表現の分布を一様に分布させたい…🤔 4 導入:文表現学習と異方性 少年が折り紙を折っている 少年が赤い折り紙を折っている 少年が町を歩いている [1]Ethayarajh.:How contextual are contextualized word representations? comparing the geometry of bert, elmo, and gpt-2 embeddings(EMNLP’19)
[2]Li et al.: On the sentence embeddings from pre-trained language models (EMNLP’20) [3]Su et al.: Whitening sentence representations for better semantics and faster retrieval (arXiv’21) [4]Giorgi et al.:Declutr: Deep contrastive learning for unsupervised textual representations(ACL’21) [5]Gao et al.: Simcse: Simple contrastive learning of sentence embeddings(EMNLP’21)
6 導入:対照学習 [5]Gao et al.: Simcse: Simple contrastive learning of sentence embeddings(EMNLP’21) [6]Su et al.: A Contrastive Framework for Learning Sentence Representations from Pairwise and Triple-wise Perspective in Angular Space(ACL’22) [7] Zhuo et al.:WhitenedCSE: Whitening-based Contrastive Learning of Sentence Embeddings (ACL’23)