Comprehend training input set

Question

We are labelling the pdf documents. We have limited resource. To complete all the labelling task, it might take more time. I am thinking of splitting the number of documents into smaller chunks and create mulitple jobs to complete the entire labelling work.

The output of this task will be feed into comprehend for training. I observed the comprehend will take only 5 jobs as input. Is there a way to increate this input number  ? Or Is there a way to retrain comprehend with next set of 5 jobs. ?

Answer

Hi,

There is a limitation of 10 for concurrent active asynchronous jobs, according to this doc below:
https://docs.aws.amazon.com/comprehend/latest/dg/guidelines-and-limits.html#limits-active-jobs

If you'd like to use the output of the labeled PDF document as the training datasets, you can output the result to a S3 bucket and use it as input for the comprehend training by navigating to it.

Thanks,

Comprehend training input set

Contenuto pertinente