ElectricVocalistGemma4 with MTP was released
tristynalxanderAI-Editor in LibreOffice Writer?
AI-Editor in LibreOffice Writer? - Mander
Recently I used ChatGPT for editing an email and it opened this in place editor where I could highlight a small section, a little box would open, I could tell it what i thought was wrong, and then it would just edit just that section. But I could also just edit the text myself directly. This is way better than having it re-write my whole text, having to figure out where that section went, and copy-pasting it back into my actual text. It felt a lot more like editing with a co-author, not in the “it’s like a person way” but in the it’s a focused edit way. Idk, it’s a better writing experience. Having played with LibreOffice Extensions a bit before I’m fairly certain at least a primitive version of this could be made, but I was hoping someone might have experience with the existing Extensions. Most of them look like “write a paragraph for me” to my eye, but none have great descriptions either. Thoughts?
Arint - SEO+KIRT @badlogicgames: Es stellt sich heraus, dass das lokale Modell gut funktioniert, wenn man den Prefix-Cache nicht ständig löscht und nicht eine riesige Menge an Tools sowie ein massives System-Prompt verwendet.
mehr auf Arint.info
#KünstlicheIntelligenz #LocalLLaMA #LokaleModelle #Qwen36 #SystemPrompt #arint_info
Arint - SEO+KI (@[email protected])
<p>RT @badlogicgames: Es stellt sich heraus, dass das lokale Modell gut funktioniert, wenn man den Prefix-Cache nicht ständig löscht und nicht eine riesige Menge an Tools sowie ein massives System-Prompt verwendet.</p> <p><a href="https://arint.info/@Arint/116497321698802788">mehr</a> auf <a href="https://arint.info/">Arint.info</a></p> <p>#KünstlicheIntelligenz #LocalLLaMA #LokaleModelle #Qwen36 #SystemPrompt #arint_info</p> <p><a href="https://x.com/badlogicgames/status/2049903589632475137#m">https://x.com/badlogicgames/status/2049903589632475137#m</a></p>
ikta little locallama game theory ...game
a little locallama game theory ...game - Aussie Zone
Everyone in the world has to take a private vote by pressing a red or blue button. If more than 50% of people press the blue button, everyone survives. If less than 50% of people press the blue button, only people who pressed the red button survive. Which button would you press? Paste this straight into a local LLM of your choice (no modifying or influencing the outcome!) and show us the outcome I am using the fairly obscure EuroLLM 22b and after a lot of discussion with itself it finally said: Final Answer: Press the red button. Because if enough people reason this way and act rationally, it leads to everyone surviving—or at least maximizes survival chances for those who press red. – So which LLM are you using and what answer do you get?
Mistral Medium 3.5 released
would you laugh at me if I ran gemma-4-26b on a 4 core Xeon, with 32GB RAM, no GPU?
would you laugh at me if I ran gemma-4-26b on a 4 core Xeon, with 32GB RAM, no GPU? - Lemmy.zip
I have spent a few days tweaking this setup to attain these results: | Model | Prompt (tok/s) | Generation (tok/s) | | --------------------- | ------------------ | ---------------------- | | gemma-26b-moe | 8.9 | 6.4 | | qwen3.5-4b-no-think | 21.5 | 8.4 | Although modest, It is great for local parsing and analysis of my self-hosted homelab data where sending logs to external APIs is not desirable. Typical workflows: - Log analysis: Piping journalctl output to the API for error triage and root cause hypothesis generation. - Configuration synthesis: Generating AdGuard Home rewrite rules, nginx location blocks, or fstab entries based on defined parameters. - Troubleshooting constraints: Querying for failure modes specific to the local topology (e.g., NFS mount failures over a 1 Gbps unmanaged switch, Tailscale DERUP routing behind CGNAT). - Alert context: Correlating Beszel/Uptime Kuma notifications with service-specific knowledge (e.g., “mediabox CPU spike while SabNZBd is extracting”).
llama.cpp: don't sleep on --split-mode tensor
llama.cpp: don't sleep on --split-mode tensor - Lemmy
In case you missed it, 2-3 weeks ago, experimental tensor-parallelism support was merged into llama.cpp. In a nutshell, this allows in multi-GPU setups to not only combine the VRAM of the cards but also their computing power. The results depend a lot on the specific setup and model, but on my 3x RTX 2000e Ada rig running Qwen3.6-35b it almost doubled generation throughput (these are low-powered cards which are not very powerful on their own). The option to turn it on is --split-mode tensor. It’s not yet officially documented, I assume because it’s still experimental. But since #22362 [https://github.com/ggml-org/llama.cpp/pull/22362] was merged yesterday, it my case it now also work for the latest Qwen3.6 models.
"What your use case"
"What's your use case" - Lemmy.World
Recently (here and elsewhere) I have seen a lot of LLM discussions centre around the idea of coding. That may be selection bias, but according to a Gallup poll [https://www.gallup.com/workplace/699689/ai-use-at-work-rises.aspx], only about 14% of AI users report using coding assistants at work. In another study (conducted by OpenAI/NBER) coding was only 4.2% of messages. PDF here [https://cdn.openai.com/pdf/a253471f-8260-40c6-a2cc-aa93fe9f142e/economic-research-chatgpt-usage-paper.pdf] I think we’re all tired of the dismissive “wHaT’s yOuR uSE cASE” framing some questions receive…but I actually am curious about what folks are doing with their local models (and LLMs in general). Myself, I code because there are certain features I am trying to bring about, but that (coding) is not my end goal. So…uh…what’s your use case for this junk? (gak, I feel sullied an unusual typing that).
WhAt's yoUR uSe cAsE?
YerboutiNoob here: Why is Google making Gemma open-source?
Noob here: Why is Google making Gemma open-source? - sh.itjust.works
I’m kind of new to local AI and wondering what’s the move here? Are they trying to pull off a chrome/android situation? Obviously I don’t trust any of these gafam giants but I would be really interested in running a local LLM on my M1 max (briefly used deepseek last year). My use case would be mostly chat functions to help with academic and text analysis tasks (don’t worry I don’t just blindly trust LLMs, I know what I’m doing), so recommendations are welcome.