char-fork/llama.cpp

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

Johannes GäßlerCUDA: refactor and deduplicate vector FA kernels (#16208)75a3a6c2c

char/aliceai-t5-moe
brancheschar/aliceai-t5-moeall refs →
287 B7 linesraw
1// This file has been autogenerated by generate_cu_files.py, do not edit manually.
2
3#include "../fattn-vec.cuh"
4
5DECL_FATTN_VEC_CASE( 64, GGML_TYPE_Q4_0, GGML_TYPE_Q4_1);
6DECL_FATTN_VEC_CASE(128, GGML_TYPE_Q4_0, GGML_TYPE_Q4_1);
7DECL_FATTN_VEC_CASE(256, GGML_TYPE_Q4_0, GGML_TYPE_Q4_1);