Decodes data using run-length coding.
tfc.ops.run_length_decode(
code,
shape,
run_length_code,
magnitude_code,
use_run_length_for_non_zeros,
name=None
)
This is the inverse operation to RunLengthEncode. The shape of the tensor
that was encoded must be known by the caller.
This op implements a superset of RunLengthGammaDecode, which is equivalent to
calling RunLengthDecode with run_length_code = -1, magnitude_code = -1, and
use_run_length_for_non_zeros = false.
Args |
|---|
code
|
A Tensor of type string.
An encoded scalar string as returned by RunLengthEncode.
|
shape
|
A Tensor of type int32.
An int32 vector giving the shape of the encoded data.
|
run_length_code
|
An int.
If >= 0, use Rice code with this parameter to decode run
lengths, else use Elias gamma code.
|
magnitude_code
|
An int.
If >= 0, use Rice code with this parameter to decode magnitudes,
else use Elias gamma code.
|
use_run_length_for_non_zeros
|
A bool.
If true, alternate between coding run lengths of
zeros and non-zeros. If false, only decode run lengths of zeros, and decode
non-zeros one by one.
|
name
|
A name for the operation (optional).
|
Returns |
|---|
A Tensor of type int32.
An int32 tensor of decoded values, with shape shape.
|
