Do Foundation Models Know Geometry? Probing Frozen Features for Continuous Physical Measurement

• URL: http://arxiv.org/abs/2603.06459v1
• Date: Fri, 06 Mar 2026 16:48:27 GMT
• Title: Do Foundation Models Know Geometry? Probing Frozen Features for Continuous Physical Measurement
• Abstract summary: Vision-language models encode geometry that their text pathway fails to express.<n>LoRA fine-tuning (r=16, 2,000 images) narrows this gap to 6.5 degrees.<n>These findings enable a single frozen backbone to function as a multi-task geometric sensor.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Vision-language models encode continuous geometry that their text pathway fails to express: a 6,000-parameter linear probe extracts hand joint angles at 6.1 degrees MAE from frozen features, while the best text output achieves only 20.0 degrees -- a 3.3x bottleneck. LoRA fine-tuning (r=16, 2,000 images) narrows this gap to 6.5 degrees, providing evidence for a pathway-training deficit rather than a representational one. Training objective determines accuracy more than architecture: five encoders spanning self-supervised, contrastive, and hybrid paradigms converge to statistically equivalent accuracy (R^2 approximately 0.55, TOST-equivalent at delta=0.03) despite sharing as little as CKA=0.41 representational similarity -- functional convergence without representational convergence. Autoregressive generation damages geometric fidelity, but the damage originates in the generation process, not in language alignment: Qwen2.5-VL's LLM layers actually improve probe accuracy over its raw vision encoder. Layer-wise analysis reveals a universal mid-network accuracy peak across all architectures, with attention heads in layers 18-22 carrying disproportionate geometric signal. These findings enable a single frozen backbone to function as a multi-task geometric sensor through lightweight probes, without fine-tuning or text generation.

Related papers

• LoRC: Detecting AI-Generated Images via Low-Rank Collapse in Semantic Residuals [50.157719466558696]
  We show that modern generators exhibit low-rank collapse in the semantic-residual subspace while preserving the dominant semantic direction.<n>We propose textbfLoRC, a framework that decouples semantic dominance to capture the collapsed residual geometry.<n>Our method improves accuracy by an average of 7.0% across multiple benchmarks and achieves 97.0% accuracy on 39 unseen generators.
  arXiv  Detail & Related papers  (2026-08-21T08:58:47Z)
• Architecture-Dependent Causal Transfer of Activation States Across Large Language Models [51.56484100374058]
  Internal language activation states can be transferred causally between different large language models.<n>Injecting projected activations into a target model during generation produces a statistically significant, preregistered causal effect.<n>We conclude that end-to-end activation-state transfer between LLMs, as currently implemented, is architecture-dependent rather than universal.
  arXiv  Detail & Related papers  (2026-08-17T09:53:27Z)
• GLID: Gated Local Intrinsic Dimension Repairs the Blind Spots of Face-Forgery Detectors [4.564848621539382]
  Fine-tuned foundation-model detectors dominate face-forgery benchmarks, yet they stay blind to generator families absent from training.<n>We present GLID, a detector that repairs this blind spot with geometry instead of data.<n>GLID treats the patch tokens of a single image as a sample from a manifold and estimates their local intrinsic dimension (LID) at several depths of a frozen vision transformer.
  arXiv  Detail & Related papers  (2026-07-21T06:52:46Z)
• FoundationGeo: Learning Spatial Pixel-Wise Fields for Monocular Metric Geometry [52.232826445242644]
  FoundationGeo bridges relative and metric prediction via spatial calibration and principled data design.<n>Stage 1 learns a high-fidelity, affine-invariant geometry model by initializing with DINOv3.<n>Stage 2 moves beyond global scaling by introducing pixel-wise calibration fields for metric estimation.
  arXiv  Detail & Related papers  (2026-07-13T14:10:01Z)
• Trajectory Geometry of Transformer Representations Across Layers [0.5599792629509229]
  We recast the transformer forward pass as a discrete population trajectory through a high-dimensional representation manifold.<n>We characterize trajectory geometry using five metrics computed directly in the ambient space.
  arXiv  Detail & Related papers  (2026-06-08T09:54:07Z)
• Convergence Theory for Iterative LLM-Based Neural Architecture Search: A Parametric Cross-Entropy Framework with Closed-Form Proxy Reliability [48.83701310501069]
  Large language models (LLMs) are increasingly used as generators in iterative neural architecture search (NAS)<n>We model iterative LLM-NAS as a parametric Cross-Entropy (CE) method over executable programs.<n>We prove six results: (1) iterative LLM fine-tuning on elite architectures is equivalent to the CE update restricted to the LLM family; (2) expected architecture quality is monotonically non-decreasing across cycles; and (3) elite-set probability converges to a fixed point at a geometric rate.
  arXiv  Detail & Related papers  (2026-05-28T15:45:19Z)
• Amplifying, Not Learning: Fine-Tuned AI Text Detectors Amplify a Pretrained Direction [51.56484100374058]
  Text detectors amplify a pretrained typicality axis; they do not construct an AI-vs-human boundary.<n>On raw encoders before any task supervision, projecting onto centroid(AI)-centroid(HC3) achieves NYT-vs-HC3 AUROC 0.806/0.944/0.834 across three architectures.<n>On RoBERTa-base, raw projection exceeds fine-tuning; on RoBERTa-base, full fine-tuning reduces discrimination below raw on both fluent-formal populations tested.
  arXiv  Detail & Related papers  (2026-05-20T19:08:38Z)
• Global and Local Topology-Aware Attention with Persistent Homology and Euler Biases for Time-Series Forecasting [0.0]
  Time series often encode predictive geometric structure, including connectivity, cycles, shell-like geometry, directional changes, and nonlinear neighborhoods.<n>We introduce a topology-aware attention framework that adds such structure to attention logits using persistent homology (H0-H2)<n>We evaluate guarded topology-aware variants across three architecture families: lightweight attention/Ridge, PatchTSTForRegression, and TimeSeriesTransformerForPrediction.
  arXiv  Detail & Related papers  (2026-05-04T21:19:12Z)
• ATLAS: Constitution-Conditioned Latent Geometry and Redistribution Across Language Models and Neural Perturbation Data [0.0]
  Constitution-conditioned post-training can be analysed as a structured perturbation of a model's learned representational geometry.<n>We introduce ATLAS, a geometry-first program that traces constitution-induced hidden-state structure across charts, models, and substrates.
  arXiv  Detail & Related papers  (2026-04-19T23:26:02Z)
• HeRO: Hierarchical 3D Semantic Representation for Pose-aware Object Manipulation [54.325346533275074]
  HeRO is a diffusion-based policy that couples geometry and semantics via hierarchical semantic fields.<n>In various tests, HeRO establishes a new state-of-the-art, improving success on Place Dual Shoes by 12.3% and averaging 6.5% gains across six challenging pose-aware tasks.
  arXiv  Detail & Related papers  (2026-02-21T12:29:10Z)
• GeoFocus: Blending Efficient Global-to-Local Perception for Multimodal Geometry Problem-Solving [55.14836667214487]
  GeoFocus is a novel framework comprising two core modules.<n>GeoFocus achieves a 4.7% accuracy improvement over leading specialized models.<n>It demonstrates superior robustness in MATHVERSE under diverse visual conditions.
  arXiv  Detail & Related papers  (2026-02-09T11:15:01Z)
• The Confidence Manifold: Geometric Structure of Correctness Representations in Language Models [1.45309944076563]
  We characterize correctness representations across 9 models from 5 architecture families.<n>We validate causally through activation steering.<n> correctness signal exists internally but is not expressed in outputs.
  arXiv  Detail & Related papers  (2026-02-08T23:27:10Z)
• Tethered Reasoning: Decoupling Entropy from Hallucination in Quantized LLMs via Manifold Steering [0.0]
  Quantized language models face a fundamental dilemma: low sampling temperatures yield repetitive, mode-collapsed outputs, while high temperatures (T > 2.0) cause trajectory divergence and semantic incoherence.<n>We present HELIX, a framework that decouples output entropy from hallucination by tethering hidden-state trajectories to a pre-computed truthfulness manifold.
  arXiv  Detail & Related papers  (2026-02-06T06:24:37Z)
• On the Relationship Between Representation Geometry and Generalization in Deep Neural Networks [0.0]
  We show that effective dimension -- an unsupervised geometric metric -- strongly predicts accuracy.<n>We establish causality: degrading geometry via noise causes accuracy loss (r=-0.94, $p 10(9)$), while improving geometry via PCA maintains accuracy across architectures (-0.03pp at 95% variance)<n>These results establish that effective dimension provides domain-agnostic predictive and causal information about neural network performance, computed entirely without labels.
  arXiv  Detail & Related papers  (2026-01-28T04:33:41Z)
• Robust Mesh Saliency GT Acquisition in VR via View Cone Sampling and Geometric Smoothing [59.12032628787018]
  3D mesh saliency ground truth is essential for human-centric visual modeling in virtual reality (VR)<n>Current VR eye-tracking pipelines rely on single ray sampling and Euclidean smoothing, triggering texture attention and signal leakage across gaps.<n>This paper proposes a robust framework to address these limitations.
  arXiv  Detail & Related papers  (2026-01-06T05:20:12Z)
• Geometric Latent Space Tomography with Metric-Preserving Autoencoders [0.0]
  We introduce geometric latent space tomography, combining classical neural encoders with parameterized quantum circuit decoders trained via a metric-preservation loss.<n>Our geometry-aware latent space enables direct state discrimination, fidelity estimation from Euclidean distances, and interpretable error for quantum error mitigation without repeated full tomography.
  arXiv  Detail & Related papers  (2025-12-16T20:18:10Z)
• SpatialActor: Exploring Disentangled Spatial Representations for Robust Robotic Manipulation [63.48859753472547]
  SpatialActor is a framework for robust robotic manipulation that explicitly decouples semantics and geometry.<n>It achieves state-of-the-art performance with 87.4% on RLBench and improves by 13.9% to 19.4% under varying noisy conditions.
  arXiv  Detail & Related papers  (2025-11-12T18:59:08Z)
• GeLoc3r: Enhancing Relative Camera Pose Regression with Geometric Consistency Regularization [44.00455492098006]
  Prior ReLoc3R achieves breakthrough performance with fast 25ms inference and state-of-the-art regression accuracy.<n>We present GeLoc3r, a novel approach to relative camera pose estimation that enhances pose regression methods through geometric consistency regularization.
  arXiv  Detail & Related papers  (2025-09-27T01:21:38Z)
• LaRI: Layered Ray Intersections for Single-view 3D Geometric Reasoning [75.9814389360821]
  layered ray intersections (LaRI) is a new method for unseen geometry reasoning from a single image.<n>Benefiting from the compact and layered representation, LaRI enables complete, efficient, and view-aligned geometric reasoning.<n>We build a complete training data generation pipeline for synthetic and real-world data, including 3D objects and scenes.
  arXiv  Detail & Related papers  (2025-04-25T15:31:29Z)
• Analytic Signal Phase in $N-D$ by Linear Symmetry Tensor--fingerprint modeling [69.35569554213679]
  We show that the Analytic Signal phase, and its gradient have a hitherto unstudied discontinuity in $2-D $ and higher dimensions. This shortcoming can result in severe artifacts whereas the problem does not exist in $1-D $ signals. We suggest the use of Linear Symmetry phase, relying on more than one set of Gabor filters, but with a negligible computational add-on.
  arXiv  Detail & Related papers  (2020-05-16T21:17:26Z)

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.