ggml
CUDA/HIP: honor GGML_PREC_F32 in the flash-attention tile kernel
#1536

Open

CUDA/HIP: honor GGML_PREC_F32 in the flash-attention tile kernel #1536

RapidMark wants to merge 1 commit into ggml-org:master from CloudhandsAI:cloudhands/fattn-tile-prec-f32

CUDA/HIP: honor GGML_PREC_F32 in the flash-attention tile kernel

7ba34e6d

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Milestone

No milestone