For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /max/get-started.md).

Mojo function

dispatch_mask

def dispatch_mask[mask_type: String, callback_fn: def[mask_t: MHAMask](mask: mask_t) raises capturing thin -> None, local_window_size: Int = Int(-1)]()

Instantiate an MHAMask by name and invoke a callback with it.

Resolves the mask string to one of the built-in MHAMask implementations at compile time and calls callback_fn with a concrete mask instance. This lets callers write mask-agnostic kernels while still specialising the generated code per mask type.

Parameters:

mask_type (String): Name of the mask (e.g. "causal", "null", "sliding_window_causal"). Must match one of the MaskName constants.
callback_fn (def[mask_t: MHAMask](mask: mask_t) raises capturing thin -> None): Parametric callback invoked with the resolved mask.
local_window_size (Int): Sliding-window or chunk size for masks that require it. Must be -1 for masks that ignore it and positive for masks that require it.

Raises:

Compile-time assertion error if mask_type is unrecognised or if local_window_size is inconsistent with the selected mask.