Repeating documents N number of times

We have multiple people viewing a set of documents and annotating them via prodigy, these users are given separate sessions . Currently all documents are viewed by all n number of users using prodigy, this is because the feed_overlap configuration is set true. Is there a way to configure prodigy such that, each document is only viewed by m number (where m<n) of users and not all users.
Thank you!

Hi! There's currently no built-in setting for this type of fine-grained overlap, but it's definitely something we want to add (including being able to define percentages of overlap etc.). This is also very relevant for the upcoming Prodigy teams.

One approximate workaround could be to write a custom stream generator that periodically checks the hashes present in the dataset and doesn't send an example out again if it's already present m times. You could then use a low batch size to ensure that answers are sent back to the server sooner and fewer examples are sent out at once, to minimise the risk of sending examples out too often (because the answer is still pending, the hashes haven't been updated yet etc.).