NER tool throwing 'not enough entities' error

[Manojkumar]

Hi Team,

I am getting the following error when i ran the NER tool configured to train with new entities.

As you can see in the screenshot i have more than 500 entities for that label.

I have attached the workflow too. Please help

Error: " Named Entity Recognition (5) Some labels don’t have enough data points to train an accurate model. Each of these labels has less than 20 data points: ['NAME']."

Thanks

[trsali]

Hi @Manojkumar You may want to increase the Epoch to 10 since that is the recommended option, your workflow has it set to 5.

If you are still running into an issue, you may want to export your workflow into a .yxzp so that the input attachment can be read.

Hope this helps!

[Manojkumar]

Hi @trsali 

Thanks for your response, i did increase the epochs and ran but still ran in to error.

I have the shared the workflow as package and attached it now.

Please let me know if you figure out the issue

[trsali]

Hi @Manojkumar For your workflow to run I found that you need to increase the rows of text streaming into the D anchor, at least 100 rows with matches to run and get a warning, around 300 rows with matches to get the warning to stop (these were just the numbers of rows I experimented with. The NER tool will not work if there are no certain matches under a threshold. Please see my attached workflow.

Hope this helps