cells-developmental / queue /cifar_scans.log
bgradowhite's picture
backup: 126 files (93 MB)
58c21d4 verified
Raw History Blame Contribute Delete
1.67 kB
[2026-09-04 19:18:36] queue out/queue/cifar_scans.txt started, pid 30252
[2026-09-04 19:18:36] start [1]: .venv/bin/python scripts/build_cifar_parts.py --net resnet --device mps --threads 4 --batch-size 32 --bank 256
[19:18:39] bank (256, 3, 32, 32) from cifar100(cifar100)[train,n=256,seed=0,norm(0.5071,0.4867,0.4408|0.2675,0.2565,0.2761)]
[19:18:42] zoo_resnet18_map:seed387 at epoch 60 of (0, 1, 2, 5, 10, 20, 30, 45, 60)
[19:18:44] overview -> out/parts/cifar/zoo_resnet18_map-seed387__overview.json
[19:18:44] training curve -> out/parts/cifar/zoo_resnet18_map-seed387__training_curve.json (61 epochs)
[19:18:44] removing zoo_resnet18-seed387__overview.json: it describes zoo_resnet18:seed387, not zoo_resnet18_map:seed387
[19:18:44] removing zoo_resnet18-seed387__training_curve.json: it describes zoo_resnet18:seed387, not zoo_resnet18_map:seed387
[2026-09-04 19:18:44] exit [1]: status 0 after 8s
[2026-09-04 19:18:44] start [2]: .venv/bin/python scripts/build_cifar_scans.py --net resnet --device mps --threads 4 --batch-size 32 --bank 256 --families covariance_gaussian --base-families covariance_gaussian real_activation --frames origin --seeds 0 1 2 --resume
[19:18:46] bank (256, 3, 32, 32) from cifar100(cifar100)[train,n=256,seed=0,norm(0.5071,0.4867,0.4408|0.2675,0.2565,0.2761)] | explicit
[19:18:46] directions ['covariance_gaussian'] | bases ['covariance_gaussian', 'real_activation'] | ['origin']
[19:18:46] epochs [0, 1, 2, 5, 10, 20, 30, 45, 60] | seeds [0, 1, 2] | 40 scans per ensemble
[19:18:48] zoo_resnet18_map:seed387 passes the contract; cuts ('layer1.0', 'layer1.1', 'layer2.0', 'layer2.1', 'layer3.0', 'layer3.1', 'layer4.0', 'layer4.1')