Truth as a Trajectory: What Internal Representations Reveal About Large Language Model Reasoning

• URL: http://arxiv.org/abs/2603.01326v1
• Date: Sun, 01 Mar 2026 23:52:10 GMT
• Title: Truth as a Trajectory: What Internal Representations Reveal About Large Language Model Reasoning
• Abstract summary: We introduce Truth as a Trajectory (TaT) for Large Language Models.<n>TaT shifts analysis from static activations to layer-wise geometric displacement.<n>We show that TaT effectively mitigates reliance on static lexical confounds, outperforming conventional probing.
• Score: 19.48395840455837
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Existing explainability methods for Large Language Models (LLMs) typically treat hidden states as static points in activation space, assuming that correct and incorrect inferences can be separated using representations from an individual layer. However, these activations are saturated with polysemantic features, leading to linear probes learning surface-level lexical patterns rather than underlying reasoning structures. We introduce Truth as a Trajectory (TaT), which models the transformer inference as an unfolded trajectory of iterative refinements, shifting analysis from static activations to layer-wise geometric displacement. By analyzing displacement of representations across layers, TaT uncovers geometric invariants that distinguish valid reasoning from spurious behavior. We evaluate TaT across dense and Mixture-of-Experts (MoE) architectures on benchmarks spanning commonsense reasoning, question answering, and toxicity detection. Without access to the activations themselves and using only changes in activations across layers, we show that TaT effectively mitigates reliance on static lexical confounds, outperforming conventional probing, and establishes trajectory analysis as a complementary perspective on LLM explainability.

Related papers

• Latent Fact-Checking: Detecting Misinformation through Activation Engineering [1.3299946892361476]
  We introduce a misinformation detection framework grounded in activation engineering.<n>Our approach elicits a misinformation direction in the residual stream by contrasting activations from paired truthful and false statements.<n>We evaluate the method across 11 models on three fact-checking benchmarks: AVeriTeC, LIAR, and FACTors.
  arXiv  Detail & Related papers  (2026-08-05T18:00:14Z)
• How Hard Does It Think? Analyzing Step-Aware Reasoning Energy in LLM Chain-of-Thought Trajectories [51.16651846289446]
  We propose Step-Aware Reasoning Energy (SARE), a geometric framework that quantifies effort at the granularity of individual CoT steps.<n>We find that reasoning energy is highly non-uniform across step types, exhibiting phase-like transitions invisible to trajectory-level metrics.<n>SARE-based features match or outperform output-based confidence baselines in most settings.
  arXiv  Detail & Related papers  (2026-07-28T06:15:51Z)
• Invariant Learning Dynamics of Transformers in Inductive Reasoning Tasks [50.79340930913805]
  We study a class of inductive tasks that unifies several synthetic tasks known in the literature, including in-context n-grams and multi-hop reasoning.<n>By casting circuit formation as a low-dimensional dynamical phenomenon, we take a step toward a predictive theory of how Transformers learn.
  arXiv  Detail & Related papers  (2026-07-13T17:56:11Z)
• Prototype-Anchored Generalized Manifold Regression for Unknown-Domain Object Detection [58.25418970608328]
  We study Single-Domain Generalized Object Detection (Single-DGOD), which aims to transfer a detector trained on a single source domain to multiple unseen domains.<n>Existing methods mainly rely on simulation-driven strategies, such as data augmentation or textual prompts, to enlarge the training distribution.<n>We propose Manifold Regression with Visual-Text Dual Chain-of-Thought (MR-DCoT), which formulates unknown-domain generalization as a manifold regression problem.
  arXiv  Detail & Related papers  (2026-07-08T09:27:15Z)
• Revisiting Anisotropy in Language Transformers: The Geometry of Learning Dynamics [5.987889368693271]
  Transformer architectures have dominated Natural Language Processing (NLP)<n>Recent research has highlighted an inherent anisotropy phenomenon in these models.<n>We use concept-based mechanistic interpretability during training, rather than only post hoc, to fit activation-derived low-rank tangent proxies.<n>We find that these activation-derived directions capture both unusually large gradient energy and a substantially larger share of gradient anisotropy than matched-rank normal controls.
  arXiv  Detail & Related papers  (2026-04-09T21:02:20Z)
• LAG-XAI: A Lie-Inspired Affine Geometric Framework for Interpretable Paraphrasing in Transformer Latent Spaces [1.3788139387418392]
  This paper introduces LAG-XAI, a geometric framework that models paraphrasing as a structured affine transformation within the embedding space.<n>Experiments on the noisy PIT-2015 Twitter corpus, encoded with Sentence-BERT, reveal a "linear transparency" phenomenon.<n>The model captures approximately 80% of the non-linear baseline's effective classification capacity (AUC 0.8405), offering explicit interpretability in exchange for a marginal drop in absolute accuracy.
  arXiv  Detail & Related papers  (2026-04-07T17:02:21Z)
• The Geometric Price of Discrete Logic: Context-driven Manifold Dynamics of Number Representations [6.414826816896125]
  Large language models (LLMs) generalize smoothly across continuous semantic spaces, yet strict logical reasoning demands the formation of discrete decision boundaries.<n>We argue that task context operates as a non-isometric dynamical operator that enforces a necessary "topological distortion"<n>We validate this geometric evolution across a gradient of tasks, from simple mapping to complex primality testing.
  arXiv  Detail & Related papers  (2026-03-24T13:41:57Z)
• FreeAct: Freeing Activations for LLM Quantization [89.97086263978058]
  Quantization is pivotal for mitigating the significant memory and computational overhead of Large Language Models.<n>FreeAct is a novel quantization framework that relaxes the static one-to-one constraint to accommodate dynamic activation disparities.<n>Experiments across dLLMs and MLLMs demonstrate that FreeAct significantly outperforms baselines, up to 5.3% performance improvement.
  arXiv  Detail & Related papers  (2026-03-02T12:02:17Z)
• Structural Disentanglement in Bilinear MLPs via Architectural Inductive Bias [0.0]
  We argue that failures arise from how models structure their internal representations during training.<n>We show analytically that bilinear parameterizations possess a non-mixing' property under gradient flow conditions.<n>Unlike pointwise nonlinear networks, multiplicative architectures are able to recover true operators aligned with the underlying algebraic structure.
  arXiv  Detail & Related papers  (2026-02-05T13:14:01Z)
• Latent Object Permanence: Topological Phase Transitions, Free-Energy Principles, and Renormalization Group Flows in Deep Transformer Manifolds [0.5729426778193398]
  We study the emergence of multi-step reasoning in deep Transformer language models through a geometric and statistical-physics lens.<n>We formalize the forward pass as a discrete coarse-graining map and relate the appearance of stable "concept basins" to fixed points of this renormalization-like dynamics.<n>The resulting low-entropy regime is characterized by a spectral tail collapse and by the formation of transient, reusable object-like structures in representation space.
  arXiv  Detail & Related papers  (2026-01-16T23:11:02Z)
• Introspection in Learned Semantic Scene Graph Localisation [7.222321327403328]
  This work investigates how semantics influence localisation performance and robustness in a self-supervised, contrastive semantic localisation framework.<n>We conduct a thorough post-hoc introspection analysis to probe whether the model filters environmental noise and prioritises distinctive landmarks over routine clutter.<n>Overall, the results indicate that the model learns noise-robust, semantically salient relations about place definition, thereby enabling explainable registration under challenging visual and structural variations.
  arXiv  Detail & Related papers  (2025-10-08T14:21:45Z)
• Large Language Models Encode Semantics in Low-Dimensional Linear Subspaces [31.401762286885656]
  Understanding the latent space geometry of large language models (LLMs) is key to their behavior and alignment.<n>We conduct a large-scale study in 11 empirical models across 6 scientific topics.
  arXiv  Detail & Related papers  (2025-07-13T17:03:25Z)
• How do Transformers Learn Implicit Reasoning? [67.02072851088637]
  We study how implicit multi-hop reasoning emerges by training transformers from scratch in a controlled symbolic environment.<n>We find that training with atomic triples is not necessary but accelerates learning, and that second-hop generalization relies on query-level exposure to specific compositional structures.
  arXiv  Detail & Related papers  (2025-05-29T17:02:49Z)
• The Shape of Adversarial Influence: Characterizing LLM Latent Spaces with Persistent Homology [4.280045926995889]
  This study focuses on how adversarial inputs systematically affect the internal representation spaces of Large Language Models.<n>By quantifying the shape of activations and neuronal information flow, our architecture-agnostic framework reveals fundamental invariants of representational change.
  arXiv  Detail & Related papers  (2025-05-26T18:31:49Z)
• TRACE for Tracking the Emergence of Semantic Representations in Transformers [10.777646083061395]
  We introduce TRACE, a diagnostic framework combining geometric, informational, and linguistic signals to detect phase transitions in Transformer-based LMs.<n>Experiments reveal that phase transitions align with clear intersections between curvature collapse and dimension stabilisation; these geometric shifts coincide with emerging syntactic and semantic accuracy.<n>This work advances our understanding of how linguistic abstractions emerge in LMs, offering insights into model interpretability, training efficiency, and compositional generalisation.
  arXiv  Detail & Related papers  (2025-05-23T15:03:51Z)
• On the Dynamics Under the Unhinged Loss and Beyond [104.49565602940699]
  We introduce the unhinged loss, a concise loss function, that offers more mathematical opportunities to analyze closed-form dynamics. The unhinged loss allows for considering more practical techniques, such as time-vary learning rates and feature normalization.
  arXiv  Detail & Related papers  (2023-12-13T02:11:07Z)
• Latent Traversals in Generative Models as Potential Flows [113.4232528843775]
  We propose to model latent structures with a learned dynamic potential landscape. Inspired by physics, optimal transport, and neuroscience, these potential landscapes are learned as physically realistic partial differential equations. Our method achieves both more qualitatively and quantitatively disentangled trajectories than state-of-the-art baselines.
  arXiv  Detail & Related papers  (2023-04-25T15:53:45Z)
• Instance-Dependent Label-Noise Learning with Manifold-Regularized Transition Matrix Estimation [172.81824511381984]
  The transition matrix T(x) is unidentifiable under the instance-dependent noise(IDN) We propose assumption on the geometry of T(x) that "the closer two instances are, the more similar their corresponding transition matrices should be" Our method is superior to state-of-the-art approaches for label-noise learning under the challenging IDN.
  arXiv  Detail & Related papers  (2022-06-06T04:12:01Z)
• Learning Conditional Invariance through Cycle Consistency [60.85059977904014]
  We propose a novel approach to identify meaningful and independent factors of variation in a dataset. Our method involves two separate latent subspaces for the target property and the remaining input information. We demonstrate on synthetic and molecular data that our approach identifies more meaningful factors which lead to sparser and more interpretable models.
  arXiv  Detail & Related papers  (2021-11-25T17:33:12Z)
• Did the Cat Drink the Coffee? Challenging Transformers with Generalized Event Knowledge [59.22170796793179]
  Transformers Language Models (TLMs) were tested on a benchmark for the textitdynamic estimation of thematic fit Our results show that TLMs can reach performances that are comparable to those achieved by SDM. However, additional analysis consistently suggests that TLMs do not capture important aspects of event knowledge.
  arXiv  Detail & Related papers  (2021-07-22T20:52:26Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.