#Apertus prompt with the discussion https://chat.publicai.co/s/1ea429d6-507b-47f3-8023-19cb29ec3b89
#Apertus prompt with the discussion https://chat.publicai.co/s/1ea429d6-507b-47f3-8023-19cb29ec3b89
"Diktiergerät eines hybriden Schwyzerlandes e.V"? 📼 🙋🥲 #Apertus #AI_quirks
Context: https://chat.publicai.co/s/f4034d1e-81d7-4d9a-983c-44e7df989b75
For now #Apertus is in the top-5 ranking of European open source models. Expect the OLMo 3 release and other updates to shuffle the deck in the days ahead. Screenshot of https://osai-index.eu/
Φασιστικές επιθέσεις και προκλήσεις στην κατάληψη Apertus
https://infolibre.gr/2025/11/23/fasistikes-epitheseis-kai-prokliseis-stin-katalipsi-apertus/
# # # # # # # # #Apertus # # # #
For more info on #Apertus see also my blog post https://hachyderm.io/@loleg/115259229881276635
stepping stone provides secure, Swiss-based AI infrastructure and HPC, supporting industrial science projects with complete data sovereignty (US-hardware excepting, as an audience question pointed out) and security requirements. Excellent input from Michael Eichenberger, CEO
#DINAconCH #Apertus #MistralAI
Main takeaways from the closing panel:
- #Apertus needs to be used to explain how AI works across the production chain, used as a model of models, and encourage more local “baby Apertus” quantized and embedded versions;
- There is a strong interest in using academia-approved LLM in education and science communications;
- AI is still seen as a curious toy in much of industry, not delivering clear ROI, threatening to pull various carpets out from under feet;
- Research on the training side is needed to establish not just competitiveness, also dataset bias and ethical guardrails;
- Vibe coding is here to stay, but probably not really where we would expect;
- Public AI is a movement to even the playing field that we should embrace.
With thanks to the organizers particularly @barochat for a spectacular #SmartCityDay
A short poem I asked #Apertus to generate in closing:
Reforesting the Mind
Beneath the canopy's green embrace,
I walk where ancient trees stand tall,
Each step a seed of mindful grace,
Answering nature's silent call.
The forest breathes, and so do I,
As sunlight filters through the leaves,
My scattered thoughts begin to fly,
Like birds released from winter's grief.
Rooted deep in present time,
My mind grows branches, strong and new,
Each breath a rhythm, soft as chime,
As forest whispers speak anew.
The undergrowth of worry clears,
Wild thoughts bloom like flowers bright,
In this cathedral of the years,
My soul finds its true light.
So walk I do, and mind I mend,
In nature's gentle, healing way—
For in the woods, the heart can tend
The gardens of a brighter day.
https://chat.publicai.co/s/ee26f0c7-acfb-4ad8-ab8e-ae5566303325
Hey early adopters! 🤞🏻🤖🇨🇭Are you using #apertus & care to share an experience ? Please fill out the Apertus User Satisfaction Survey (and pass it on to other users!) https://forms.gle/9TGSfTK1RfKj2b6X9
🤖 Now, the unavoidable topic: Artificial Intelligence. But where do we stand?
Fun fact: AI-generated images of Lavaux (on the left) without Lake Geneva prove humans still have an edge. 😉
He traces AI’s #evolution, from #LLMs genesis to OpenAi and ChatGPT, and of course highlighting #APERTUS, the open, compliant model developed by @EPFL @ethz.ch, CSCS and the help of @swisscom
They have not tested #Apertus yet, due to some negative perceptions of the initial release. I invited everyone to join us for the #OpenParlData challenge and give it a try at the DINAcon #HACKnight https://hacknight.dinacon.ch/event/12/challenges
Researchers surprised that with AI, toxicity is harder to fake than intelligence https://arstechni.ca/M5rn #computationalTuringtest #UniversityofAmsterdam #largelanguagemodels #UniversityofZurich #machinelearning #DukeUniversity #AIsycophancy #generativeai #AIalignment #AIdetection #emotionalAI #socialmedia #AIbehavior #AIresearch #chatbots #deepseek #Apertus #Mistral #Biz&IT #gemma #LLaMA #Qwen #NYU #AI
@spotbot2k @golem #Apertus ist m.E. das, was der Linux Kernel 0.12 war: Ein Ansatz, der zeigt, was möglich ist. Teuken ist komplett unbrauchbar. Das einzige vergleichbare Modell wäre #Olmo2 und das ist Apertus unterlegen. Natürlich wird aktuell jedes opensource LLM schlechter sein als jedes nur open weight LLM. Aber darum geht es nicht. Wenn sich um solche einzelnen Initiativen wie Apertus eine Community versammelt, die die Philosophie praktisch umsetzen wollen, hat das eine Chance. Und ich glaube, dass der Erfolg dieser Idee enorm wichtig werden wird für uns als Gesellschaft. Letztlich geht es um epistemische Autonomie, Mündigkeit und KI als gesellschaftliches Gut.
A recent blog post on #huggingface pointed me to aisheets project: very alpha, but ambitious and usable. Here you can see it connect to my local #ollama and wrangle through some open data with #apertus https://huggingface.co/blog/aisheets-unlock-images
I'm sick of people calling open weights AI models open source. There is a huge difference. The only real decent open source and ethical AI model to my knowledge is Apertus AI. We can't allow companies to convert open source into another marketing buzzword. Be careful with "open source" check first if it really is.
#opensource #openweight #ai #apertus #apertusai #opensourceai
"I am a schoolteacher at heart and #Apertus reminds me of something my favorite schoolteacher Shirley Chisholm used to say about herself: “unbought and unbossed.”
Intentionally open and transparent, this LLM is designed with Data for Good in mind. It is a great example of how support for basic research in science and technology at a federal level, untethered to commercial demands and concerns, produces tools and applications that provide superior support for projects for the public good."
-- David Corliss
https://magazine.amstat.org/blog/2025/10/01/apertus-llm-data-for-good/
Per a executar #Apertus en local (sistema operatiu Gentoo) tenint una CPU AMD Ryzen 7 i 32GB de RAM, em recomanen una targeta gràfica AMD Radeon RX 7900 XT (20 GB VRAM), uns 750€ 😳
Amb 20 GB VRAM podria executar Apertus-70B en Q4_K_M a ~15–20 tokens/segon.
La gràfica que tinc ara, una Nvidia GT730 (molt vella), no em serveix per a res. 😅
Wie geht digitale Souveränität? Imanol Schlag und Maria-Grazia Giuffreda sprechen im @nzz.ch-Podcast über die KI #Apertus:
www.nzz.ch/podcast/die-...
Schweizer KI gegen Silicon Val...
Note of course the #vLLM support was available from the launch of #Apertus - and there may be good strategic reasons to prioritize this in your stack, especially when hosting larger groups of users.
See my blog post for more details & instructions https://log.alets.ch/110/#using-ollama