First switch to `exp` folder, then:


For CIFAR100 + VGG16:

	python train.py --adjust d --data_path [your path]

	python train.py --adjust bna --data_path [your path]


For CIFAR10 + ResNet19:

	python -m torch.distributed.launch --nproc_per_node 4 --nnode 1 distrain_10.py --adjust d --data_path [your path] --device_ids=0,1,2,3

	python -m torch.distributed.launch --nproc_per_node 4 --nnode 1 distrain_10.py --adjust bna --data_path [your path] --device_ids=0,1,2,3


For CIFAR10-DVS + ResNet18:

	Please use the cumulative processing method for the original cifar10-dvs dataset, and divide it into train folder and test folder.

	python -m torch.distributed.launch --nproc_per_node 2 --nnode 1 for_dspike.py --adjust bna --data_path [your path] --device_ids=0,1


For CIFAR10-DVS + vgg11 + tet:

	Please use the cumulative processing method for the original cifar10-dvs dataset, and divide it into train folder and test folder.

	python -m torch.distributed.launch --nproc_per_node 2 --nnode 1 tet_vgg11.py --adjust bna --data_path [your path] --device_ids=0,1 --back_method=trainEn

For ImgeNet + Sew-ResNet34:
	
	For faster training, we use the same processing method as Sew-ResNet forImageNet, more details please refer to "https://github.com/fangwei123456/Spike-Element-Wise-ResNet".

	If you want to get started quickly, replace train_dir (line 227), val_dir (line 228) load_data's cache_path (line 230) in trainImg.py with "yourPath/ImageNet/train", "yourPath/ImageNet/val", "yourPath/ImageNet/tmp")

	python -m torch.distributed.launch --nproc_per_node 8 --nnode 1 trainImg.py --device_ids=0,1,2,3,4,5,6,7
