TY  - RPRT
TI  - Multi-armed bandits for resource efficient, online optimization of language model pre-training: the use case of dynamic masking
AU  - Iñigo Urteaga
AU  - Moulay-Zaïdane Draïdia
AU  - Tomer Lancewicki
AU  - Shahram Khadivi
PY  - 2023
UR  - https://arxiv.org/abs/2203.13151
ID  - 2203.13151
ER  -