llama.cpp
21c84b5d - CUDA: fix Volta FlashAttention logic (#11615)
Commit
View On GitHub
Commit
145 days ago
CUDA: fix Volta FlashAttention logic (#11615)
Parents
d92cb67e
  • ggml/src/ggml-cuda
    • File
      fattn-wmma-f16.cu
    • File
      fattn.cu
FAQ Terms Privacy Refunds Impressum