Revisiting Distance Metric Learning for Few-Shot Natural Language Classification (2211.15202v1)

Abstract: Distance Metric Learning (DML) has attracted much attention in image processing in recent years. This paper analyzes its impact on supervised fine-tuning LLMs for NLP classification tasks under few-shot learning settings. We investigated several DML loss functions in training RoBERTa LLMs on known SentEval Transfer Tasks datasets. We also analyzed the possibility of using proxy-based DML losses during model inference. Our systematic experiments have shown that under few-shot learning settings, particularly proxy-based DML losses can positively affect the fine-tuning and inference of a supervised LLM. Models tuned with a combination of CCE (categorical cross-entropy loss) and ProxyAnchor Loss have, on average, the best performance and outperform models with only CCE by about 3.27 percentage points -- up to 10.38 percentage points depending on the training dataset.