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_shape

def max_pool_shape[dtype: DType, int_type: DType](input: ManagedTensorSlice[IOSpec[_, _].Input, static_spec=input.static_spec], filter: ManagedTensorSlice[IOSpec[_, _].Input, static_spec=filter.static_spec], strides: ManagedTensorSlice[IOSpec[_, _].Input, static_spec=strides.static_spec], dilations: ManagedTensorSlice[IOSpec[_, _].Input, static_spec=dilations.static_spec], paddings: ManagedTensorSlice[IOSpec[_, _].Input, static_spec=paddings.static_spec]) -> IndexList[Int(4)]

Computes the output shape for the mo.max_pool graph op.

Parameters:

• ​dtype (DType): Element type of the pooling input tensor.
• ​int_type (DType): Element type of the filter, strides, dilations, and paddings tensors.

Args:

• ​input (ManagedTensorSlice[IOSpec[_, _].Input, static_spec=input.static_spec]): Rank-4 batched image input to the pooling operator.
• ​filter (ManagedTensorSlice[IOSpec[_, _].Input, static_spec=filter.static_spec]): One-dimensional tensor of filter sizes on the height and width dimensions, (filter_h, filter_w).
• ​strides (ManagedTensorSlice[IOSpec[_, _].Input, static_spec=strides.static_spec]): One-dimensional tensor of strides on the height and width dimensions, (stride_h, stride_w).
• ​dilations (ManagedTensorSlice[IOSpec[_, _].Input, static_spec=dilations.static_spec]): One-dimensional tensor of dilations on the height and width dimensions, (dilation_h, dilation_w).
• ​paddings (ManagedTensorSlice[IOSpec[_, _].Input, static_spec=paddings.static_spec]): One-dimensional tensor of paddings on the height and width dimensions, (pad_h_before, pad_h_after, pad_w_before, pad_w_after).

Returns:

IndexList[Int(4)]: The output shape of the max pooling operation.