<feed xmlns="http://www.w3.org/2005/Atom"> <id>https://enklypesalt.com/</id><title>En Klype Salt</title><subtitle>Dette er en blogg om maskinlæring, statistikk og kunstig intelligens, med innslag av mine egne tanker.</subtitle> <updated>2026-07-30T10:58:20+02:00</updated> <author> <name>Håkon Måløy</name> <uri>https://enklypesalt.com/</uri> </author><link rel="self" type="application/atom+xml" href="https://enklypesalt.com/feed.xml"/><link rel="alternate" type="text/html" hreflang="en" href="https://enklypesalt.com/"/> <generator uri="https://jekyllrb.com/" version="4.4.1">Jekyll</generator> <rights> © 2026 Håkon Måløy </rights> <icon>/assets/img/favicons/favicon.ico</icon> <logo>/assets/img/favicons/favicon-96x96.png</logo> <entry><title>Context Collapse, Part 3 - AI Worming through Word</title><link href="https://enklypesalt.com/posts/context-collapse-part3-ai-worming-through-word/" rel="alternate" type="text/html" title="Context Collapse, Part 3 - AI Worming through Word" /><published>2026-07-28T14:00:00+02:00</published> <updated>2026-07-30T10:58:03+02:00</updated> <id>https://enklypesalt.com/posts/context-collapse-part3-ai-worming-through-word/</id> <content type="text/html" src="https://enklypesalt.com/posts/context-collapse-part3-ai-worming-through-word/" /> <author> <name>Håkon Måløy</name> </author> <category term="Coordinated Vulnerability Disclosures" /> <summary>I would like to thank Microsoft product teams and Microsoft Security Response Center (MSRC) for collaborating with me on this technical analysis and mitigation of the disclosed vulnerabilities. The editorial opinions reflected below are solely the author’s and do not necessarily reflect those of the organizations I collaborated with. Summary The findings described in this post are part of a ...</summary> </entry> <entry><title>Context Collapse, Part 2 - When Emails Instruct</title><link href="https://enklypesalt.com/posts/context-collapse-part2-when-emails-instruct/" rel="alternate" type="text/html" title="Context Collapse, Part 2 - When Emails Instruct" /><published>2026-07-14T08:00:00+02:00</published> <updated>2026-07-28T14:27:01+02:00</updated> <id>https://enklypesalt.com/posts/context-collapse-part2-when-emails-instruct/</id> <content type="text/html" src="https://enklypesalt.com/posts/context-collapse-part2-when-emails-instruct/" /> <author> <name>Håkon Måløy</name> </author> <category term="Coordinated Vulnerability Disclosures" /> <summary>I would like to thank Microsoft product teams and Microsoft Security Response Center (MSRC) for collaborating with me on this technical analysis and mitigation of the disclosed vulnerabilities. The editorial opinions reflected below are solely the author’s and do not necessarily reflect those of the organizations I collaborated with. Summary The findings described in this post are part of a ...</summary> </entry> <entry><title>Multippel regressjon</title><link href="https://enklypesalt.com/posts/separate-genders/" rel="alternate" type="text/html" title="Multippel regressjon" /><published>2026-07-10T06:57:00+02:00</published> <updated>2026-07-28T17:17:22+02:00</updated> <id>https://enklypesalt.com/posts/separate-genders/</id> <content type="text/html" src="https://enklypesalt.com/posts/separate-genders/" /> <author> <name>Håkon Måløy</name> </author> <category term="Maskinlæring og statistikk" /> <summary>I forrige post så vi på hvordan vi kan evaluere hvor godt en modell passer til dataen den er trent på. Vi fikk også evaluert modellen vi har så langt ved å bruke denne metoden, men kan vi gjøre en enda bedre jobb? La oss igjen se på dataen vi har: Her plotter vi vekt (x-aksen) mot høyde (y-aksen). Så langt har vi jo sett på dataen vår uten å ta hensyn til at vi også har informasjon om kjønne...</summary> </entry> <entry><title>Context Collapse, Part 1 - Poisoning Copilot Memory</title><link href="https://enklypesalt.com/posts/context-collapse-part1-poisoning-copilot-memory/" rel="alternate" type="text/html" title="Context Collapse, Part 1 - Poisoning Copilot Memory" /><published>2026-06-22T08:00:00+02:00</published> <updated>2026-07-28T14:27:01+02:00</updated> <id>https://enklypesalt.com/posts/context-collapse-part1-poisoning-copilot-memory/</id> <content type="text/html" src="https://enklypesalt.com/posts/context-collapse-part1-poisoning-copilot-memory/" /> <author> <name>Håkon Måløy</name> </author> <category term="Coordinated Vulnerability Disclosures" /> <summary>I would like to thank Microsoft product teams and Microsoft Security Response Center (MSRC) for collaborating with me on this technical analysis and mitigation of the disclosed vulnerabilities. The editorial opinions reflected below are solely the author’s and do not necessarily reflect those of the organizations I collaborated with. Summary The findings described in this post are part of a ...</summary> </entry> <entry><title>Hvor bra passer modellen?</title><link href="https://enklypesalt.com/posts/how-good-is-the-fit/" rel="alternate" type="text/html" title="Hvor bra passer modellen?" /><published>2026-01-16T05:57:00+01:00</published> <updated>2026-07-10T12:04:55+02:00</updated> <id>https://enklypesalt.com/posts/how-good-is-the-fit/</id> <content type="text/html" src="https://enklypesalt.com/posts/how-good-is-the-fit/" /> <author> <name>Håkon Måløy</name> </author> <category term="Maskinlæring og statistikk" /> <summary>I forrige post kom vi ett steg nærmere å ha en modell som kan hjelpe oss med å avgjøre om vi kan la en person ta karusellen vår kun ved å måle personens vekt og bruke denne til å forutsi hvor høy personen er. Men vi er fortsatt ikke helt fornøyde. La oss først, igjen, ta en kikk på dataen vår: Her plotter vi vekt (x-aksen) mot høyde (y-aksen). I forrige post fant vi at linjen $Y = 71.94 + 1....</summary> </entry> </feed>
