
Requirments:
tensorflow-addons==0.8.3
libcudnn8=8.5.0.96-1
cuda11.7

To run the code on one/several GPUs:

!python run.py --color_jitter_strength=0.5 \
  --train_batch_size=320 --train_epochs=1 \
  --learning_rate=1.5 --weight_decay=1e-4 --temperature=0.3 \
  --dataset=cifar10 --image_size=32 --eval_split=test \
  --use_blur=False --resnet_depth=18 \
  --model_dir=/tmp/simclr_test --tekrar=3 \
  --NumofWorkers=10  --epochsm=1 --serBatch=32\
  --AutoBatch=64 --testBatch=320 --norm=GN --CPC=10


--norm: to choose the normalization techniquees: 'PN', 'GN', 'BN'
-CPC: number of class per client 1 for 1 class, 2 for 2 classes, 10 for 10 classes