llama.cpp
Assume tied weights if lm_head/output weights is missing.
#5824
Merged

Loading