Use this endpoint after predict record pairs to generate high-impact training clusters for users to curate. This step is optional and only used when users are still training the model. Tamr Core uses high-impact clusters to signal which clusters are most likely to impact the model in subsequent rounds of learning. Users can filter to high-impact clusters on the Tamr Core UI to review and verify them.
On success, this call returns an operation object describing the operation to generate training clusters and materialize the associated dataset.