I am attempting to load a model that I trained locally to QC Cloud, but am having some difficulties deserializing it. Following the instructions in the docs here, I am doing something like::

  1. import pickle
  2. import base64
  4. # Stuff to be saved
  5. MyClass:
  6.     def __init__(self):
  7.         pass
  8.     def foo(self):
  9.         return 1
  10. myObj = MyClass()
  11. myDat = {'a' : 1}
  12. stuff = {'myObj' : myObj, 'myDat' : myDat }
  14. # Serialize as string
  15. stuff_serialized = pickle.dumps( stuff )
  16. stuff_serialized = base64.b64encode( stuff_serialized ).decode('utf-8')
  18. # Save to local ObjectStore
  19. self.ObjectStore.Save('Stuff', stuff_serialized)
  21. ## Upload the file to Dropbox, then on QC Cloud:
  23. # Download the file
  24. stuff_serialized = self.Download('PATH_TO_FILE')
  26. # Deserialize
  27. stuff_serialized = base64.b64decode(stuff_serialized.encode('utf-8'))
  28. stuff = pickle.loads(stuff_serialized)
+ Expand

But I am receiving the error in the pickle.loads step:

  1. could not find MARK

Note that I'm using utf-8 instead of ascii as in the docs due to some invalid character errors. 

Any suggestions?

Author

Adam W

June 2023