whisper.cpp/ggml
Johannes Gäßler d052e64d42 CUDA: batched+noncont MMQ, refactor bs>1 MoE code (llama/13199) 2025-05-01 13:29:02 +03:00
..
cmake
include CUDA: fix q_nope_absorbed prec for DS 2 Lite f16 (llama/13137) 2025-05-01 13:29:02 +03:00
src CUDA: batched+noncont MMQ, refactor bs>1 MoE code (llama/13199) 2025-05-01 13:29:02 +03:00
.gitignore
CMakeLists.txt whisper : add check that target name exists (#3103) 2025-05-01 10:05:24 +02:00