Neat! I'm already using openwebui/ollama with a 7900 xtx but the STT and TTS parts don't seem to work with it yet:
2025-05-05 20:53:15,808] [WARNING] [real_accelerator.py:194:get_accelerator] Setting accelerator to CPU. If you have GPU or other accelerator, we were unable to detect it.
Error loading model for checkpoint ./models/Lasinya: This op had not been implemented on CPU backend.
Basically anything llama.cpp (Vulkan backend) should work out of the box w/o much fuss (LM Studio, Ollama, etc).
The HIP backend can have a big prefill speed boost on some architectures (high-end RDNA3 for example). For everything else, I keep notes here: https://llm-tracker.info/howto/AMD-GPUs
Main Kiwix dev in charge of scrapers (tools to create ZIM files, even if we do not really scrape technically speaking) here.
We are working hard toward upgrading the Wikipedia ZIMs, but it is far from being an easy feat. I'm mostly solo on this, and far from dedicating 100% of my time to this, so it does not move very fast. We are quite close to being able to reach the goal however, probably only a matter of weeks now.
Bonus: the tool will now get pretty good at making a ZIM of any Mediawiki, not only Wikimedia ones, we expect for instance to work on all Fandom wikis somewhere this year since there is significant knowledge over there.
Hey, thank you for what you do! Kiwix was the first project that made me feel like using Github Sponsors to support it; your work is wrapped into countless other educational projects like IIAB.
Is there any specific help you need, or where could folks get involved if they wanted to?
I looked into it once, I think the script or system that built the larger dumps broke and no one fixed it. I started working on it but other stuff got in the way.
Searx is great (I use SearXNG personally) but be aware that it's something that lifts along on the main search engines. It's not a search engine, just a meta one. It still depends on the big US ones for its results. Just like Kagi for that matter (though they do have a small crawler themselves, their main results are metasourced from various large engines)
They're not using/advocating to use Signal for their military control/communication:
> This week, Brigadier General Mattias Hanson, the Swedish Armed Forces' CIO (Chief Information Officer), decided that calls and text messages that do not concern classified information should, as far as possible, be made using the Signal app. The decision aims to make it more difficult to intercept calls and messages sent via the telephone network.
Seems people were using SMS for those messages they are now advocating to use Signal for.
Also, seems they've done a review (obviously) but unclear if they had access to something internal from Signal to do the review, feels like they had to:
> The Signal application has been deemed by the Swedish Armed Forces to have sufficient security to make it difficult to intercept calls and messages.
Any decent military will be using multiple forms of communication systems.
I was a communications specialist for the Swedish Armed forces 10+ years ago, including a tour in Afghanistan and a tour in Kosovo.
We used radio links for internet that I can tell you, were more adversarial than friendly.
The Swedish military is highly capable when it comes to network communications. A small nation will have to think differently.
You could potentially use an instant messaging system in control by someone else, if you are willing and capable of sharing encryption keys with whomever you are going to communicate with beforehand.
>Because everything in Signal is end-to-end encrypted, we can rent server infrastructure from a variety of providers like Amazon AWS, Google Compute Engine, Microsoft Azure, and others while ensuring that your messages and calls remain private and secure.
Your source doesn't support your claim. The exact snippet you quoted, interpreted strictly, only means they have the option to host it across providers, not that they actually do so. It also doesn't say anything about where it's hosted. It can be hosted in AWS, GCP, and azure, but all in the US, for instance.
Last time I checked it (a few years ago), it was quite buggy and crashed often for me (self-hosted).
For example, what I remember, there was a module to move photos from Google Photos, but it always stopped at 10% photos without any errors. Then there was also some document suite module and it would hang my entire machine regularly, for some reason (had to restart it).
