Viewing profile — ssivark

Here's a thought: once could derive the spectrum of the Markov transition matrix, and assign an entropy to each of the eigenvectors. The dominant eigenvector (highest entropy) woul…

If you want something that's like every other harness then what's the point of considering pi? It's specifically for people who are dissatisfied with other harnesses. Also, you can…

It's already in that minimal but useful state so you can just start and ask it to write code. If you want a more tricked out "starter pack" there's oh-my-pi or lazypi and maybe a f…

Every once in a while I see some feature like app XYZ getting dark mode or spell check pop up on the HN front page. I don't understand why. These are absolutely basic features ever…

> people only look at the form and ignore the substance of the content. This is largely because most people are incapable of thinking critically and evaluating substance. So they r…

The core thesis of this essay is reminiscent of the Lisp Curse [1] / Bipolar Lisp Programmer [2]. It's been a few years since I read these, but if I recall the argument there, it w…

> All I can say is skill issue I know nothing about you or your code, but till you make your code + workflows public and have all of it reviewed / critiqued (so that other can repl…

Imagine I want to attend a conference in a different country. Google maps might give turn by turn navigation but that is an overwhelming and largely irrelevant mess of details for …

Ha, interesting. I wasn't aware of Sutton's blog post, but if I might make a shameless plug, we demonstrated [1] exactly this problem (see section 4.4.3), and how multi-step world …

It'll get paid from revenue , not by redirecting employee salaries. All that AI+compute is literally what customers pay Anthropic for. Big AI labs are not software companies where …

I doubt they're the first solution to use coordinate based editing, or even the best one right now. Eg: Check out hash-anchored editing. The first place where I recall seeing this …

The analysis and synthesis approaches to understanding systems have respectively been the driving forces for two major breakthroughs in 20th century physics: reductionist and emerg…

If density is the primary factor, why doesn't an American city like NYC have faster/cheaper internet than Switzerland?

> Without the price cut, Deepseek V4-pro tokens would have cost more than resold Opus 4.8 tokens. You mean it's functionally as if American tokens are being price dumped in China a…

Qualcomm seems to be assembling a whole portfolio of technologies/products aimed at 1. Moving beyond ARM to RISC-V 2. Being competitive for AI/could needs instai of just chips for …

I don't understand the justification for local hardware with cost as the motivation. The same (or bigger/better) open weights models can served by third parties at much higher reso…

You need a full ring of sensors to be receiving at all times. The Caltech design has a rotating element that emits once at each location on the ring. The Midjourney design instead …

When we measure the average experience, it's crucial what we are sampling/measuring uniformly to construct that experience. The service provider is choosing to weight all requests …

I did a tech deep-dive into the Midjourney tank, and this is basically the origin of that (the first author David Garrett worked at Midjourney for a while) so I have some thoughts …

Yes. IIRC David Garrett (the first author of the Nature paper) also worked at Midjourney in the intervening period.

When doing auto regressive inference, how often do you do a CUDA kernel call? What is the main bottleneck at the throughputs you're operating?

When aiming for 100k tok/s, you would still have CUDA overheads (on the order of microseconds) -- which might become the bottleneck, even if you do everything else right with the i…

My apologies... I was responding to the above comment / ranting about the general trend and got carried away. Wasn't directed at specifically at your post. I love your second graph…

I don't specifically care about Claude -vs- GPT, but comparing models at different amounts of test time compute is a gaping hole. It also means that any unreasonably-expensive toke…