Are Transformers Able to Reason by Connecting Separated Knowledge in Training Data?

Yutong Yin, Zhaoran Wang

Browse the full ICLR paper archive.