Open weights are quietly closing up – and that's a problem
7 points
1/21/1970
13 hours ago
by maxloh
Comments
anthonycoslett
12 hours ago
gnabgib
Original source please, that just links to: https://martinalderson.com/posts/open-weights-are-quietly-cl...
Which was submitted 10 hours ago.. perhaps upvote that submission? https://news.ycombinator.com/item?id=48036924
12 hours ago
goodmythical
I was genuinely confused to be linked from a community discussion of a link to another community discussion of a link.
Let's see...if I linked to a lobsters post that links to tumblr post that links to a personal blog that links to a substack that links an arxive of a blog that mentions an arxiv that...
It just feels as if there can be only one reason to link to the lobsters discussion instead of the content at hand, no?
9 hours ago
samdhar
[flagged]
11 hours ago
Interesting article and I'm a big proponent of sounding the alarm bells even if it might seem the noise is being swallowed by the vacuum of space. I'd love to see your take on ways in which we can avoid the outcome of a complete siloing of open weights - a "and here's what we can do or need to do about it before it's too late." I'd also love to see your take on the fact that enterprise organizations are seemingly shifting to open weights so ironically could be the industry's salvation (fine tuning massive open models - Cursor with Kimi 2, etc.) or other examples? Perhaps you might find it's not so bleak! Here's to hoping.