I have been looking for the solution for this error for a whole morning. I created an separate environment for python 3.6 and I still got this error. I am using anacondas. So i am so frustrated.
ModuleNotFoundError: No module named 'mxnet'
from gluonts.model.deepar import DeepAREstimator from gluonts.trainer import Trainer import mxnet as mx import numpy as np np.random.seed(7) mx.random.seed(7) estimator = DeepAREstimator( prediction_length=28, context_length=100, freq='H', trainer=Trainer(ctx="gpu", # remove if running on windows epochs=5, learning_rate=1e-3, num_batches_per_epoch=100 ) ) predictor = estimator.train(train_ds) --------------------------------------------------------------------------- ModuleNotFoundError Traceback (most recent call last) <ipython-input-14-d803033f31d5> in <module> ----> 1 from gluonts.model.deepar import DeepAREstimator 2 from gluonts.trainer import Trainer 3 import mxnet as mx 4 import numpy as np 5 ~\miniconda3\envs\deepar\lib\site-packages\gluonts\model\deepar\__init__.py in <module> 12 # permissions and limitations under the License. 13 ---> 14 from ._estimator import DeepAREstimator 15 16 __all__ = ["DeepAREstimator"] ~\miniconda3\envs\deepar\lib\site-packages\gluonts\model\deepar\_estimator.py in <module> 16 17 import numpy as np ---> 18 from mxnet.gluon import HybridBlock 19 20 from gluonts.core.component import DType, validated ModuleNotFoundError: No module named 'mxnet' I installed both gluonts successfully. then I tried to install mxnet by conda install mxnet, there are so many conflicts, not sure why, here are just part of conflicts. Thanks for your help
Package notebook conflicts for: widgetsnbextension -> notebook[version='>=4.4.1'] jupyter -> notebook ipywidgets -> widgetsnbextension[version='>=3.5.0,<3.6.0'] -> notebook[version='>=4.4.1'] Package matplotlib-base conflicts for: matplotlib -> matplotlib-base[version='3.1.2|3.1.2|3.1.2|3.1.3|3.1.3|3.1.3|>=3.2.1,<3.2.2.0a0|>=3.2.2,<3.2.3.0a0|>=3.3.1,<3.3.2.0a0|>=3.4.3,<3.4.4.0a0|>=3.4.2,<3.4.3.0a0|>=3.3.4,<3.3.5.0a0|>=3.3.2,<3.3.3.0a0',build='py36h64f37c6_1|py36h64f37c6_0|py37h64f37c6_0|py38h64f37c6_0|py37h64f37c6_1|py38h64f37c6_1'] gluonts -> matplotlib~=3.0 -> matplotlib-base[version='3.1.2|3.1.2|3.1.2|3.1.3|3.1.3|3.1.3|>=3.2.1,<3.2.2.0a0|>=3.2.2,<3.2.3.0a0|>=3.3.1,<3.3.2.0a0|>=3.4.3,<3.4.4.0a0|>=3.4.2,<3.4.3.0a0|>=3.3.4,<3.3.5.0a0|>=3.3.2,<3.3.3.0a0',build='py36h64f37c6_1|py36h64f37c6_0|py37h64f37c6_0|py38h64f37c6_0|py37h64f37c6_1|py38h64f37c6_1'] Package pandocfilters conflicts for: notebook -> nbconvert -> pandocfilters[version='>=1.4.1'] nbconvert -> pandocfilters[version='>=1.4.1'] jupyter -> nbconvert -> pandocfilters[version='>=1.4.1'] Package fonttools conflicts for: matplotlib-base -> fonttools[version='>=4.22.0'] matplotlib -> matplotlib-base[version='>=3.4.3,<3.4.4.0a0'] -> fonttools[version='>=4.22.0'] Package pywinpty conflicts for: notebook -> terminado[version='>=0.8.1'] -> pywinpty terminado -> pywinpty Package sip conflicts for: matplotlib -> pyqt -> sip[version='4.18.*|>=4.19.4|>=4.19.4,<=4.19.8|4.19.13.*|>=4.19.13,<=4.19.14'] pyqt -> sip[version='4.18.*|>=4.19.4|>=4.19.4,<=4.19.8|4.19.13.*|>=4.19.13,<=4.19.14'] qtconsole -> pyqt -> sip[version='4.18.*|>=4.19.4|>=4.19.4,<=4.19.8|4.19.13.*|>=4.19.13,<=4.19.14'] Package scandir conflicts for: importlib_metadata -> pathlib2 -> scandir testpath -> pathlib2 -> scandir ipython -> pathlib2 -> scandir pickleshare -> pathlib2 -> scandir Package olefile conflicts for: pillow -> olefile matplotlib-base -> pillow[version='>=6.2.0'] -> olefile Package pandoc conflicts for: nbconvert -> pandoc[version='>=1.12.1|>=1.12.1,<2.0.0'] jupyter -> nbconvert -> pandoc[version='>=1.12.1|>=1.12.1,<2.0.0'] notebook -> nbconvert -> pandoc[version='>=1.12.1|>=1.12.1,<2.0.0'] Package async_generator conflicts for: nbclient -> async_generator nbconvert -> nbclient[version='>=0.5.0,<0.6.0'] -> async_generator 2 Answers
Conda is more usable when we want to install something that is not written in python. It is not the case in Mxnet. I would suggest using pip install for libraries in python.
You may take a look at this link to better understand how to use conda environments.
What is the difference between pip and conda?
Also here's the official documentation of anaconda:
You may go through these. One common trick to understand if we need pip install or conda install, we check if the source code of the library in question is written in python or in another language.
If it is in python, then it's highly recommended to use pip install
Otherwise, conda install.
Here, in it has been mentioned that the latest version of Mxnet works in all python version from 3.5 onwards, so pip install mxnet should work.
use pip install mxnet. don't use conda install mxnet. if there is an error about permission, then use pip install mxnet --user. It worked for me.
