Dense retrieval을 위한 completely hands-off pretrain method 제안
Abstract
Recent research demonstrates the effectiveness of using fine-tuned language models~(LM) for dense retrieval. However, dense retrievers are hard to train, typically requiring heavily engineered fine-tuning pipelines to realize their full potential. In this paper, we identify and address two underlying problems of dense retrievers: i) fragility to training data noise and ii) requiring large batches to robustly learn the embedding space. We use the recently proposed Condenser pre-training architecture, which learns to condense information into the dense vector through LM pre-training. On top of it, we propose coCondenser, which adds an unsupervised corpus-level contrastive loss to warm up the passage embedding space. Retrieval experiments on MS-MARCO, Natural Question, and Trivia QA datasets show that coCondenser removes the need for heavy data engineering such as augmentation, synthesis, or filtering, as well as the need for large batch training. It shows comparable performance to RocketQA, a state-of-the-art, heavily engineered system, using simple small batch fine-tuning.
Keywords
Condenser; Pre-training for Retriever; Contrastive learning; Natural Question; Trivia QA; MS-MARCO;
TL;DR
TL;DR
Condenser를 기반으로 추가 pretraining을 진행하는 모델
문서로부터 특정 span을 추출하고 contrastive learning에 사용함
비교적 간단한 학습 전략으로 방식으로 준수한 성능을 기록함
Contribution
Abstract
Recent research demonstrates the effectiveness of using fine-tuned language models~(LM) for dense retrieval. However, dense retrievers are hard to train, typically requiring heavily engineered fine-tuning pipelines to realize their full potential. In this paper, we identify and address two underlying problems of dense retrievers: i) fragility to training data noise and ii) requiring large batches to robustly learn the embedding space. We use the recently proposed Condenser pre-training architecture, which learns to condense information into the dense vector through LM pre-training. On top of it, we propose coCondenser, which adds an unsupervised corpus-level contrastive loss to warm up the passage embedding space. Retrieval experiments on MS-MARCO, Natural Question, and Trivia QA datasets show that coCondenser removes the need for heavy data engineering such as augmentation, synthesis, or filtering, as well as the need for large batch training. It shows comparable performance to RocketQA, a state-of-the-art, heavily engineered system, using simple small batch fine-tuning.
Paper link
https://arxiv.org/abs/2108.05540
Presentation link
https://drive.google.com/file/d/1sJzZ5_TvaIiSE4fG2YaEyrmo62N2PaH1/view?usp=sharing
Reference papers for presentation
video link
https://youtu.be/G7Yxz0_QkWk