Pdf Automation AI Center On-Prem Orchestrator Learning Help

Hi Team

I have committed into pdf automation projects. I want some clarity about the AI Center Process.
If possible share the link for step by step procedure to built the automation

Totally I have 50 pdf files, First labelled the twenty documents that time I create the Datasets, train Pipelines, and ML Skills. At the same projects, I have added the another 30 documents and imported, Create the new train pipeline and ML Packages but not create a ML Skills due to enable the auto update options.

If I tried the full pipeline, It shows the error logs as ML Package Issue and failed the full pipeline but the train pipeline was successful.

I need help for successfully complete the full pipeline. Kindly guide me. How to resolve the conflict.

Thanks in advance

@Iswarya_P1

Can you please tell the error you are getting…

Also did you relabel the documents?

Which dataset did you uplod the new files?

Did you create a separate dataset for evaluation?we cannot use same for train and evaluation

Cheers

  1. I have create one dataset
  2. Yes, I relabeled the documents
  3. No, I have used the same dataset for Train path and evaluation path. guide me what should I mentioned in the train dataset and what should I mentioned in the evaluation dataset

Attached the screenshot of error logs, while using the full pipeline.

@Iswarya_P1

Create one more dataset similar to train dataset and you will have a check box for selecting evaluation dataset in it…upload all or few documents in it and use that as evaluation dataset when running full pipeline

we should not use same dataset for bot

cheers

From this reply, I shared my understanding. kindly correct if it was anything wrong.

In dataset option, need to create the another dataset like Datasets–> create new → New dataset creation–>upload files(some raw pdf documents must be uploaded here).

Then, We create the full pipeline.

kindly rectify the above flows.

@Iswarya_P1

Yes this is correct…one more thing is once dataset is created you will have a check box in that dataset select it to make it an evaluation set

cheers


I have shared the screenshots for full pipeline creation as well as evaluation dataset upload files details. Could you please verify and correct the flows and could you please guide me how I select the check box in a dataset using the screenshot

Thanks in advance

@Iswarya_P1

I dont have access to ai center as of now…click on edit dataset then you can find it.or dhow a screenshot…

Remaining everything is as is as the training dataset

Also evaluation pipeline is not mandatory…as the name suggest it is to evaluvate the trained skill and show the results

Cheers

This topic was automatically closed 3 days after the last reply. New replies are no longer allowed.