5,337
edits
Line 571: | Line 571: | ||
Generally two-stage models perform better. Everything is trained end-to-end. | Generally two-stage models perform better. Everything is trained end-to-end. | ||
===Homework=== | |||
Read these 4 papers and also read the professor's online hard mining paper (OHEM). | |||
==Will be on the exam== | ==Will be on the exam== |