alimagic opened a new issue, #13062:
URL: https://github.com/apache/tvm/issues/13062
In python code, use some List[torch.Tensor] = [], then the mode convert from
torch to onnx， show in graph using SequenceEmpty. And now, use code as follow
in xmir/python/tvm/relay/frontend/onnx.py:
class SequenceEmpty(OnnxOpConverter):
"""Operator converter for sequence construction op."""
@classmethod
def _impl_v11(cls, inputs, attr, params):
# Construct a tuple from input tensors.
# const0 = _expr.const(0, dtype="int64")
return _expr.Tuple([])
However， error occur at"get_const_tuple", as follow:
def get_var(name, val, scan=False):
checked_type = infer_type(val)
if hasattr(checked_type, "type_annotation"):
checked_type = checked_type.type_annotation
if hasattr(checked_type, "checked_type"):
checked_type = checked_type.checked_type
shape = get_const_tuple(checked_type.shape)
I print the "val" and "checked_type" and all is show "()". So，has anyone
encountered a similar situation?
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
