You’d expect the same answer each time. It’s the same photo, the same model, the same question. But you won’t get the same answer.
I don’t know what ads show that, but anyone who knows the first thing about LLMs knows you don’t get the same answer twice.
I’d get this expectation 5 years ago when most people weren’t familiar with it, but come on… you don’t need to feed it an image 500 times to see that.
Waste of energy. It’s like asking a person to estimate a non-trivial angle. Either use a model trained for that task, or don’t bother.
Exactly, it’s only an improvement until they’re bought and we’re all in the same boat again. We need a federated forge and open standards.
wtf
An unprivileged local user can write 4 controlled bytes into the page cache of any readable file on a Linux system, and use that to gain root.
If your kernel was built between 2017 and the patch — which covers essentially every mainstream Linux distribution — you’re in scope.
how does that only get a CVE score of 7.8, the impact of this is huge
tl;dr clickbaity title for an article that cites xitter and tries to sell their security solutions
yeah, that was cringy
the world’s most-followed person
most followed on his own echo chamber :clap: :clap: :clap:
they are open weight and have a whitepaper, that’s already vastly better than whatever openai and anthropic are doing
it is, but it’s initially opt-in; meaning the system won’t have AI features, but once it does, you’ll remove them by removing snaps
it’s confusing because the paragraphs they talk about opt-in and removing snaps are different, but 26.10 won’t ship with AI features in a fresh install.
why would that only affect “some” countries
at this rate, it will be
I’m interested in setting it up, are you using vs code? Which extension or editor?
watch me go back to debugging like a real engineer: copying and pasting from stack overflow
Users on annual Pro or Pro+ plans will remain on their existing plan with premium request-based pricing until their plan expires, however, model multipliers will increase on June 1 (see table).
holy shit, 9x the previous cost. which was already not great. I was on the fence about cancelling it, but thanks for making up my mind, MS
ok, to start with, if you need a POSIX interface to the filesystem, you already have an SSH connection to that server, and don’t need much stability across multiple clients, SSHFS may do just fine. For a homelab, that is likely the case.
now, if you’re hosting a web server that needs data distributed across drives/nodes, data redundancy, and the usage is primarily programmatic, closer to a CDN’s or machine learning pipeline than a single user browsing files; then you want an S3-compatible solution. The S3 API makes it easier to plug it into your application, while allowing you to migrate to a different one - which I’m actually currently doing for a MinIO deployment at work.
SSHFS is a hack and has nothing to do with the proposal of S3 compatible backends
you got a pizza box? Fancy
It’s still matrix multiplication. Running it on a general purpose CPU is inefficient.
Barely. Even with the code and seeds, it’s still a struggle to do that. There’s plenty of questions from people running pytorch and tensorflow models that can’t reproduce results. Maybe you isolate enough variables that consecutive runs actually produce the same output, but the study is about commercial models. You’ll never get deterministic output from those.