Viewing profile — TheCog

I wrote a speculative little story about the recent going ons at Anthropic and the launch of Mythos.

I wrote a... slightly substantial writeup: https://hallofdreams.org/posts/agents-and-ants/

A retrospective on the SWARM challenge originally posted in this thread: https://news.ycombinator.com/item?id=47270807 . It's very long, and goes into topics from design challenge …

I thought I should solve Advent of Code in assembly. This proved to be a... poor choice, for a variety of reasons. A retrospective on registers and having to hand roll everything.

The main impetus was trying to show that you could do AlphaZero on a regular computer. It didn't pan out that way.

> It's also fascinating to me that the author attempted to train an ML model to play the game. My initial thought was that an exhaustive search was the only way to approach this (b…

From Dave: Short answer: SAT solvers are hard. Long answer: I actually discussed it with Tim once, long after part 2 of our blog post and after the whole thing settled down. Tim wa…

Passing this on from Dave, since he doesn't have a HN account: He recommends trying NNUE on a CPU: https://www.chessprogramming.org/NNUE Mostly because he hasn't seen anyone try it…

I think one of the general takeaways about ML is that barring a few experts, its really challenging to reason about your system. Like, yes, you might expect that a convolutional la…