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

pack_filter

def pack_filter(filter: TileTensor[Storage=filter.Storage, address_space=filter.address_space, linear_idx_type=filter.linear_idx_type], packed_filter: TileTensor[Storage=packed_filter.Storage, address_space=packed_filter.address_space, linear_idx_type=packed_filter.linear_idx_type], num_groups: Int)

This packs the filter form RSCF to FRSCf. Use the default micro kernel size for dynamic shapes.

Args:

filter (TileTensor[Storage=filter.Storage, address_space=filter.address_space, linear_idx_type=filter.linear_idx_type]): The unpacked filter tensor in RSCF layout.
packed_filter (TileTensor[Storage=packed_filter.Storage, address_space=packed_filter.address_space, linear_idx_type=packed_filter.linear_idx_type]): The destination tensor for the packed filter in FRSCf layout.
num_groups (Int): Number of convolution groups for grouped convolution.