+ Start a Discussion
Daniel BallingerDaniel Ballinger 

Failure creating or updating model using feedback

I'm attempting to retrain an evisting vision model using additional feedback. I've also tried creating a new model for the dataset.

In both cases I was using a trainParams of:
'{"withFeedback": true}'
The existing model has a SUCCEEDED status but shows no sign that the feedback was incoperated.

The new model ends up with the failureMsg:
"java.lang.IllegalArgumentException: The dataset example referenced examples/1010273-4e2abc4183fd732aae779a1a3a3cb088de02898ec2fce61aee921df31675a676 does not exist"
Has something got corrupted within my dataset? Should I just start again with a new dataset?
Alain CabonAlain Cabon

It is not the solution at all. I just tried to understand where your problem is (Analytics or Einstein).

I have a minimal knowledge of Einstein vision (trailhead).

Introduction to Salesforce Einstein Vision

JUNE 27, 2017​

NEW Optimize your model using feedback. Use the feedback API to add a misclassified image with the correct label to the dataset from which the model was created.
Use the new API call to add a feedback example. See Create a Feedback Example.
The call to get all examples now has three new query parameters: feedback, upload, and all. Use these query parameters to refine the examples that are returned. See Get All Examples.
The call to train a dataset and create a model now takes the trainParams object {"withFeedback": true}. This option specifies that the feedback examples are used during the training process. By default, the feedback examples aren't used during training if you don't specify this value. See Train a Dataset.

{"withFeedback": true}—Lets you specify that feedback examples are included in the data to be trained to create the model. If you omit this parameter, feedback examples aren't used in training. Available in Einstein Vision API version 2.0 and later.


It seems possible to reproduce your problem with a free org for everyone here (curl requests)

What You Need to Call the API
Before you can access the Einstein Platform Services APIs, you first create an account and download your key. Then you use your key to generate an OAuth token. You can use your key to access both the Einstein Vision and Einstein Language APIs.

Get an account

Generate a token

Daniel BallingerDaniel Ballinger
Yes, I'm using the Train a Dataset (https://metamind.readme.io/docs/train-a-dataset) and Retrain a Dataset (https://metamind.readme.io/docs/retrain-a-dataset) API calls. I'm not specifiying the epochs or learningRate, so they should be the default values. I'm only setting the trainParams as above. 
Michael Machado 22Michael Machado 22
Hi Daniel, That is not a typical error from our API, so I am wondering if this is coming from a different application error.  To continue progressing- you can try reuploading the entire dataset but happy to help you troubleshoot the reason behind the error too.