Gluon Reshape does in-place for NDArray and out-of-place for Symbol

[gautham-kollu]

Description

Gluon Reshape() behaves differently for NDArray and Symbol.
NDArray -> Does the reshape IN-place
Symbol -> Does the reshape Out-of-place

Error Message

(Paste the complete error message. Please also include stack trace by setting environment variable DMLC_LOG_STACK_TRACE_DEPTH=10 before running your script.)

To Reproduce

import mxnet as mx
from mxnet.gluon import nn

class HybridNet(nn.HybridBlock):
def init(self, **kwargs):
super(HybridNet, self).init(**kwargs)

def hybrid_forward(self, F, x):
    y = x.reshape(-1)
    y = F.square(y, out = y)
    return x+y

model = HybridNet()
model.initialize()

x = mx.nd.array([1,2,3])
print(model(x))

model.hybridize()
x = mx.nd.array([1,2,3])
print(model(x))

Running the above on CPU

[ 2. 8. 18.]
<NDArray 3 @cpu(0)>

[ 2. 6. 12.]
<NDArray 3 @cpu(0)>

Steps to reproduce

(Paste the commands you ran that produced the error.)

What have you tried to solve it?

Environment

We recommend using our script for collecting the diagnositc information. Run the following command and paste the outputs below:

curl --retry 10 -s https://raw.githubusercontent.com/dmlc/gluon-nlp/master/tools/diagnose.py | python

# paste outputs here

[szha]

In Gluon one should not specify where the output is stored using out= argument. In the future, Gluon should automatically detect and warn (or throw execption) on such case.

cc @leezu

[gautham-kollu]

Would Gluon not support in-place ops ?

[leezu]

Not when hybridizing. Ideally the framework can optimize the computational plan in the background and decide when in-place can be used.