Related papers: Conditional probing: measuring usable information …

When evaluating causal influence from one time series to another in a multivariate dataset it is necessary to take into account the conditioning effect of the other variables. In the presence of many variables, and possibly of a reduced…

Interpretability benefits the theoretical understanding of representations. Existing word embeddings are generally dense representations. Hence, the meaning of latent dimensions is difficult to interpret. This makes word embeddings like a…

By introducing a small set of additional parameters, a probe learns to solve specific linguistic tasks (e.g., dependency parsing) in a supervised manner using feature representations (e.g., contextualized embeddings). The effectiveness of…

When people interpret text, they rely on inferences that go beyond the observed language itself. Inspired by this observation, we introduce a method for the analysis of text that takes implicitly communicated content explicitly into…

We propose a new framework for reasoning about information in complex systems. Our foundation is based on a variational extension of Shannon's information theory that takes into account the modeling power and computational constraints of…

The question of how to probe contextual word representations for linguistic structure in a way that is both principled and useful has seen significant attention recently in the NLP literature. In our contribution to this discussion, we…

Multiple studies have probed representations emerging in neural networks trained for end-to-end NLP tasks and examined what word-level linguistic information may be encoded in the representations. In classical probing, a classifier is…

Many machine learning algorithms represent input data with vector embeddings or discrete codes. When inputs exhibit compositional structure (e.g. objects built from parts or procedures from subroutines), it is natural to ask whether this…

The dominant probing approaches rely on the zero-shot performance of image-text matching tasks to gain a finer-grained understanding of the representations learned by recent multimodal image-language transformer models. The evaluation is…

Language decoding studies have identified word representations which can be used to predict brain activity in response to novel words and sentences (Anderson et al., 2016; Pereira et al., 2018). The unspoken assumption of these studies is…

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…

Edge probing tests are classification tasks that test for grammatical knowledge encoded in token representations coming from contextual encoders such as large language models (LLMs). Many LLM encoders have shown high performance in EP…

Understanding which information is encoded in deep models of spoken and written language has been the focus of much research in recent years, as it is crucial for debugging and improving these architectures. Most previous work has focused…

As large and powerful neural language models are developed, researchers have been increasingly interested in developing diagnostic tools to probe them. There are many papers with conclusions of the form "observation X is found in model Y",…

The pervasive use of distributional semantic models or word embeddings in a variety of research fields is due to their remarkable ability to represent the meanings of words for both practical application and cognitive modeling. However,…

We provide explicit, finite-sample guarantees for learning causal representations from data with a sublinear number of environments. Causal representation learning seeks to provide a rigourous foundation for the general representation…

We give a finite-sample analysis of predictive inference procedures after model selection in regression with random design. The analysis is focused on a statistically challenging scenario where the number of potentially important…

Transformer-based language models benefit from conditioning on contexts of hundreds to thousands of previous tokens. What aspects of these contexts contribute to accurate model prediction? We describe a series of experiments that measure…

Learning word embeddings using distributional information is a task that has been studied by many researchers, and a lot of studies are reported in the literature. On the contrary, less studies were done for the case of multiple languages.…

Assisted text input techniques can save time and effort and improve text quality. In this paper, we investigate how grounded and conditional extensions to standard neural language models can bring improvements in the tasks of word…