Beyond State Consistency: Behavior Consistency in Text-Based World Models

• URL: http://arxiv.org/abs/2604.13824v1
• Date: Wed, 15 Apr 2026 12:56:45 GMT
• Title: Beyond State Consistency: Behavior Consistency in Text-Based World Models
• Abstract summary: We introduce a new behavior-aligned training paradigm aimed at improving the functional consistency between the world model and the real environment.<n> Experiments on WebShop and TextWorld show that BehR-based training improves long-term alignment in several settings.
• Score: 58.216587360435305
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: World models have been emerging as critical components for assessing the consequences of actions generated by interactive agents in online planning and offline evaluation. In text-based environments, world models are typically evaluated and trained with single-step metrics such as Exact Match, aiming to improve the similarity between predicted and real-world states, but such metrics have been shown to be insufficient for capturing actual agent behavior. To address this issue, we introduce a new behavior-aligned training paradigm aimed at improving the functional consistency between the world model and the real environment. This paradigm focuses on optimizing a tractable step-level metric named Behavior Consistency Reward (BehR), which measures how much the likelihood of a logged next action changes between the real state and the world-model-predicted state under a frozen Reference Agent. Experiments on WebShop and TextWorld show that BehR-based training improves long-term alignment in several settings, with the clearest gains in WebShop and less movement in near-ceiling regimes, while preserving or improving single-step prediction quality in three of four settings. World models trained with BehR also achieve lower false positives in offline surrogate evaluation and show modest but encouraging gains in inference-time lookahead planning.

Related papers

• IMPACT: Attention Is the Interaction Map for Scalable Interaction-Aware World Model Training [18.292611164084537]
  We develop IMPACT, a scalable Interaction-aware framework with Prior-guided Attention and Targeting.<n> IMPACT consistently outperforms the corresponding MSE-trained baselines, improving interaction fidelity, physical plausibility, and visual quality.
  arXiv  Detail & Related papers  (2026-08-31T18:00:36Z)
• Do Robotic World Models Really Follow Actions? Diagnosing and Aligning Action-Conditioned Generation for Policy Learning [53.77164209208635]
  We introduce WorldEcho, which probes action following over a broader action distribution using visual integrity and trajectory alignment.<n>We propose WorldSync, which strengthens action following along three axes: distributional coverage, representational grounding, and intervention-effect alignment.<n>Experiments on RoboTwin benchmarks and real-robot tasks show that WorldSync improves WorldEcho metrics and serves as a more reliable simulator for iterative policy improvement.
  arXiv  Detail & Related papers  (2026-08-25T17:59:49Z)
• Self-Evolving World Models for LLM Agent Planning [72.99782619992361]
  We introduce WorldEvolver, a self-evolving world model framework that revises its deployment-time context while keeping the downstream agent and all model parameters frozen.<n>WorldEvolver achieves the highest prediction accuracy across three backbones and leads other world model baselines on downstream agent success rate.
  arXiv  Detail & Related papers  (2026-06-29T17:58:43Z)
• COMAP: Co-Evolving World Models and Agent Policies for LLM Agents [14.918267305899619]
  COMAP is a novel framework that co-evolves textual world models and agent policies through closed-loop interaction.<n>At each decision step, the world model predicts future state feedback for candidate actions, and the agent performs future-aware reflection by estimating the reliability of this feedback.<n>The resulting on-policy trajectories are then used to update the world model via self-distillation, allowing it to better match the agent's evolving interaction distribution.
  arXiv  Detail & Related papers  (2026-06-01T15:21:17Z)
• Masked Diffusion Language Models are Strong and Steerable Text-Based World Models for Agentic RL [0.8546394944138487]
  We formalize text-based world modeling as a steerable transition-dynamics problem decomposed into initial state.<n>We curate 239,403 grounded state-action trajectories spanning nine open-source environments and twelve frontier model families.<n>We introduce a plug-and-play GRPO training framework with deterministic state checks, and perform zero-shot transfer ablations on three OOD environments.
  arXiv  Detail & Related papers  (2026-05-07T00:40:32Z)
• World Action Verifier: Self-Improving World Models via Forward-Inverse Asymmetry [82.93104394404781]
  General-purpose world models promise scalable policy evaluation, optimization, and planning.<n>We propose World Action Verifier (WAV), a framework that enables world models to identify their own prediction errors and self-improve.
  arXiv  Detail & Related papers  (2026-04-02T12:48:36Z)
• Self-Correcting VLA: Online Action Refinement via Sparse World Imagination [55.982504915794514]
  We propose Self-Correcting VLA (SC-VLA), which achieve self-improvement by intrinsically guiding action refinement through sparse imagination.<n>SC-VLA achieve state-of-the-art performance, yielding the highest task throughput with 16% fewer steps and a 9% higher success rate than the best-performing baselines.
  arXiv  Detail & Related papers  (2026-02-25T06:58:06Z)
• Test-Time Mixture of World Models for Embodied Agents in Dynamic Environments [29.514831254621438]
  Test-time Mixture of World Models (TMoW) is a framework that enhances adaptability to unseen and evolving domains.<n>TMoW updates its routing function over world models at test time, unlike conventional MoE where the function remains fixed.<n>We evaluate TMoW on VirtualHome, ALFWorld, and RLBench benchmarks, demonstrating strong performance in both zero-shot adaptation and few-shot expansion scenarios.
  arXiv  Detail & Related papers  (2026-01-30T07:06:40Z)
• Agent2World: Learning to Generate Symbolic World Models via Adaptive Multi-Agent Feedback [51.22403664895878]
  Agent2World is a tool-augmented multi-agent framework that achieves strong inference-time world-model generation.<n>It also serves as a data engine for supervised fine-tuning, by grounding generation in multi-agent feedback.
  arXiv  Detail & Related papers  (2025-12-26T18:54:14Z)
• Generative Actor Critic [74.04971271003869]
  Generative Actor Critic (GAC) is a novel framework that decouples sequential decision-making by reframing textitpolicy evaluation as learning a generative model of the joint distribution over trajectories and returns.<n>Experiments on Gym-MuJoCo and Maze2D benchmarks demonstrate GAC's strong offline performance and significantly enhanced offline-to-online improvement compared to state-of-the-art methods.
  arXiv  Detail & Related papers  (2025-12-25T06:31:11Z)
• nuPlan-R: A Closed-Loop Planning Benchmark for Autonomous Driving via Reactive Multi-Agent Simulation [2.585002881750625]
  We present nuPlan-R, a new reactive closed-loop planning benchmark.<n>Our benchmark replaces the rule-based IDM agents with noise-decoupled diffusion-based reactive agents.<n>We extend the benchmark with two additional metrics to enable a more comprehensive assessment of planning performance.
  arXiv  Detail & Related papers  (2025-11-13T15:23:30Z)
• Adapting World Models with Latent-State Dynamics Residuals [10.892848566977369]
  ReDRAW is a latent-state autoregressive world model pretrained in simulation and calibrated to target environments.<n>It enables RL agents to be optimized with imagined rollouts under corrected dynamics and then deployed in the real world.
  arXiv  Detail & Related papers  (2025-04-03T03:41:30Z)
• STAT: Towards Generalizable Temporal Action Localization [56.634561073746056]
  Weakly-supervised temporal action localization (WTAL) aims to recognize and localize action instances with only video-level labels. Existing methods suffer from severe performance degradation when transferring to different distributions. We propose GTAL, which focuses on improving the generalizability of action localization methods.
  arXiv  Detail & Related papers  (2024-04-20T07:56:21Z)
• ReCoRe: Regularized Contrastive Representation Learning of World Model [21.29132219042405]
  We present a world model that learns invariant features using contrastive unsupervised learning and an intervention-invariant regularizer. Our method outperforms current state-of-the-art model-based and model-free RL methods and significantly improves on out-of-distribution point navigation tasks evaluated on the iGibson benchmark.
  arXiv  Detail & Related papers  (2023-12-14T15:53:07Z)

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.