ZheyuYe opened a new issue #17638: [Numpy] unknown type_flag=7 URL: https://github.com/apache/incubator-mxnet/issues/17638 ## Description A series of issues related to `kBool` occured afther the pull #17438 and [#4571](https://github.com/apache/incubator-tvm/pull/4571/) in tvm. This is, after all, a serious problem that makes many of deep numpy's features unusable. Here is a simple error case. ### Error Message ``` --------------------------------------------------------------------------- MXNetError Traceback (most recent call last) <ipython-input-2-48e3c90810f4> in <module> 17 foo = Foo() 18 foo.hybridize() ---> 19 out = foo(mx.np.ones((10,), ctx=mx.gpu())) ~/incubator-mxnet/python/mxnet/gluon/block.py in __call__(self, *args) 680 hook(self, args) 681 --> 682 out = self.forward(*args) 683 684 for hook in self._forward_hooks.values(): ~/incubator-mxnet/python/mxnet/gluon/block.py in forward(self, x, *args) 1175 'Find all contexts = {}'.format(ctx_set)) 1176 with ctx: -> 1177 return self._call_cached_op(x, *args) 1178 with ctx: 1179 try: ~/incubator-mxnet/python/mxnet/gluon/block.py in _call_cached_op(self, *args) 1022 cargs = [args_without_none[i] if is_arg else i.data() 1023 for is_arg, i in self._cached_op_args] -> 1024 out = self._cached_op(*cargs) 1025 if isinstance(out, NDArray): 1026 out = [out] ~/incubator-mxnet/python/mxnet/_ctypes/ndarray.py in __call__(self, *args, **kwargs) 167 ctypes.byref(num_output), 168 ctypes.byref(output_vars), --> 169 ctypes.byref(out_stypes))) 170 171 if original_output is not None: ~/incubator-mxnet/python/mxnet/base.py in check_call(ret) 244 """ 245 if ret != 0: --> 246 raise get_last_ffi_error() 247 248 MXNetError: Traceback (most recent call last): File "../src/nnvm/plan_memory.cc", line 58 MXNetError: unknown type_flag=7 ``` ## To Reproduce ``` import mxnet as mx import numpy as np from numpy.testing import assert_allclose from mxnet.gluon import HybridBlock mx.npx.set_np() class Foo(HybridBlock): def __init__(self, prefix=None, params=None): super(Foo, self).__init__(prefix=prefix, params=params) def hybrid_forward(self, F, valid_length): mask = (F.np.ones((10,)) < valid_length).astype(np.float32) mask2 = (F.np.ones((10,)) < valid_length).astype(np.float32) mask = mask * F.np.expand_dims(mask2, axis=-1) return mask foo = Foo() foo.hybridize() out = foo(mx.np.ones((10,), ctx=mx.gpu())) ``` ## Comments @sxjscience @yzhliu
---------------------------------------------------------------- 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
