whisper.cpp
3d54b68e - CUDA: noncont MMVQ + batched bs1 MUL_MAT_ID (llama/13014)

Commit
1 year ago
CUDA: noncont MMVQ + batched bs1 MUL_MAT_ID (llama/13014) * CUDA: noncont MMVQ + batched bs1 MUL_MAT_ID * fix logic for RoPE support, CUDA graphs
Committer
Parents
Loading