Classifier object corrupted when compressed with joblib.

classifier_train.py

from sklearn.linear_model import SGDClassifier
from sklearn.externals import joblib


#load data vectors (vectorized with Tfidf) and target array
def train(documents,target):
  vectorizer = TfidfVectorizer(stop_words='english', ngram_range=(1,2),
                smooth_idf=True, sublinear_tf=True, max_df=0.5, token_pattern=ur'\b(?!\d)\w\w+\b', use_idf=False)
  data_vectors = vectorizer.fit_transform(documents)
  
  joblib.dump("vectorizer.joblib", compress=9) 
  joblib.dump("data_vectors.joblib", compress=9)
  joblib.dump("target.joblib", compress=9) 
  #all the vectors above load correctly, can be used with uncompressed classifier

  #Fit using SGD
  clf = SGDClassifier(loss="log", n_iter=35, alpha=0.00001)
  clf.fit(data_vectors, target)
  joblib.dump(clf, "compressedclassifier.joblib", compress=6)

predict.py

from sklearn.externals import joblib

#load the vectorizer and classifier

vectorizer = joblib.load("vectorizer.joblib")
clf = joblib.load("compressedclassifier.joblib")

#Vectorize the document to predict
with open("topredict.txt", "rb") as f:
  doc=f.read()
document_vector = vectorizer.transform(doc)

prediction = clf.predict(document_vector)

#process further using the category determined

python

$ ipython
Python 2.6.6 (r266:84292, Sep 11 2012, 08:34:23) 
Type "copyright", "credits" or "license" for more information.

IPython 0.13 -- An enhanced Interactive Python.
?         -> Introduction and overview of IPython's features.
%quickref -> Quick reference.
help      -> Python's own help system.
object?   -> Details about 'object', use 'object??' for extra details.

In [1]: import sklearn

In [2]: sklearn.__version__
Out[2]: '0.14-git'

In [3]: from sklearn.externals import joblib

In [4]: clf=joblib.load("compressedclassifier.joblib")
---------------------------------------------------------------------------
AssertionError                            Traceback (most recent call last)
<ipython-input-5-ad6b23335871> in <module>()
----> 1 clf=joblib.load("compressedclassifier.joblib")

/home/n7/newenv/lib/python2.6/site-
packages/sklearn/externals/joblib/numpy_pickle.pyc in load(filename, mmap_mode)
    422 
    423     try:
--> 424         obj = unpickler.load()
    425     finally:
    426         if hasattr(unpickler, 'file_handle'):

/usr/lib64/python2.6/pickle.pyc in load(self)
    856             while 1:
    857                 key = read(1)
--> 858                 dispatch[key](self)
    859         except _Stop, stopinst:
    860             return stopinst.value

/home/n7/newenv/lib/python2.6/site-
packages/sklearn/externals/joblib/numpy_pickle.pyc in load_build(self)
    291                         "but numpy didn't import correctly")
    292             nd_array_wrapper = self.stack.pop()
--> 293             array = nd_array_wrapper.read(self)
    294             self.stack.append(array)
    295 

/home/n7/newenv/lib/python2.6/site-
packages/sklearn/externals/joblib/numpy_pickle.pyc in read(self, unpickler)
    157         filename = os.path.join(unpickler._dirname, self.filename)
    158         array = 
unpickler.np.core.multiarray._reconstruct(*self.init_args)
--> 159         data = read_zfile(open(filename, 'rb'))
    160         state = self.state + (data,)
    161         array.__setstate__(state)

/home/n7/newenv/lib/python2.6/site-
packages/sklearn/externals/joblib/numpy_pickle.pyc in read_zfile(file_handle)
     69     assert len(data) == length, (
     70         "Incorrect data length while decompressing %s."
---> 71         "The file could be corrupted." % file_handle)
     72     return data
     73 

AssertionError: Incorrect data length while decompressing <open file 
'compressedclassifier.joblib_01.npy.z', mode 'rb' at 0x2d5b5d0>.The file could 
be corrupted.

In [5]:

v Further Details

Vectorizer:
TfidfVectorizer(analyzer=word, binary=False, charset=utf-8,
        charset_error=strict, dtype=<type 'long'>, input=content,
        lowercase=True, max_df=0.5, max_features=None, max_n=None,
        min_df=2, min_n=None, ngram_range=(1, 2), norm=l2,
        preprocessor=None, smooth_idf=True, stop_words=english,
        strip_accents=None, sublinear_tf=True,
        token_pattern=\b(?!\d)\w\w+\b, tokenizer=None, use_idf=False,
        vocabulary=None).

Number of samples in training data: 12440
Number of features: 484762

Classifier:
SGDClassifier(alpha=1e-05, class_weight=None, epsilon=0.1, eta0=0.0,
       fit_intercept=True, l1_ratio=0.15, learning_rate=optimal, loss=log,
       n_iter=35, n_jobs=1, penalty=l2, power_t=0.5, random_state=None,
       rho=None, shuffle=False, verbose=0, warm_start=False)