How to import a module from a py file from gdrive in colabratory?

Question

I have a .py file which is located in my gdrive here:

content/gdrive/My Drive/Colab Notebooks/Data/pySEN.py

I am trying to import module ESN from this file and I fail.
My code is as follows:

from google.colab import drive 
drive.mount('/content/gdrive')
sys.path.append('/content/gdrive/My Drive/Colab Notebooks/Data/')
from pySEN import ESN

I am getting:

ModuleNotFoundError: No module named 'pySEN'

I tried this solutions and failed:
Importing a python module from the python script from drive in colab Is it possible to import my own modules into a google-colaboratory notebook?

Please advice. thanks

Can you please tell us the output of `! ls "/content/gdrive/My Drive/Colab Notebooks/Data"` — Anubhav Singh, Jul 02 '19 at 11:01

Anubhav Singh · Answer 1 · 2019-07-02T12:34:28.807

0

So, the output of ! ls "/content/gdrive/My Drive/Colab Notebooks/Data" is

20190618.csv.gz order_books_data_archive.zip pyESN.py

This means that pySEN.py actually doesnot exists. It's actually pyESN.py which you want to import. So, replace from pySEN import ESN with from pyESN import ESN

Check this:

Folder structure on Google Drive:

Code:

edited Jul 02 '19 at 12:34

answered Jul 02 '19 at 10:25

Anubhav Singh

8,321
4
25
43

that's a 'typo'. still error after fixed - ImportError: cannot import name 'ESN' – Giladbi Jul 02 '19 at 12:18
Have you checked the output of `! ls "/content/gdrive/My Drive/Colab Notebooks/Data"` this time ? – Anubhav Singh Jul 02 '19 at 12:19
If everything is right as you are saying, then check if `ESN` module actually exists in the file `pyESN.py` or not. – Anubhav Singh Jul 02 '19 at 12:26
@Giladbi, check my updated explaination with code and output results. – Anubhav Singh Jul 02 '19 at 12:35
class ESN(): def __init__(self, n_inputs, n_outputs, n_reservoir=200, spectral_radius=0.95, sparsity=0, noise=0.001, input_shift=None, input_scaling=None, teacher_forcing=True, feedback_scaling=None, teacher_scaling=None, teacher_shift=None, out_activation=identity, inverse_out_activation=identity, random_state=None, silent=True): """ – Giladbi Jul 02 '19 at 12:36
----> 3 from pyESn import ESN ModuleNotFoundError: No module named 'pyESn' – Giladbi Jul 02 '19 at 12:38
Isn't it pyESN ? – Anubhav Singh Jul 02 '19 at 12:42
the .py file is pyESN.py and the class inside is ESN – Giladbi Jul 02 '19 at 12:47
But, in the above error `----> 3 from pyESn import ESN ModuleNotFoundError: No module named 'pyESn' `, here filename is `pyESn.py`, not `pyESN.py`. So, correct your code. – Anubhav Singh Jul 02 '19 at 12:52
fixed again and stil error: ----> 3 from pyESN import ESN ImportError: cannot import name 'ESN' – Giladbi Jul 02 '19 at 13:05

score 0 · Answer 2 · answered Aug 13 '21 at 16:26

Not sure this is still helpful, but I had the very same problem, I was trying to load pyESN by doing:

!git clone https://github.com/cknd/pyESN.git
!cp /content/pyESN/pyESN.py .
from pyESN import ESN

And I got the very same

ModuleNotFoundError: No module named 'pyESN'

I loaded pyESN on google drive, mounted the drive and followed the methods above, same result.

Eventually I restarted the Runtime, and now it seems to be working. In particular if I do:

import pyESN
help(pyESN)

I get:

Help on module pyESN:

NAME
    pyESN

CLASSES
    builtins.object
        ESN
    
    class ESN(builtins.object)
     |  ESN(n_inputs, n_outputs, n_reservoir=200, spectral_radius=0.95, sparsity=0, noise=0.001, input_shift=None, input_scaling=None, teacher_forcing=True, feedback_scaling=None, teacher_scaling=None, teacher_shift=None, out_activation=<function identity at 0x7f16e6596b00>, inverse_out_activation=<function identity at 0x7f16e6596b00>, random_state=None, silent=True)
     |  
     |  Methods defined here:
     |  
     |  __init__(self, n_inputs, n_outputs, n_reservoir=200, spectral_radius=0.95, sparsity=0, noise=0.001, input_shift=None, input_scaling=None, teacher_forcing=True, feedback_scaling=None, teacher_scaling=None, teacher_shift=None, out_activation=<function identity at 0x7f16e6596b00>, inverse_out_activation=<function identity at 0x7f16e6596b00>, random_state=None, silent=True)
     |      Args:
     |          n_inputs: nr of input dimensions
     |          n_outputs: nr of output dimensions
     |          n_reservoir: nr of reservoir neurons
     |          spectral_radius: spectral radius of the recurrent weight matrix
     |          sparsity: proportion of recurrent weights set to zero
     |          noise: noise added to each neuron (regularization)
     |          input_shift: scalar or vector of length n_inputs to add to each
     |                      input dimension before feeding it to the network.
     |          input_scaling: scalar or vector of length n_inputs to multiply
     |                      with each input dimension before feeding it to the netw.
     |          teacher_forcing: if True, feed the target back into output units
     |          teacher_scaling: factor applied to the target signal
     |          teacher_shift: additive term applied to the target signal
     |          out_activation: output activation function (applied to the readout)
     |          inverse_out_activation: inverse of the output activation function
     |          random_state: positive integer seed, np.rand.RandomState object,
     |                        or None to use numpy's builting RandomState.
     |          silent: supress messages
     |  
     |  fit(self, inputs, outputs, inspect=False)
     |      Collect the network's reaction to training data, train readout weights.
     |      
     |      Args:
     |          inputs: array of dimensions (N_training_samples x n_inputs)
     |          outputs: array of dimension (N_training_samples x n_outputs)
     |          inspect: show a visualisation of the collected reservoir states
     |      
     |      Returns:
     |          the network's output on the training data, using the trained weights
     |  
     |  initweights(self)
     |  
     |  predict(self, inputs, continuation=True)
     |      Apply the learned weights to the network's reactions to new input.
     |      
     |      Args:
     |          inputs: array of dimensions (N_test_samples x n_inputs)
     |          continuation: if True, start the network from the last training state
     |      
     |      Returns:
     |          Array of output activations
     |  
     |  ----------------------------------------------------------------------
     |  Data descriptors defined here:
     |  
     |  __dict__
     |      dictionary for instance variables (if defined)
     |  
     |  __weakref__
     |      list of weak references to the object (if defined)

FUNCTIONS
    correct_dimensions(s, targetlength)
        checks the dimensionality of some numeric argument s, broadcasts it
           to the specified length if possible.
        
        Args:
            s: None, scalar or 1D array
            targetlength: expected length of s
        
        Returns:
            None if s is None, else numpy vector of length targetlength
    
    identity(x)

FILE
    /content/pyESN.py

Notice the

FILE
    /content/pyESN.py

at the end, which to me means that my method eventually worked.

I have no idea how reproducible this solution is, but it really sound just a problem with Google Colab

How to import a module from a py file from gdrive in colabratory?

2 Answers2