For my #FIAF presentation & workshop in Bologna end of the week, I've put together a clean "#PlugAndPlayAI" setup:
#USB #LiveDistro: #MXLinux #GNU #Debian with #Ollama, #LLava and #OpenWebUI in #Docker and #LLM files on another USB3 #nvme
Fast like a local install. Given enough RAM and CPU/GPU.
Scary how easy it is to do that.
Bonus: Creates self-replicating #ISO boot images of itself (in current configuration) in less than 5 minutes. 
.
Wow.
@Triffen Okay I might try it out 
But I don't know how good #llava is at detecting whether something was made with #generativeAI...
Being constantly annoyed by all the generated images for blog posts and articles (seriously, STOP!) I thought about making something with #ollama #llava that identifies, then hides them from me - when I realized that the solution to less #generativeAI can't be *more* #LLM's 
Wie kann #KI Museen dabei helfen, Sammlungen zu erschließen? Sebastian Ruff vom Stadtmuseum Berlin erzählt von seinen Erfahrungen mit Tools zur automatischen Schlagwortgenerierung. Sein Fazit: die Arbeit damit kostet erstmal Zeit & die Tools haben ihre Grenzen, aber sie haben Potenzial. Wichtig am Anfang: vollständige Thesauri mit Normdaten & Fehlstellenanalyse im Datenbestand 
#Datenqualität
https://www.kultur-b-digital.de/digitale-kultur/impulse/ki-im-stadtmuseum-interview-mit-sebastian-ruff/
#LLM #DataScience #AI #CulturalHeritage #ChatGPT #llava #Bilderkennung @museum
I realized that #llava-next is incredibly good in describing images in prose, but when I prompt it to deliver some delimiter separated tags, it usually screws up at some point.
Therefore, my current approach is to use that neat prose text and throw some #NLP on it. So far it looks promising.
The computer scientist in me is screaming, though. So much superfluous text that gets thrown away. Maybe I should keep the whole description as well and simply support some full-text search... hmm....
For 10 days, #QNAP support is trying to help me get #QuMagie to show thumbnails. The communication is viscous at best with almost only one response per day.
In the meanwhile I set up #photoprism in a docker container on the same #NAS, which already indexed all 60K images and ran facial recognition once. The labeling is not working that great, but I will try to run my own algorithm on that and add my tags using #llava-next.
At the AI Sauna hackathon today, we experimented with generating alt text for historical photos using the LLaVA-13B visual LLM running on the LUMI supercomputer. The results were surprisingly good!
Project page: https://meta.wikimedia.org/wiki/AI_Sauna/Generate_alt-texts_for_historical_images
Code/results: https://github.com/osma/llava-alttext-hkm
Encaramos el final de marzo con otro nuevo episodio de "Accesibilidad con Tecnologías libres", #ATL para los amigues.
En https://accesibilidadtl.gitlab.io/05 tenéis las notas del programa con los temas tratados y los enlaces.
Si seguís a este alias @6706483 , tendréis automáticamente las publicaciones del podcast dentro del Fediverso
Y por si aún no lo tenéis el feed RSS para vuestras aplicaciones es:
https://accesibilidadtl.gitlab.io/feed
En este episodio 05 participan entre otras personas:
- @raivenra
-@jonathanchacon
- @DavidMarzalC
Esperamos que os resulte interesante.
#AIworm infects users via #AI-enabled #emailclients — #MorrisII #generativeAI worm steals confidential data as it spreads
Named 'Morris II' after original Morris, first computer worm that created a worldwide nuisance online in 1988. This worm targets AI apps and AI-enabled email assistants that generate text and images using models like #GeminiPro, #ChatGPT4.0, and #LLaVA. Researchers successfully tested this Morris II #worm and published its findings using two methods.
https://www.tomshardware.com/tech-industry/artificial-intelligence/ai-worm-infects-users-via-ai-enabled-email-clients-morris-ii-generative-ai-worm-steals-confidential-data-as-it-spreads
Gemini, ChatGPT und LLaVA: Neuer Wurm verbreitet sich in KI-Ökosystemen selbst - Golem.de
https://www.golem.de/news/gemini-chatgpt-und-llava-neuer-wurm-verbreitet-sich-in-ki-oekosystemen-selbst-2403-182790.html #Malware #Computerwurm #KI #Gemini #ChatGPT #LLaVA
If true and replicated this offline (!) and opensource LLaVa model on a raspberry Pi is very impressive.
#AI #generativeAI #LLM #LLaVa
Just came across the Llava - Large Language and Vision. Impressive demo on image understanding. Upload a picture and ask questions about it.
Demo: https://llava.hliu.cc/
GitHub: https://llava-vl.github.io/
Paper: https://arxiv.org/abs/2304.08485
#creativeAI #llava #llm
#Microsoft Open-Sources 13 Billion Parameter Language and Vision #Chatbot #LLaVA
https://www.infoq.com/news/2023/05/microsoft-llava-chatbot/ #LLM #ai
