3 releases
0.1.5-alpha.0 | Mar 25, 2023 |
---|---|
0.1.4-alpha.0 | Mar 3, 2023 |
0.1.3-alpha.0 | Mar 2, 2023 |
#37 in #digital-signal-processing
700KB
5K
SLoC
The caffe2op-accum
crate provides an operator
for accumulating values in networked digital
signal processing and deep learning
computations. This crate defines the
AccumulateOp
which performs accumulation on the
input tensors according to a specified axis. The
resulting tensor is returned as the output. The
Accumulated
token is used to refer to the output
tensor.
Note: This crate is currently being translated from C++ to Rust, and some function bodies may still be in the process of translation.
The operation performed by AccumulateOp
can be
represented mathematically as follows:
If the input tensor is of shape [a, b, c, ..., m, n]
and the accumulation axis is k
, then the
output tensor will be of shape [a, b, c, ..., l, n]
where l
is the size of the tensor along the
accumulation axis. The elements of the output
tensor are calculated as follows:
output[i1, i2, ..., il, j]
= sum(input[i1, i2, ..., i_{k-1}, l, i_{k+1}, ..., i_{m}, j])
where the sum is over all l
along the
accumulation axis.
The caffe2op-accum
crate also provides tokens
such as Accumulation
, accumulates
,
accumulations
, depends
, fiddles
, interim
,
and reshaped
, which are used within the
implementation of the accumulation operation.
Dependencies
~36MB
~406K SLoC