Multi-Horizon Consistency as Geometry: When Latent Dynamics Contract, and When They Do Not

• URL: http://arxiv.org/abs/2607.21645v1
• Date: Tue, 21 Jul 2026 23:50:10 GMT
• Title: Multi-Horizon Consistency as Geometry: When Latent Dynamics Contract, and When They Do Not
• Abstract summary: Multi-step latent agreement is a common training knob in video predictors and world models.<n>We measure an empirical expansion proxy L20,q95 together with horizon-20 prediction error E20.<n>Four of six seeds cross L1 at=0.8. The same loss does not produce population L1 on action-v1 or CartPole-v1, nor on KTH video, even when E20 improves.
• Score: 0.3058685580689604
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Multi-horizon latent consistency is a common training knob in video predictors and world models, but practitioners rarely know what it does to transition geometry. We treat lambda, the weight on multi-step latent agreement, as a diagnostic control and measure an empirical expansion proxy L20,q95 together with horizon-20 prediction error E20. On Moving-MNIST (n=6 seeds at the critical pair), raising lambda from 0 to 0.8 cuts L20 from 4.96 +/- 2.01 to 1.01 +/- 0.06 (paired t p=0.005, Wilcoxon p=0.031) and halves E20 (0.365 to 0.177, paired t p=1.1e-13). Four of six seeds cross L<1 at lambda=0.8. The same loss does not produce population L<1 on action-conditioned Pendulum-v1 or CartPole-v1, nor on KTH Actions video, even when E20 improves. An associational mediation analysis on MMNIST gives r-hat=0.94 (95% CI [0.88, 1.00], n=27, B=2000); lambda was not randomized. Defensive checks (architectural baselines, exogenous stress, WorldTest, MPC, scaling) mostly support a narrow claim: soft consistency can push passive video toward a near-contractive band, and that band is domain-limited. A stochastic-forcing law L20 ~ 1.23 + 1.82 eta at lambda=0.8 (bootstrap slope CI [1.73, 1.92], R^2=0.96) unifies control domains on the same curve via calibrated eta_eff. Complete joint slices at lambda in {0.4, 1.2} (30/30 cells, 5 eta x 3 seeds) show comparable linear L20(eta) slopes (~1.69 and ~2.00); we do not fit a continuous (lambda, eta) surface. We do not report DreamerV3 or TD-MPC2 returns.

Related papers

• Deposon: An Auditable, Conservation-Guaranteed, Game-Theoretically Tested Scattering Layer over LLM Reasoning Paths [0.0]
  Multi-step LLM reasoning lacks a machine-recheckable ledger.<n>We propose the Deposon scattering layer, which binds each node of an LLM-generated concept-decomposition graph.
  arXiv  Detail & Related papers  (2026-09-08T16:40:31Z)
• Replicating TRACE: A Practitioner's Guide to Its Threshold and Particle Budget [0.10923877073891443]
  TRACE reads causal graphs out of a pretrained autoregressive sequence model.<n>Mean per-sequence F1 against exact interventional truth reaches 0.90-0.91 at vocabulary size 1000.
  arXiv  Detail & Related papers  (2026-09-01T11:48:16Z)
• An Omitted Mode Is a Rare Rule: The Sampling-Verification Danger Law in Continuous Code World Models [0.0]
  GPT-5.x repairs an omitted 1D clamp in 105 of 111 mode-containing synthesis draws.<n>On 2D regions no artifact recovers the rule (0/156); eight targeted interventions leave the failure in place, and positive controls locate it.<n>We prove a class of entry rules exactly consistent with every sample yet harmless at play, so identifiability is a measurable property of the instrument.
  arXiv  Detail & Related papers  (2026-08-18T16:09:51Z)
• WDL-OPD: Weak-Driven On-Policy Distillation via Mixture-Constrained Co-Training [48.290951447286744]
  WDL-OPD is a mixture-constrained co-training method with two trainable policies.<n>An anchor policy generates every rollout, an auxiliary policy evaluates the same visited states, and a geometric mixture of their token distributions is matched to a frozen teacher by reverse KL.<n>We show that freezing the auxiliary recovers an anchor-plus-contrast proxy target closely related to OPD$2$ and W2S-OPD, whereas joint training creates branch-level degrees of freedom that a static delta cannot express.
  arXiv  Detail & Related papers  (2026-08-10T11:22:53Z)
• Fixed-Protocol Amortized MPS Tomography with Conformalized Predictive Uncertainty [31.082191748525137]
  Quantum state tomography is sample-starved, and the states one prepares live on a narrow, learnable manifold.<n>A prior estimate is already near-optimal, so high fidelity at few measurements'' can be family memorization rather than tomography.<n>We propose a emph-protocol amortized MPS estimator trained once with a gauge-invariant fidelity loss.
  arXiv  Detail & Related papers  (2026-07-13T08:57:23Z)
• Complex spacing ratio statistics in the partially open asymmetric quantum baker map [36.94429692322632]
  We study the complex eigenvalue statistics of the asymmetric quantum baker map with partial projective openings.<n>We find a smooth crossover from a quasi-1D spectral regime, where eigenvalues cluster near the unit circle and the phase distribution of $z$ is peaked.<n>Both the number of open channels $M$ and the reflectivity $$ modulate this crossover, and $$ provides an additional continuous control even at fixed opening size.
  arXiv  Detail & Related papers  (2026-07-08T19:24:35Z)
• Stable Geometry, Reversing Poles: The Bipolar Structure of AI Occupational Substitutability and Its Decade-Scale Inversion [0.0]
  We decompose 1,961 O*NET Detailed Work Activities into 15,817 micro-actions using a multi-agent LLM pipeline with 31-expert HITL calibration.<n>We project the DWA-level Occupational Automation Index from our prior work onto a 7-macro semantic typology.
  arXiv  Detail & Related papers  (2026-06-06T02:00:32Z)
• 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)
• Measuring Maximum Activations in Open Large Language Models [60.3514350516308]
  We measure global and layerwise maxima on 27 checkpoints from 8 open families spanning dense, MoE, vision-language, intermediate-training, and instruction-tuned variants.<n>We conclude that maximum activation magnitude is a model property tied to family, architecture, and training stage - not a simple byproduct of size.
  arXiv  Detail & Related papers  (2026-05-15T03:31:51Z)
• Reframing preprocessing selection as model-internal calibration in near-infrared spectroscopy: A large-scale benchmark of operator-adaptive PLS and Ridge models [0.2609784101826761]
  This paper studies the case where that search can be collapsed into one calibration step.<n> Linear operator-adaptive calibration gives comparable prediction quality to exhaustive preprocessing screening.
  arXiv  Detail & Related papers  (2026-05-13T14:23:00Z)
• The Geometry of Forgetting: Temporal Knowledge Drift as an Independent Axis in LLM Representations [50.43168858368539]
  Large language models confidently produce outdated answers, and no existing method can detect them.<n>We show this is not an engineering failure but a structural one: temporal drift is encoded as a direction in the residual stream geometrically to both correctness and uncertainty.
  arXiv  Detail & Related papers  (2026-05-09T22:27:31Z)
• The Extrapolation Cliff in On-Policy Distillation of Near-Deterministic Structured Outputs [52.709361620508595]
  ListOPD brings a student to in-domain with an 8B-SFT baseline at one-fifth the parameters.<n>On Amazon Fashion, three pre-registered tests--a fine-grid cliff interval test, and a small-clip cross-prediction--fall within their locked prediction windows--with the small-clip value matching the closed-form prediction below grid resolution.
  arXiv  Detail & Related papers  (2026-05-09T06:48:00Z)
• Affinity Is Not Enough: Recovering the Free Energy Principle in Mixture-of-Experts [0.0]
  Sfinity MoE fails at domain transitions, where the current token belongs to one distribution and the next to another.<n>Three lightweight gate modifications raise this to 0.748 +/- 0.002 (124x)<n>We show that beta-routing reduces transition-step B from 6.56 +/- 0.01 to 4.01 +/- 0.15 (beta-MoE)
  arXiv  Detail & Related papers  (2026-05-01T12:18:55Z)
• Zoom Consistency: A Free Confidence Signal in Multi-Step Visual Grounding Pipelines [0.880899367147235]
  Multi-step zoom-in pipelines are widely used for GUI grounding.<n> intermediate outputs contain a useful confidence signal for free: zoom consistency.<n>We prove this quantity is a linear estimator of step-1 spatial error under idealized conditions.
  arXiv  Detail & Related papers  (2026-04-15T20:47:08Z)
• Unregularized Linear Convergence in Zero-Sum Game from Preference Feedback [50.89125374999765]
  We provide the first convergence guarantee for Optimistic Multiplicative Weights Update ($mathtOMWU$) in NLHF.<n>Our analysis identifies a novel marginal convergence behavior, where the probability of rarely played actions grows exponentially from exponentially small values.
  arXiv  Detail & Related papers  (2025-12-31T12:08:29Z)
• Identity-Link IRT for Label-Free LLM Evaluation: Preserving Additivity in TVD-MI Scores [3.959606869996232]
  We show that averaging TVD-MI's binary trials yields centered-probability scores with additive structure suitable for item-response theory (IRT) without nonlinear link functions.<n>We derive this clipped-linear evaluations from Gini entropy, yielding a box-constrained least-squares formulation that handles boundary saturation.
  arXiv  Detail & Related papers  (2025-10-16T17:59:25Z)

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.