Phase 3 - DWIE
Created on October 13|Last edited on October 14
Comment
Maybe DWIE gives a different set of results for all our experiments than ontonotes?
CR
10
CRNR
11
Trim 100 LR Sweep
We vary just the LR here.
Add a comment
Priyansh Trivedi •
# LR/ELR
So it seems that DWIE runs prefer encoder learning rate to be 1e-5, and learning rate to vary. We stick with 2e-4, 1e-4 then.
Reply