Prompt Perturbation Consistency Learning for Robust Language Models
This addresses robustness issues in LLMs for personal assistant systems, offering an efficient mitigation method, though it is incremental as it builds on fine-tuning and regularization techniques.
The paper tackles the problem of large language models (LLMs) underperforming on sequence labeling tasks like intent classification and slot filling (IC-SF) and being vulnerable to input perturbations, showing that fine-tuning can match discriminative models and proposing Prompt Perturbation Consistency Learning (PPCL) to recover 59% and 69% of performance drops for IC and SF, respectively, with ten times fewer augmented samples.
Large language models (LLMs) have demonstrated impressive performance on a number of natural language processing tasks, such as question answering and text summarization. However, their performance on sequence labeling tasks such as intent classification and slot filling (IC-SF), which is a central component in personal assistant systems, lags significantly behind discriminative models. Furthermore, there is a lack of substantive research on the robustness of LLMs to various perturbations in the input prompts. The contributions of this paper are three-fold. First, we show that fine-tuning sufficiently large LLMs can produce IC-SF performance comparable to discriminative models. Next, we systematically analyze the performance deterioration of those fine-tuned models due to three distinct yet relevant types of input perturbations - oronyms, synonyms, and paraphrasing. Finally, we propose an efficient mitigation approach, Prompt Perturbation Consistency Learning (PPCL), which works by regularizing the divergence between losses from clean and perturbed samples. Our experiments demonstrate that PPCL can recover on average 59% and 69% of the performance drop for IC and SF tasks, respectively. Furthermore, PPCL beats the data augmentation approach while using ten times fewer augmented data samples.