Over ¾ of a century old, been cutting code since 1979, and still love doing it. Currently working on a stand-alone, EU AI Compliant Knowledge Management System.
Bass Player, Hotrod Builder, ex Sailor (Yachtmaster Offshore), lover of food and wine, grumpy old fart with a heart.
WARNING !!! .... I have identified a serious flaw in the data generation process.
Invoices and other routine documents are being created as boilerplates only. Lacking the "real" data. I am Working right now on a fix and should be able to replace the defective code, and the 30,000 document prompt list fairly soon.
My most sincere apologies for this, but unfortunately, with such a large dataset, it only became apparent when i used it to benchmark my our Knowledge Management tools this week.
@biytelum That's an interresting take ..
but in many ways, it is much easier to make "little, microservice-like tool AIs" accountable and transparent.
Certainly compared with enforcing compliance on the monolitihic "do it all" agentic monsters that do everything internally and never explain anything.
When I get tired of waiting for my two RTX3060 to crunch data, I sometimes open up the Document Audit pane on my UI and sse how "Forbee", my little 4B model, is handling his tagging job ...
The excuse I use is that I am evaluating the UI as the transparency/auditability tool for EU AI Act compliance .. but actually ...
I'm just amused by how 4B thinks .. He's kinda cute :)
#LocalAI #EUAIAct #SovereignAI #SmallLLM #XAI #BuildInPublic #KnowledgeManagement
WARNING - IF YOU DOWNLOADED THIS REPO BEFORE 12:45 on 31 March 2026
*there was a flaw in the prompt generation process ... This issue causes ever document to be tied to a hard coded statement, which killed the timeline by injecting 2026 concerns into every prompt .... it is now fixed ... apologies for the hassle - it is now back online at:
A 60k document test dataset for AI development and Prompt Engineering, representing 4 years in the life of a fictitious tech company, All data is artificial so no GDPR worries. Read the "dna" files to get an idea of what it is. The repo includes python scripts to create and generate your own test...
Well i finally did it. I just released my test dataset for AI Evaluation. Its a simulated company, represented by 60,000 documents, the readme in the image explains it all ... If you are interested, its at https://codeberg.org/Lorenz_Systems/Company_Sim.git
#EUAIAct #DigitalSovereignty #SovereignCloud #FOSS #FLOSS #Codeberg #Forgejo #OpenSource #DataGovernance #Auditability #ForensicAI #EUTech #PrivacyByDesign #InformationRetrieval #KnowledgeManagement #DeterministicAI #EUPL
Massive apologies for the 404s, to anyone who tried to access my Codeberg repo .... I messed up my visibilities (new to Codeberg, stepped over from Github) ...
Anyway .. the repo is now open and should be accessible .. please let me know if it isn't
#EUAIAct
#DigitalSovereignty
#SovereignCloud
#FOSS
#FLOSS
#Codeberg
#OpenSource
#DataGovernance
#EUTech
Well i finally did it. I just released my test dataset for AI Evaluation. Its a simulated company, represented by 60,000 documents, the readme in the image explains it all ... If you are interested, its at https://codeberg.org/Lorenz_Systems/Company_Sim.git
#EUAIAct #DigitalSovereignty #SovereignCloud #FOSS #FLOSS #Codeberg #Forgejo #OpenSource #DataGovernance #Auditability #ForensicAI #EUTech #PrivacyByDesign #InformationRetrieval #KnowledgeManagement #DeterministicAI #EUPL
OS-SCI