ONE - On-device Neural Engine
|
Decribe how to build a (convolution) filter from a tensor. More...
#include <FilterCodec.h>
Public Member Functions | |
virtual | ~FilterEncoder ()=default |
virtual FilterShape | shape (const TensorShape &shape) const =0 |
virtual TensorIndex | value (const FilterIndex &index) const =0 |
Decribe how to build a (convolution) filter from a tensor.
Let us assume that "enc" is a filter encoder.
Given a tensor "inp" and its shape "inp.shape", "enc" builds a filter "out" as follows:
for each valid filter index (referred to as filter_index below) for enc.shape(inp.shape) out.at(filter_index) = inp.at(enc.value(filter_index))
Definition at line 42 of file FilterCodec.h.
|
virtualdefault |
|
pure virtual |
|
pure virtual |
Implemented in loco::PermutingEncoder< Domain::Filter >.