请尝试以下操作：进口 Torch

A = torch.tensor([1, 2, 3, 4, 5, 6, 7, 8, 9, 10])
B = torch.tensor([2, 5, 6, 8, 12, 15, 16])

# Convert tensors to sets
setA = set(A.numpy())
setB = set(B.numpy())

# Find intersection and difference
intersection = setA & setB
difference = setB - setA

# Calculate the counts
exists = len(intersection)
not_exist = len(difference)

print(f"Exists: {exists}")
print(f"Do not exist: {not_exist}")

更新：

你可以坚持使用原生的PyTorch方法，比如广播。这种方法可能会占用更多内存，但对于大型Tensor更有效，特别是在使用GPU加速时。

import torch

A = torch.tensor([1, 2, 3, 4, 5, 6, 7, 8, 9, 10])
B = torch.tensor([2, 5, 6, 8, 12, 15, 16])

# Use broadcasting to compare each element of B with all of A
# The result will be a 2D tensor with shape (len(B), len(A))
comparison_matrix = B[:, None] == A

# Sum along the second axis will give a tensor where each element
# indicates how many times an element from B was found in A
matches_per_element = comparison_matrix.sum(dim=1)

# Calculate the counts
exists = (matches_per_element > 0).sum().item()
not_exist = len(B) - exists

print(f"Exists: {exists}")
print(f"Do not exist: {not_exist}")