llama.cpp
Only use Q6_K for output weights if tensor size is multiple of 256
#1932
Merged
Merged
Only use Q6_K for output weights if tensor size is multiple of 256 #1932
ggerganov
merged
2 commits into
master
from
ik/check_Q6_K_is_OK
Only use Q6_K for output weights if tensor size is multiple of 256
510b537d
KerfuffleV2
commented
on 2023-06-19
Fixed copy/paste mistake
986a56e2
ggerganov
approved these changes
on 2023-06-19
Login to write a write a comment.
Login via GitHub