View Source EXLA.Lib (EXLA v0.4.1)

High-level operations built on top of EXLA.Op.

Link to this section Summary

Functions

Computes the argmax of the given operation.

Computes the argmin of the given operation.

Sorts a tensor and returns the corresponding indices in the new positions.

Builds iota along axis.

Returns a maximum value scalar operator for the given type.

Returns a minimum value scalar operator for the given type.

Element-wise tangent function.

Link to this section Functions

Link to this function

argmax(builder, op, opts \\ [])

View Source

Computes the argmax of the given operation.

options

Options

  • :axis - the axis to reduce on
  • :keep_axis - whether or not to keep reduced axis
  • :tie_break - how to break ties
Link to this function

argmin(builder, op, opts \\ [])

View Source

Computes the argmin of the given operation.

options

Options

  • :axis - the axis to reduce on
  • :keep_axis - whether or not to keep reduced axis
  • :tie_break - how to break ties
Link to this function

argsort(builder, operand, dimension, comparator, iota_type)

View Source

Sorts a tensor and returns the corresponding indices in the new positions.

Link to this function

iota(builder, shape, axis)

View Source

Builds iota along axis.

Link to this function

max_number(builder, type)

View Source

Returns a maximum value scalar operator for the given type.

Maximum values are defined in Nx.Type.max_finite_binary/1.

Link to this function

min_number(builder, type)

View Source

Returns a minimum value scalar operator for the given type.

It will be negative infinity for floating point types.

Element-wise tangent function.