llama.cpp
658987cf - CUDA: noncont MMVQ + batched bs1 MUL_MAT_ID (#13014)

Commit

143 days ago

CUDA: noncont MMVQ + batched bs1 MUL_MAT_ID (#13014) * CUDA: noncont MMVQ + batched bs1 MUL_MAT_ID * fix logic for RoPE support, CUDA graphs

References

Author

JohannesGaessler

Parents