chirospasm

At a load of shit

Hello! I recently deployed GPUStack, a self-hosted GPU resource manager.

It helps you deploy AI models across clusters of GPUs, regardless of network or device. Got a Mac? It can toss a model on there and route it into an interface. Got a VM on a sever somewhere? Same. How about your home PC, with that beefy gaming GPU? No prob. GPUStack is great at scaling what you have on hand, without having to deploy a bunch of independent instances of ollama, llama.ccp, etc.

I use it to route pre-run LLMs into Open WebUI, another self-hosted interface for AI interactions, via the OpenAI API that both GPUStack and Open WebUI support!

I remember reading somewhere that drivers beyond 550* may have some issues with linux right now. I might consider rolling them back to a version that was already working.

*Citation needed, but it was a lower version than what is currently the latest available

If you have a smart TV, you’re already at a disadvantage.

One solution to consider might be a black hole DNS on your local network, like Pi-Hole, that can target this device and prevent all Google requests.

Another, unfortunately, might be to get a dumb TV and use an HTPC as your streaming solution for the content you already watch.

And another might be to look into custom TV OS options out in the wild.

What’s your hypervisor manager? Or are you just bare metal?

For VMWare and Proxmox both, I would recommend the community edition of Veeam. It can handle up to 10 VMs for free.

If you’ve got the funds as a small-to-large business, Veeam’s first paid tier, on a yearly basis, is a solid option to backup even more.

Caveat emptor if you buy a license (or not): Veeam runs on Windows only. I have used, like, a single internal network Windows VM dedicated just to Veeam before. It has an easy to pick up UX after a little research, and the UI is clean.

Bacula is deprecated, unfortunately.

I have looked for something similar. There are a number of spaces where FOSS project lists are maintained, but they are often focused on a singular topics like ‘privacy’ or something akin, and they aren’t often parts of larger lists that can be sorted based on the conditions you mentioned above.

The closest thing, if you are interested in other possible tools that might help: Alternative.to, a crowdsourced software searching tool, which has a means of filtering to show only, say, open source projects, or sort by tags that denote stacks used, languages used, etc. (see screenshot of tags I added). It has been useful enough for my own needs when looking for what you’ve been looking for.

Either way, best of luck! I haven’t been able to find something yet, myself.

Less of an axis and more of general left-center-right, all with regards to which news outlets tend to lean one way in tone and language choice vs. another. You can select summaries of each bias to understand those choices in the app. It also helps break down a few other items of note:

While this may be beyond the scope of your efforts, it does do some solid highlighting of news sources for me.

There are a few Ground.news bots floating around Lemmy – or at least there used to be – that would comment on posts to provide some or all of the above.

Although this is getting some downvotes – likely because of the ‘AI’ and ‘bot’ nature of it – I can image the benefits of running this on your own personal Lemmy instance, leveraging it as a sort of RSS skimmer to determine which article were worth diving into or not.

In the roadmap of this project, there looks to be a political alignment feature, which is the big benefit of services like Ground.news and why I subscribe to it as a news service. As well: a feature to summarize a day, week, a month, etc., of news, which may well have the ability to be topical.

I try to bring as much of my reading into an RSS app as possible, rather than leverage algos on social to spoonfeed it to me. And while I love Mastodon, I also have to do a lot of scrolling and manual visiting of profiles to catch up. The same applies to Lemmy.

This may well be a tooling to make the kind of RSS experience I have been wanting, so kudos to the author.

As another consideration, this guy’s app is also available on F-droid, which means you can avoid updates via Play Store

Ooh, neat! This feels like Folding@Home for AI tasks.

There’s a handful of approaches that may be helpful for this. I don’t want to make any assumptions on what you’ve tried, yet, though. Would you mind clarifying what you’ve given a go at, so that we don’t offer something that doesn’t do what you’re looking for?

TrailSense, an easy to use, comprehensive wilderness tool.

The goals of the developer are fun to consider:

Goals

• Trail Sense must not use the Internet in any way, as I want the entire app usable when there is no Internet connection

• Features must provide some benefits to people using the app while hiking, in a survival situation, etc.

• Features should make use of the sensors on a phone rather than relying on stored information such as guides

• Features must be based on peer-reviewed science or be verified against real world data

Likewise, the features being developed under those goals are great for getting outside:

Features

• Designed for hiking, backpacking, camping, and geocaching
• Place beacons and navigate to them
• Follow paths
• Retrace your steps with backtrack
• Use a photo as a map
• Plan what to pack
• Be alerted before the sun sets
• Predict the weather
• Use your phone for astronomy
• And more