Reithan

Sorry if this comment/ask is out of line or place, but I've been loving Lamarck and evangelizing it all over. One thing I'd love to see, given the fact that Lamarck has R1 as an element, is a bit more consistency with it's output of <think> (specifically, without a <think> prefill, it's not super consistent in adding one, and sometimes with prefill it forgets to close it with </think>.)

It's use of it's think block, and the way it incorporates its thought into the post-think output though, as honestly better than I've seen even with base R1. It makes logical and mathematical jumps from one step to another that I haven't seen R1 do, and it's better at error-checking itself without vomiting out 5 paragraphs of "well let me double check". Not to mention it's superior prose means the output actually explains what it thought up far better and more clearly.