ALR$^2$: A Retrieve-then-Reason Framework for Long-context Question Answering

Open in new window