char-fork/llama.cpp

git clone https://git.t4t.associates/char-fork/llama.cpp

Johannes GäßlerCUDA: MMQ code deduplication + iquant support (#8495)69c487f4e

char/aliceai-t5-moe
139 B5 linesraw
1// This file has been autogenerated by generate_cu_files.py, do not edit manually.
2
3#include "../mmq.cuh"
4
5DECL_MMQ_CASE(GGML_TYPE_IQ2_S);