samskalicky edited a comment on issue #19354: URL: https://github.com/apache/incubator-mxnet/issues/19354#issuecomment-708849369
It further compounds in Gluon where creating the new Parameter doesnt set the dtype explicitly: https://github.com/apache/incubator-mxnet/blob/6729cf3bf4edd6837b0feb6417691ce2e00dbcee/python/mxnet/gluon/block.py#L1002-L1003 This leads to the default dtype `mx_real_t` which is float32: https://github.com/apache/incubator-mxnet/blob/ce1e68260eb3cf9219ae4d59df8bdde7361802ec/python/mxnet/gluon/parameter.py#L106 with this error: ``` Traceback (most recent call last): File "test_subgraph.py", line 155, in <module> test("myProp") File "test_subgraph.py", line 113, in test clear=False, backend_opts={'dedup_subgraph':True}) File "/home/ubuntu/incubator-mxnet/python/mxnet/gluon/block.py", line 1119, in optimize_for self._build_cache(x, *args) File "/home/ubuntu/incubator-mxnet/python/mxnet/gluon/block.py", line 1003, in _build_cache param._load_init(param_data, args[0].context) File "/home/ubuntu/incubator-mxnet/python/mxnet/gluon/parameter.py", line 302, in _load_init self.name, str(self.dtype), str(data.dtype)) AssertionError: Failed loading Parameter '_op0_input' from saved params: dtype incompatible expected <class 'numpy.float32'> vs saved <class 'numpy.float16'>. Set cast_dtype=True to cast the dtype of saved params. ``` ---------------------------------------------------------------- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
