Why I am getting the following AttributeError in Python? - python-2.7

Question

I am using the sklearn's GradientBoostingRegression method. So after fitting it with 2000 estimators, I wanted to add more estimators to it. Since it is taking too long to rerun the entire fitting process, I used the set_params() method. Note that it is a multi-target problem, meaning, I have 3 targets to fit. So I am using the following code to add more estimators.
'''parameters: models (list of length 3 in our case )
train_X, train_y [n_samples x 3], test
n_estimators : previous + 500 (default) [additional estimators]
warm_start : True (default)
'''
def addMoreEstimators(train_X, train_y, test, models, n_estimators = 500, warm_start=True):
params = {'n_estimators':n_estimators, 'warm_start':warm_start}
gbm_pred= pd.DataFrame()
for (i,stars),clf in zip(enumerate(['*','**','***']), models):
clf.set_params(**params)
%time clf.fit(train_X.todense(),train_y[stars])
%time gbm_pred[stars] = clf.predict(test.todense())
gbm_pred = gbm_pred.as_matrix()
gbm_dict ={'model': gbm, 'prediction': gbm_pred}
return gbm_dict
Note: the models parameter is a list of 3 fitted models for the 3 targets.
When I ran it for the first time using 2500 (originally I had 2000 estimators), it ran fine and gave me an output.
When, I am running the same function using 3000 estimators, I am getting an AttributeError (see the traceback of the error below). Here the models contained the 3 fitted models. Below is the traceback of the error: (it's kinda long)
AttributeError Traceback (most recent call last)
<ipython-input-104-9418ada3b36f> in <module>()
7 test = val_X_tfidf[:,shortened_col_index],
8 models = models,
----> 9 n_estimators = 3000)
10
11 reduced_features_gbm_pred_3000_2_lr_1_msp_2 = reduced_features_gbm_model_3000_2_lr_1_msp_2['prediction']
<ipython-input-103-e15a4fb70b50> in addMoreEstimators(train_X, train_y, test, models, n_estimators, warm_start)
15
16 clf.set_params(**params)
---> 17 get_ipython().magic(u'time clf.fit(train_X.todense(),train_y[stars])')
18 print 'starting prediction'
19
//anaconda/lib/python2.7/site-packages/IPython/core/interactiveshell.pyc in magic(self, arg_s)
2305 magic_name, _, magic_arg_s = arg_s.partition(' ')
2306 magic_name = magic_name.lstrip(prefilter.ESC_MAGIC)
-> 2307 return self.run_line_magic(magic_name, magic_arg_s)
2308
2309 #-------------------------------------------------------------------------
//anaconda/lib/python2.7/site-packages/IPython/core/interactiveshell.pyc in run_line_magic(self, magic_name, line)
2226 kwargs['local_ns'] = sys._getframe(stack_depth).f_locals
2227 with self.builtin_trap:
-> 2228 result = fn(*args,**kwargs)
2229 return result
2230
//anaconda/lib/python2.7/site-packages/IPython/core/magics/execution.pyc in time(self, line, cell, local_ns)
//anaconda/lib/python2.7/site-packages/IPython/core/magic.pyc in <lambda>(f, *a, **k)
191 # but it's overkill for just that one bit of state.
192 def magic_deco(arg):
--> 193 call = lambda f, *a, **k: f(*a, **k)
194
195 if callable(arg):
//anaconda/lib/python2.7/site-packages/IPython/core/magics/execution.pyc in time(self, line, cell, local_ns)
1160 if mode=='eval':
1161 st = clock2()
-> 1162 out = eval(code, glob, local_ns)
1163 end = clock2()
1164 else:
<timed eval> in <module>()
//anaconda/lib/python2.7/site-packages/sklearn/ensemble/gradient_boosting.pyc in fit(self, X, y, sample_weight, monitor)
973 self.estimators_.shape[0]))
974 begin_at_stage = self.estimators_.shape[0]
--> 975 y_pred = self._decision_function(X)
976 self._resize_state()
977
//anaconda/lib/python2.7/site-packages/sklearn/ensemble/gradient_boosting.pyc in _decision_function(self, X)
1080 # not doing input validation.
1081 score = self._init_decision_function(X)
-> 1082 predict_stages(self.estimators_, X, self.learning_rate, score)
1083 return score
1084
sklearn/ensemble/_gradient_boosting.pyx in sklearn.ensemble._gradient_boosting.predict_stages (sklearn/ensemble/_gradient_boosting.c:2502)()
AttributeError: 'int' object has no attribute 'tree_'
Sorry for the long traceback, but I think it wouldn't be possible to provide me with meaningful feedback.
Again, why am I getting this feedback ?
Any help would be greatly appreciated.
Thanks
Edit
Below is the code that generates the models that was one of the inputs in the above function.
from sklearn import ensemble
def updated_runGBM(train_X, train_y, test,
n_estimators =100,
max_depth = 1,
min_samples_split=1,
learning_rate=0.01,
loss= 'ls',
warm_start=True):
'''train_X : n_samples x m_features
train_y : n_samples x k_targets (multiple targets allowed)
test : n_samples x m_features
warm_start : True (originally the default is False, but I want to add trees)
'''
params = {'n_estimators': n_estimators, 'max_depth': max_depth, 'min_samples_split': min_samples_split,
'learning_rate': learning_rate, 'loss': loss,'warm_start':warm_start}
gbm1 = ensemble.GradientBoostingRegressor(**params)
gbm2 = ensemble.GradientBoostingRegressor(**params)
gbm3 = ensemble.GradientBoostingRegressor(**params)
gbm = [gbm1,gbm2,gbm3]
gbm_pred= pd.DataFrame()
for (i,stars),clf in zip(enumerate(['*','**','***']), gbm):
%time clf.fit(train_X.todense(),train_y[stars])
%time gbm_pred[stars] = clf.predict(test.todense())
gbm_pred = gbm_pred.as_matrix()
gbm_pred = np.clip(gbm_pred,0,np.inf)
gbm_dict ={'model': gbm, 'prediction': gbm_pred}
return gbm_dict
NOTE In the code above, I have removed some of the print statements to reduce clutter.
These are the two functions I am using, nothing else (apart from the code to split up the data).