sxjscience commented on a change in pull request #16465: [Gluon] [Fix] [WIP]
Fix HybridBlock when hybridize is not called
URL: https://github.com/apache/incubator-mxnet/pull/16465#discussion_r334711549
##########
File path: python/mxnet/gluon/block.py
##########
@@ -1054,34 +1098,16 @@ def register_op_hook(self, callback,
monitor_all=False):
def forward(self, x, *args):
"""Defines the forward computation. Arguments can be either
:py:class:`NDArray` or :py:class:`Symbol`."""
- flatten_args = _flatten([x] + list(args), 'inputs')[0]
- is_ndarray = None
- ctx = None
- exist_sym_nd = False
- for ele in flatten_args:
- if isinstance(ele, NDArray):
- if is_ndarray is False:
- raise ValueError('In HybridBlock, we do not support mixed
NDArrays and Symbols'
- ' types for the input.\n'
- 'Received types are: {}.'
- .format([type(ele) for ele in
flatten_args]))
- is_ndarray = True
- exist_sym_nd = True
- ctx = ele.context
Review comment:
Actually, all ctxs are supposed to be the same. For example, we should not
allow the mixing of cpu and gpu contexts. However, we currently allow to do so
because we will need to mix `cpu`, `cpu_pinned`, and `cpu_shared`.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to 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
