Closed czhaneva closed 2 years ago
Thanks for your awesome work, I'd like to know does this method use multi encoder when doing down stream tasks? or just use one encoder?
We use single encoder representations during the downstream for all experiments except Table 10 (in appendix).
Thanks for your awesome work, I'd like to know does this method use multi encoder when doing down stream tasks? or just use one encoder?