# Static Embeddings as Efficient Knowledge Bases?

arXiv: https://arxiv.org/abs/2104.07094 · HTML: https://mlsanity.com/paper/2104.07094/

## Key facts

- Cited by 11 papers in the map (data through 14 Oct 2026).
- Builds on 29 of its 34 references that are in the map.
- First posted on arXiv 14 Apr 2021.

## Builds on (most cited first, 29 of 29)

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Deep contextualized word representations](https://mlsanity.com/paper/1802.05365/) (2018)
- [REALM: Retrieval-Augmented Language Model Pre-Training](https://mlsanity.com/paper/2002.08909/) (2020)
- [Language Models as Knowledge Bases?](https://mlsanity.com/paper/1909.01066/) (2019)
- [Enriching Word Vectors with Subword Information](https://mlsanity.com/paper/1607.04606/) (2016)
- [Energy and Policy Considerations for Deep Learning in NLP](https://mlsanity.com/paper/1906.02243/) (2019)
- [Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering](https://mlsanity.com/paper/2007.01282/) (2020)
- [How Can We Know What Language Models Know?](https://mlsanity.com/paper/1911.12543/) (2019)
- [How Much Knowledge Can You Pack Into the Parameters of a Language Model?](https://mlsanity.com/paper/2002.08910/) (2020)
- [ERNIE: Enhanced Language Representation with Informative Entities](https://mlsanity.com/paper/1905.07129/) (2019)
- [COMET: Commonsense Transformers for Automatic Knowledge Graph Construction](https://mlsanity.com/paper/1906.05317/) (2019)
- [Measuring and Improving Consistency in Pretrained Language Models](https://mlsanity.com/paper/2102.01017/) (2021)
- [What BERT is not: Lessons from a new suite of psycholinguistic diagnostics for language models](https://mlsanity.com/paper/1907.13528/) (2019)
- [K-Adapter: Infusing Knowledge into Pre-Trained Models with Adapters](https://mlsanity.com/paper/2002.01808/) (2020)
- [K-BERT: Enabling Language Representation with Knowledge Graph](https://mlsanity.com/paper/1909.07606/) (2019)
- [Knowledge Enhanced Contextual Word Representations](https://mlsanity.com/paper/1909.04164/) (2019)
- [Negated and Misprimed Probes for Pretrained Language Models: Birds Can Talk, But Cannot Fly](https://mlsanity.com/paper/1911.03343/) (2019)
- [How Context Affects Language Models' Factual Predictions](https://mlsanity.com/paper/2005.04611/) (2020)
- [Multilingual LAMA: Investigating Knowledge in Multilingual Pretrained Language Models](https://mlsanity.com/paper/2102.00894/) (2021)
- [Language Models as Knowledge Bases: On Entity Representations, Storage Capacity, and Paraphrased Queries](https://mlsanity.com/paper/2008.09036/) (2020)
- [When Do You Need Billions of Words of Pretraining Data?](https://mlsanity.com/paper/2011.04946/) (2020)
- [Pretrained Encyclopedia: Weakly Supervised Knowledge-Pretrained Language Model](https://mlsanity.com/paper/1912.09637/) (2019)
- [E-BERT: Efficient-Yet-Effective Entity Embeddings for BERT](https://mlsanity.com/paper/1911.03681/) (2019)
- [Pre-training via Paraphrasing](https://mlsanity.com/paper/2006.15020/) (2020)
- [Inducing Relational Knowledge from BERT](https://mlsanity.com/paper/1911.12753/) (2019)
- [JAKET: Joint Pre-training of Knowledge Graph and Language Understanding](https://mlsanity.com/paper/2010.00796/) (2020)
- [BERT-kNN: Adding a kNN Search Component to Pretrained Language Models for Better QA](https://mlsanity.com/paper/2005.00766/) (2020)
- [Corpus-level Fine-grained Entity Typing](https://mlsanity.com/paper/1708.02275/) (2017)

## Built on by (most cited first, 11 of 11)

- [Language Models Implement Simple Word2Vec-style Vector Arithmetic](https://mlsanity.com/paper/2305.16130/) (2023)
- [Measuring Causal Effects of Data Statistics on Language Model's `Factual' Predictions](https://mlsanity.com/paper/2207.14251/) (2022)
- [Give Me the Facts! A Survey on Factual Knowledge Probing in Pre-trained Language Models](https://mlsanity.com/paper/2310.16570/) (2023)
- [Challenges in Generalization in Open Domain Question Answering](https://mlsanity.com/paper/2109.01156/) (2021)
- [P-Adapters: Robustly Extracting Factual Information from Language Models with Diverse Prompts](https://mlsanity.com/paper/2110.07280/) (2021)
- [Rewire-then-Probe: A Contrastive Recipe for Probing Biomedical Knowledge of Pre-trained Language Models](https://mlsanity.com/paper/2110.08173/) (2021)
- [Transparency Helps Reveal When Language Models Learn Meaning](https://mlsanity.com/paper/2210.07468/) (2022)
- [GrEmLIn: A Repository of Green Baseline Embeddings for 87 Low-Resource Languages Injected with Multilingual Graph Knowledge](https://mlsanity.com/paper/2409.18193/) (2024)
- [Multi-Sense Embeddings for Language Models and Knowledge Distillation](https://mlsanity.com/paper/2504.06036/) (2025)
- [Probing Commonsense Knowledge in Pre-trained Language Models with Sense-level Precision and Expanded Vocabulary](https://mlsanity.com/paper/2210.06376/) (2022)
- [Explaining Black-box Language Models with Knowledge Probing Systems: A Post-hoc Explanation Perspective](https://mlsanity.com/paper/2508.16969/) (2025)

## Abstract (arXiv)

Recent research investigates factual knowledge stored in large pretrained language models (PLMs). Instead of structural knowledge base (KB) queries, masked sentences such as "Paris is the capital of [MASK]" are used as probes. The good performance on this analysis task has been interpreted as PLMs becoming potential repositories of factual knowledge. In experiments across ten linguistically diverse languages, we study knowledge contained in static embeddings. We show that, when restricting the output space to a candidate set, simple nearest neighbor matching using static embeddings performs better than PLMs. E.g., static embeddings perform 1.6% points better than BERT while just using 0.3% of energy for training. One important factor in their good comparative performance is that static embeddings are standardly learned for a large vocabulary. In contrast, BERT exploits its more sophisticated, but expensive ability to compose meaningful representations from a much smaller subword vocabulary.