TY  - RPRT
TI  - How Infinitely Wide Neural Networks Can Benefit from Multi-task Learning -- an Exact Macroscopic Characterization
AU  - Jakob Heiss
AU  - Josef Teichmann
AU  - Hanna Wutte
PY  - 2022
DO  - 10.3929/ethz-b-000550890
UR  - https://arxiv.org/abs/2112.15577
ID  - 2112.15577
ER  -