'_SingleProcessDataLoaderIter' object has no attribute '_num_workers'

[tiaotiaosong]

Environment:

• NNI version:2.0
• NNI mode (local|remote|pai):local
• Client OS:MAC
• Server OS (for remote mode only):
• Python version:3.7
• PyTorch/TensorFlow version:
• Is conda/virtualenv/venv used?:
• Is running in Docker?:

Log message:

• nnimanager.log:
• dispatcher.log:
• nnictl stdout and stderr: /Users/admin/anaconda3/lib/python3.7/site-packages/nni/algorithms/feature_engineering/gradient_selector/learnability.py:118: RuntimeWarning: divide by zero encountered in true_divide a = coeff / scipy.special.binom(Nminibatch, np.arange(coeff.size) + 2) Traceback (most recent call last): File "/Users/admin/PycharmProjects/sd/first_breach_case/test.py", line 37, in fgs.fit(X_train, y_train) File "/Users/admin/anaconda3/lib/python3.7/site-packages/nni/algorithms/feature_engineering/gradient_selector/gradient_selector.py", line 260, in fit self._fit(X, y, groups=groups) File "/Users/admin/anaconda3/lib/python3.7/site-packages/nni/algorithms/feature_engineering/gradient_selector/gradient_selector.py", line 318, in _fit soft_groups=groups if self.soft_grouping else None) File "/Users/admin/anaconda3/lib/python3.7/site-packages/nni/algorithms/feature_engineering/gradient_selector/fgtrain.py", line 202, in _train S.train(f_stop=f_stop, f_callback=f_callback) File "/Users/admin/anaconda3/lib/python3.7/site-packages/nni/algorithms/feature_engineering/gradient_selector/learnability.py", line 468, in train xsub, ysub = next(dataloader_iterator) File "/Users/admin/anaconda3/lib/python3.7/site-packages/nni/algorithms/feature_engineering/gradient_selector/fginitialize.py", line 600, in next if self.iter._num_workers == 0: # same-process loading AttributeError: '_SingleProcessDataLoaderIter' object has no attribute '_num_workers'

What issue meet, what's expected?:

How to reproduce it?:

Additional information:

[tiaotiaosong]

`from sklearn.model_selection import train_test_split

from nni.algorithms.feature_engineering.gradient_selector import FeatureGradientSelector

x, y = [[1, 0.4], [1, 1], [2, 2], [2, 3], [1, 2], [1, 3]], [0, 1, 0, 1, 1, 1]

X_train, X_test, y_train, y_test = train_test_split( np.array(x), np.array(y), test_size=0.33, random_state=42 ) fgs = FeatureGradientSelector(n_features=1) fgs.fit(X_train, y_train) print(fgs.get_selected_features())`

[scarlett2018]

@tiaotiaosong - are you still facing the issue with latest nni release? thanks.