daniel han and unsloth once again provides great content. while llama.cpp is easily set up as a server to be hooked into openai-api compatible...
as of late 2025, embeddinggemma a 300m model is likely the 'best' low resource embedding model. now, voyage mongodb team released a series of models...
a reminder to use more efficient models where possible. gliner supports fast training and inference of NER models tailored towards things like...
Quoting Will McGugan Sometimes, developers get hyper-focused on tackling the immediate technical challenges in front of them that it prevents them...
testing a simple blog cms with copy-paste image functionality screenshot of cms
Siterer fra Matias Jentoft: La oss fortsatt ha et kritisk blikk på språk-KIs kunnskap om språk. Aftenposten og VG skrev forrige uke om hvordan...
for some reason, after 10 years of programming, i never thought it would be neat to list files by their last updated time through git vc, rather than...
sammendrag Moderne språkmodeller kan effektivt prosessere ustrukturert og støyfylt data, slik man ofte finner i samtalebasert tekst. Denne egenskapen...
karpathy says it again, on the dwarkesh podcast: On LLM agents. My critique of the industry is more in overshooting the tooling w.r.t. present...
as ai-generated tools become more-or-less enforced in the workflow of a modern developer, having the correct mindset when dealing with these tools is...
wanting to move to pure html, embarrassingly from the website justfuckingusehtml, I ended up moving from the typical jekyll gh-pages setup, to my own...
These are a few notes from experiences and random readings online as of late. prompt chaining monolith: small discrete steps with individual prompts....
The last weeks have been extremely turbulent, with one side doing massive development on with Open AI apis, such as AgentGPT...
spaCys coref model is intriguing -- although it is largely based on WL-coref
... While it is impossible to catch up with the current state of things, here are a few ramblings from the last couple of days: - BloombergGPT, for...
As a test/demo of the my first blog entry for this site, a few findings from the past few weeks and update on my current projects seemed...