Strange behavior of python memoization

Question

When playing arround with python memoization, I wrote the following code.

def can_sum_rec(target, nums):
    if target < 0:
        return False
    if target == 0:
        return True
    for i in nums:
        if can_sum_rec(target - i, nums):
            return True
    return False

def can_sum_mem(target, nums, mem={}):
    if target in mem:
        return mem[target]
    if target < 0:
        return False
    if target == 0:
        return True
    for i in nums:
        if can_sum_mem(target - i, nums, mem):
            mem[target] = True
            return True
    mem[target] = False
    return False

However, I am getting some confusing test results.

tests = [(7, [3, 2]), (7, [2, 4]), (7, [5, 4, 7, 3]), (8, [2, 3, 5])]
for target, nums in tests:
    print(can_sum_mem(target, nums))

True True True True

Obviously, 2 and 4 cannot sum to 7. However, when testing the cases separately the results are fine.

tests = [(7, [2, 4])]
for target, nums in tests:
    print(can_sum_mem(target, nums))

False

I can't seem to find any global variable. So I would really appreciate it if someone could point out the issue here.