Implement technique for adapting the auxiliary loss based on the similarity between the main gradient and the auxiliary gradient. The method is described in this paper by Du et. al. I actually like the slight modification described in this [blog post(https://vivien000.github.io/blog/journal/learning-though-auxiliary_tasks.html) by Tran-Thien, so I will be implementing this version.
Implement technique for adapting the auxiliary loss based on the similarity between the main gradient and the auxiliary gradient. The method is described in this paper by Du et. al. I actually like the slight modification described in this [blog post(https://vivien000.github.io/blog/journal/learning-though-auxiliary_tasks.html) by Tran-Thien, so I will be implementing this version.