Abstract
Improving our understanding of how information is encoded in vector space can yield valuable interpretability insights. Alongside vector dimensions, we argue that it is possible for the vector norm to also carry linguistic information. We develop a method to test this: an extension of the probing framework which allows for relative intrinsic interpretations of probing results. It relies on introducing noise that ablates information encoded in embeddings, grounded in random baselines and confidence intervals. We apply the method to well-established probing tasks and find evidence that confirms the existence of separate information containers in English GloVe and BERT embeddings. Our correlation analysis aligns with the experimental findings that different encoders use the norm to encode different kinds of information: GloVe stores syntactic and sentence length information in the vector norm, while BERT uses it to encode contextual incongruity.- Anthology ID:
- 2022.blackboxnlp-1.34
- Volume:
- Proceedings of the Fifth BlackboxNLP Workshop on Analyzing and Interpreting Neural Networks for NLP
- Month:
- December
- Year:
- 2022
- Address:
- Abu Dhabi, United Arab Emirates (Hybrid)
- Venue:
- BlackboxNLP
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 404–417
- Language:
- URL:
- https://aclanthology.org/2022.blackboxnlp-1.34
- DOI:
- 10.18653/v1/2022.blackboxnlp-1.34
- Cite (ACL):
- Filip Klubicka and John Kelleher. 2022. Probing with Noise: Unpicking the Warp and Weft of Embeddings. In Proceedings of the Fifth BlackboxNLP Workshop on Analyzing and Interpreting Neural Networks for NLP, pages 404–417, Abu Dhabi, United Arab Emirates (Hybrid). Association for Computational Linguistics.
- Cite (Informal):
- Probing with Noise: Unpicking the Warp and Weft of Embeddings (Klubicka & Kelleher, BlackboxNLP 2022)
- PDF:
- https://preview.aclanthology.org/remove-xml-comments/2022.blackboxnlp-1.34.pdf