方法证据记录
Contrastive Learning for NLP
Contrastive learning for NLP is a representation-learning technique — popularised by SimCSE (Gao et al., 2021) and Supervised Contrastive Learning (Khosla et al., 2020) — that trains a text encoder by pulling embeddings of similar text pairs together while pushing embeddings of dissimilar pairs apart. The result is a dense, high-quality embedding space that can be learned with no labels at all, or with minimal supervision, making it especially valuable when annotated data are scarce.
源记录
引文逐字复制自方法源记录。这些引文不代表任何层级的验证。
Contrastive Learning for Natural Language Processing
分类方法记录 · process-pipeline / text-mining
- Gao, T., Yao, X., & Chen, D. (2021). SimCSE: Simple Contrastive Learning of Sentence Embeddings. Proceedings of EMNLP 2021. · URL
- Khosla, P., et al. (2020). Supervised Contrastive Learning. Advances in Neural Information Processing Systems (NeurIPS) 33. · URL
精选声明
声明已持久化到证据分类账中，每个声明都有自己的评估。
尚无精选声明
当分类账中没有声明时，此视图不会自行创建声明评估。
相关方法
从方法图中生成，显示为机器建议的关系 — 不推断任何证据声明。