How to create a pipeline?

[ext-o]

Question about creating a pipeline in AI Center… As “Input dataset” is supposed to place the original data (without any labeling), right? And what should be placed as a “Evaluation dataset”. I tried placing as a Evaluation set the whole export folder with its content (after the original data has been trained), but it gives me a failed status => “No data found”. I appreciate any help with this.

[Anil_G]

@ext-o

Evaluation data will genrally be the data containing the input required again…

You can as well run only training pipeline…

Depending on what model you are using you can send the evaluation dataset…

Say for text classification model…a csv file with teo columns description and the category is needed…

So accordingly depending on the model you are using you can choose the datset…

And it is not the output that you need to give for evaluation…even for evaluation you need to provide similar input as training dataset

Cheers

[supermanPunch (Arpan)]

@ext-o

Hi @ext-o ,

The Input dataset chosen is training_data2, but we would require to use the folder under the export folder i.e., Export_dd_iban_....

However, It is not very clear as to what you want to perform, If you would want to just Perform an Evaluation Pipeline, you don’t need to select the Full Pipeline option. If you are trying to perform both Training and Evaluation Pipelines then you would need to input the Training input Dataset as well.

Also note that the Input Dataset provided for Training pipeline should not be the same as the Dataset to be supplied for Evaluation Pipeline, if you would require to get a better evaluation results (It can be the same, However we’ll not be able to understand the Model accuracy better).

Let us know if the configuration does not work, We however do not have a full view of the Configuration done that made you get the Errored Output. It would be helpful if you’ll be able to provide that as well.