Probing Out-of-Distribution Robustness of Language Models with Parameter-Efficient Transfer Learning
This addresses the problem of ensuring language models handle distributional shifts effectively for users relying on robust AI systems, but it is incremental as it builds on existing PETL methods.
The study systematically evaluated how parameter-efficient transfer learning methods and model size affect the out-of-distribution robustness of language models on intention classification tasks, finding that performance varies with techniques and scales.
As the size of the pre-trained language model (PLM) continues to increase, numerous parameter-efficient transfer learning methods have been proposed recently to compensate for the tremendous cost of fine-tuning. Despite the impressive results achieved by large pre-trained language models (PLMs) and various parameter-efficient transfer learning (PETL) methods on sundry benchmarks, it remains unclear if they can handle inputs that have been distributionally shifted effectively. In this study, we systematically explore how the ability to detect out-of-distribution (OOD) changes as the size of the PLM grows or the transfer methods are altered. Specifically, we evaluated various PETL techniques, including fine-tuning, Adapter, LoRA, and prefix-tuning, on three different intention classification tasks, each utilizing various language models with different scales.