Will Smith eating spaghetti is the only true test of an AI video creator.

Also, a bit impressed the girl at the top isn’t insanely gorgeous like most AI generations. Making mundane or ugly things is usually a struggle for most models that are built on high quality, studio imagery.

Spoiler: Mac and Windows only. Was hoping for a Flatpak or something but not yet.

Hoping it remains viable for a long time without updates. Syncing my KeePass database is really key for me. I need to fluidly add and read passwords from at least 3 devices.

Don’t know but copyright holders have demonstrated a few cases where they got AI to blatantly rip off copyrighted pictures or music.

Usually it doesn’t solve my problems but it gives me a few places to start looking. I know some models are capable of this but to get a perfectly accurate and useful response would probably require it to recall a specific piece of input it was given and not just an “average” of the inputs.

I’d agree with this recommendation. I believe there were multiple occasions where my router assigned a dynamic IP the same as some other reserved IP. Hard as hell to diagnose. Key indicator was that roughly half the packets were being lost.

Don’t remember the tool, maybe someone here does, but there’s some web service out there that boasts a “no storage” approach. You provide some URI and some other value (maybe username) and it makes a password for you, but it’s always the same for a given combination. Basically it’s a purely functional generator.

Downside would be forgetting a minor detail (Did it end with a slash or not? What was the username?) or the site going down. You can achieve the same thing yourself with a hash calculator but those passwords are a bitch to type in.

tl;dr just use KeePass

Never tried Authelia or Authentik but I’ve heard good things about them. I’m sure one of them will integrate with a reverse proxy.

Agreed, OpenWRT is for something with limited resources like an OTS router.

I didn’t for the longest time but now I use Traefik for this. It can automatically add services (i.e. containers) to it’s routing list so the overhead is low and since I also run openwrt on my router I setup *. localhost to point to 127.0.0.1 so I don’t have to remember what ports I’m using for which service (e.g. jellyfin.localhost). You can also setup DNS entries using something like PiHole.

I put the sample template (https://yacht.sh/docs/Templates/Templates/) into a file named docker-compose.yml and Docker said the syntax was invalid. Are you saying I can give Yacht a compose file and it’s cool with it?

Used it for a bit but I didn’t like how you have to deploy things from templates which are basically compose files that don’t look like compose files.

This is the kind of AI stuff that really annoys me. Looking at one of the mutation examples I didn’t see anything that wouldn’t normally be tested by a typical mutation tool. You took a simple, idempotent process and you got an llm to do it slower, less accurately, and using more resources.

If you wanted to marry the two in a new and possibly useful fashion I would say use an llm to analyze the results of a standard mutation test and give guidance on what issues should be acted upon first. An off-by-one calculation could mean somebody loses a million dollars or it could mean a button is grayed out. Standard mutation tools don’t give you that context.

Other than the low chance of you being targeted I would say only expose your services through something like Wireguard. Other than the port being open attackers won’t know what it’s for. Wireguard doesn’t respond if you don’t immediately authenticate.

There’s a little overlap with things like Terraform but it’s not as bad as if they bought the companies that owned Chef or Puppet.

Can’t believe that’s gone through. They took JBoss when they bought RedHat so now it doesn’t have to compete with Websphere and when they bought HashiCorp Openshift doesn’t have to compete with Nomad. At this rate they’ll buy CyberArk and then that’s no more competition with Vault.

If OP has a thrift store nearby it’s pretty likely they can get both for under $30.

This is actually great because not only does it share the hosting load but models occasionally get pulled from sites so they can be exclusive to some AI subscription service.

I haven’t looked into the built in features in a while but I think comfy needs to have more basic nodes. I’ve picked up workflows with dead nodes and it’s not easy to see what used to be there and fill in the gap. Sometimes it’s something complex like an animation node but I’ve had basic text box nodes go dead.

DebOps my dude.