llama.cpp
llama : use n_embd_head_v instead of n_embd_head_k when reshaping kqv
#7327
Merged
Merged
llama : use n_embd_head_v instead of n_embd_head_k when reshaping kqv #7327
ggerganov
merged
2 commits into
ggml-org:master
from
fairydreaming:llm_build_kqv_fix
llama : use n_embd_head_v instead of n_embd_head_k when reshaping kqv
f15e933f
llama : use n_embd_v_gqa and n_embd_head_v instead of n_embd_k_gqa an…
886f89da
ggerganov
approved these changes
on 2024-05-17
Reviewers
Assignees
No one assigned
Labels
bugfix
Review Complexity : Medium
Milestone
No milestone
Login to write a write a comment.
Login via GitHub