Amazing. Thanks for your detailed posts on the bake-off between the Mac and GB10, Daniel, and on your learnings. I had trying similar on both compute platforms on my to-do list. Your post should save me a lot of debugs, sweat, and tears.
Whether by design or accident, the way Anthropic is managing the Claude Mythos release is creating a Glomar Trap for customers, competitors, and investors.
Do you want to define the seasons by temperature? Or by lengthening of the day? Because, to me, seasons are tied to weather in general, but temperature specifically. And temperature seems to correlates to length of day, but trails it by about a month. Which makes sense, since it takes time to heat/cool the enormous thermal mass. So, if weather is how you track the changing of the seasons, it’s close enough to correct as-is.
CLaRa (Continuous Latent Reasoning) is an approach Retrieval-Augmented Generation (RAG) that shortens context, reduces double-encoding, and improves quality of responses by compressing documents into a small set of "continuous memory tokens" that preserve the key information in the documents, and optimizes and performs retrieval and generation out of that shared latent space.
Because, the programming languages best matched to a (natural human language-based) declarative programming paradigm (e.g., vibe coding) would be declarative programming languages, not imperative programming languages.
These instructions are to achieve a much more modest goal than that!