Comment by Der_Einzige

Peer reviews NeurIPS caliber paper to prove that? Because I can show you one titled "Long context generation is a sampling problem"...