Unknown input/output type: %s of type %s.

Package:

torch

50580

Exception Class:

TypeError

Raise code

        rectified.append(blob)
            else:
                raise TypeError(
                    "I/O blob #{} of unsupported type: {} of type {}"
                    .format(len(rectified), str(blob), type(blob)))
        return rectified
    else:
        raise TypeError(
            "Unknown input/output type: %s of type %s." %
            (str(blobs), type(blobs))
        )


def CreateOperator(
    operator_type,
    inpu

Links to the raise (1)

https://github.com/pytorch/pytorch/blob/e56d3b023818f54553f2dc5d30b6b7aaf6b6a325/caffe2/python/core.py#L357

Ways to fix

Error code:

from caffe2.python import core
op = core.CreateOperator(
  "GaussianFill",
  5,               #Error occurs because of 5 is integer 
  ["Z"],
  shape=[100, 100], # shape argument as a list of ints.
  mean=1.0, # mean as a single float
  std=1.0, # std as a single float
)
print(str(op))

CreateOperator gets string,binary,list, tuple as an input/output arguments. In the above we try to assign an integer as an input, That's why we get an error.

Fix code:

from caffe2.python import core
op = core.CreateOperator(
  "GaussianFill",
  [], # GaussianFill does not need any parameters.
  ["Z"],
  shape=[100, 100], # shape argument as a list of ints.
  mean=1.0, # mean as a single float
  std=1.0, # std as a single float
)
print(str(op))

Jun 05, 2021

anonim 13.0k

Add a possible fix

Please authorize to post fix