Yes it's between this and Gemma 4 31B which is much slower, but looks like it won't ever get an upgrade. I have to conclude that the MoE variants are unreliable, and MTP sometimes just can't get tricky formatting right.
Alan Turing already proved with the Halting Problem that reasoning about program correctness is not possible. But we still try.
Wikipedia: [1] Turing proved no algorithm exists that always correctly decides whether, for a given arbitrary program and input, the program halts when run with that input. The essence of Turing's proof is that any such algorithm can be made to produce contradictory output and therefore cannot be correct.
I'm very confused because there are 2 Andrews, the author in the blog post only states "Andrew", and by the list of Authors the author seems to be Andrew Gelman, but the slug in the first link is "aking", and then there is also Andrew King, lol.
I’m an early fan (Polymer, anyone?) but somehow the mindshare is just not there and trying to evangelize it to mainstream was too much. So now it just kinda there for people to slowly discover when they run into niche use cases.