diff options
| author |  Allen Lavoie <allenl@google.com> | 2018-07-30 11:28:32 -0700 |
|---|---|---|
| committer |  TensorFlower Gardener <gardener@tensorflow.org> | 2018-07-30 11:32:37 -0700 |
| commit | 7bab62beeb5d3cabfee9be1f62e3f57d6788a658 (patch) | |
| tree | b27227cff4459e8dbe9515cca5ab5858d1b23248 /tensorflow/contrib/timeseries | |
| parent | b3df17e99995cce2c0c3c7a84d6d113a8965b36a (diff) | |
TFTS: axis=<list> -> axis=<tuple> in numpy.squeeze
Apparently lists and tuples behave differently in some numpy versions.
PiperOrigin-RevId: 206616759
Diffstat (limited to 'tensorflow/contrib/timeseries')
| -rw-r--r-- | tensorflow/contrib/timeseries/examples/multivariate.py | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/tensorflow/contrib/timeseries/examples/multivariate.py b/tensorflow/contrib/timeseries/examples/multivariate.py index ed799542fd..e81cb18ad7 100644 --- a/tensorflow/contrib/timeseries/examples/multivariate.py +++ b/tensorflow/contrib/timeseries/examples/multivariate.py @@ -80,8 +80,8 @@ def multivariate_train_and_sample( session=session, steps=1)) next_sample = numpy.random.multivariate_normal( # Squeeze out the batch and series length dimensions (both 1). - mean=numpy.squeeze(current_prediction["mean"], axis=[0, 1]), - cov=numpy.squeeze(current_prediction["covariance"], axis=[0, 1])) + mean=numpy.squeeze(current_prediction["mean"], axis=(0, 1)), + cov=numpy.squeeze(current_prediction["covariance"], axis=(0, 1))) # Update model state so that future predictions are conditional on the # value we just sampled. filtering_features = { |