Meaningless question, fit will put everything on the gpu if it fits. Fa is default on. No-mmap is not an inference tradeoff and if you do turn it off you need to turn on direct io via -dio

What he should actually do is enable speculative decoding

Most normal people: "Only you decide what you feel."

HN busybodies: "You have forced me to hate you, im an incapable child"

They said earned, not forced

It's not trust. There are plenty of major projects that only have or strongly prefer curl | bash. Trusting a vendor is fine.

Have you missed the Trump presidencies?

If there was a pill for that, I would take it

That's incredibly stupid. Every part of the nsa, to include its progenitor organization went through congressional review and lawmaking. Incredibly stupid

Maybe the NSA they are aware of is just the facade. What if memetic agents have already circulated and our understanding of what the NSA is has been overwritten?

The dod was granted an exemption from the termination of anthropic contracts. It has several more months to divest. Poor reporting by axios, and poor skepticism by HN

The crazy thing is how effort posts went from the most valuable part of this site to the most hateful part of the site, by the very people claiming to be protecting the site

Qwen max has always been cloud only. And its a 1T+ model so it would be expensive

That's not necessarily true, that they have to out innovate Adobe. There's an old aphorism that you can always displace an incumbent with a product if it's 80% as good, at 20% the cost of the incumbent product (ibm/ms, cable/netflix, etc). As adobe increases the price, that 20% gets easier to hit. I suspect (well hope) that we'll see this happen to a bunch of incumbents.