Just yesterday I was reading through this five year old post on triton by its creator. Triton was their PHD thesis and they coined the name before the inference server was renamed to it… if this is what you are referring to?
Sometimes a cosmic ray might hit the sign bit of the register and flip it to a negative value. So it is useful to pass it through a rectifier to ensure it's never negative, even in this rare case.
Indeed, we should call all idempotent functions twice just in case the first incantation fails to succeed.
In all seriousness, this is not at all how resilience to cosmic interference works in practice, and the probability of any executed instruction or even any other bit being flipped is far greater than the one specific bit you are addressing.
Humans do not write conclusions? As someone who went to college, that is a natural way to end a long essay. True mark of higher education would be writing the conclusion at the top.
Exactly, it's a natural way to write a college essay. I've never not cringed reading an article/blog post that is structured that way, it comes across very contrived. I've also noticed that LLMs tend to prefer it, and humans tend to avoid it in general.
I get a 401 error with that link, but when I visit it from a search engine it's fine. I think they may have blocked referrals from HN, so if you see that message, try copying-and-pasting the URL into a new tab / window / incognito window. I read this a while ago, and as other say, it's worth it.
Given the design, I reckon they might well be on a tiny hand-crafted server, in which case it makes sense. I seem to remember something similar the last time this was posted.
