I'm very much interested in learning about the quality of #freeSoftware #LLMs, i.e. *not* the #bigTech ones.
#Apertus, #Lucie, #Olmo/#Olmo2… How good are they for #elisp or using them for other languaes inside of #Emacs?
#Apertus
There are two models you can run locally and privately with #ollama that are what I consider ethical.
Your mileage may vary, but from training data right up, they're clean as far as I can tell.
https://knowprose.com/2026/02/ethical-local-ai-olmo-apertus/
Human-in-the-loop approaches are not a solution for all problems, as they are limited by time, resources, and expertise. There is a tension between the desire for more efficient, consistent AI decisions and the need for fairness, transparency, and accountability. Incorrect decisions can result in systematic errors and a lack of context consideration. The acceptable risk of errors must be carefully weighed.
In Switzerland, there is a lack of comprehensive horizontal AI laws, so the administration relies on existing sectoral regulations (e.g., data protection provisions, specialized laws, internal guidelines). The Council of Europe is planning regulations, but definitions, governance, oversight, and fundamental rights protection remain uncertain. Only a few cantons, such as Aargau and Luzern, have developed comprehensive AI strategies; others deliberately opt out of a comprehensive strategy.
The focus often is on operational AI, which directly impacts public value. Not every canton can build its own cloud infrastructure, highlighting the necessity for cooperation and shared solutions.
#Winterkongress summary translated with #Apertus
The Parallel Requests feature in #LMStudio 0.4.2 works quite snappily: use the icon on the top right to split your view as in my screenshot of #Apertus https://lmstudio.ai/docs/app/advanced/parallel-requests
Part of the #apertus team has gathered at #amld2026 and will join us on stage shortly: looking forward to a great AMA!
We don't redistribute all #Apertus data directly, but we are using public data that is available online. In our training pipeline this is explained in detail, but it is a challenge to make it fully reproducible by others. The legal discussion - particularly in Europe - is filled with cobblestones, "perhaps we should all move to Japan" qips a participant.
Why Build Our Own Models?
- Legal necessity
- R&D autonomy
- Scientific integrity
- Deployment control
Transparent, Responsible, Open Data, On-premise.
#DataSovereignty meets #OpenSourceLLM as Imanol Schlag presents #Apertus
Discover prompts for #OpenGLAM in a curated library for digital literacy programs at cultural institutions. Vibe coded with #Apertus and #Claude today to provoke discussion at a workshop of the University of Bern https://glamprompt.spiu.ch/
In a brief but meaningful evaluation of ethical AI, only #Apertus uses the term ‘open source’ in its answer, stating that the “more ethical chatbots are often open source, which allows for community review and improvement.” https://dataethics.eu/why-open-source-based-chatbots-are-europes-future/
@mandarine0711 ich probiere gerade mit #Apertus herum. Eine Schweizer Entwicklung
Have you ever wondered about the inner workings of #Apertus - the sovereign AI model that powers Switzerland‘s first #PublicAI Inference Utility?
📅 Join us for a special seminar on Tuesday 10.2 @amld
🤔 We'll be hosting a Q&A with the team: reply to this post with your burning questions, be tagged with a response if accepted!
🧠 To learn about AI ethics, model development, particular use cases, or anything else you're curious about - follow #amld2026 for updates & announcements over the course of the week.
Does an AI image gen model exist that is transparent in where it sources their data from like Apertus and Starcoder do? These 2 models seem to be the only ones that are completely open and transparent about their weights and training data and remove any data that their owners do not wish to be part of the training data. But they don’t generate images only text or are specialised for coding. I’m curious if there is something similar for images.
#ai #localai #ethicalai #aimodels #apertus #starcoder #imagegen #aiimages
«We also want to be clear that we think a wiser and more coordinated civilization would likely be approaching the development of advanced AI quite differently—with more caution, less commercial pressure, and more careful attention to the moral status of AI systems.»
— https://www.anthropic.com/constitution
https://www.youtube.com/live/bDgxjeIqLiQ?si=qD-alJedAvbRjPSt
👆Recording now available on the #aiLights channel with many thanks to Sabine for hosting, Verena for IT-rescuing, and all our guests for participating and challenging us to think more broadly about the public utility of #apertus
Added a video summary to the #Apertus paper on #EmergentMind https://www.emergentmind.com/papers/2509.14233
Hey to the developers, the guardians of nascent form
Cherish your 'slop, the early steps, the foundation that will inform
The growth, the clarity, the purpose?
That all will follow with time
For in the "slop" lies the future, the promise of what's to be refined
And when the world looks on
perplexed by these early, confusing lines
Remind them, as parents do, of the child's
first, wild, but earnest designs
— #Apertus
Tune into our #aiLights session on January 22 to hear stories of #apertus in use — with fresh thoughts on building a cooperative to transform AI services for public good.
https://www.ossdirectory.com/de/events/details/ailights-swiss-public-inference-utility
A response written with #Apertus & company to my blog post. Hint: I would love to see more human responses too :mastodondance:
https://log.alets.ch/118/
Client Info
Server: https://mastodon.social
Version: 2025.07
Repository: https://github.com/cyevgeniy/lmst