一作ACL 2026Findings· Findings of ACL
MIPIC : Matryoshka Representation Learning via Self-Distilled Intra-Relational and Progressive Information Chaining
Phung Gia Huy · Hai An Vu · Minh-Phuc Truong · Thang Duc Tran · Linh Ngo Van · Thanh Hong Nguyen · Trung Le
摘要
Representation learning is fundamental to NLP, but building embeddings that work well at different computational budgets is challenging. Matryoshka Representation Learning (MRL) offers a flexible inference paradigm through nested embeddings; however, learning such structures requires explicit coo...