Attention Padding Mask

Build the padding mask for attention: given a row of token ids, mark every position that holds a real token (True) and every position equal to pad_id (False).

Implement solve(ids, pad_id).

Returns. Return a boolean NumPy array of the same length as ids.

Examples

Example 1

Input:

solve([5, 6, 0, 0], 0)

Output:

[True, True, False, False]

Example 2

Input:

solve([[3, 0], [4, 5]], 0)

Output:

[[True, False], [True, True]]