First Train to Generate, then Generate to Train: UnitedSynT5 for Few-Shot NLI
This work addresses dataset constraints for NLI researchers, offering an incremental improvement over previous state-of-the-art models.
The paper tackled the problem of limited dataset diversity in Natural Language Inference (NLI) by proposing UnitedSynT5, which uses synthetic data augmentation to enhance training data, achieving new benchmarks of 94.7% accuracy on SNLI, 94.0% on E-SNLI, and 92.6% on MultiNLI.
Natural Language Inference (NLI) tasks require identifying the relationship between sentence pairs, typically classified as entailment, contradiction, or neutrality. While the current state-of-the-art (SOTA) model, Entailment Few-Shot Learning (EFL), achieves a 93.1% accuracy on the Stanford Natural Language Inference (SNLI) dataset, further advancements are constrained by the dataset's limitations. To address this, we propose a novel approach leveraging synthetic data augmentation to enhance dataset diversity and complexity. We present UnitedSynT5, an advanced extension of EFL that leverages a T5-based generator to synthesize additional premise-hypothesis pairs, which are rigorously cleaned and integrated into the training data. These augmented examples are processed within the EFL framework, embedding labels directly into hypotheses for consistency. We train a GTR-T5-XL model on this expanded dataset, achieving a new benchmark of 94.7% accuracy on the SNLI dataset, 94.0% accuracy on the E-SNLI dataset, and 92.6% accuracy on the MultiNLI dataset, surpassing the previous SOTA models. This research demonstrates the potential of synthetic data augmentation in improving NLI models, offering a path forward for further advancements in natural language understanding tasks.