Related papers: Disentangling Direction and Magnitude in Transform…

Transformer-based language models excel at both recall (retrieving memorized facts) and reasoning (performing multi-step inference), but whether these abilities rely on distinct internal mechanisms remains unclear. Distinguishing recall…

Large Language Models (LLMs) frequently prioritize conflicting in-context information over pre-existing parametric memory, a phenomenon often termed sycophancy or compliance. However, the mechanistic realization of this behavior remains…

Linguistic representation learning in deep neural language models (LMs) has been studied for decades, for both practical and theoretical reasons. However, finding representations in LMs remains an unsolved problem, in part due to a dilemma…

Since their introduction, Transformer architectures have dominated Natural Language Processing (NLP). However, recent research has highlighted an inherent anisotropy phenomenon in these models, presenting a significant challenge to their…

The usage-based constructionist (UCx) approach to language posits that language comprises a network of learned form-meaning pairings (constructions) whose use is largely determined by their meanings or functions, requiring them to be graded…

Large language models (LLMs) are known to exhibit brittle behavior under adversarial prompts and jailbreak attacks, even after extensive alignment and fine-tuning. This fragility reflects a broader challenge of modern neural language…

Symbolic perturbations offer a novel approach for influencing neural representations without requiring direct modification of model parameters. The recursive regeneration of symbolic structures introduces structured variations in latent…

Large language models (LLMs) achieve remarkable fluency across linguistic and reasoning tasks but remain systematically prone to hallucination. Prevailing accounts attribute hallucinations to data gaps, limited context, or optimization…

Contextual hallucinations -- statements unsupported by given context -- remain a significant challenge in AI. We demonstrate a practical interpretability insight: a generator-agnostic observer model detects hallucinations via a single…

Due to the unidirectional masking mechanism, Decoder-Only models propagate information from left to right. LVLMs (Large Vision-Language Models) follow the same architecture, with visual information gradually integrated into semantic…

Transformer-based language models exhibit complex and distributed behavior, yet their internal computations remain poorly understood. Existing mechanistic interpretability methods typically treat attention heads and multilayer perceptron…

Large language models have been successful at tasks involving basic forms of in-context reasoning, such as generating coherent language, as well as storing vast amounts of knowledge. At the core of the Transformer architecture behind such…

In this paper, we investigate how model distillation impacts the development of reasoning features in large language models (LLMs). To explore this, we train a crosscoder on Qwen-series models and their fine-tuned variants. Our results…

Multimodal large reasoning models (MLRMs) often suffer from hallucinations that stem not only from insufficient visual grounding but also from imbalanced allocation between perception and reasoning processes. Building upon recent…

Large Language Models with transformer architecture have revolutionized the domain of text generation, setting unprecedented benchmarks. Despite their impressive capabilities, LLMs have been criticized for generating outcomes that deviate…

Architectural obfuscation - e.g., permuting hidden-state tensors, linearly transforming embedding tables, or remapping tokens - has recently gained traction as a lightweight substitute for heavyweight cryptography in privacy-preserving…

Recently, the attention-enhanced multi-layer encoder, such as Transformer, has been extensively studied in Machine Reading Comprehension (MRC). To predict the answer, it is common practice to employ a predictor to draw information only from…

The incredible success of transformers on sequence modeling tasks can be largely attributed to the self-attention mechanism, which allows information to be transferred between different parts of a sequence. Self-attention allows…

Vision Transformers face a fundamental limitation: standard self-attention jointly processes spatial and channel dimensions, leading to entangled representations that prevent independent modeling of structural and semantic dependencies.…

In this work, we revisit the Transformer-based pre-trained language models and identify two different types of information confusion in position encoding and model representations, respectively. Firstly, we show that in the relative…