TY  - RPRT
TI  - Improving task-specific representation via 1M unlabelled images without any extra knowledge
AU  - Aayush Bansal
PY  - 2020
UR  - https://arxiv.org/abs/2006.13919
ID  - 2006.13919
ER  -