Fangyu Liu on X: Want to bootstrap your sentence similarity model

2022 Acl-Long 331, PDF

Fangyu Liu on X: Surprisingly, on a wide range of tasks including sentence similarity, Mirror-BERT is as good as its supervised counterparts. On word similarity, it is usually better than the strong

PDF) Enhancing Out-of-Distribution Detection in Natural Language Understanding via Implicit Layer Ensemble

PDF) Modality-Aware Contrastive Instance Learning with Self-Distillation for Weakly-Supervised Audio-Visual Violence Detection

Fangyu Liu on X: Want to bootstrap your sentence similarity model for free? Check out our #ICLR2022 paper Trans-Encoder – jointly learning dual- and cross-encoders through self-distillation, achieving a new SotA on

ACL 2022, 60th Annual Meeting of the Association for Computational Linguistics

Pre-trained Language Models in Biomedical Domain: A Systematic Survey

ICML 2022 Spotlights

PDF) Modality-Aware Contrastive Instance Learning with Self-Distillation for Weakly-Supervised Audio-Visual Violence Detection

Fangyu Liu on X: So you have all heard of how contrastive learning can bootstrap sentence embeddings to be as good as supervised sentence-bert. We discovered so too, and we also found

Fangyu Liu 🏖️ (@hardy_qr) / X

ICML 2022 Spotlights