Summary of Contextual Document Embeddings, by John X. Morris et al.
Contextual Document Embeddings
by John X. Morris, Alexander M. Rush
First submitted to arxiv on: 3 Oct 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper argues that current dense document embeddings are out-of-context for targeted retrieval tasks and proposes two methods for contextualized document embeddings. The authors suggest an alternative contrastive learning objective that incorporates neighboring documents into the loss function, as well as a new architecture that encodes neighbor information into the encoded representation. Experimental results show that both methods perform better than biencoders in several settings, especially out-of-domain. This approach achieves state-of-the-art results on the MTEB benchmark without relying on specialized techniques like hard negative mining or large batch sizes. The proposed method can be applied to improve performance on any contrastive learning dataset and biencoder. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about improving how computers understand documents by taking into account what’s around them, just like we do when reading a book. Currently, computers use embeddings that are made without considering the context of other documents. The authors propose two new ways to create embeddings that consider this context. They show that these new methods work better than old ones in certain situations. This is important because it can help improve how computers find relevant information. |
Keywords
» Artificial intelligence » Loss function
