Probing with Noise: Unpicking the Warp and Weft of Embeddings
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.
Code (1)
Tasks
SentenceMethods 이 논문이 사용한 방법론
Similar Papers 제목 키워드 기반
Image-based Textile Decoding
A textile fabric consists of countless parallel vertical yarns (warps) and horizontal yarns (wefts). While common looms can weave repetitive patterns, Jacquard looms can weave the patterns without repetition restrictions…
Warp and Weft Wiring method for rapid, modifiable, self-aligned, and bonding-free fabrication of multi electrodes microfluidic sensors
The need for rapid fabrication of microfluidic devices has become increasingly critical as microfluidics become part of biomedical sensors. Using Warp and Weft Wiring (WWW) of copper wires, this paper presents a novel lo…
ManagementAutomatic Defect Detection of Print Fabric Using Convolutional Neural Network
Automatic defect detection is a challenging task because of the variability in texture and type of fabric defects. An effective defect detection system enables manufacturers to improve the quality of processes and produc…
Defect DetectionMultiscale Analysis of Woven Composites Using Hierarchical Physically Recurrent Neural Networks
Multiscale homogenization of woven composites requires detailed micromechanical evaluations, leading to high computational costs. Data-driven surrogate models based on neural networks address this challenge but often suf…
TACos: Learning Temporally Structured Embeddings for Few-Shot Keyword Spotting with Dynamic Time Warping
To segment a signal into blocks to be analyzed, few-shot keyword spotting (KWS) systems often utilize a sliding window of fixed size. Because of the varying lengths of different keywords or their spoken instances, choosi…
Dynamic Time WarpingKeyword Spotting