For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /get-started.md).
Mojo function
max_pool
def max_pool[dtype: DType, int_type: DType, target: StringSpan[ImmStaticOrigin] = StringSpan("cpu")](input: TileTensor[dtype, Storage=input.Storage, address_space=input.address_space, linear_idx_type=input.linear_idx_type], filter: TileTensor[int_type, Storage=filter.Storage, address_space=filter.address_space, linear_idx_type=filter.linear_idx_type], strides: TileTensor[int_type, Storage=strides.Storage, address_space=strides.address_space, linear_idx_type=strides.linear_idx_type], dilations: TileTensor[int_type, Storage=dilations.Storage, address_space=dilations.address_space, linear_idx_type=dilations.linear_idx_type], paddings: TileTensor[int_type, Storage=paddings.Storage, address_space=paddings.address_space, linear_idx_type=paddings.linear_idx_type], output: TileTensor[dtype, Storage=output.Storage, address_space=output.address_space, linear_idx_type=output.linear_idx_type], ceil_mode: Bool = False, ctx: Optional[DeviceContext] = None)
Dispatches the max pooling operation to the CPU or GPU backend based on the target.
Parameters:
- dtype (
DType): Data type of the input and output tensors. - int_type (
DType): Data type of the filter, strides, dilations, and paddings tensors. - target (
StringSpan[ImmStaticOrigin]): Execution target, either "cpu" or "gpu".
Args:
- input (
TileTensor[dtype, Storage=input.Storage, address_space=input.address_space, linear_idx_type=input.linear_idx_type]): Batched image input to the pool2d operator. - filter (
TileTensor[int_type, Storage=filter.Storage, address_space=filter.address_space, linear_idx_type=filter.linear_idx_type]): Filter size on height and width dimensions with assumed tuple (filter_h, filter_w). - strides (
TileTensor[int_type, Storage=strides.Storage, address_space=strides.address_space, linear_idx_type=strides.linear_idx_type]): Strides on height and width dimensions with assumed tuple (stride_h, stride_w). - dilations (
TileTensor[int_type, Storage=dilations.Storage, address_space=dilations.address_space, linear_idx_type=dilations.linear_idx_type]): Dilations on height and width dimensions with assumed tuple (dilation_h, dilation_w). - paddings (
TileTensor[int_type, Storage=paddings.Storage, address_space=paddings.address_space, linear_idx_type=paddings.linear_idx_type]): Paddings on height and width dimensions with assumed tuple (pad_h_before, pad_h_after, pad_w_before, pad_w_after)). - output (
TileTensor[dtype, Storage=output.Storage, address_space=output.address_space, linear_idx_type=output.linear_idx_type]): Pre-allocated output tensor space. - ceil_mode (
Bool): Ceiling mode defines the output shape and implicit padding. - ctx (
Optional[DeviceContext]): The DeviceContext to use for GPU execution.