Skip to content
Better HN
Top
Best
Ask
Show
New
Jobs
Search
⌘K
0 points
archerx
5mo ago
0 comments
Save
Share
People already do this with multi agent workflows. I kind of do this with local models, I get a smaller model to do the hard work for speed and use a bigger model to check its work and improve it.
0 comments
1 comments · 1 top-level
top
newest
oldest
lukan
5mo ago
The tech surely has lots of potential, but my point was just, that self improvement does not really work yet unsupervised.
j
/
k
navigate · click thread line to collapse