2 ms·
This can be then quantized to the llama.cpp/gpt4all format, right? Specifically, this only tweaks the existing weights slightly, without changing the structure?
by viraptor 4y 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?