Very interesting and very good (though, let us say, much more recent that ancient): only, it paradoxically does suggest that the computational cost (when compar

Very interesting and very good (though, let us say, much more recent that ancient): only, it paradoxically does suggest that the computational cost (when compared either to a simple "search by vectors distance" or to the task failures that we can have witnessed by chief implementations*) strongly suggest the benefit of 1000x speed boosting, energy conservative Taalas board. *(Of course it has "always" worked well for «simple high contrast Q&A», ever since the base embeddings technology worked properly: that is almost by definition; it is on real world use cases, where the nuances of reality are present, that it failed miserably.)