Impossible do write model

[gustavosol]

Dear all, thank you in advance.

I'm trying to write the model using ann and svm objects but it's being impossible. The path is correct but is not working. I also tried a suggestion you've made in another issue closed before. This is the massage: error: ml.svm: unable to write model to path: /Users/ML.model

I'm using two Mac computers and the error is basically the same: Max 8.1.8 in Catalina v.10.15.7 Max 8.1.5 in High Sierra v.10.13.6

SVMFaceOSC_Biofilia_Face_SIGRaDi_04_11_2020_WORKING@.maxpat.zip

Any help would be great! Thanks Sol

[NiccoloGranieri]

Hi @gustavosol, I just downloaded your patch and successfully wrote and read a saved model with ml.svm. Could you please try to change your path to somewhere inside your User folder? For example, for me something like this worked like a charm.

/Users/Nicco/Desktop

You should be then able to see your ml.data file appear in the location, and you should be able to read it by using the correct name in the read message. Please let me know if you encounter any other errors.

[gustavosol]

Dear Nocco

I did your suggestion with the same result. I've sent you a video (via wetransfer to this email) showing what is happening. Thank you in advance. It seems to regard Catalina, no?

Sol

On Thu, Nov 19, 2020 at 5:24 AM Niccolò notifications@github.com wrote:

Hi @gustavosol https://github.com/gustavosol, I just downloaded your patch and successfully wrote and read a saved model with ml.svm. Could you please try to change your path to somewhere inside your User folder? For example, for me something like this worked like a charm.

/Users/Nicco/Desktop

You should be then able to see your ml.data file appear in the location, and you should be able to read it by using the correct name in the read message. Please let me know if you encounter any other errors.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/irllabs/ml-lib/issues/171#issuecomment-730210754, or unsubscribe https://github.com/notifications/unsubscribe-auth/AC7TZZ6QGT7JQZDK3UJCFSLSQTI3FANCNFSM4T2TPZGA .

-- http://www.facebook.com/gustavo.sol.77 https://www.gustavosol.com.br/

Gustavo Sol 011-9614-6692

[NiccoloGranieri]

Hi @gustavosol, I have not received your video but I am finally able to reproduce your bug. I was not encountering the same error whilst trying to save the training data, but once the model is trained I encounter your same issue if I try to save the model. I will proceed with some extensive testing and let you know as soon as we fix the issue.

In the meantime, the best solution is to save the untrained data, and train before using the object. Sorry for the inconvenience.

[gustavosol]

No worries, thanks for the update! Yesterday I was trying other solutions reinstalling older versions of Max and the behavior was even more wierd, nearly what you described. rsrsrs So, how could I save it without using the object? The idea is to be able to reinsert the collected data again to train, without using the read command? Anyway, here is the link for the video showing the problem if you need: https://youtu.be/mf5W_D1d3Yw Thank you for all your help. Sol

On Fri, Nov 20, 2020 at 5:50 AM Niccolò notifications@github.com wrote:

Hi @gustavosol https://github.com/gustavosol, I have not received your video but I am finally able to reproduce your bug. I was not encountering the same error whilst trying to save the training data, but once the model is trained I encounter your same issue if I try to save the model. I will proceed with some extensive testing and let you know as soon as we fix the issue.

In the meantime, the best solution is to save the untrained data, and train before using the object. Sorry for the inconvenience.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/irllabs/ml-lib/issues/171#issuecomment-731032933, or unsubscribe https://github.com/notifications/unsubscribe-auth/AC7TZZZU5RJF7O6EYKYHRZDSQYUV7ANCNFSM4T2TPZGA .

-- http://www.facebook.com/gustavo.sol.77 https://www.gustavosol.com.br/

Gustavo Sol 011-9614-6692

[NiccoloGranieri]

Hi @gustavosol, the solution in the meantime is to do exactly the same thing but before training the model. This will create a ".data" file that can be read with a traditional (read) message. The only inconvenience is that using this method, the model needs to be re-trained everything with the saved data.

[gustavosol]

Thanks for the tip! I'll try and wait for the news. All the best

On Fri, Nov 20, 2020 at 4:31 PM Niccolò notifications@github.com wrote:

Hi @gustavosol https://github.com/gustavosol, the solution in the meantime is to do exactly the same thing but before training the model. This will create a ".data" file that can be read with a traditional (read) message. The only inconvenience is that using this method, the model needs to be re-trained everything with the saved data.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/irllabs/ml-lib/issues/171#issuecomment-731365980, or unsubscribe https://github.com/notifications/unsubscribe-auth/AC7TZZ4VJN3SYXOIIVMV4CTSQ27YTANCNFSM4T2TPZGA .

-- http://www.facebook.com/gustavo.sol.77 https://www.gustavosol.com.br/

Gustavo Sol 011-9614-6692

[i-n-g-o]

176 should fix this

[jamiebullock]

Closing as this fix is now in the public release.