*former

Good catch. OP should probably update the article

No, the latter. To make a lattice, we usually want to decompose x to floor(x) and whatever's left over, which is x - floor(x) and always nonnegative. fract(x) gives the leftover of trunc(x) instead and is negative when x is negative, and it's usually a logic error for a lerp variable to be outside [0, 1]. In this case, it creates artifacts along the coordinate axes.

underrated comment

Would you mind expanding on this? I thought Google changed the name to 'Gemini Enterprise Agent Platform', and altered focus to 'agent governance' workflows, but that there were no breaking changes from what was offered with Vertex AI.

I was more saying that nobody should use it. AI Studio, while also terrible, is better and does (AFAIK) the exact same thing.

exactly. all of these people complaining are using it wrong. you cant just expect a layer of abstraction to perfectly fix all the layers below it! plus, open router does a great job of handling a lot of the provider problems, but it literally cannot fix the output of the provider

exactly. you can specify a provider in your request to openrouter, or, better yet, use the :exacto endpoint to make open router automatically choose one that is good at tool calls.

That is a very common title format for HN, and has been for a while. It wasn't just since AI has evolved—it has always been a load bearing part of HN

100% Human written slop

Human written slop?

I wrote that slop comment

Exactly! Composer 2/2.5 were amazing, cheap, and fast. Everyone else was Gaga about GPT 5.5 and such, while we were over here doing the work with less cost and more speed

Composer 2.5 is phenomenal for so many tasks!

I don't think it was necessary for two different AI buttons, but whatever

This actually looks sick