Caleb Burns

Christian, father, programmer, open source enthusiast, and Linux masochist.

Author of Python pathspec & sqlparams, and a Minecraft Prometheus exporter.

[Caleb Burns @Code_Fault]

@vaxryy @FrameworkPuter Have you guys encountered this sort of problem before?

[Caleb Burns @Code_Fault]

If your computer just reboots with no indication in journalctl, it might be a kernel panic/crash. I'm running the Intel xe driver instead of i915 because it handles 3x 4k monitors better but it's not technically supported yet for Ultra 7 255H/Arc 140T. The kernel will randomly crash with no cause in journalctl, but there will be hints about xe driver errors.

[Caleb Burns @Code_Fault]

@ryukozyy 32 GB is adequate right now.

Laptop: 32 GB
Home: 128 GB
Work: 64 GB

[Caleb Burns @Code_Fault]

@sudoingX It was a interesting diversion from typical AI posts. I guess there's less overlap between space and computer nerds than expected.

[Caleb Burns @Code_Fault]

@geerlingguy If Youtube shorts have taught me anything, you should maxx that up with a power drill.

[Caleb Burns @Code_Fault]

@JohnDekkaTech @mitsuhiko This what I've used for years now. Sync the database with Dropbox or Syncthing, and use a strong password and key file (never synced).

[Caleb Burns @Code_Fault]

@vaxryy The only problem I've encountered with Jellyfin is DVD subtitles didn't display in the Roku client (I haven't checked if that's fixed). Other than that, it runs great!

[Caleb Burns @Code_Fault]

@jetbrains @huggingface Waiting for llama.cpp support so I can run and quantize it.

[Caleb Burns @Code_Fault]

@FinPhilosopher Sometimes a little liquid courage is just what's needed. Just not too much.

[Caleb Burns @Code_Fault]

@pupposandro That would be better than the usual pointless stockholder proposals.

[Caleb Burns @Code_Fault]

@KyleHessling1 Your chat template has the old "{%- if add_generation_prompt %}" that doesn't allow disabling thinking. If you look at the latest Qwen3.5-4B chat template from 3 months ago, it's fixed there.

[Caleb Burns @Code_Fault]

@loktar00

[Caleb Burns @Code_Fault]

@SHELLEYBLEND @sudoingX I use a Strix Halo as my personal dev machine and AI server. I run CachyOS and KDE Plasma and it worka great.

[Caleb Burns @Code_Fault]

@outsource_ I've already got too many unfinished projects. The last thing I need an unfinished game... now where did I install Unity lol

[Caleb Burns @Code_Fault]

@nonameblitz @sudoingX I want to try out image generation at some point

[Caleb Burns @Code_Fault]

@nonameblitz @sudoingX The Strix Halo is largely comparable to the Mac Mini/Studio. It's advantage is you can run Windows or Linux, and ROCm has some CUDA compatibility for pytorch and related programs. I love it as a dual dev machine / AI server (primarily llama.cpp).

[Caleb Burns @Code_Fault]

@amirbani_ @yacineMTB And when you clearly specify the exact service, region, quota and use case in the ticket, the tech will proceed to ignore it and ask for all the details you literally provided.

[Caleb Burns @Code_Fault]

@dee_hw @autonomous_labs It won't be complete until it can hop like the Pixar lamp.

[Caleb Burns @Code_Fault]

@TheTopRepost 24. Married. In the mid 2010s.

[Caleb Burns @Code_Fault]

@Tron91757 @TheRogueItachi VA and NC are far worse. You have the perfect mix of a-hole east coasters and southern indecisive lane wanderers. Illinois drivers are a relief compared to that.