fmthoker / skeleton-contrast

41 stars 5 forks source link

Question about downstream tasks #4

Closed czhaneva closed 2 years ago

czhaneva commented 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?

fmthoker commented 2 years ago

We use single encoder representations during the downstream for all experiments except Table 10 (in appendix).