Probing Classifiers: Promises, Shortcomings, and Advances
Open accessThe DOI matches its Crossref record; 17 of the 17 references it cites were found in a registry or an index.
Abstract
Probing classifiers have emerged as one of the prominent methodologies for interpreting and analyzing deep neural network models of natural language processing. The basic idea is simple—a classifier is trained to predict some linguistic property from a model’s representations—and has been used to examine a wide variety of models and properties. However, recent studies have demonstrated various methodological limitations of this approach. This squib critically reviews the probing classifiers framework, highlighting their promises, shortcomings, and advances.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2018) · arXiv · doi:10.48550/arxiv.1810.04805 · Cited by 73,173
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1423 · Cited by 54,325
Representational similarity analysis-connecting the branches of systems neuroscience
Kriegeskorte, N. (2008) · Frontiers in Systems Neuroscience · doi:10.3389/neuro.06.004.2008 · Cited by 3,355
A primer in BERTology: What we know about how BERT works
Rogers, A., Kovaleva, O., & Rumshisky, A. (2020) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00349 · Cited by 1,552 · cited 1×
“For recent analyses focusing on the BERT model (Devlin et al. 2019), see Rogers, Kovaleva, and Rumshisky (2020).”
MethodEdmonds, J. (1967) · Journal of Research of the National Bureau of Standards Section B Mathematics and Mathematical Physics · doi:10.6028/jres.071b.032 · Cited by 731
Understanding intermediate layers using linear classifier probes
Alain, G., & Bengio, Y. (2016) · arXiv · doi:10.48550/arxiv.1610.01644 · Cited by 581
Designing and Interpreting Probes with Control Tasks
Hewitt, J., & Liang, P. (2019) · Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP) · doi:10.18653/v1/d19-1275 · Cited by 393 · cited 4×
“They found that MDL provides more information about how a probe g works, for instance by revealing differences in com-5 plexity of probes when performing control tasks from D P,Rand , as in Hewitt and Liang (2019).”
BackgroundInfluentialWhat do Neural Machine Translation Models Learn about Morphology?
Belinkov, Y., Durrani, N., Dalvi, F., Sajjad, H., & Glass, J. (2017) · Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/p17-1080 · Cited by 307 · cited 8×
“7 “evaluate the quality of the trained classifier on the given task as a proxy to the quality of the extracted representations” (Belinkov et al. 2017a).”
BackgroundResultInfluentialSelf-citationAn Analysis of Encoder Representations in Transformer-Based Machine Translation
Raganato, A., & Tiedemann, J. (2018) · Proceedings of the 2018 EMNLP Workshop BlackboxNLP: Analyzing and Interpreting Neural Networks for NLP · doi:10.18653/v1/w18-5431 · Cited by 277 · cited 1×
“These scores may come from attention weights ( Raganato and Tiedemann 2018; Clark et al. 2019; Mareˇcek and Rosa 2019; Htut et al. 2019) or from distances between word representations, perhaps including perturbations of the input sentence (Wu et al. 2020).”
BackgroundWhat you can cram into a single vector: Probing sentence embeddings for linguistic properties
Conneau, A., Kruszewski, G., Lample, G., Barrault, L., & Baroni, M. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1805.01070 · Cited by 263
Information-Theoretic Probing with Minimum Description Length
Voita, E., & Titov, I. (2020) · Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2020.emnlp-main.14 · Cited by 251
Probing the Probing Paradigm: Does Probing Accuracy Entail Task Relevance?
Ravichander, A., Belinkov, Y., & Hovy, E. (2021) · Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume · doi:10.18653/v1/2021.eacl-main.295 · Cited by 132
Self-citationAmnesic Probing: Behavioral Explanation with Amnesic Counterfactuals
Elazar, Y., Ravfogel, S., Jacovi, A., & Goldberg, Y. (2021) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00359 · Cited by 126 · cited 5×
“Figure 1 Components constituting the probing classifiers framework. stable form by several groups who studied sentence embeddings (Ettinger, Elgohary, and Resnik 2016; Adi et al. 2017; Conneau et al. 2018) and recurrent/recursive neural networks (Belinkov et al. 2017a; Hupkes, Veldhoen, and Zuidema 2018).3 The same id…”
BackgroundInfluential(2019) · doi:10.18653/v1/n19-1002 · Cited by 118 · cited 2×
“11 Other studies that perform interventions to interpret NLP models without involving probing classifiers (e.g., Bau et al. 2019; Lakretz et al. 2019; Vig et al. 2020) are left out of the present scope.”
BackgroundWhen Do You Need Billions of Words of Pretraining Data?
Zhang, Y., Warstadt, A., Li, X., & Bowman, S. R. (2021) · Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) · doi:10.18653/v1/2021.acl-long.90 · Cited by 85 · cited 1×
“Concerning the latter, Zhang et al. (2021) found that models require less data to encode syntactic and semantic properties compared to commonsense knowledge.”
BackgroundDistributional vectors encode referential attributes
Gupta, A., Boleda, G., Baroni, M., & Padó, S. (2015) · Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/d15-1002 · Cited by 79
What if This Modified That? Syntactic Interventions with Counterfactual Embeddings
Tucker, M., Qian, P., & Levy, R. (2021) · Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021 · doi:10.18653/v1/2021.findings-acl.76 · Cited by 16
Cited by
Explainability for Large Language Models: A Survey
Zhao, H., Chen, H., Yang, F., Liu, N., Deng, H., Cai, H., Wang, S., Yin, D., & Du, M. (2024) · ACM Transactions on Intelligent Systems and Technology · doi:10.1145/3639372 · Cited by 681
“Although high probing performance is often attributed to the quality and interpretability of representations [Belinkov 2022], this assumption remains largely unproven and difficult to validate.”
BackgroundMethodInfluentialDissociating language and thought in large language models
Mahowald, K., Ivanova, A. A., Blank, I. A., Kanwisher, N., Tenenbaum, J. B., & Fedorenko, E. (2024) · Trends in Cognitive Sciences · doi:10.1016/j.tics.2024.01.011 · Cited by 345
Open Problems in Mechanistic Interpretability
Sharkey, L., Chughtai, B., Batson, J., Lindsey, J., Wu, J., Bushnaq, L., Goldowsky-Dill, N., Heimersheim, S., Ortega, A., Bloom, J., et al. (2025) · arXiv · doi:10.48550/arxiv.2501.16496 · Cited by 291
Acquisition of chess knowledge in AlphaZero
McGrath, T., Kapishnikov, A., Tomašev, N., Pearce, A., Wattenberg, M., Hassabis, D., Kim, B., Paquet, U., & Kramnik, V. (2022) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2206625119 · Cited by 166
Toward Transparent AI: A Survey on Interpreting the Inner Structures of Deep Neural Networks
Räuker, T., Ho, A., Casper, S., & Hadfield-Menell, D. (2023) · 2023 IEEE Conference on Secure and Trustworthy Machine Learning (SaTML) · doi:10.1109/satml54575.2023.00039 · Cited by 122
Metadata
- Authors
- Belinkov, Y.
- Venue
- Computational Linguistics
- Publisher
- MIT Press - Journals
- Published
- 2022
- Volume
- 48
- Issue
- 1
- Pages
- 207-219
- ISSN
- 0891-2017 (ISSN-L)
- 1530-9312
- OpenAlex
- W3193068792
- Cited by
- 249
- References
- 17
- FWCI
- 11.73
- Type
- journal article
- Language
- en
- Open access
- diamond
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.