Hacker News
new
past
comments
ask
show
jobs
points
by
bicepjai
5 hours ago
|
comments
by
esseph
4 hours ago
|
next
[-]
Who was training LLMs on residential AMD GPUs?
reply
by
MrDrMcCoy
2 hours ago
|
parent
|
[-]
I am. Most models that can be run locally have extraordinarily out of date training data for certain frameworks that predate breaking changes and important features, so their docs and API schemas need to be put into LoRAs.
reply
by
amelius
5 hours ago
|
prev
|
[-]
can't or can?
reply