offline-policy-evaluation
offline-policy-evaluation copied to clipboard
Support running OPE on batches
Currently OPE is conducted on datasets that fit into memory and is done on all the data at once. We should support running OPE on batches of data and combining the results across batches.