summaryrefslogtreecommitdiff
path: root/results/resnet_crossover/r2/resnet-r2-pepita-d20-lr0p003.json
diff options
context:
space:
mode:
Diffstat (limited to 'results/resnet_crossover/r2/resnet-r2-pepita-d20-lr0p003.json')
-rw-r--r--results/resnet_crossover/r2/resnet-r2-pepita-d20-lr0p003.json1565
1 files changed, 1565 insertions, 0 deletions
diff --git a/results/resnet_crossover/r2/resnet-r2-pepita-d20-lr0p003.json b/results/resnet_crossover/r2/resnet-r2-pepita-d20-lr0p003.json
new file mode 100644
index 0000000..1bbd577
--- /dev/null
+++ b/results/resnet_crossover/r2/resnet-r2-pepita-d20-lr0p003.json
@@ -0,0 +1,1565 @@
+{
+ "architecture": {
+ "base_width": 16,
+ "depth": 20,
+ "family": "CIFAR 6n+2 ResNet, option-A shortcuts",
+ "forward_parameter_count": 269722,
+ "normalization": "batchnorm",
+ "residual_scale": 1.0
+ },
+ "args": {
+ "augment_train": 1,
+ "batch_size": 128,
+ "data_dir": "/orion/u/oscarwan/scratch/sdil-data",
+ "depth": 20,
+ "device": "cuda",
+ "dp_alpha": 0.0,
+ "dp_beta": 0.1,
+ "dp_inference_passes": 16,
+ "ep_beta": 0.5,
+ "ep_dt": 0.5,
+ "ep_free_steps": 20,
+ "ep_nudge_steps": 4,
+ "epochs": 100,
+ "eval_every": 1,
+ "eval_split": "validation",
+ "feedback_seed": 1729,
+ "ff_score_from_layer": 1,
+ "ff_threshold": 2.0,
+ "loader_seed": 0,
+ "lr": 0.003,
+ "lr_gamma": 0.1,
+ "lr_milestones": "60,90",
+ "lr_schedule": "pepita",
+ "max_steps": 0,
+ "method": "pepita",
+ "momentum": 0.9,
+ "out": "/orion/u/oscarwan/scratch/sdil-work/sdil-a6000-full-rerun/results/resnet_crossover/r2/resnet-r2-pepita-d20-lr0p003.json",
+ "output_lr": 0.003,
+ "pepita_projection_scale": 0.05,
+ "seed": 0,
+ "split_seed": 2027,
+ "train_limit": 0,
+ "val_examples": 5000,
+ "weight_decay": 0.0001,
+ "width": 16
+ },
+ "epochs": [
+ {
+ "epoch": 1,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.2540517320844864,
+ "train_seconds": 5.846177339553833,
+ "validation": {
+ "accuracy": 0.2154,
+ "examples": 5000,
+ "loss": 2.17969853515625
+ },
+ "validation_seconds": 0.12939953804016113
+ },
+ {
+ "epoch": 2,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.170380486128065,
+ "train_seconds": 5.499620676040649,
+ "validation": {
+ "accuracy": 0.2038,
+ "examples": 5000,
+ "loss": 2.134754833984375
+ },
+ "validation_seconds": 0.10496997833251953
+ },
+ {
+ "epoch": 3,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.1359980241563585,
+ "train_seconds": 5.497474431991577,
+ "validation": {
+ "accuracy": 0.225,
+ "examples": 5000,
+ "loss": 2.09927607421875
+ },
+ "validation_seconds": 0.10551118850708008
+ },
+ {
+ "epoch": 4,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.1131656904008653,
+ "train_seconds": 5.495712995529175,
+ "validation": {
+ "accuracy": 0.2192,
+ "examples": 5000,
+ "loss": 2.096121337890625
+ },
+ "validation_seconds": 0.10497379302978516
+ },
+ {
+ "epoch": 5,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.096841954845852,
+ "train_seconds": 5.500979900360107,
+ "validation": {
+ "accuracy": 0.2506,
+ "examples": 5000,
+ "loss": 2.0687662109375
+ },
+ "validation_seconds": 0.10506653785705566
+ },
+ {
+ "epoch": 6,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.08288049314287,
+ "train_seconds": 5.502379894256592,
+ "validation": {
+ "accuracy": 0.2552,
+ "examples": 5000,
+ "loss": 2.0530152099609373
+ },
+ "validation_seconds": 0.10521101951599121
+ },
+ {
+ "epoch": 7,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.073017496617635,
+ "train_seconds": 5.497807741165161,
+ "validation": {
+ "accuracy": 0.2366,
+ "examples": 5000,
+ "loss": 2.05641904296875
+ },
+ "validation_seconds": 0.10515880584716797
+ },
+ {
+ "epoch": 8,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0659207894643146,
+ "train_seconds": 5.5049827098846436,
+ "validation": {
+ "accuracy": 0.257,
+ "examples": 5000,
+ "loss": 2.03794013671875
+ },
+ "validation_seconds": 0.10519862174987793
+ },
+ {
+ "epoch": 9,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0593762793223065,
+ "train_seconds": 5.4969775676727295,
+ "validation": {
+ "accuracy": 0.259,
+ "examples": 5000,
+ "loss": 2.0241550048828123
+ },
+ "validation_seconds": 0.10501766204833984
+ },
+ {
+ "epoch": 10,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0526318064371747,
+ "train_seconds": 5.5005998611450195,
+ "validation": {
+ "accuracy": 0.2634,
+ "examples": 5000,
+ "loss": 2.0274732177734376
+ },
+ "validation_seconds": 0.10511279106140137
+ },
+ {
+ "epoch": 11,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.047295862748888,
+ "train_seconds": 5.514113903045654,
+ "validation": {
+ "accuracy": 0.2494,
+ "examples": 5000,
+ "loss": 2.034632568359375
+ },
+ "validation_seconds": 0.10521459579467773
+ },
+ {
+ "epoch": 12,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.046622137429979,
+ "train_seconds": 5.512402534484863,
+ "validation": {
+ "accuracy": 0.2686,
+ "examples": 5000,
+ "loss": 2.011696142578125
+ },
+ "validation_seconds": 0.10520362854003906
+ },
+ {
+ "epoch": 13,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0403066360897486,
+ "train_seconds": 5.509901523590088,
+ "validation": {
+ "accuracy": 0.258,
+ "examples": 5000,
+ "loss": 2.012770654296875
+ },
+ "validation_seconds": 0.10524630546569824
+ },
+ {
+ "epoch": 14,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0382033768547907,
+ "train_seconds": 5.508563995361328,
+ "validation": {
+ "accuracy": 0.2696,
+ "examples": 5000,
+ "loss": 2.0040715087890626
+ },
+ "validation_seconds": 0.10525345802307129
+ },
+ {
+ "epoch": 15,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0371743593004017,
+ "train_seconds": 5.505054712295532,
+ "validation": {
+ "accuracy": 0.2662,
+ "examples": 5000,
+ "loss": 2.0049404296875
+ },
+ "validation_seconds": 0.1051628589630127
+ },
+ {
+ "epoch": 16,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0296005568610296,
+ "train_seconds": 5.501487493515015,
+ "validation": {
+ "accuracy": 0.2676,
+ "examples": 5000,
+ "loss": 1.9989622802734375
+ },
+ "validation_seconds": 0.10553455352783203
+ },
+ {
+ "epoch": 17,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.028386363093058,
+ "train_seconds": 5.505746603012085,
+ "validation": {
+ "accuracy": 0.2682,
+ "examples": 5000,
+ "loss": 2.0050867919921873
+ },
+ "validation_seconds": 0.10517692565917969
+ },
+ {
+ "epoch": 18,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0280178286234536,
+ "train_seconds": 5.502475023269653,
+ "validation": {
+ "accuracy": 0.2634,
+ "examples": 5000,
+ "loss": 2.004136962890625
+ },
+ "validation_seconds": 0.10525631904602051
+ },
+ {
+ "epoch": 19,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.022250784280565,
+ "train_seconds": 5.5056681632995605,
+ "validation": {
+ "accuracy": 0.2744,
+ "examples": 5000,
+ "loss": 1.9959945556640626
+ },
+ "validation_seconds": 0.10529446601867676
+ },
+ {
+ "epoch": 20,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0207669247521296,
+ "train_seconds": 5.503343820571899,
+ "validation": {
+ "accuracy": 0.272,
+ "examples": 5000,
+ "loss": 1.990040185546875
+ },
+ "validation_seconds": 0.10525321960449219
+ },
+ {
+ "epoch": 21,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0204448586357966,
+ "train_seconds": 5.503366708755493,
+ "validation": {
+ "accuracy": 0.2566,
+ "examples": 5000,
+ "loss": 2.01042177734375
+ },
+ "validation_seconds": 0.10516881942749023
+ },
+ {
+ "epoch": 22,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0200392368740507,
+ "train_seconds": 5.5079567432403564,
+ "validation": {
+ "accuracy": 0.2672,
+ "examples": 5000,
+ "loss": 1.998391650390625
+ },
+ "validation_seconds": 0.10535097122192383
+ },
+ {
+ "epoch": 23,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0179887710571287,
+ "train_seconds": 5.508881568908691,
+ "validation": {
+ "accuracy": 0.2746,
+ "examples": 5000,
+ "loss": 1.9914981689453124
+ },
+ "validation_seconds": 0.1053166389465332
+ },
+ {
+ "epoch": 24,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.012541169950697,
+ "train_seconds": 5.511257648468018,
+ "validation": {
+ "accuracy": 0.2688,
+ "examples": 5000,
+ "loss": 1.9919697998046875
+ },
+ "validation_seconds": 0.10536813735961914
+ },
+ {
+ "epoch": 25,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0146305094189114,
+ "train_seconds": 5.506036281585693,
+ "validation": {
+ "accuracy": 0.2658,
+ "examples": 5000,
+ "loss": 1.9949030029296875
+ },
+ "validation_seconds": 0.10530686378479004
+ },
+ {
+ "epoch": 26,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.010540577167935,
+ "train_seconds": 5.503469944000244,
+ "validation": {
+ "accuracy": 0.2656,
+ "examples": 5000,
+ "loss": 1.995817578125
+ },
+ "validation_seconds": 0.10513448715209961
+ },
+ {
+ "epoch": 27,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.011019580438402,
+ "train_seconds": 5.504186153411865,
+ "validation": {
+ "accuracy": 0.2672,
+ "examples": 5000,
+ "loss": 1.9920932861328124
+ },
+ "validation_seconds": 0.10550999641418457
+ },
+ {
+ "epoch": 28,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0107565660264757,
+ "train_seconds": 5.501965761184692,
+ "validation": {
+ "accuracy": 0.2562,
+ "examples": 5000,
+ "loss": 2.00138916015625
+ },
+ "validation_seconds": 0.10556864738464355
+ },
+ {
+ "epoch": 29,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0093017225901284,
+ "train_seconds": 5.500823736190796,
+ "validation": {
+ "accuracy": 0.2646,
+ "examples": 5000,
+ "loss": 1.9926986083984375
+ },
+ "validation_seconds": 0.1052248477935791
+ },
+ {
+ "epoch": 30,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0063924302630953,
+ "train_seconds": 5.504383563995361,
+ "validation": {
+ "accuracy": 0.2624,
+ "examples": 5000,
+ "loss": 1.9840763671875
+ },
+ "validation_seconds": 0.10538268089294434
+ },
+ {
+ "epoch": 31,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.005855017513699,
+ "train_seconds": 5.50493311882019,
+ "validation": {
+ "accuracy": 0.266,
+ "examples": 5000,
+ "loss": 1.9905255859375
+ },
+ "validation_seconds": 0.1052553653717041
+ },
+ {
+ "epoch": 32,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0042439441257054,
+ "train_seconds": 5.5027971267700195,
+ "validation": {
+ "accuracy": 0.2678,
+ "examples": 5000,
+ "loss": 1.970855615234375
+ },
+ "validation_seconds": 0.10536980628967285
+ },
+ {
+ "epoch": 33,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0055182597054375,
+ "train_seconds": 5.503118991851807,
+ "validation": {
+ "accuracy": 0.2644,
+ "examples": 5000,
+ "loss": 1.999814013671875
+ },
+ "validation_seconds": 0.1053767204284668
+ },
+ {
+ "epoch": 34,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0039610786014133,
+ "train_seconds": 5.502718448638916,
+ "validation": {
+ "accuracy": 0.2558,
+ "examples": 5000,
+ "loss": 1.9934460693359375
+ },
+ "validation_seconds": 0.10529065132141113
+ },
+ {
+ "epoch": 35,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.002173393376668,
+ "train_seconds": 5.504491090774536,
+ "validation": {
+ "accuracy": 0.2718,
+ "examples": 5000,
+ "loss": 1.983355615234375
+ },
+ "validation_seconds": 0.10515451431274414
+ },
+ {
+ "epoch": 36,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 1.999710160975986,
+ "train_seconds": 5.503596782684326,
+ "validation": {
+ "accuracy": 0.2654,
+ "examples": 5000,
+ "loss": 1.9914493896484375
+ },
+ "validation_seconds": 0.10546326637268066
+ },
+ {
+ "epoch": 37,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.002970977168613,
+ "train_seconds": 5.50464129447937,
+ "validation": {
+ "accuracy": 0.2574,
+ "examples": 5000,
+ "loss": 1.9867867919921876
+ },
+ "validation_seconds": 0.10530948638916016
+ },
+ {
+ "epoch": 38,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0020395337846546,
+ "train_seconds": 5.5078301429748535,
+ "validation": {
+ "accuracy": 0.2706,
+ "examples": 5000,
+ "loss": 1.9826528076171874
+ },
+ "validation_seconds": 0.10545110702514648
+ },
+ {
+ "epoch": 39,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 1.9987340057584975,
+ "train_seconds": 5.504313707351685,
+ "validation": {
+ "accuracy": 0.2714,
+ "examples": 5000,
+ "loss": 1.9860088134765626
+ },
+ "validation_seconds": 0.10541486740112305
+ },
+ {
+ "epoch": 40,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 1.9994566903856066,
+ "train_seconds": 5.507040977478027,
+ "validation": {
+ "accuracy": 0.2682,
+ "examples": 5000,
+ "loss": 1.9782498291015624
+ },
+ "validation_seconds": 0.10532188415527344
+ },
+ {
+ "epoch": 41,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0013496548334757,
+ "train_seconds": 5.507043361663818,
+ "validation": {
+ "accuracy": 0.2496,
+ "examples": 5000,
+ "loss": 1.9953440185546876
+ },
+ "validation_seconds": 0.10550904273986816
+ },
+ {
+ "epoch": 42,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 1.9970213481055366,
+ "train_seconds": 5.501050710678101,
+ "validation": {
+ "accuracy": 0.2582,
+ "examples": 5000,
+ "loss": 1.980520703125
+ },
+ "validation_seconds": 0.10538554191589355
+ },
+ {
+ "epoch": 43,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0023596636454264,
+ "train_seconds": 5.504518508911133,
+ "validation": {
+ "accuracy": 0.276,
+ "examples": 5000,
+ "loss": 1.9724589599609375
+ },
+ "validation_seconds": 0.1053304672241211
+ },
+ {
+ "epoch": 44,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 1.9987859845691256,
+ "train_seconds": 5.504904270172119,
+ "validation": {
+ "accuracy": 0.267,
+ "examples": 5000,
+ "loss": 1.991623046875
+ },
+ "validation_seconds": 0.10524177551269531
+ },
+ {
+ "epoch": 45,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0013457901000975,
+ "train_seconds": 5.505333662033081,
+ "validation": {
+ "accuracy": 0.2622,
+ "examples": 5000,
+ "loss": 1.987554638671875
+ },
+ "validation_seconds": 0.10548973083496094
+ },
+ {
+ "epoch": 46,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 1.9991893960740832,
+ "train_seconds": 5.507464647293091,
+ "validation": {
+ "accuracy": 0.2706,
+ "examples": 5000,
+ "loss": 1.9903315673828126
+ },
+ "validation_seconds": 0.10526585578918457
+ },
+ {
+ "epoch": 47,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0000709929996066,
+ "train_seconds": 5.5044050216674805,
+ "validation": {
+ "accuracy": 0.2728,
+ "examples": 5000,
+ "loss": 1.9653149658203124
+ },
+ "validation_seconds": 0.10531759262084961
+ },
+ {
+ "epoch": 48,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.0002532496558296,
+ "train_seconds": 5.506645917892456,
+ "validation": {
+ "accuracy": 0.2778,
+ "examples": 5000,
+ "loss": 1.9617325439453126
+ },
+ "validation_seconds": 0.10522890090942383
+ },
+ {
+ "epoch": 49,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 2.000072623655531,
+ "train_seconds": 5.511117935180664,
+ "validation": {
+ "accuracy": 0.2652,
+ "examples": 5000,
+ "loss": 1.982528076171875
+ },
+ "validation_seconds": 0.10529708862304688
+ },
+ {
+ "epoch": 50,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 1.9980969663831922,
+ "train_seconds": 5.5062291622161865,
+ "validation": {
+ "accuracy": 0.2728,
+ "examples": 5000,
+ "loss": 1.97162822265625
+ },
+ "validation_seconds": 0.1052713394165039
+ },
+ {
+ "epoch": 51,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 1.994790125126309,
+ "train_seconds": 5.508324146270752,
+ "validation": {
+ "accuracy": 0.2596,
+ "examples": 5000,
+ "loss": 1.9888105224609376
+ },
+ "validation_seconds": 0.10519123077392578
+ },
+ {
+ "epoch": 52,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 1.9956360507965087,
+ "train_seconds": 5.5053040981292725,
+ "validation": {
+ "accuracy": 0.2598,
+ "examples": 5000,
+ "loss": 1.987352001953125
+ },
+ "validation_seconds": 0.10519528388977051
+ },
+ {
+ "epoch": 53,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 1.999503876071506,
+ "train_seconds": 5.503100395202637,
+ "validation": {
+ "accuracy": 0.2622,
+ "examples": 5000,
+ "loss": 1.9869154541015626
+ },
+ "validation_seconds": 0.10521578788757324
+ },
+ {
+ "epoch": 54,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 1.9960996131049262,
+ "train_seconds": 5.509793996810913,
+ "validation": {
+ "accuracy": 0.2582,
+ "examples": 5000,
+ "loss": 1.99108720703125
+ },
+ "validation_seconds": 0.10523152351379395
+ },
+ {
+ "epoch": 55,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 1.9974913773854575,
+ "train_seconds": 5.506419658660889,
+ "validation": {
+ "accuracy": 0.2644,
+ "examples": 5000,
+ "loss": 1.9845252685546875
+ },
+ "validation_seconds": 0.10523486137390137
+ },
+ {
+ "epoch": 56,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 1.9962633374955918,
+ "train_seconds": 5.5040602684021,
+ "validation": {
+ "accuracy": 0.263,
+ "examples": 5000,
+ "loss": 1.9842421875
+ },
+ "validation_seconds": 0.10512018203735352
+ },
+ {
+ "epoch": 57,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 1.9980382384406197,
+ "train_seconds": 5.508415460586548,
+ "validation": {
+ "accuracy": 0.2684,
+ "examples": 5000,
+ "loss": 1.9884821533203125
+ },
+ "validation_seconds": 0.10535335540771484
+ },
+ {
+ "epoch": 58,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 1.9994783101823594,
+ "train_seconds": 5.509284257888794,
+ "validation": {
+ "accuracy": 0.2598,
+ "examples": 5000,
+ "loss": 1.9920580078125
+ },
+ "validation_seconds": 0.10525155067443848
+ },
+ {
+ "epoch": 59,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 1.994873679860433,
+ "train_seconds": 5.507220506668091,
+ "validation": {
+ "accuracy": 0.2646,
+ "examples": 5000,
+ "loss": 1.984413818359375
+ },
+ "validation_seconds": 0.10529899597167969
+ },
+ {
+ "epoch": 60,
+ "lr": 0.003,
+ "output_lr": 0.003,
+ "train_examples": 45000,
+ "train_loss": 1.9963021400875516,
+ "train_seconds": 5.505999326705933,
+ "validation": {
+ "accuracy": 0.2654,
+ "examples": 5000,
+ "loss": 1.97801142578125
+ },
+ "validation_seconds": 0.10530734062194824
+ },
+ {
+ "epoch": 61,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9900015704684788,
+ "train_seconds": 5.507628679275513,
+ "validation": {
+ "accuracy": 0.2764,
+ "examples": 5000,
+ "loss": 1.9735885986328126
+ },
+ "validation_seconds": 0.10532093048095703
+ },
+ {
+ "epoch": 62,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9889358832465278,
+ "train_seconds": 5.505821466445923,
+ "validation": {
+ "accuracy": 0.2746,
+ "examples": 5000,
+ "loss": 1.9731400634765626
+ },
+ "validation_seconds": 0.10542631149291992
+ },
+ {
+ "epoch": 63,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9910470107184517,
+ "train_seconds": 5.504168510437012,
+ "validation": {
+ "accuracy": 0.2706,
+ "examples": 5000,
+ "loss": 1.9786871826171875
+ },
+ "validation_seconds": 0.10542702674865723
+ },
+ {
+ "epoch": 64,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9888048695670233,
+ "train_seconds": 5.503101110458374,
+ "validation": {
+ "accuracy": 0.2722,
+ "examples": 5000,
+ "loss": 1.9800283447265625
+ },
+ "validation_seconds": 0.10519576072692871
+ },
+ {
+ "epoch": 65,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9899456718656752,
+ "train_seconds": 5.503441572189331,
+ "validation": {
+ "accuracy": 0.2718,
+ "examples": 5000,
+ "loss": 1.9744058837890626
+ },
+ "validation_seconds": 0.10547184944152832
+ },
+ {
+ "epoch": 66,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9914917253070408,
+ "train_seconds": 5.5037009716033936,
+ "validation": {
+ "accuracy": 0.2768,
+ "examples": 5000,
+ "loss": 1.970571484375
+ },
+ "validation_seconds": 0.10534429550170898
+ },
+ {
+ "epoch": 67,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9919023951212564,
+ "train_seconds": 5.503073453903198,
+ "validation": {
+ "accuracy": 0.2702,
+ "examples": 5000,
+ "loss": 1.9756346435546874
+ },
+ "validation_seconds": 0.10526371002197266
+ },
+ {
+ "epoch": 68,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9882253373675876,
+ "train_seconds": 5.506524085998535,
+ "validation": {
+ "accuracy": 0.2782,
+ "examples": 5000,
+ "loss": 1.9718321044921876
+ },
+ "validation_seconds": 0.10534501075744629
+ },
+ {
+ "epoch": 69,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9925236448076036,
+ "train_seconds": 5.504960536956787,
+ "validation": {
+ "accuracy": 0.2716,
+ "examples": 5000,
+ "loss": 1.9783636962890625
+ },
+ "validation_seconds": 0.10545825958251953
+ },
+ {
+ "epoch": 70,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9897801370620727,
+ "train_seconds": 5.505310773849487,
+ "validation": {
+ "accuracy": 0.269,
+ "examples": 5000,
+ "loss": 1.9767344482421876
+ },
+ "validation_seconds": 0.10528755187988281
+ },
+ {
+ "epoch": 71,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9880535870022245,
+ "train_seconds": 5.503842830657959,
+ "validation": {
+ "accuracy": 0.2676,
+ "examples": 5000,
+ "loss": 1.9813978271484376
+ },
+ "validation_seconds": 0.10532116889953613
+ },
+ {
+ "epoch": 72,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9871454840554132,
+ "train_seconds": 5.511318206787109,
+ "validation": {
+ "accuracy": 0.2758,
+ "examples": 5000,
+ "loss": 1.97433564453125
+ },
+ "validation_seconds": 0.10527610778808594
+ },
+ {
+ "epoch": 73,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9925464500427246,
+ "train_seconds": 5.5087480545043945,
+ "validation": {
+ "accuracy": 0.2724,
+ "examples": 5000,
+ "loss": 1.978366552734375
+ },
+ "validation_seconds": 0.10521245002746582
+ },
+ {
+ "epoch": 74,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9864715037875704,
+ "train_seconds": 5.508193492889404,
+ "validation": {
+ "accuracy": 0.2686,
+ "examples": 5000,
+ "loss": 1.9754033935546875
+ },
+ "validation_seconds": 0.10522222518920898
+ },
+ {
+ "epoch": 75,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9897771712409125,
+ "train_seconds": 5.50649881362915,
+ "validation": {
+ "accuracy": 0.2734,
+ "examples": 5000,
+ "loss": 1.974899462890625
+ },
+ "validation_seconds": 0.10533857345581055
+ },
+ {
+ "epoch": 76,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9886419744915433,
+ "train_seconds": 5.505634069442749,
+ "validation": {
+ "accuracy": 0.2692,
+ "examples": 5000,
+ "loss": 1.977445703125
+ },
+ "validation_seconds": 0.10537195205688477
+ },
+ {
+ "epoch": 77,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9885343943066067,
+ "train_seconds": 5.503732204437256,
+ "validation": {
+ "accuracy": 0.2682,
+ "examples": 5000,
+ "loss": 1.9802877685546876
+ },
+ "validation_seconds": 0.10513567924499512
+ },
+ {
+ "epoch": 78,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.993202943250868,
+ "train_seconds": 5.506843566894531,
+ "validation": {
+ "accuracy": 0.2686,
+ "examples": 5000,
+ "loss": 1.9763575439453125
+ },
+ "validation_seconds": 0.10518097877502441
+ },
+ {
+ "epoch": 79,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9885233533859252,
+ "train_seconds": 5.502397775650024,
+ "validation": {
+ "accuracy": 0.2736,
+ "examples": 5000,
+ "loss": 1.971972607421875
+ },
+ "validation_seconds": 0.10559296607971191
+ },
+ {
+ "epoch": 80,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9916276458104452,
+ "train_seconds": 5.503775358200073,
+ "validation": {
+ "accuracy": 0.2702,
+ "examples": 5000,
+ "loss": 1.983391943359375
+ },
+ "validation_seconds": 0.10515046119689941
+ },
+ {
+ "epoch": 81,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9912612048678928,
+ "train_seconds": 5.502410888671875,
+ "validation": {
+ "accuracy": 0.2722,
+ "examples": 5000,
+ "loss": 1.9768909912109376
+ },
+ "validation_seconds": 0.10518693923950195
+ },
+ {
+ "epoch": 82,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9933317521413167,
+ "train_seconds": 5.505985498428345,
+ "validation": {
+ "accuracy": 0.2738,
+ "examples": 5000,
+ "loss": 1.975612646484375
+ },
+ "validation_seconds": 0.10521507263183594
+ },
+ {
+ "epoch": 83,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.992124314159817,
+ "train_seconds": 5.503434896469116,
+ "validation": {
+ "accuracy": 0.2694,
+ "examples": 5000,
+ "loss": 1.9838414794921875
+ },
+ "validation_seconds": 0.10535001754760742
+ },
+ {
+ "epoch": 84,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9881560234917535,
+ "train_seconds": 5.5040600299835205,
+ "validation": {
+ "accuracy": 0.2718,
+ "examples": 5000,
+ "loss": 1.981240087890625
+ },
+ "validation_seconds": 0.10520529747009277
+ },
+ {
+ "epoch": 85,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9894730068630642,
+ "train_seconds": 5.501953840255737,
+ "validation": {
+ "accuracy": 0.2672,
+ "examples": 5000,
+ "loss": 1.9822439208984375
+ },
+ "validation_seconds": 0.10531735420227051
+ },
+ {
+ "epoch": 86,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9889087203979492,
+ "train_seconds": 5.50821590423584,
+ "validation": {
+ "accuracy": 0.2714,
+ "examples": 5000,
+ "loss": 1.9796527099609376
+ },
+ "validation_seconds": 0.10519123077392578
+ },
+ {
+ "epoch": 87,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9899914241366916,
+ "train_seconds": 5.50803017616272,
+ "validation": {
+ "accuracy": 0.27,
+ "examples": 5000,
+ "loss": 1.981403369140625
+ },
+ "validation_seconds": 0.1051938533782959
+ },
+ {
+ "epoch": 88,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.989610216055976,
+ "train_seconds": 5.505961179733276,
+ "validation": {
+ "accuracy": 0.2696,
+ "examples": 5000,
+ "loss": 1.976666259765625
+ },
+ "validation_seconds": 0.10532689094543457
+ },
+ {
+ "epoch": 89,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.9914015740712483,
+ "train_seconds": 5.505902290344238,
+ "validation": {
+ "accuracy": 0.2742,
+ "examples": 5000,
+ "loss": 1.968590283203125
+ },
+ "validation_seconds": 0.10518217086791992
+ },
+ {
+ "epoch": 90,
+ "lr": 0.00030000000000000003,
+ "output_lr": 0.00030000000000000003,
+ "train_examples": 45000,
+ "train_loss": 1.991297045856052,
+ "train_seconds": 5.503058910369873,
+ "validation": {
+ "accuracy": 0.2686,
+ "examples": 5000,
+ "loss": 1.97153271484375
+ },
+ "validation_seconds": 0.10520219802856445
+ },
+ {
+ "epoch": 91,
+ "lr": 3.0000000000000004e-05,
+ "output_lr": 3.0000000000000004e-05,
+ "train_examples": 45000,
+ "train_loss": 1.9905024444156223,
+ "train_seconds": 5.507503986358643,
+ "validation": {
+ "accuracy": 0.2682,
+ "examples": 5000,
+ "loss": 1.9812239013671875
+ },
+ "validation_seconds": 0.10525727272033691
+ },
+ {
+ "epoch": 92,
+ "lr": 3.0000000000000004e-05,
+ "output_lr": 3.0000000000000004e-05,
+ "train_examples": 45000,
+ "train_loss": 1.9875384762658013,
+ "train_seconds": 5.504276514053345,
+ "validation": {
+ "accuracy": 0.271,
+ "examples": 5000,
+ "loss": 1.9784233642578124
+ },
+ "validation_seconds": 0.10520076751708984
+ },
+ {
+ "epoch": 93,
+ "lr": 3.0000000000000004e-05,
+ "output_lr": 3.0000000000000004e-05,
+ "train_examples": 45000,
+ "train_loss": 1.9886907052993774,
+ "train_seconds": 5.507157325744629,
+ "validation": {
+ "accuracy": 0.2704,
+ "examples": 5000,
+ "loss": 1.9788807373046875
+ },
+ "validation_seconds": 0.10531425476074219
+ },
+ {
+ "epoch": 94,
+ "lr": 3.0000000000000004e-05,
+ "output_lr": 3.0000000000000004e-05,
+ "train_examples": 45000,
+ "train_loss": 1.9844931491851807,
+ "train_seconds": 5.5087127685546875,
+ "validation": {
+ "accuracy": 0.2722,
+ "examples": 5000,
+ "loss": 1.9788918701171876
+ },
+ "validation_seconds": 0.10535192489624023
+ },
+ {
+ "epoch": 95,
+ "lr": 3.0000000000000004e-05,
+ "output_lr": 3.0000000000000004e-05,
+ "train_examples": 45000,
+ "train_loss": 1.9873190053304037,
+ "train_seconds": 5.506003379821777,
+ "validation": {
+ "accuracy": 0.2748,
+ "examples": 5000,
+ "loss": 1.9743620849609376
+ },
+ "validation_seconds": 0.10539937019348145
+ },
+ {
+ "epoch": 96,
+ "lr": 3.0000000000000004e-05,
+ "output_lr": 3.0000000000000004e-05,
+ "train_examples": 45000,
+ "train_loss": 1.984961536640591,
+ "train_seconds": 5.514376163482666,
+ "validation": {
+ "accuracy": 0.2724,
+ "examples": 5000,
+ "loss": 1.9778021240234376
+ },
+ "validation_seconds": 0.10521578788757324
+ },
+ {
+ "epoch": 97,
+ "lr": 3.0000000000000004e-05,
+ "output_lr": 3.0000000000000004e-05,
+ "train_examples": 45000,
+ "train_loss": 1.9883242013295492,
+ "train_seconds": 5.5068089962005615,
+ "validation": {
+ "accuracy": 0.2696,
+ "examples": 5000,
+ "loss": 1.9805218017578126
+ },
+ "validation_seconds": 0.10537600517272949
+ },
+ {
+ "epoch": 98,
+ "lr": 3.0000000000000004e-05,
+ "output_lr": 3.0000000000000004e-05,
+ "train_examples": 45000,
+ "train_loss": 1.9905134502834745,
+ "train_seconds": 5.501478672027588,
+ "validation": {
+ "accuracy": 0.272,
+ "examples": 5000,
+ "loss": 1.977109130859375
+ },
+ "validation_seconds": 0.10539984703063965
+ },
+ {
+ "epoch": 99,
+ "lr": 3.0000000000000004e-05,
+ "output_lr": 3.0000000000000004e-05,
+ "train_examples": 45000,
+ "train_loss": 1.9883133581373427,
+ "train_seconds": 5.503832578659058,
+ "validation": {
+ "accuracy": 0.2712,
+ "examples": 5000,
+ "loss": 1.97808974609375
+ },
+ "validation_seconds": 0.10522580146789551
+ },
+ {
+ "epoch": 100,
+ "lr": 3.0000000000000004e-05,
+ "output_lr": 3.0000000000000004e-05,
+ "train_examples": 45000,
+ "train_loss": 1.9903361009385852,
+ "train_seconds": 5.5033111572265625,
+ "validation": {
+ "accuracy": 0.2726,
+ "examples": 5000,
+ "loss": 1.9752171875
+ },
+ "validation_seconds": 0.10524916648864746
+ }
+ ],
+ "evaluation_protocol": {
+ "split": "validation",
+ "test_evaluations": 0,
+ "test_used_for_selection": false
+ },
+ "final": {
+ "accuracy": 0.2726,
+ "evaluation_seconds": 0.10563945770263672,
+ "examples": 5000,
+ "finite": true,
+ "loss": 1.9752171875
+ },
+ "first_nonfinite_step": null,
+ "hardware": {
+ "cuda_device_name": "NVIDIA RTX A6000",
+ "cuda_visible_devices": "0",
+ "device": "cuda",
+ "device_total_memory_bytes": 50896568320,
+ "peak_memory_allocated_bytes": 2008731648,
+ "peak_memory_reserved_bytes": 2543845376,
+ "torch_version": "2.3.1+cu118"
+ },
+ "layers": [],
+ "protocol_family": "resnet_local_learning_crossover",
+ "provenance": {
+ "git_commit": "f55872beaee658ff76ac15fd421b73017820c320",
+ "git_tracked_dirty": false
+ },
+ "schema_version": 1,
+ "split": {
+ "cifar_source_files": [
+ {
+ "bytes": 31035704,
+ "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/data_batch_1",
+ "sha256": "54636561a3ce25bd3e19253c6b0d8538147b0ae398331ac4a2d86c6d987368cd"
+ },
+ {
+ "bytes": 31035320,
+ "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/data_batch_2",
+ "sha256": "766b2cef9fbc745cf056b3152224f7cf77163b330ea9a15f9392beb8b89bc5a8"
+ },
+ {
+ "bytes": 31035999,
+ "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/data_batch_3",
+ "sha256": "0f00d98ebfb30b3ec0ad19f9756dc2630b89003e10525f5e148445e82aa6a1f9"
+ },
+ {
+ "bytes": 31035696,
+ "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/data_batch_4",
+ "sha256": "3f7bb240661948b8f4d53e36ec720d8306f5668bd0071dcb4e6c947f78e9682b"
+ },
+ {
+ "bytes": 31035623,
+ "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/data_batch_5",
+ "sha256": "d91802434d8376bbaeeadf58a737e3a1b12ac839077e931237e0dcd43adcb154"
+ },
+ {
+ "bytes": 31035526,
+ "path": "/orion/u/oscarwan/scratch/sdil-data/cifar-10-batches-py/test_batch",
+ "sha256": "f53d8d457504f7cff4ea9e021afcf0e0ad8e24a91f3fc42091b8adef61157831"
+ }
+ ],
+ "dataset": "cifar10",
+ "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": 45000,
+ "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"
+ },
+ "total_wall_seconds": 561.5238916873932,
+ "work": {
+ "candidate_label_evaluation_presentations": 0,
+ "completed_global_epochs": 100,
+ "feedforward_example_passes": 14005000,
+ "forward_macs_per_example": 40551040,
+ "forward_parameter_count": 269722,
+ "local_target_backward_example_evaluations": 0,
+ "local_vjp_example_evaluations": 0,
+ "logical_task_loss_queries": 0,
+ "num_trainable_layers": 20,
+ "ordinary_training_examples": 4500000,
+ "ordinary_validation_examples": 505000,
+ "relaxation_example_passes": 0,
+ "training_example_presentations": 9000000
+ }
+}