Viewing profile — derefr

Only if the end-user is the one compiling the software, on the same very system they'll be running it on. Which is true of GPU shader kernels, due to how GPU drivers work; but isn'…

That's an good answer to the question when taken in strict terms of a model version = weights. But, insofar as: 1. "a model" as presented to the user, isn't just its weights, but a…

This and https://en.wikipedia.org/wiki/Paris_syndrome make me feel like Japanese culture has somehow induced a unique kind of coddling within their population. People there seem to…

I think the problem with considering it a "pure optimization" is that code that is written to use tail-calls, if not optimized, is almost always unbounded recursive code. And moder…

I think what GP is saying is that, upon Fraunhofer's licensing of their technologies to Thomson, Thomson demanded (perhaps contractually) that Fraunhofer defend their patents so th…

"A game" is a different abstraction layer from "a piece of software", though. A game has designed mechanics, a scenario (level design, etc.), art/music assets, writing, and so on. …

I would point out that there's a middle way between those stances: GLP-1 receptor agonism has anti-auto-immune effects. (I don't know yet if anyone has determined whether this is a…

> If you are willing to sell, the interested buyer has no way to learn that short of a cold email to a WHOIS contact that privacy redaction has probably removed. Has everybody forg…

> The model weights and inference code need to be contained within 25KB of user-space memory Wouldn’t it be era-appropriate to allow relying on banked memory? You’d still need to h…

If you didn't care about having a fully-redistributable image, but were okay with doing things more like modern old-console-game decompilation projects do (where compiling the proj…

You wouldn’t be able to guarantee execution of arbitrary ELF, but couldn’t you intentionally build non-PIC-compiled ELF executables where the section base addresses as defined in t…

I could imagine true WYSIWYG politics, but it would need a very different temporal organizational structure than current politics. Currwntly, politicians each have a series of meet…

Is that with or without the LLM being able to assemble + profile its hypotheses itself? Because “optimize this code to execute in fewer CPU cycles under the test harness” seems lik…

An agent doesn’t know what “job” it’s being asked to do until you tell it. Imagine you’re blind and deaf and have temporary retrograde amnesia. You “wake up” one moment with a memo…

I never said the VC's timeline is arbitrary! They're ultimately based in loan interest rates / bond yields / etc — as you say, the "cost of money." But the timelines that founders …

The religion of speed is the religion of VC investment backing, because VCs have set time horizons for delivering returns to their own investors. You can only get their interest if…

> The lady is using some sort of tool made out of hand-cut bamboo. I still haven’t figured out exactly how they work, but every time I’ve seen this process they are using what look…

Long-context models do work as advertised... just not when combined with multi-turn conversation sessions. It's my understanding that, for all current models having long-context ca…

I get the sense that it's mostly that institutional HF/liquidity traders can trust that the system isn't unfairly advantaging anyone. In a centralized system where trades are accep…

The "reusable for free" aspect is about building ecosystems. Abstractions that can be freely integrated with have low barriers to entry to existing complementary systems being upda…

That's a fun thought. A gradual division between "institutional models" trained to have capabilities that solve for the needs of corporations, vs. "personal models" trained to have…

I think GP meant “collaboratively” as in collaborating online through the game itself. The same way you might e.g. collaborate on a Google Doc. “Sit down together” might be impract…

True. The real trick, if you have a client-side agent framework to hand, is to prompt it once as "gently" as possible to "just solve the problem"; and then, after its response to t…

It would (and does), yes; but this takes a lot more output tokens than asking for a summary would. The summary approach is only helpful insofar as it can be cheaper than using the …

The "auto" mode is (AFAICT) a per-conversation-turn router. (Presumably via a preliminary pass through a very fast tiny model that spits out an number for how challenging it thinks…