llama.cpp
Make IQ1_M work for QK_K = 64
#6327
Merged

Make IQ1_M work for QK_K = 64 #6327

ikawrakow merged 3 commits into master from ik/iq1m_64
ikawrakow
iq1_m: make it work for QK_K = 64 (WIP)
e1939bc8
iq1_m: make it work for QK_K = 64 (scalar and AVX2)
5c953a1a
iq1_m: QK_K = 64 seems to work on Metal and ARM_NEON
b0d0bdd0
ggerganov
ggerganov approved these changes on 2024-03-27
ikawrakow ikawrakow merged cbc83436 into master 1 year ago
ikawrakow ikawrakow deleted the ik/iq1m_64 branch 1 year ago
ikawrakow
ggerganov
SomeoneSerge
mscheong01

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
Milestone