| import os | |
| import random | |
| import signal | |
| import numpy as np | |
| import torch | |
| import torch.distributed as dist | |
| from tqdm import tqdm | |
| import random | |
| def generate_mask(bz, ch_num, patch_num, mask_ratio, device): | |
| mask = torch.zeros((bz, ch_num, patch_num), dtype=torch.long, device=device) | |
| mask = mask.bernoulli_(mask_ratio) | |
| return mask | |
| def to_tensor(array): | |
| return torch.from_numpy(array).float() | |
| if __name__ == '__main__': | |
| a = generate_mask(192, 32, 15, mask_ratio=0.5, device=None) | |
| print(a) |