lr": 0.09045084971874738,
+      "step": 395,
+      "train_examples": 10000,
+      "train_loss": 2.186990054321289
+    },
+    {
+      "calibration": {
+        "calibration_mse": 1382.679324004961,
+        "prediction_target_cosine": -1.069098398475288e-05,
+        "target_power": 1382.6792525332908
+      },
+      "epoch": 6,
+      "lr": 0.008535533905932738,
+      "output_lr": 0.08535533905932738,
+      "step": 474,
+      "train_examples": 10000,
+      "train_loss": 2.075714948654175
+    },
+    {
+      "calibration": {
+        "calibration_mse": 462.8764529534434,
+        "prediction_target_cosine": -1.6088110219793693e-05,
+        "target_power": 462.876455851375
+      },
+      "epoch": 7,
+      "lr": 0.007938926261462366,
+      "output_lr": 0.07938926261462366,
+      "step": 553,
+      "train_examples": 10000,
+      "train_loss": 2.0236370288848877
+    },
+    {
+      "calibration": {
+        "calibration_mse": 1054.5301591010266,
+        "prediction_target_cosine": 4.050277694471746e-05,
+        "target_power": 1054.5301153976257
+      },
+      "epoch": 8,
+      "lr": 0.007269952498697734,
+      "output_lr": 0.07269952498697733,
+      "step": 632,
+      "train_examples": 10000,
+      "train_loss": 2.0253488752365114
+    },
+    {
+      "calibration": {
+        "calibration_mse": 1057.2615307163528,
+        "prediction_target_cosine": 5.20554017652448e-05,
+        "target_power": 1057.2614973764055
+      },
+      "epoch": 9,
+      "lr": 0.006545084971874737,
+      "output_lr": 0.06545084971874737,
+      "step": 711,
+      "train_examples": 10000,
+      "train_loss": 1.9688177600860595
+    },
+    {
+      "calibration": {
+        "calibration_mse": 1072.6804250551306,
+        "prediction_target_cosine": -6.634325282023175e-05,
+        "target_power": 1072.6804443732553
+      },
+      "epoch": 10,
+      "lr": 0.0057821723252011546,
+      "output_lr": 0.05782172325201155,
+      "step": 790,
+      "train_examples": 10000,
+      "train_loss": 1.8554277042388916
+    },
+    {
+      "calibration": {
+        "calibration_mse": 978.8598406502088,
+        "prediction_target_cosine": 3.40096549432178e-05,
+        "target_power": 978.8598520575655
+      },
+      "epoch": 11,
+      "lr": 0.005,
+      "output_lr": 0.05,
+      "step": 869,
+      "train_examples": 10000,
+      "train_loss": 1.8643826206207275
+    },
+    {
+      "calibration": {
+        "calibration_mse": 600.056727797202,
+        "prediction_target_cosine": -1.9106017742333596e-05,
+        "target_power": 600.0567101223791
+      },
+      "epoch": 12,
+      "lr": 0.004217827674798845,
+      "output_lr": 0.04217827674798845,
+      "step": 948,
+      "train_examples": 10000,
+      "train_loss": 1.7651144916534425
+    },
+    {
+      "calibration": {
+        "calibration_mse": 732.2157620286812,
+        "prediction_target_cosine": 4.7248998125412554e-05,
+        "target_power": 732.2157623713431
+      },
+      "epoch": 13,
+      "lr": 0.003454915028125263,
+      "output_lr": 0.03454915028125263,
+      "step": 1027,
+      "train_examples": 10000,
+      "train_loss": 1.7445065725326538
+    },
+    {
+      "calibration": {
+        "calibration_mse": 976.8885286455568,
+        "prediction_target_cosine": 6.757763090613125e-05,
+        "target_power": 976.8885265516197
+      },
+      "epoch": 14,
+      "lr": 0.0027300475013022664,
+      "output_lr": 0.027300475013022664,
+      "step": 1106,
+      "train_examples": 10000,
+      "train_loss": 1.7122911861419678
+    },
+    {
+      "calibration": {
+        "calibration_mse": 668.0147250166406,
+        "prediction_target_cosine": 4.402830101716635e-05,
+        "target_power": 668.0147206327189
+      },
+      "epoch": 15,
+      "lr": 0.0020610737385376348,
+      "output_lr": 0.02061073738537635,
+      "step": 1185,
+      "train_examples": 10000,
+      "train_loss": 1.6635886901855468
+    },
+    {
+      "calibration": {
+        "calibration_mse": 980.3828986875064,
+        "prediction_target_cosine": 4.302361982883913e-05,
+        "target_power": 980.382911251547
+      },
+      "epoch": 16,
+      "lr": 0.0014644660940672626,
+      "output_lr": 0.014644660940672627,
+      "step": 1264,
+      "train_examples": 10000,
+      "train_loss": 1.6665239530563354
+    },
+    {
+      "calibration": {
+        "calibration_mse": 675.8823695696565,
+        "prediction_target_cosine": -6.293169594075907e-05,
+        "target_power": 675.882374613732
+      },
+      "epoch": 17,
+      "lr": 0.0009549150281252633,
+      "output_lr": 0.009549150281252633,
+      "step": 1343,
+      "train_examples": 10000,
+      "train_loss": 1.6579075477600098
+    },
+    {
+      "calibration": {
+        "calibration_mse": 1099.5418591521516,
+        "prediction_target_cosine": 3.818202317773011e-05,
+        "target_power": 1099.5419457962148
+      },
+      "epoch": 18,
+      "lr": 0.0005449673790581611,
+      "output_lr": 0.005449673790581611,
+      "step": 1422,
+      "train_examples": 10000,
+      "train_loss": 1.6407965618133544
+    },
+    {
+      "calibration": {
+        "calibration_mse": 371.8435573219928,
+        "prediction_target_cosine": -4.679937512933992e-05,
+        "target_power": 371.843570496725
+      },
+      "epoch": 19,
+      "lr": 0.00024471741852423234,
+      "output_lr": 0.0024471741852423235,
+      "step": 1501,
+      "train_examples": 10000,
+      "train_loss": 1.6256537380218505
+    },
+    {
+      "calibration": {
+        "calibration_mse": 1096.4412992398852,
+        "prediction_target_cosine": 2.999884061467783e-05,
+        "target_power": 1096.44131298087
+      },
+      "epoch": 20,
+      "lr": 6.15582970243117e-05,
+      "output_lr": 0.0006155829702431171,
+      "step": 1580,
+      "train_examples": 10000,
+      "train_loss": 1.6222436019897462
+    }
+  ],
+  "evaluation_protocol": {
+    "test_evaluations": 0,
+    "test_used_for_selection": false,
+    "validation_evaluations": 1
+  },
+  "final": {
+    "accuracy": 0.4054,
+    "epoch": 20,
+    "evaluation_split": "validation",
+    "finite": true,
+    "loss": 1.6269997802734375,
+    "step": 1580
+  },
+  "hardware": {
+    "cuda_device_name": "NVIDIA GeForce GTX 1080",
+    "cuda_visible_devices": "5",
+    "device": "cuda:0",
+    "device_total_memory_bytes": 8507949056,
+    "peak_memory_allocated_bytes": 1733492736,
+    "peak_memory_reserved_bytes": 2858418176,
+    "torch_version": "2.3.1+cu118"
+  },
+  "protocol_family": "oral_a_cifar_local_resnet_development",
+  "provenance": {
+    "git_commit": "a1d060e6d96d4ab973d16cd75149dd61899c7c95",
+    "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": 6.557543039321899,
+    "evaluation_wall_s": 0.36567163467407227,
+    "predictor_warmup_wall_s": 0.0,
+    "timing_excludes_data_loading_hashing_and_model_construction": true,
+    "total_timed_wall_s": 101.1947500705719,
+    "train_wall_s": 94.26846575737
+  },
+  "work": {
+    "apical_macs_per_example": 202176,
+    "causal_scalar_observations": 990,
+    "components": {
+      "apical_projection_macs": 43000409088,
+      "apical_regression_macs": 12674009088,
+      "bp_reverse_macs_estimate": 0,
+      "local_weight_correlation_macs": 8110208000000,
+      "ordinary_forward_macs": 8110208000000,
+      "perturbation_forward_macs": 5084127191040,
+      "warmup_clean_forward_macs": 514511595520
+    },
+    "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": 990,
+    "per_example_cross_entropy_terms": 125376,
+    "total_clean_forward_examples": 212688,
+    "total_forward_equivalent_examples": 338064,
+    "total_macs_estimate": 21874729204736
+  }
+}