> deterministic lossy compression function This would only be true if it didn't mix or 'forget' any of the training data for each prompt, and only then for the

> deterministic lossy compression function This would only be true if it didn't mix or 'forget' any of the training data for each prompt, and only then for the exact same inputs of the training data. That is a lot of qualifiers that aren't practically applicable. Slight side note: a deterministic lossy compression function should not always be treated the same as jpeg/zlib, as hash functions are technically also in that category.