Related papers: Momentum Point-Perplexity Mechanics in Large Langu…
We seek to understand how the representations of individual tokens and the structure of the learned feature space evolve between layers in deep neural networks under different learning objectives. We focus on the Transformers for our…
While large language models (LLMs) excel in mathematical and code reasoning, we observe they struggle with social reasoning tasks, exhibiting cognitive confusion, logical inconsistencies, and conflation between objective world states and…
The rapid advancement of large language models (LLMs) has made machine-generated text increasingly difficult to distinguish from human-written text. While recent studies explore leveraging internal representations of language models to…
This paper describes the conversion of a Hidden Markov Model into a sequential transducer that closely approximates the behavior of the stochastic model. This transformation is especially advantageous for part-of-speech tagging because the…
Tracking the internal states of large language models across conversations is important for safety, interpretability, and model welfare, yet current methods are limited. Linear probes and other white-box methods compress high-dimensional…
Transformers underpin modern large language models (LLMs) and are commonly assumed to be behaviorally unstructured at random initialization, with all meaningful preferences emerging only through large-scale training. We challenge this…
Recent reasoning language models, particularly those that employ long latent chains of thought, achieve strong performance on complex agentic tasks. However, as these models operate over increasingly long time horizons, their internal…
It has long been established that predictive models can be transformed into lossless compressors and vice versa. Incidentally, in recent years, the machine learning community has focused on training increasingly large and powerful…
Large Language Models (LLMs) exhibit impressive capabilities yet suffer from sensitivity to slight input context variations, hampering reliability. Conventional metrics like accuracy and perplexity fail to assess local prediction…
Large language models (LLMs) tend to verbalize confidence scores that are largely detached from their actual accuracy, yet the geometric relationship governing this behavior remain poorly understood. In this work, we present a mechanistic…
The choice of tokenizer can profoundly impact language model performance, yet accessible and reliable evaluations of tokenizer quality remain an open challenge. Inspired by scaling consistency, we show that smaller models can accurately…
Bidirectional Encoder Representations from Transformers (BERT) reach state-of-the-art results in a variety of Natural Language Processing tasks. However, understanding of their internal functioning is still insufficient and unsatisfactory.…
Augmenting large language models (LLMs) with auxiliary tokens has emerged as a promising strategy for enhancing model performance. In this work, we introduce a lightweight method termed latent tokens; these are dummy tokens that may be…
Can large language models introspect, that is, accurately detect perturbations to their own internal states? We systematically investigate this question using activation steering in Meta-Llama-3.1-8B-Instruct. First, we show that the binary…
Transformer architectures are the backbone of most modern language models, but understanding the inner workings of these models still largely remains an open problem. One way that research in the past has tackled this problem is by…
We consider the context of molecular motors modelled by a diffusion process driven by the gradient of a weakly periodic potential that depends on an internal degree of freedom. The switch of the internal state, that can freely be…
Language models (LMs) automatically learn word embeddings during pre-training on language corpora. Although word embeddings are usually interpreted as feature vectors for individual words, their roles in language model generation remain…
Recent advances in interpretability suggest we can project weights and hidden states of transformer-based language models (LMs) to their vocabulary, a transformation that makes them more human interpretable. In this paper, we investigate LM…
Intrinsic self-correction refers to the phenomenon where a language model refines its own outputs purely through prompting, without external feedback or parameter updates. While this approach improves performance across diverse tasks, its…
Translation into morphologically-rich languages challenges neural machine translation (NMT) models with extremely sparse vocabularies where atomic treatment of surface forms is unrealistic. This problem is typically addressed by either…