Rollout

Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as a promising paradigm for training agentic retrieval-augmented generation (RAG) systems from outcome-only supervision. Most existing methods optimize policies from uniformly sampled rollouts, implicitly treating all trajectories as equally informative. However, trajectories differ substantially in search depth and are therefore not equally informative: deeper-search trajectories contain more retrieval decision points and provide denser direct supervision for the retrieval sub-policy. Moreover, this heterogeneity grows over training as the within-batch depth distribution shifts toward higher values, yet uniform rollout sampling remains blind to this shift. To address this, we propose CuSearch, a curriculum rollout sampling framework built on Search-Depth Greedy Allocation (SDGA), a batch-level operator that reallocates a fixed update budget toward deeper-search trajectories. SDGA-Auto always targets the deepest available trajectories in the current batch, yielding an implicit training-aligned curriculum as the depth distribution shifts upward. SDGA-Phase explicitly advances the curriculum threshold as deeper trajectories become sufficiently abundant. Experiments across model types and retrieval frameworks show that CuSearch consistently improves performance, achieving up to 11.8 exact-match points over standard GRPO on ZeroSearch. These results establish per-trajectory search depth as a reliable, annotation-free proxy for retrieval supervision density in RLVR-based agentic RAG training.

Closed-loop driving simulation requires real-time interaction beyond short offline clips, pushing current driving world models toward autoregressive (AR) rollout. Existing AR distillation approaches typically rely on frame sinks or student-side degradation training. The former transfers poorly to driving due to fast ego-motion and rapid scene changes, while the latter remains bounded by the teacher's single-pass output length and thus provides only a limited supervision horizon. A natural question is: can the teacher itself be extended via AR rollout to provide unbounded-horizon supervision at bounded memory cost? The key difficulty is that a standard teacher drifts under its own predictions, contaminating the supervision it provides. Our key insight is to make the teacher rollout-capable, ensuring reliable supervision from its own AR rollouts. This is instantiated as HorizonDrive, an anti-drifting training-and-distillation framework for AR driving simulation. First, scheduled rollout recovery (SRR) trains the base model to reconstruct ground-truth future clips from prediction-corrupted histories, yielding a teacher that remains stable across long AR rollouts. Second, the rollout-capable teacher is extended via AR rollout, providing long-horizon distribution-matching supervision under bounded memory, while a short-window student aligns to it with teacher rollout DMD (TRD) for efficient real-time deployment. HorizonDrive natively supports minute-scale AR rollout under bounded memory; on nuScenes, HorizonDrive reduces FID by 52% and FVD by 37%, and lowers ARE and DTW by 21% and 9% relative to the strongest long-horizon streaming baselines, while remaining competitive with single-pass driving video generators.

Learning-assisted hyper-heuristics can select among dispatching rules while preserving the feasibility and interpretability of constructive Job Shop Scheduling Problem (JSSP) heuristics. Their main computational cost lies in label generation rather than model fitting, since each supervised label usually requires rolling out candidate rules from a partial schedule. We study this label-cost problem together with a reliability problem: a learned selector should not switch away from a strong default rule unless the predicted gain is credible. The proposed selector uses regret-normalized rollout labels, a contextual KNN uncertainty estimate, and a gate that acts only when the predicted improvement exceeds an uncertainty-adjusted margin. We also vary rollout depth and breadth to measure the cost-quality trade-off. On synthetic JSSP instances, the gated selector achieves the lowest mean RPD among learned selectors, remains close to the best fixed dispatching rule, and reduces Random-HH mean RPD by more than an order of magnitude.

Vision-Language-Action (VLA) models remain brittle in long-horizon, contact-rich manipulation because success-only imitation provides little supervision for execution drift, while failed rollouts are often discarded. We introduce RePO-VLA, a recovery-driven policy optimization framework that assigns distinct roles to success, recovery, and failure trajectories. RePO-VLA first applies Recovery-Aware Initialization (RAI), slicing recovery segments and resetting history so corrective actions depend on the current adverse state rather than the preceding failure. It then learns a Progress-Aware Semantic Value Function (PAS-VF), aligning spatiotemporal trajectory features with instructions and successful references. The resulting labels salvage useful failure prefixes via reliability decay, while low-value labels mark drift and terminal breakdowns, teaching differences among nominal, failed, and corrective actions. The data engine turns adverse states into planner-generated or human-collected corrective rollouts, teaching recovery to the success manifold. Value-Conditioned Refinement (VCR) trains the policy to prefer high-progress actions. At deployment, a fixed high value ($v=1.0$) biases actions toward the learned success manifold without online failure detectors or heuristic retries. We introduce FRBench, with standardized error injection and recovery-focused evaluation. Across simulated and real-world bimanual tasks, RePO-VLA improves robustness, raising adversarial success from 20% to 75% on average and up to 80% in scaled real-world trials.

Self-evolving agents present a promising path toward continual adaptation by distilling task interactions into reusable knowledge artifacts. In practice, this paradigm remains hindered by two coupled bottlenecks: data inefficiency, where costly rollout effort is disproportionately spent on low-value samples rather than informative ones, and knowledge interference, where heterogeneous knowledge stored in shared repositories leads to noisy retrieval and task-misaligned guidance. Together, these issues form a self-reinforcing failure loop in which uninformative rollouts yield noisy knowledge, which in turn degrades subsequent rollouts. In this work, we introduce Ace-Skill, a co-evolutionary framework that jointly optimizes rollout allocation and knowledge organization for self-evolving multimodal agents. Specifically, Ace-Skill combines aprioritized sampler with lazy-decay proficiency tracking to focus rollouts on informative and insufficiently mastered samples, and a clustered organizer that semantically clusters knowledge for cleaner retrieval and more reliable adaptation. By improving sampling and organization together, Ace-Skill turns self-evolution into a virtuous cycle in which more informative rollouts produce higher-quality knowledge that supports stronger subsequent rollouts. Across four multimodal tool-use benchmarks, Ace-Skill delivers strong gains (e.g., +35.46% relative improvement in Avg@4 accuracy), enabling an opensource 35B MoE model to match or surpass proprietary models. The acquired knowledge also transfers effectively in a zero-shot manner to smaller 9B and 4B models, allowing resource-constrained agents to inherit advanced capabilities without additional training. The code has been publicly available at https://github.com/AMAP-ML/Ace-Skill.

Reinforcement Learning (RL) has become a cornerstone for improving the performance of Large Language Models (LLMs). However, its rollout phase constitutes a significant efficiency bottleneck, mainly arising from the long-tail bubbles across data parallel ranks, particularly in long-context scenarios where faster GPUs remain idle while waiting for stragglers. Existing solutions, such as partial rollout or asynchronous RL, mitigate these bubbles by compromising the algorithm's strict synchronous nature. Instead, we propose BubbleSpec, a novel framework that accelerates RL rollouts while strictly keeping the mathematical exactness. Instead of attempting to eliminate bubbles, BubbleSpec exploits them. We exploit the idle time windows of faster ranks to pre-generate rollout results for subsequent steps, serving as drafts for speculative decoding. Unlike prior speculative methods that rely on historical epoch similarity and warm-ups, BubbleSpec is agnostic to dataset size and provides immediate acceleration from the onset of training. Extensive evaluations demonstrate that BubbleSpec reduces decoding steps by 50% and increases rollout throughput by up to 1.8x. Critically, BubbleSpec is seamlessly compatible with various RL frameworks and strategies as it sustains the strict synchronous property of RL algorithms.

Autoregressive neural simulators now match classical solvers on short-horizon prediction of physical systems, yet their accuracy degrades rapidly when rolled out over long horizons. In this work, we identify transient amplification of perturbations around rollout trajectories as a structural mechanism driving rollout error. Using a linearization analysis we show that when the Jacobians along an autoregressive trajectory are non-normal and non-commuting, the model amplifies errors transiently, resulting in model rollout drift even when the overall system is asymptotically stable. Building on the analysis, we propose commutativity regularization: a combination of two penalties designed to reduce the normality defect of individual Jacobians and the commutator norm of Jacobians across steps. The penalties are estimated with Jacobian-vector products and have no inference-time cost. We show a propagator bound that quantifies rollout error under approximate commutativity and normality. We evaluate UNet and FNO variants with commutativity regularization on 1D and 2D spatio-temporal data in synthetic and real settings, showing successful long-horizon rollouts over thousands of steps. Further, we show that the method improves FourCastNet climate forecasts on ERA5 without using any new data. The gain is most pronounced out-of-distribution: trained on trajectories of a few hundred steps, regularized models remain in-distribution for thousands of rollout steps on initial conditions where baselines diverge.

Code-agent RL often receives weak feedback: rollout-time signals are reliable and executable, but capture only necessary or surface conditions for task success rather than the target semantic predicate. Using agentic compile-fix as the setting, we study signal reshaping for standard GRPO under such feedback. Our central claim is that GRPO's within-group comparison is meaningful only after three kinds of signals are reshaped: outcome rewards recover semantic ranking, process signals localize intra-trajectory credit, and rollouts from the same prompt remain execution-comparable. We operationalize these conditions with a minimal signal-reshaping construction that leaves GRPO's group-normalized advantage construction unchanged: compile-and-semantic layered rewards reshape trajectory ranking, step-level process scores outside group reward normalization reshape within-trajectory update strength, and failure-cause-aware rollout governance reshapes within-group comparability. Experiments show a clear end-to-end gain: full signal-reshaped GRPO improves strict compile-and-semantic accuracy from the base model's zero-shot $0.385$ to $0.535$. Controlled comparisons further explain the source of this gain: binary rewards remove the compile-only middle tier and degrade trajectory control; on top of layered rewards, process-score weighting further improves accuracy from $0.48$ to $0.53$ and reduces average evaluation steps from $23.50$ to $17.02$. As a boundary comparison, privileged-prompt token-level distillation mainly optimizes local distributional alignment; in long tool-use trajectories, this signal is diluted by non-critical tokens and cannot replace outcome semantics, process credit, or within-group comparability.

Reinforcement learning with verifiable rewards (RLVR) has emerged as a central paradigm for improving the reasoning capabilities of large language models. Group-based policy optimization methods, such as GRPO, typically allocate a fixed number of rollouts to every prompt. This uniform allocation can be inefficient: it over-allocates compute to prompts whose sampled groups are already saturated while under-exploring prompts for which additional samples may reveal useful correct trajectories. To address this limitation, we introduce hit utility, the posterior probability that at least one rollout in a proposed additional allocation for a prompt will be correct. Building on this notion, we propose Hit-Utility Optimal Rollout Allocation (HORA), a learning-free rollout allocation policy that maximizes total posterior hit utility within each allocation batch. HORA adaptively reallocates rollout budgets while leaving the downstream reward evaluation and group-based advantage estimator unchanged. Across four mathematical reasoning benchmarks and three model scales, HORA preserves comparable Pass@1 and improves Pass@K over compute-matched GRPO in ten of twelve model--benchmark configurations, with one tie and one saturated exception. It is also drop-in compatible with other group-based estimators such as RLOO. Ablation studies indicate that the uniform prior used by HORA is competitive with five prompt-conditioned learned-prior alternatives.

State-of-the-art model-based reinforcement learning methods train policies on imagined rollouts. These rollouts are trajectories generated by a learned dynamics model and are scored by a learned reward model, but without querying the true environment during policy updates. We study this training paradigm by quantifying how errors in learned dynamics and reward models affect returns and policy optimization. First, we extend the analysis of Asadi et al. (2018) to MDPs with learned reward models, and derive the optimal sample allocation--the ratio of dynamics samples to reward samples that minimizes a bound on return error under power-law scaling assumptions. We identify lower Lipschitz constants of the learned dynamics, reward, and policy as a representation desideratum that tightens this bound, and we connect this perspective to the temporal-straightening objective of Wang et al. (2026). Second, we examine how policy optimization with REINFORCE tolerates noisy rewards, which are often cheaper to obtain. We show that zero-mean reward noise leaves the gradient estimator unbiased and adds at most a variance term that decreases with the number of rollouts. This introduces a practical tradeoff: given a fixed budget, should one buy more rollouts with cheaper but noisier rewards, or fewer rollouts with more expensive but less noisy rewards? We reduce this choice to a one-dimensional optimization problem and characterize the optimum.

Group-relative RL training (GRPO) samples a small group of parallel rollouts for every training prompt and uses their within-group reward spread to compute per-trajectory advantages. In agentic environments each rollout is a long multi-turn dialogue with one LLM call per step, so this multi-sample multiplier dominates the total training cost. When every rollout of a prompt ends with the same reward, the group has zero reward variance and contributes no gradient, so the extra rollouts add no information; such groups are common in practice (typically around 40% of all groups), so the wasted-compute fraction is substantial rather than marginal. Existing methods filter such groups at the prompt level, either after their rollouts are paid for or before any rollout begins, but both decide without using information that becomes available during the rollout itself. We instead ask whether the in-group divergence between the partial trajectories at an intermediate step can already predict that the group will be zero-variance: when the parallel rollouts have already converged on the same action prefix, the group is on track to produce a single reward, and we can stop early. We propose a one-parameter gate that stops a group when the mean pairwise prefix edit distance between its partial action sequences falls below a threshold. On a 60-iteration on-policy GRPO run on ALFWorld with Qwen2.5-7B, averaged over four random seeds, the gated arm finishes 10.7% faster in wall-clock (bootstrap 95% CI excludes 0) and shifts held-out success rate on 50 unseen tasks by +2.5 pp, with the held-out gain tracing to a measurable reduction in zero-advantage gradient-batch dilution. Code is available at https://github.com/zhiyuanZhai20/selective-rollout.

Agentic reinforcement learning (RL) for software engineering spends much of its compute on stateful trajectories whose grouped binary rewards are highly skewed and weakly contrastive. We frame this as pass-rate control and show that the binary reward-side signal is strongest near a 50% rollout pass rate under four criteria: reward entropy, group-filtering survival, leave-one-out (RLOO) advantage energy under Group Relative Policy Optimization (GRPO), and success-failure pair count. We propose Prefix Sampling (PS), which replays self-generated trajectory prefixes to steer skewed groups toward this regime: successful prefixes give mostly failing groups a head start, while failing prefixes handicap mostly passing groups. Replayed states are reconstructed through the existing rollout path, and replayed tokens are masked from the loss so optimization applies only to current-policy continuations. On SWE-bench Verified, PS reaches the baseline high-score regime within evaluation variability while delivering 2.01x and 1.55x end-to-end wall-clock speedups on Qwen3-14B and Qwen3-32B; the 14B peak improves from 0.274 to 0.295. AIME 2025 experiments on 4B and 8B show the same pass-rate-control pattern, and 4B ablations attribute gains to replay, bidirectional coverage, and adaptive control.

Simulation-based planning with rollouts is a widely-deployed technique for decision making in stochastic environments. The primary instrument of simulation-based planning is a sampling model, which is repeatedly called to generate trajectories and estimate the utilities of available actions. Among the actions thus explored, one with the maximum estimated utility is then executed. In this paper, we examine the effect of using common random numbers in the simulation process. We obtain a simple recipe for (provably) reducing variance in relative utility when simulations invoke a rollout policy beyond some depth. Experiments on synthetic tasks confirm that our scheme improves task performance. The broader significance of our innovation is apparent from two practical applications: (1) single-step lookahead planning in a pension-disbursement task, and (2) a deployment of the well-known UCT algorithm for the game of Ludo.

We formalize Rollout Informativeness under a Fixed Budget (RIFB) as the expected non-vanishing policy-gradient mass that a tool-use rollout set injects into Group Relative Policy Optimization (GRPO). We prove that any budget-agnostic independent sampler suffers a collapse rate bounded away from zero for hard prompts regardless of the budget. Motivated by this, we recast intermediate state selection as a monotone submodular maximization problem, where a greedy one-step selector enjoys a 1 minus 1/e approximation guarantee. Our Uncertainty-aware Upper Confidence Bound (UUCB) terms arise as closed-form marginal gains of this objective. This turns the token-level entropy bonus from an empirical trick into an analytic consequence of the formulation. We present InfoTree, a training-time tree-search framework coupling UUCB with a learned Adaptive Budget Allocator (ABA) and an asynchronous Speculative Expansion scheme. ABA rescues prompts whose initial tree is wasted on uniform outcomes, lifting the mixed-outcome ratio from 58.1 percent to 76.3 percent with less than 5 percent budget overhead. Speculative Expansion reduces wall-clock overhead from 14.3 percent to 4.8 percent by tolerating bounded staleness in UUCB scores. Across nine benchmarks spanning math reasoning (AIME 2024 and 2025, MATH-500, OlympiadBench, USAMO), web-search agents (GAIA, HLE-100, BrowseComp-lite), and tool-rich coding and OS agents (APPS-verified, AgentBench-OS), InfoTree outperforms flat GRPO, DeepSearch, Tree-GRPO, AT2PO, CW-GRPO, and RC-GRPO. Head-to-head compositions with Tree-GRPO prefix sharing and CW-GRPO contribution weights deliver further gains, confirming that our selector operates orthogonally to rollout reuse and trajectory re-weighting. A 5 by 5 by 5 robustness grid reveals that over three quarters of the hyperparameter space lies on a performance plateau, confirming UUCB robustness.

Large-scale ranking systems depend on thousands of features derived from user behavior across multiple time horizons. Typically requires model retraining -- resulting in long iteration cycles (3--6 months), substantial GPU resource consumption, and limited rollout throughput. We introduce Intelligent Elastic Feature Fading (IEFF), a production infrastructure system that enables retrain-free feature efficiency rollouts by elastically controlling feature coverage and distribution at serving time. IEFF supports incremental feature coverage adjustments while models adapt through recurring training, eliminating dependencies on explicit retraining cycles. The system incorporates strict safety guardrails, reversibility mechanisms, and comprehensive monitoring to ensure stability at scale. Across multiple production use cases, IEFF accelerates efficiency-related rollouts by 5$\times$, eliminates retraining-related GPU overhead, and enables faster capacity recycling. Extensive offline and online experiments demonstrate that gradual feature fading prevents 50--55% of online performance degradation compared to abrupt feature removal, while maintaining stable model behavior. These results establish elastic, system-level feature fading as a practical and scalable approach for managing feature efficiency in modern industrial ranking systems.

Dense, dynamic crowds pose a persistent challenge for autonomous mobile robots. Purely reactive planning methods, such as Model Predictive Path Integral (MPPI) control, often fail to escape local minima in complex scenarios due to their limited prediction horizon. To bridge this gap, we propose Ray-based Task-Oriented Latent Dynamics (RAY-TOLD), a hybrid control architecture that integrates obstacle information into latent dynamics and utilizes the robustness of physics-based MPPI with the long-horizon foresight of reinforcement learning. RAY-TOLD leverages a LiDAR-centric latent dynamics model to encode high-dimensional sensor data into a compact state representation, enabling the learning of a terminal value function and a policy prior. We introduce a policy mixture sampling strategy that augments the MPPI candidate population with trajectories derived from the learned policy, effectively guiding the planner towards the goal while maintaining kinematic feasibility. Extensive tests in a stochastic environment with high-density dynamic obstacles demonstrate that our method outperforms the MPPI baseline, reducing the collision rate. The results confirm that blending short-horizon physics-based rollouts with learned long-horizon intent significantly enhances navigation reliability and safety.

RL post-training of frontier language models is increasingly bottlenecked by autoregressive rollout generation, making rollout acceleration a central systems challenge. Many existing efficiency methods improve throughput by changing the rollout or optimization regime, for example, through off-policy execution, replay, or lower-precision generation. We study speculative decoding as a lossless acceleration primitive for RL rollouts that preserves the target model's output distribution. We implement speculative decoding in NeMo-RL with a vLLM backend, supporting both synchronous and asynchronous pipelines and enabling speculation during RL rollouts. This benefit is realizable across speculation mechanisms, such as pretrained MTP heads, small external draft models or even techniques such as Eagle3, which are traditionally applied after RL phase. This yields a deployment path for state-of-the-art speculative decoding inside RL training. In a reasoning post-training workload at 8B scale under synchronous RL, speculative decoding improves rollout throughput by 1.8x. Using a high-fidelity performance simulator, we project that combining speculative decoding with asynchronous RL yields up to 2.5x end-to-end training speedup at 235B scale.

Large language models (LLMs) are increasingly deployed as customer-facing agents, yet evaluating their reliability remains challenging due to stochastic, multi-turn interactions. Current evaluation protocols rely on linear Monte Carlo rollouts of complete agent-user conversations to estimate success. However, this approach is computationally inefficient, repeatedly regenerating identical early prefixes, and often fails to uncover deep failure modes that arise from rare user behaviors. We introduce DIVERT (Diversity-Induced Evaluation via Branching of Trajectories), an efficient, snapshot-based, coverage-guided user simulation framework for systematic exploration of agent-user interactions. DIVERT captures the full agent-environment state at critical decision points and resumes execution from these snapshots, enabling reuse of shared conversation prefixes and reducing redundant computation. From each junction, the framework branches using targeted, diversity-inducing user responses, allowing directed exploration of alternative interaction paths. By focusing evaluation on semantically diverse and underexplored trajectories, DIVERT improves both efficiency and coverage. Empirical results show that it discovers more failures per token compared to standard linear rollout protocols, while expanding the set of tasks on which failures are identified.

Deep learning surrogates for CFD flow-field prediction often rely on large, complex models, which can be slow and fragile when data are noisy or incomplete. We introduce FlowForge, a staged local rollout engine that predicts future flow fields by compiling a locality-preserving update schedule and executing it with a shared lightweight local predictor. Rather than producing the next frame in a single global pass, FlowForge rewrites spatial sites stage by stage so that each update conditions only on bounded local context exposed by earlier stages. This compile-execute design aligns inference with short-range physical dependence, keeps latency predictable, and limits error amplification from global mixing. Across PDEBench, CFDBench, and BubbleML, FlowForge matches or improves upon strong baselines in pointwise accuracy, delivers consistently better robustness to noise and missing observations, and maintains stable multi-step rollout behavior while reducing per-step latency.

Test-time scaling has become a powerful way to improve large language models. However, existing methods are best suited to short, bounded outputs that can be directly compared, ranked or refined. Long-horizon coding agents violate this premise: each attempt produces an extended trajectory of actions, observations, errors, and partial progress taken by the agent. In this setting, the main challenge is no longer generating more attempts, but representing prior experience in a form that can be effectively selected from and reused. We propose a test-time scaling framework for agentic coding based on compact representations of rollout trajectories. Our framework converts each rollout into a structured summary that preserves its salient hypotheses, progress, and failure modes while discarding low-signal trace details. This representation enables two complementary forms of inference-time scaling. For parallel scaling, we introduce Recursive Tournament Voting (RTV), which recursively narrows a population of rollout summaries through small-group comparisons. For sequential scaling, we adapt Parallel-Distill-Refine (PDR) to the agentic setting by conditioning new rollouts on summaries distilled from prior attempts. Our method consistently improves the performance of frontier coding agents across SWE-Bench Verified and Terminal-Bench v2.0. For example, by using our method Claude-4.5-Opus improves from 70.9% to 77.6% on SWE-Bench Verified (mini-SWE-agent) and 46.9% to 59.1% on Terminal-Bench v2.0 (Terminus 1). Our results suggest that test-time scaling for long-horizon agents is fundamentally a problem of representation, selection, and reuse.

Ambiguity is an inherent property of natural-language agent specifications. When a system prompt leaves behaviour underdetermined, identical inputs follow divergent execution paths and produce inconsistent outcomes. The standard remedy is prompt optimisation: propose candidate prompts, run the agent to score them, and keep the best. This loop pays for the agent twice: once to generate candidates and again to rank them. On a tool-using agent whose rollouts cost dollars and minutes, the ranking cost dominates and budget-constrained optimisers routinely fail to find improvements. We present Semantic Feature Analysis (SFA), a pipeline that repairs agent specifications without running any search. SFA reads execution traces the agent has already produced, clusters the outputs of each workflow node, decomposes them into semantic feature classes using an extended subject-verb-object schema, ranks those features by their contribution to outcome separation using a decision tree, and injects the surviving features as corrective statements into the affected node's system prompt. Because it never ranks candidate prompts, it never spends a rollout on selection. We evaluate SFA against five prompt optimisers (GEPA, MIPROv2, SIMBA, BootstrapFewShot with random search, InferRules) and a single-reflection control, budget-matched in dollars at three budget levels across four benchmarks (IF-Bench, HotpotQA, HoVer, and GAIA). SFA consistently improves over the unmodified agent across benchmarks and budget levels, with the largest gains where rollouts are most expensive. On GAIA, where budget-constrained optimisers cannot afford to score even one candidate, SFA improves accuracy while other arms return their seed unchanged.

Video generative models (VGMs) pretrained on large-scale internet data can produce temporally coherent rollout videos that capture rich object dynamics, offering a compelling foundation for zero-shot robotic manipulation. However, VGMs often produce physically implausible rollouts, and converting their pixel-space motion into robot actions through geometric retargeting further introduces cumulative errors from imperfect depth estimation and keypoint tracking. To address these challenges, we present EmboAlign, a data-free framework that aligns VGM outputs with compositional constraints generated by vision-language models (VLMs) at inference time. The key insight is that VLMs offer a capability complementary to VGMs: structured spatial reasoning that can identify the physical constraints critical to the success and safety of manipulation execution. Given a language instruction, EmboAlign uses a VLM to automatically extract a set of compositional constraints capturing task-specific requirements, which are then applied at two stages: (1) constraint-guided rollout selection, which scores and filters a batch of VGM rollouts to retain the most physically plausible candidate, and (2) constraint-based trajectory optimization, which uses the selected rollout as initialization and refines the robot trajectory under the same constraint set to correct retargeting errors. We evaluate EmboAlign on six real-robot manipulation tasks requiring precise, constraint-sensitive execution, improving the overall success rate by 43.3 percentage points over the strongest baseline without any task-specific training data.

Existing reinforcement learning methods for LLM reasoning implicitly assume that the policy generating training trajectories should coincide with the one producing inference responses. We argue that this is a misleading inductive bias: the optimization-optimal trajectory distribution favors informative gradients, whereas the inference-optimal response distribution emphasizes accuracy and consistency. Forcing both into a single policy entangles their gradients and suppresses exploration. We propose R$^2$PO (Residual Rollout Policy Optimization), which attaches a lightweight Residual Rollout-Head atop the policy to decouple training trajectories from inference responses, diversifying rollouts during training while keeping inference generation intact. Experiments show that R$^2$PO consistently outperforms baselines, with average accuracy gains of 3.4% on MATH-500 and 1.3% on APPS, alongside more diverse rollouts and reduced length bias. Our code is available at https://github.com/RRPO-ARR/Code.

Mobile GUI agents have shown strong potential in real-world automation and practical applications. However, most existing agents remain reactive, making decisions mainly from current screen, which limits their performance on long-horizon tasks. Building a world model from repeated interactions enables forecasting action outcomes and supports better decision making for mobile GUI agents. This is challenging because the model must predict post-action states with spatial awareness while remaining efficient enough for practical deployment. In this paper, we propose MobileDreamer, an efficient world-model-based lookahead framework to equip the GUI agents based on the future imagination provided by the world model. It consists of textual sketch world model and rollout imagination for GUI agent. Textual sketch world model forecasts post-action states through a learning process to transform digital images into key task-related sketches, and designs a novel order-invariant learning strategy to preserve the spatial information of GUI elements. The rollout imagination strategy for GUI agent optimizes the action-selection process by leveraging the prediction capability of world model. Experiments on Android World show that MobileDreamer achieves state-of-the-art performance and improves task success by 5.25%. World model evaluations further verify that our textual sketch modeling accurately forecasts key GUI elements.

Determining where to execute the manipulation policy is a fundamental challenge in mobile manipulation. Most approaches have formulated this as a geometric search problem, prioritizing physical reachability. However, given the high sensitivity of modern learning-based manipulation policies, geometric criteria alone are insufficient. Optimal performance requires base positioning that is aware of the policy's preference. While recent works have attempted to address this, they remain limited in practicality due to reliance on pre-built scene reconstruction and slow inference. In this work, we introduce N2M that systematically reformulates the approach to base positioning problem, naturally overcoming limitations of previous methods. Our key insight is that policy preferences are inherent to the local scene structure and can be effectively learned from the policy rollouts. Technically, we propose a novel viewpoint augmentation strategy that enables the model to learn robust, viewpoint-invariant pose preferences with remarkable data efficiency. Extensive experiments demonstrate that N2M achieves state-of-the-art performance, outperforming both non-policy-aware baselines and recent policy-aware alternatives. Furthermore, we provide a comprehensive analysis highlighting N2M's broad applicability, generalization capabilities, and data efficiency. Project website: https://clvrai.github.io/N2M/