Why do I get the error "The value of 'CheckpointPath' is invalid. Checkpoint path must be an existing directory with write permission or empty."?

Hi everyone!

I got some issue when I run this code:

checkpointPath = pwd;
options = trainingOptions('adam', ...
    'InitialLearnRate',0.1, ...
    'MaxEpochs',20, ...
    'Verbose',false, ...
    'Plots','training-progress', ...
    'Shuffle','every-epoch', ...
    'CheckpointPath',checkpointPath);

Matlab says that some error occurs:

Error using nnet.cnn.TrainingOptionsADAM
The value of 'CheckpointPath' is invalid. Checkpoint path must be an existing directory with write permission or empty.
Error in trainingOptions (line 342)
opts = nnet.cnn.TrainingOptionsADAM(varargin{:});

I can save .mat file at this path, but I get this error when I use the function trainingOptions with it. I can run this code on C drive without error but cannot do that on the other drives. Thanks everyone.

[Stephen23]

What OS? What value does PWD() return?

Is there anything special about that folder (e.g. cloud drive, installation folder, etc).

[Duc-Toan]

Thanks for replying, bro.

1/ I am using matlab 2022b on win10 and installed matlab on drive "I".

2/ This folder is synchronized with google cloud.

3/ I can run this code on any other drive except the folders on drive "I" (including folders not synchronized with cloud).

4/ The return value of pwd is as follows 'I:\Desktop\work data\Programming\Matlab\Neural network\Neural network for classification'

Answer 1

[Manish]

Hi,

I understand that you'd like to save the checkpoints in the current working directory. However, the issue you're encountering might be due to write permissions on your C drive.

As a workaround, you can create an empty folder in the current working directory. Once the folder is created, use its path as the value for the CheckpointPath.

Here is the sample code:

path = "C:\Users\Demo"; % Demo is the new empty folder
options = trainingOptions("sgdm", ...
    MaxEpochs=15, ...
    InitialLearnRate=0.01, ...
    Plots="training-progress", ...
    Metrics="accuracy", ...
     CheckpointPath=path,...
    Verbose=0);

Hope this helps!