pub unsafe extern "C" fn ggml_flash_attn_ext(
ctx: *mut ggml_context,
q: *mut ggml_tensor,
k: *mut ggml_tensor,
v: *mut ggml_tensor,
mask: *mut ggml_tensor,
scale: f32,
max_bias: f32,
logit_softcap: f32,
) -> *mut ggml_tensor