Asian supermarket and specifically the toasted kind

You need a few drops of it, the taste is pretty intense

Toasted sesame oil, sunflower oil, a bit of salt and pepper and fry it a bit to get it a bit crispy - or just warm it up a bit if in a hurry

Broccoli

It’s super good if you just cook it right

Yea, Jellyfin might be good tech but the fans are insufferable

Maybe if you’re stupid and don’t go directly for the billionaires

Now guide your parents through installing Jellyfin on their TV so they can connect to your instance.

That’s why people get Plex.

Basically the local models don’t (and can’t) contain the full knowledge of the universe.

BUT they can call tools pretty well and if you give the harness the capability to search Wikipedia for example, it becomes a lot smarter

Qwen3.5 and Gemma4 are the best ones for tool calling that don’t need massive amounts of memory

Nothing you can run with affordable hardware. The SOTA stuff requires hundreds of gigabytes of memory - and not RAM, GPU memory.

But you can try with stuff like gpt-oss or qwen coder

His star was just rising and the way he died was so fucking stupid and preventable.

No drugs or partying, just a piece of shit American car…

I do run Jellyfin too, it’s pretty handy at getting metadata for stuff locally so I have something outside of Plex’s internal databases.

And IIRC I have used Jellyfin in the past to watch some more esoteric stuff that Plex (and its official clients) couldn’t play.

But there’s no way in hell I’m opening it to the outside world.

I paid 79€ or something for Plex Lifetime well over a decade ago and it’s by far the best bang for buck I’ve ever got.

It’s also dead simple to share it with friends and acquaintances safely. I know how to manage VPN tunnels and all that shit. I’m not setting that up for my aunt who lives 4 hours away. We did manage to get Plex running on her TV over the phone though.

A 20€ plan of Claude Code for a month will teach you all of it.

It’s pretty much just scripting ffmpeg and feeding the screenshots to a local model via an API (or use a CLIP model)

I have a script that uses yt-dlp to get subtitles off a YouTube video and summarises the main points for me with a language model so that I don’t have to watch a 20 minute top10 list video that could’ve been a buzzfeed article.

The whole thing is fully vibe engineered too.

A HIPAA violation is a death sentence to a company, along with massive fines.

There’s no incentive for them to fuck around

I built a system that translates subtitles from English to my native language and it beats cheap-ass “official” translations 9/10

It even gets colloquial terms and phrases right, adapting to the correct song for example - something a human translator working for minimum pay usually won’t bother

Multimodal local models like qwen3.5 are pretty good

Basically you grab screenshots from the video at intervals and ask the model to tag / describe them.

ChatGPT is pretty good at “what was that thing where…” type of stuff and usually gets it on the first go

Literal megacorporations have run purely on Windows since the 90s and it’s not S-tier corporate? lol

Systemd has to do EVERYTHING

They drop a systemd-kerneld one day and drop the Linux bit altogether