Here's an example script for pretraining a resnet 18 model on cifar10 dataset (run in folder /spectral_contrastive), with hyper parameter \mu=1 (see paper for more discussions):
python pretrain_no_redundent_lr_update.py -c configs/sparsecut_v23new_resnet_mlp1000_norelu_cifar10_lr003_r1.yaml --hide_progress

Here's an example script for linear evaluation on labeled data (run in folder /spectral_contrastive):
python eval/eval_run.py --dataset cifar10 --dir PATH_TO_LOG_DIR  --arch resnet18_cifar_variant1 --batch_size 256 --epochs 100 --schedule 60 80 --specific_ckpts 800.pth --opt sgd --lr 30.0 --nomlp
