Y
Hacker News
new
|
ask
|
show
|
jobs
by
viraptor
1174 days ago
This can be then quantized to the llama.cpp/gpt4all format, right? Specifically, this only tweaks the existing weights slightly, without changing the structure?