Probing with Noise: Unpicking the Warp and Weft of Embeddings
This provides interpretability insights for NLP researchers by revealing distinct information encoding patterns in embeddings, though it is incremental as it extends existing probing frameworks.
The researchers tackled the problem of understanding how linguistic information is encoded in vector embeddings by developing a noise-based probing method to test whether vector norms carry information, finding evidence that GloVe stores syntactic and sentence length information in norms while BERT uses them for contextual incongruity.
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.