在 sci-kit 中保存和加载模型优化

Question

I used Bayesian Optimization with gp_minimize and a Random search with dummy_minimize in the skopt package.

I wanted to save both models by the load option in scikit-opt:

skopt.dump(gp_100, 'gp_100.pkl')

According to my knowledge, it should be possible to load the model with the following line:

gp_100 = skopt.load('gp_100.pkl')

Unfortunately, this raises an error:

<ipython-input-18-a24ad5907175> in <module>
      1 from skopt import load
----> 2 gpall_1000 = skopt.load('gp_100.pkl')

/opt/conda/lib/python3.7/site-packages/skopt/utils.py in load(filename, **kwargs)
    170         Reconstructed OptimizeResult instance.
    171     """
--> 172     return load_(filename, **kwargs)
    173 
    174 

/opt/conda/lib/python3.7/site-packages/joblib/numpy_pickle.py in load(filename, mmap_mode)
    603                     return load_compatibility(fobj)
    604 
--> 605                 obj = _unpickle(fobj, filename, mmap_mode)
    606 
    607     return obj

/opt/conda/lib/python3.7/site-packages/joblib/numpy_pickle.py in _unpickle(fobj, filename, mmap_mode)
    527     obj = None
    528     try:
--> 529         obj = unpickler.load()
    530         if unpickler.compat_mode:
    531             warnings.warn("The file '%s' has been generated with a "

/opt/conda/lib/python3.7/pickle.py in load(self)
   1083                     raise EOFError
   1084                 assert isinstance(key, bytes_types)
-> 1085                 dispatch[key[0]](self)
   1086         except _Stop as stopinst:
   1087             return stopinst.value

/opt/conda/lib/python3.7/pickle.py in load_global(self)
   1371         module = self.readline()[:-1].decode("utf-8")
   1372         name = self.readline()[:-1].decode("utf-8")
-> 1373         klass = self.find_class(module, name)
   1374         self.append(klass)
   1375     dispatch[GLOBAL[0]] = load_global

/opt/conda/lib/python3.7/pickle.py in find_class(self, module, name)
   1425             return _getattribute(sys.modules[module], name)[0]
   1426         else:
-> 1427             return getattr(sys.modules[module], name)
   1428 
   1429     def load_reduce(self):

AttributeError: module '__main__' has no attribute 'objective'

How to load these models correctly?

I want to load them and plot both via plot_convergence.

Thank you in advance !

Kind regards

Answer 1

Before you call load, you need to import all the libraries and other class definitions that were used in the model that was pickled.

For example, if you call load inside the same function that you call dump (right after the dump), you probably won't get this error because the libraries are already loaded. And so it can identify the attributes in the object loaded.