Thanks for the feedback Ryan. Indeed, it would solve part of the problem I guess the larger point was if we can encode episodic, procedural, etc. types of memory as vectors?