Their algorithm is based on splitting the embeddings into a bit-wise representation and then sampling each bit based on the secret key, preserving the same like

Their algorithm is based on splitting the embeddings into a bit-wise representation and then sampling each bit based on the secret key, preserving the same likelihood distribution as with random sampling. (Given that the key is random) They say this works wlog for more complex embeddings, by encoding each token as a bit string. Could someone explain this generalization to me? If we have 4 tokens, 00, 01, 10, 11 with probabilities 0.5 for 00 and 11 and probability 0 for 01 and 10. Going through bit by bit, how will the algorithm guarantee not to produce 01 or 10?