+ Start a Discussion
Emilien Guichard 40Emilien Guichard 40 

Build a Cat Rescue App That Recognizes Cat Breeds - Create and Train the Dataset error

Hi,

I am stuck at the Create and Train the Dataset step on the Build a Cat Rescue App That Recognizes Cat Breeds Trailhead module.

I have successfully created the lightning component, uploaded the zip file and clicked on the train button but when I try to Verifiy I get the following error message:
"Challenge Not yet complete... here's what's wrong: 
Could not find a successful cat dataset training. The training may take a couple of minutes after the model is created."

Could you please advise ?

Thanks for your help.
Best Answer chosen by Emilien Guichard 40
muenzpraegermuenzpraeger
Let's do a quick'n'dirty approach: delete and re-upload the dataset. Once you see the dataset labels with their respective numbers (like in your screenshot, you may have to "refresh datasets") run the training again. Wait approx. 1-2 min. Then verify the challenge.

All Answers

muenzpraegermuenzpraeger
Can you please post a screenshot of the "Einstein Vision" tab? Also how long did you wait for the training - as the message says it can take a couple of minutes.
Emilien Guichard 40Emilien Guichard 40
Sure, here you go:

User-added image
also I did wait a few hours before tring to verifty the step again...
muenzpraegermuenzpraeger
Let's do a quick'n'dirty approach: delete and re-upload the dataset. Once you see the dataset labels with their respective numbers (like in your screenshot, you may have to "refresh datasets") run the training again. Wait approx. 1-2 min. Then verify the challenge.
This was selected as the best answer
Emilien Guichard 40Emilien Guichard 40
Puurfect! It worked this time, thanks for the help.
MedhanieHabteMedhanieHabte
I too get this error, I also get some lightning component load errors as I upload these dataset.
User-added image

User-added image
Aron Schor 24Aron Schor 24
MedhanieHabte try resaving the components in the prior units...I had the same issue.
MedhanieHabteMedhanieHabte
Works for me!
lela tlela t
Hi, I tried saving back the lighting component, helper, apex class... deleting the dataset, uploading a new one, refreshing, clicking on train, waiting again... nothing! I always get the same error, which is impossible to interpratate. 

Also, I downloaded the zip file and we have those 3 folder 15 files each, so the Bengal folder instead shows up as if they were 14, I checked and there are 2 identical files in that folder ( different name, but same size and possibly they have the same md5 hash! 

I am not sure if this is related somehow to this issue, but it is quite annoying that we do not get a real advice on what is going on in this step.
muenzpraegermuenzpraeger

@lela t:

When you look into the Lightning/Apex code of this step you'll see that a model training is executed. This needs to be finished as it's the precondition for a later image prediction. See also here the documentation for trainings (https://metamind.readme.io/docs/train-a-dataset).

When you execute this code as anonymous Apex code within your Developer Console you'll see in the debug logs the status of the model trainings. The training needs a status of "SUCCEEDED" with a progress of 1.0 (which are the exact counterparts of the response bode in the above mentioned documentation link).
 

EinsteinVision_PredictionService service = new EinsteinVision_PredictionService();
EinsteinVision_Dataset[] datasets = service.getDatasets();
for (EinsteinVision_Dataset dataset : datasets) {
     if (dataset.name.equals('Cats')) {
         EinsteinVision_Model[] models = service.getModels(dataset);
         for (EinsteinVision_Model model : models) {
             System.debug(model);
         }
     }
}
 

The image duplication that you mentioned has no influence on this.
 

BillPowell__cBillPowell__c
Resaved the components, replaced the code with a new copy/paste just in case I missed something, still doesnt work. 
muenzpraegermuenzpraeger
@BillPowell__c What exactly "doesn't work"? If it's the training - what's the output of the Apex code as mentioned in my post from July 3rd?
BillPowell__cBillPowell__c
Getting component errors when trying to import the dataset from the URL. Deleted the dataset (it manages to import after a few minutes anyway), I went in and ensured all the apex code was 100% precise from the module and resaved all, still doesnt work 
muenzpraegermuenzpraeger
@BillPowell__c: Can you post the component error here?
BillPowell__cBillPowell__c
User-added image
muenzpraegermuenzpraeger
A shot in the dark - are you running Lightning Debug mode? If yes - does the error occur when you switch it off?
BillPowell__cBillPowell__c
HA! That did it. Weird. Must have turned it on in a previous module and forgot to turn it off. THANK YOU!
Marcin TrofiniakMarcin Trofiniak

Similar like others i have problem with it. I am doing it same time with my friend who has same problem. We do al the copy/past staff create app upoad zip drom url and in this moment we sometimes get errors sometimes not... After it ulpoads i do train and refresh and ... nothing .... No Labels and can't pass the chalange... I checked debug i checked code ... Can't really find a rison why...

 

Answer after training

 

That what i get after trining... Wed for sometime and refreshing shows nothing new... Help me plis :)

Kyle O'Brien 7Kyle O'Brien 7
When I load the Cats.zip dataset, I get the "success" banner. However, in the developer console log I am receiving the status "attempt to de-reference a null object" - referring to this code snippet from the EinsteinVision_Admin.apxc:

  @AuraEnabled
    public static String trainDataset(Decimal datasetId) {
        EinsteinVision_PredictionService service = new EinsteinVision_PredictionService();
        EinsteinVision_Model model = service.trainDataset(Long.valueOf(String.valueOf(datasetId)), 'Training', 0, 0, '');
        return model.modelId;
    }

Trailhead won't validate that I've finished this step because it can't detect the dataset having been loaded... Any idea what's going on here? I've re-saved all the of components/classes, deleted and re-uploaded the dataset to no avail. Also, despite it saying "success" there is no adjustment to the dataset models or labels columns. 

Thanks!
Carla Silvado 15Carla Silvado 15
I am having an issue with this challenge. My Einstein Vision tab will not load. I can click on it but I get an Internal Error message and then it keeps spinning. Any thoughts?
muenzpraegermuenzpraeger
@Carla Silvado 15 - this can happen when you didn't set up the credentials correctly. Check out the possbile error logs in the Developer Console.
J C Shiva PrasadJ C Shiva Prasad
hey  guys when i put the URl and click create dataset nothing is happening can anyone suggest what must be wrong
 
Angelica Veness 12Angelica Veness 12
I'm getting this error message when I put the 'Einsteinvision_Admin_UI' component onto my new Lightning App. Then the same error if I try to go to the page.
User-added image
muenzpraegermuenzpraeger
@Angelica Veness - Please check if you've set up the einstein_platform.pem file correctly.
Tuananh Vu 9Tuananh Vu 9
i'm getting the same error message like Angelica and i think i did set up the einstein_platform.pem file correctly. Anything else i could do?
David CantorDavid Cantor
I am also getting this error anything else to fix it??