Combining cross-entropy with contrastive loss and L2 regularization reportedly raises few-shot text classification accuracy on FewRel 2.0, but the method is standard and the evaluation is under-specified.
Transformers are short-text classifiers,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Towards Robust Few-Shot Text Classification Using Transformer Architectures and Dual Loss Strategies
Combining cross-entropy with contrastive loss and L2 regularization reportedly raises few-shot text classification accuracy on FewRel 2.0, but the method is standard and the evaluation is under-specified.