SplaXBERT: Leveraging Mixed Precision Training and Context Splitting for Question Answering

Zhu Yufan, Hao Zeyu, Li Siqi +1

SplaXBERT, built on ALBERT-xlarge with context-splitting and mixed precision training, achieves high efficiency in question-answering tasks on lengthy texts. Tested on SQuAD v1.1,…