Override SSM_A op for Qwen3 Next to reduce splits #17587
pwilkin
merged
4 commits into
ggml-org:master
from
pwilkin:qwen3_next_ssm_a_override
Override SSM_A op for Qwen3 Next to reduce splits
0a60230c
New tensor mapping SSM_A_NOSCAN for SSM_A used outside of OP_SSM_SCAN…
bac8dede
Update src/llama-model.cpp
e79bae0e
Update src/llama-model.cpp
5f93ef65
Login to write a write a comment.
Login via GitHub