[jit] RuntimeError: v->type()->isSubtypeOf(elem_type) #20500
Description
🐛 Bug
To Reproduce
This code:
class MyMod2(torch.jit.ScriptModule):
__constants__ = ['mean', 'std']
def __init__(self):
super(MyMod2, self).__init__()
self.mean = [0.485, 0.456, 0.406]
self.std = [0.229, 0.224, 0.225]
@torch.jit.script_method
def forward(self, input):
mean = torch.tensor(self.mean)
std = torch.tensor(self.std)
return input.sub(mean[:, None, None]).div_(std[:, None, None])yields the following error:
Traceback (most recent call last):
File "C:/Development/_PytorchDev/pytorch-box2pix/jit.py", line 75, in <module>
test = MyMod2()
File "C:\Python36\lib\site-packages\torch\jit\__init__.py", line 1047, in init_then_register
_create_methods_from_stubs(self, methods)
File "C:\Python36\lib\site-packages\torch\jit\__init__.py", line 1012, in _create_methods_from_stubs
self._c._create_methods(self, defs, rcbs, defaults)
RuntimeError: v->type()->isSubtypeOf(elem_type) ASSERT FAILED at ..\torch\csrc\jit\ir.cpp:1272, please report a bug to PyTorch. (createList at ..\torch\csrc\jit\ir.cpp:1272)
(no backtrace available)
Expected behavior
Should work without a problem.
Environment
- PyTorch Version: 1.1.0