Supervised Contrastive Learning for Pre-trained Language Model Fine-tuning
This addresses the issue of fine-tuning inefficiency for natural language understanding tasks, offering a method to enhance model robustness and generalization with limited labeled data, though it is incremental as it builds on existing contrastive learning ideas.
The paper tackles the problem of sub-optimal generalization and instability in fine-tuning pre-trained language models by proposing a supervised contrastive learning objective, which combined with cross-entropy loss achieves significant improvements over a RoBERTa-Large baseline on multiple GLUE datasets in few-shot settings.
State-of-the-art natural language understanding classification models follow two-stages: pre-training a large language model on an auxiliary task, and then fine-tuning the model on a task-specific labeled dataset using cross-entropy loss. However, the cross-entropy loss has several shortcomings that can lead to sub-optimal generalization and instability. Driven by the intuition that good generalization requires capturing the similarity between examples in one class and contrasting them with examples in other classes, we propose a supervised contrastive learning (SCL) objective for the fine-tuning stage. Combined with cross-entropy, our proposed SCL loss obtains significant improvements over a strong RoBERTa-Large baseline on multiple datasets of the GLUE benchmark in few-shot learning settings, without requiring specialized architecture, data augmentations, memory banks, or additional unsupervised data. Our proposed fine-tuning objective leads to models that are more robust to different levels of noise in the fine-tuning training data, and can generalize better to related tasks with limited labeled data.