Revealing Language Model Trajectories via Kullback-Leibler Divergence

A recently proposed method enables efficient estimation of the KL divergence between language models, including models with different architectures, by assigning coordinates based on log-likelihood vectors. To better understand the behavior of this metric, we systematically evaluate KL divergence across a wide range of conditions using publicly available language models. Our analysis covers comparisons between pretraining checkpoints, fine-tuned and base models, and layers via the logit lens. We find that trajectories of language models, as measured by KL divergence, exhibit a spiral structure during pretraining and thread-like progressions across layers. Furthermore, we show that, in terms of diffusion exponents, model trajectories in the log-likelihood space are more constrained than those in weight space.

Likelihood Variance as Text Importance for Resampling Texts to Map Language Models

We address the computational cost of constructing a model map, which embeds diverse language models into a common space for comparison via KL divergence. The map relies on log-likelihoods over a large text set, making the cost proportional to the number of texts. To reduce this cost, we propose a resampling method that selects important texts with weights proportional to the variance of log-likelihoods across models for each text. Our method significantly reduces the number of required texts while preserving the accuracy of KL divergence estimates. Experiments show that it achieves comparable performance to uniform sampling with about half as many texts, and also facilitates efficient incorporation of new models into an existing map. These results enable scalable and efficient construction of language model maps.

Mapping 1,000+ Language Models via the Log-Likelihood Vector

To compare autoregressive language models at scale, we propose using log-likelihood vectors computed on a predefined text set as model features. This approach has a solid theoretical basis: when treated as model coordinates, their squared Euclidean distance approximates the Kullback-Leibler divergence of text-generation probabilities. Our method is highly scalable, with computational cost growing linearly in both the number of models and text samples, and is easy to implement as the required features are derived from cross-entropy loss. Applying this method to over 1,000 language models, we constructed a "model map," providing a new perspective on large-scale model analysis.

Understanding Higher-Order Correlations Among Semantic Components in Embeddings

Independent Component Analysis (ICA) is an effective method for interpreting the intrinsic geometric structure of embeddings as semantic components. While ICA theory assumes that embeddings can be linearly decomposed into independent components, real-world data often do not satisfy this assumption. Consequently, there are remaining non-independencies between the estimated components that ICA cannot eliminate. We quantified these non-independencies using higher-order correlations and demonstrated that when the higher-order correlation between two components is large, it indicates a strong semantic association between them. The entire structure was revealed through visualization using a maximum spanning tree of semantic components. These findings allow for further understanding of embeddings through ICA.

Revisiting Cosine Similarity via Normalized ICA-transformed Embeddings

Cosine similarity is widely used to measure the similarity between two embeddings, while interpretations based on angle and correlation coefficient are common. In this study, we focus on the interpretable axes of embeddings transformed by Independent Component Analysis (ICA), and propose a novel interpretation of cosine similarity as the sum of semantic similarities over axes. To investigate this, we first show experimentally that unnormalized embeddings contain norm-derived artifacts. We then demonstrate that normalized ICA-transformed embeddings exhibit sparsity, with a few large values in each axis and across embeddings, thereby enhancing interpretability by delineating clear semantic contributions. Finally, to validate our interpretation, we perform retrieval experiments using ideal embeddings with and without specific semantic components.

Predicting Drug-Gene Relations via Analogy Tasks with Word Embeddings

Natural language processing (NLP) is utilized in a wide range of fields, where words in text are typically transformed into feature vectors called embeddings. BioConceptVec is a specific example of embeddings tailored for biology, trained on approximately 30 million PubMed abstracts using models such as skip-gram. Generally, word embeddings are known to solve analogy tasks through simple vector arithmetic. For instance, $\mathrm{\textit{king}} - \mathrm{\textit{man}} + \mathrm{\textit{woman}}$ predicts $\mathrm{\textit{queen}}$. In this study, we demonstrate that BioConceptVec embeddings, along with our own embeddings trained on PubMed abstracts, contain information about drug-gene relations and can predict target genes from a given drug through analogy computations. We also show that categorizing drugs and genes using biological pathways improves performance. Furthermore, we illustrate that vectors derived from known relations in the past can predict unknown future relations in datasets divided by year.

Discovering Universal Geometry in Embeddings with ICA

This study employs Independent Component Analysis (ICA) to uncover universal properties of embeddings of words or images. Our approach extracts independent semantic components of embeddings, enabling each embedding to be represented as a composition of intrinsic interpretable axes. We demonstrate that embeddings can be expressed as a combination of a few axes and that these semantic axes are consistent across different languages, modalities, and embedding algorithms. This discovery of universal properties in embeddings contributes to model interpretability, potentially facilitating the development of highly interpretable models and the compression of large-scale models.

Norm of word embedding encodes information gain

Distributed representations of words encode lexical semantic information, but how is that information encoded in word embeddings? Focusing on the skip-gram with negative-sampling method, we show theoretically and experimentally that the squared norm of word embedding encodes the information gain defined by the Kullback-Leibler divergence of the co-occurrence distribution of a word to the unigram distribution of the corpus. Furthermore, through experiments on tasks of keyword extraction, hypernym prediction, and part-of-speech discrimination, we confirmed that the KL divergence and the squared norm of embedding work as a measure of the informativeness of a word provided that the bias caused by word frequency is adequately corrected.