@moritzkremb Has anyone tried using Jev inside a speculative decoding pipeline —

@moritzkremb Has anyone tried using Jev inside a speculative decoding pipeline — e.g. generate many candidate continuations cheaply, use Jev to rank them in parallel, and only have

@moritzkremb Has anyone tried using Jev inside a speculative decoding pipeline — e.g. generate many candidate continuations cheaply, use Jev to rank them in parallel, and only have the large LLM verify the best ones? Curious if @typesafeai / @CompleteSkeptic have explored something like this