sandeep-krishnamurthy commented on issue #11915: The behavior of np.zeros_like(x) if x is a NDArray is unexpected. URL: https://github.com/apache/incubator-mxnet/issues/11915#issuecomment-408475474 Hello @kohillyang You are mixing Numpy.NDArray and MXNet.NDArray hence, seeing a different behavior. ### Numpy only ``` import numpy as np >>> y = np.zeros(shape=(17,5)) >>> print(y.shape) (17, 5) >>> print(np.zeros_like(y).shape) (17, 5) ``` ### MXNet NDArray only ``` >>> import mxnet.ndarray as nd >>> x = nd.zeros(shape=(17,5)) >>> print(x.shape) (17, 5) >>> print(nd.zeros_like(x).shape) (17, 5) ``` You passed np.zeros_like(MXNet NDarray) which treat it as an array object. (x in your example is mxnet ndarray that cannot be directly understood by Numpy. If you want to use MXNet NDArray inside numpy, here is the modified code for your example: ``` import numpy as np import mxnet.ndarray as nd import mxnet as mx print(mx.__version__) x = nd.zeros(shape=(17,5)) print(x.shape) print(np.zeros_like(x.asnumpy()).shape) # Converting X to numpy print(type(x)) print(nd.zeros_like(x).shape) ```
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
