The class implements a layer that repeats the input sequence several times.
void SetRepeatCount(int count);
Sets the number of repetitions. For example: you set it to 2
, a {0,3,7}
sequence will be transformed into {0,3,7,0,3,7}
.
The layer has no trainable parameters.
The single input accepts a blob with a sequence of objects, of the dimensions:
BatchLength
is the sequence lengthBatchWidth * ListSize
is the number of sequences in the setHeight * Width * Depth * Channels
is the size of each object in the sequence
The single output returns a blob with the results, of the dimensions:
BatchLength
isGetRepeatCount()
times larger than the inputBatchLength
.- the other dimensions are equal to the corresponding input dimensions.