Hacker News
new
past
comments
ask
show
jobs
points
by
taskylizard
13 hours ago
|
comments
by
13 hours ago
|
next
[-]
deleted
reply
by
lossolo
13 hours ago
|
prev
|
[-]
It's fine tuned Kimi, they didn't train it from scratch.
reply
by
YmiYugy
7 hours ago
|
parent
|
[-]
Sure, but so what? It seems that model size and RL are the determining factors these days.
reply