pytoda.datasets.utils.background_tensors module

Summary

Functions:

constant_value_tensor	Returns a background tensor filled with a constant value.
range_tensor	Returns a background tensor filled with a given range of values.

Reference

range_tensor(value_range, repetitions)

Returns a background tensor filled with a given range of values.

Parameters

• value_range (Any) – Range of values to insert into each row of the background tensor.

• repetitions (Tuple[int]) – The number of repetitions of value_range along each axis.

Returns

Tensor containing repetitions of the given range of values along specified axes.

Example, value_range = [0,1,2], repetitions = (1,2) will repeat [0,1,2] once along dim 0 and twice along dim 1, i.e, tensor([[0,1,2,0,1,2]]) of size (1,6) will be the output.

NOTE: if a pattern [0,1,2] is required to fill a tensor of shape (2,5)

specify value_range as [0,1,2,0,1] and repetiitons as (2,). The value_range is filled ‘row-wise’. Simply transpose the output for a ‘column-wise’ fill.

Return type

Tensor

constant_value_tensor(value, shape)

Returns a background tensor filled with a constant value.

Parameters

• value (float) – Value to fill the background tensor with.

• shape (Tuple) – Shape of the background tensor.

Returns

Tensor of given shape filled with the given constant value.

Return type

Tensor