Unified Enhancement of the Generalization and Robustness of Language Models via Bi-Stage Optimization

Yudao Sun, Juan Yin, Juan Zhao +3

Neural network language models (LMs) are confronted with significant challenges in generalization and robustness. Currently, many studies focus on improving either generalization o…