Temporally Centered SIGReg Improves Multi-Task LeWorldModel Learning: From Analysis to Method

• URL: http://arxiv.org/abs/2607.26924v2
• Date: Fri, 31 Jul 2026 16:12:58 GMT
• Title: Temporally Centered SIGReg Improves Multi-Task LeWorldModel Learning: From Analysis to Method
• Abstract summary: We show that marginal Gaussianization compresses the separation between task-dependent latent clusters.<n>To address this problem, we apply SIGReg to temporally centered residuals rather than to the latent marginal distribution.<n>Our method improves downstream success on the long-horizon suite by 1.7x and raises the average success rate across four suites from 53.2% to 73.6%.
• Score: 35.288206438331436
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Recent work on LeWorldModel (LeWM) has shown that the Sketched Isotropic Gaussian Regularizer (SIGReg) enables stable end-to-end world-model learning from pixels by regularizing the latent marginal distribution toward an isotropic Gaussian, thereby preventing representation collapse. While effective and elegant in single-task settings, this recipe does not extend reliably to multi-task training, leading to substantially worse downstream behavior-cloning performance. In this paper, we show that marginal Gaussianization compresses the separation between task-dependent latent clusters relative to within-cluster variation. This compression introduces representation aliasing across tasks and states, and makes the learned representations highly sensitive to small visual perturbations. To address this problem, we apply SIGReg to temporally centered residuals rather than to the latent marginal distribution. This surrogate target places no direct regularization pressure on the separation among cluster centers, removes the requirement that the full latent follow a single isotropic Gaussian, and retains the anti-collapse effect of SIGReg. On the LIBERO benchmark, our method improves downstream success on the long-horizon suite by 1.7x and raises the average success rate across four suites from 53.2% to 73.6%. Without external pretraining, it slightly outperforms Diffusion Policy trained from scratch and approaches the performance of large-scale pretrained policy baselines. These results reveal a structural incompatibility between marginal Gaussian priors and multi-task latent structure, and provide a simple route toward stable and scalable end-to-end multi-task world-model learning.

Related papers

• Hierarchical Exponential-Gaussian Mixtures for Watch-Time Distribution Prediction [74.05279072166697]
  Exponential-Gaussian Mixture Network (EGMN) models the full conditional WT distribution.<n>EGMN is vulnerable to variance collapse, component redundancy, and inactive components.<n>We propose a Hierarchical Exponential-Gaussian Mixture (HEGM) model that addresses these failure modes.
  arXiv  Detail & Related papers  (2026-08-24T15:06:18Z)
• Expanding SPHERE-JEPA: A Family of Statistical Regularizers for the Hypersphere [11.087393998848254]
  In Self-Supervised Learning (SSL), preventing representation collapse by explicitly enforcing a uniform distribution on the unit hypersphere has proven to be effective.<n>Current frameworks typically rely on sliced statistical regularizers such as SIGReg (used in LeJEPA) and SUSReg (used in SPHEREJEPA)<n>This injects projection variance into the training gradients, destabilizing optimization, and hindering convergence.<n>In this work, we first show that analytically integrating out these random projections yields a deterministic Maximum Mean Discrepancy (MMD), bypassing the variance of sliced methods.
  arXiv  Detail & Related papers  (2026-06-16T07:10:05Z)
• Closed-Form Spectral Regularization for Multi-Task Model Merging [96.82449201305234]
  Model merging combines several independently fine-tuned experts into a single multi-task model without any training data.<n>State-of-the-art merging methods formulate merging as a layer-wise interference problem.<n>We propose SWUDI, a closed-form method that combines a soft exponential filter, which matches the gradient-flow trajectory of iterative descent.
  arXiv  Detail & Related papers  (2026-06-05T14:00:47Z)
• Extra-Merge: Tracing the Rank-1 Subspace of Model Merging in Language Model Pre-Training [56.323119575322146]
  Extra-Merge consistently outperforms standard merging baselines in experiments across GPT-2 and LLaMA families.<n>It yields consistent zero-shot accuracy gains on Pythia-12B downstream tasks and generalizes effectively to the Muon citepjordan2024muon.
  arXiv  Detail & Related papers  (2026-05-26T02:48:34Z)
• Too Correct to Learn: Reinforcement Learning on Saturated Reasoning Data [55.84428098924793]
  We propose Constrained Uniform Top-K Sampling (CUTS), a parameter-free decoding strategy enforcing structure-preserving exploration.<n>We integrate this into Mixed-CUTS, a training framework synergizing exploitative and exploratory rollouts to amplify intra-group advantage variance.<n> Notably, Mixed-CUTS improves Pass@1 accuracy on the challenging AIME25 benchmark by up to 15.1% over standard GRPO.
  arXiv  Detail & Related papers  (2026-04-20T16:43:28Z)
• Improving Search Agent with One Line of Code [68.58667107354253]
  Tool-based Agentic Reinforcement Learning (TARL) has emerged as a promising paradigm for training search agents to interact with external tools.<n>We propose textbfSearch textbfAgent textbfPolicy textbfOptimization (textbfSAPO), which stabilizes training via a conditional token-level KL constraint.
  arXiv  Detail & Related papers  (2026-03-10T04:07:39Z)
• Weak-SIGReg: Covariance Regularization for Stable Deep Learning [0.0]
  This work adopts Sketched Isotropic Gaussian Regularization ( SIGReg), recently introduced in the LeJEPA self-supervised framework, and repurposes it as a general optimization stabilizer for supervised learning.<n>Inspired by interacting particle systems, representation collapse is viewed as drift; SIGReg constrains the representation density towards an isotropic Gaussian, mitigating this drift.<n> Empirically, SIGReg recovers the training of a ViT on CIFAR-100 from a collapsed 20.73% to 72.02% accuracy without architectural hacks and significantly improves the convergence of deep vanillas trained with pure SGD. Code is available at
  arXiv  Detail & Related papers  (2026-03-06T05:32:40Z)
• Functionality-Oriented LLM Merging on the Fisher--Rao Manifold [14.349284217707575]
  Weight-space merging aims to combine multiple fine-tuned LLMs into a single model without retraining.<n>We derive a practical fixed-point algorithm using a lightweight spherical proxy that preserves norms and generalizes directly to multi-expert merging.
  arXiv  Detail & Related papers  (2026-03-05T09:08:38Z)
• Compound Batch Normalization for Long-tailed Image Classification [77.42829178064807]
  We propose a compound batch normalization method based on a Gaussian mixture. It can model the feature space more comprehensively and reduce the dominance of head classes. The proposed method outperforms existing methods on long-tailed image classification.
  arXiv  Detail & Related papers  (2022-12-02T07:31:39Z)
• Meta-Learning Adversarial Bandits [49.094361442409785]
  We study online learning with bandit feedback across multiple tasks, with the goal of improving average performance across tasks if they are similar according to some natural task-similarity measure. As the first to target the adversarial setting, we design a meta-algorithm that setting-specific guarantees for two important cases: multi-armed bandits (MAB) and bandit optimization (BLO) Our guarantees rely on proving that unregularized follow-the-leader combined with multiplicative weights is enough to online learn a non-smooth and non-B sequence.
  arXiv  Detail & Related papers  (2022-05-27T17:40:32Z)

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.