arXiv:2606.17195v2 Announce Type: replace
Abstract: Newly formed ions in $E \times B$ devices are rapidly accelerated by strong radial electric fields and execute large cycloidal orbits in the presence of an axial magnetic field. At locations where these orbits intersect, ions originating from different birth radii arrive with substantially different velocities, producing a non-Maxwellian velocity distribution with a large velocity variance. Through Coulomb collisions and collective interactions, this distribution relaxes toward a drifting Maxwellian in the rotating frame. Here, we present the first optical emission spectroscopy (OES) measurements of the line-of-sight-convolved ion-velocity distribution, from which an apparent ion temperature is determined, in Avalanche Energy's centrifugal mirror machine. High-resolution $H_\alpha$ spectra obtained along five chordal lines-of-sight spanning the plasma radius are analyzed using two complementary models representing limiting cases of the ion dynamics: a collisionless cycloidal model based on the ion-velocity distribution arising from deterministic single-particle orbits, and a rotating Gaussian model based on collisions and collective processes that fully randomize the cycloidal motion into a drifting Maxwellian in the rotating frame. Combined, these approaches bracket the possible degree of velocity-space relaxation and provide a stringent test of the inferred ion energies. Both models reproduce the measured spectra relatively well and yield density-weighted apparent ion temperatures of $1.40\pm0.43$ keV for the rotating Gaussian model and $1.55\pm0.24$ keV for the cycloidal model. These results provide direct spectroscopic evidence that strong $E \times B$ rotation in a device only a few centimeters in size can generate ion populations with keV energy spreads.
Science Journals
arXiv:2606.17390v2 Announce Type: replace
Abstract: We present a fully differentiable, GPU-accelerated finite element framework JAX-FEM-ANISO, for forward simulation and inverse parameter identification of finite-strain anisotropic plasticity. Built on JAX-FEM, the framework exploits modern accelerator architectures by parallelizing the three major computational bottlenecks in nonlinear FEM: elemental weak-form and tangent-stiffness evaluation, global sparse matrix assembly, and sparse linear solution. For a large-scale forward problem with 3 million degrees of freedom, JAX-FEM-ANISO on a single NVIDIA H100 GPU achieves up to 9.4$\times$ speed-up over a 24-core CPU Abaqus baseline. Automatic differentiation is applied through the constitutive update and solver workflow, providing consistent Jacobians for complex constitutive models without manual derivation and accurate gradients for PDE-constrained inverse analysis. Compared with finite differences, the JAX-AD gradients avoid step-size sensitivity and provide the required sensitivities at substantially lower computational cost. For inverse characterization, we combine information-rich, topology-optimized heterogeneous specimens with full-field displacement data to identify advanced constitutive model parameters from a single test, replacing what would otherwise require many conventional experiments. We demonstrate accurate recovery of anisotropic yield and hardening parameters in progressively challenging settings, including uniform and spatially varying material properties. The resulting AD-based formulation enables efficient optimization in high-dimensional parameter spaces where finite-difference approaches are computationally infeasible. These results establish differentiable, GPU-accelerated FEM as a practical high-throughput engine for simulation, characterization, and optimization workflows in advanced manufacturing.
arXiv:2606.18618v2 Announce Type: replace
Abstract: We study weak shock reflection from a rigid wall in the joint limit of weak shock strength and nearly glancing incidence. In the distinguished scaling $\Mach=1+\lambda\alpha^2$, the inner reflection region is governed by the unsteady transonic small-disturbance (UTSD equation and is controlled, to leading order, by the single parameter $a_0=1/(2\sqrt{\lambda})$, independent of the ratio of specific heats $\gamma$. Thus the known UTSD detachment value $a_d=\sqrt2$ corresponds in this scaling to $\lambda_d=1/8$, with Guderley--Mach reflection for $\lambda>1/8$. The physical trajectory angle is obtained by multiplying the canonical UTSD trajectory function $g(a)$ by the Mach-number strength scale $\delta=\sqrt{2(\Mach^2-1)}$, so that $\chi_{\rm phys}=\delta g(a)+O(\delta^2)=2\sqrt{\lambda}\,\alpha g(a_0)+O(\alpha^3)$. We rederive the self-similar UTSD reduction, sonic parabola, and shock polar in order to make the convention and the detachment map self-contained. We also record a formal adjoint solvability expression for the first correction $H(a;\gamma)$, while specifying the free-boundary data required to evaluate it. Finally, a time-marching solver for the full leading-order canonical UTSD system is benchmarked at $a_0=0.5$: retaining the transverse compression $u>1$ gives a $u=0.5$ contour location consistent with the Hunter--Tesdall triple-point benchmark. This computation is used only as a leading-order benchmark, not as a substitute for an adaptive self-similar Guderley free-boundary solver.
arXiv:2606.19493v3 Announce Type: replace
Abstract: The Bures--Helstrom metric is the minimal monotone Riemannian metric on the state space of a qubit. With the quantum Fisher normalization used here, it identifies the Bloch ball with a geodesic hemisphere of the unit round three--sphere. We describe its Ricci flow explicitly. In a general rotationally symmetric gauge the flow is a coupled system for the radial lapse and warping factor; a single scalar equation appears only after a Hamilton--DeTurck gauge choice. In the corresponding moving DeTurck frame the squared warping function $\Psi=\Phi^2$ satisfies the linear forced heat equation \begin{equation*} D_t\Psi=\Psi_{ss}-2, \end{equation*} while the fixed-lapse coordinate form contains the associated transport term. Since the Bures--Helstrom metric is Einstein, the geometric flow itself is the homothetic shrinker \begin{equation*} g(t)=(1-4t)g_{\mathrm{BH}}, \end{equation*} with scalar curvature $6/(1-4t)$ and extinction time $T=1/4$. Thus the metric remains inside the monotone cone for all $t<T$ and leaves the cone of nondegenerate Riemannian metrics only through the collapsed limit. We also record the volume--normalized flow, for which the Bures--Helstrom metric is a fixed point. Its linearization is the shifted round--sphere Laplacian $\Delta_{\mathbb S^3}+3$, with spectrum \begin{equation*} \sigma_\ell=-(\ell-1)(\ell+3), \end{equation*} and spectral gap $5$ after removal of the scaling mode.
arXiv:2606.20780v2 Announce Type: replace
Abstract: This report introduces the concept, prototype design, projected costs, and scientific goals of a mobile experiment for detecting geoneutrinos originating from uranium and thorium decay chains in the Earth's mantle. This will constrain the planet's radiogenic heat production and unearth its geochemical makeup. This design of a deep-ocean mobile neutrino experiment, which is not mirrored by any active or planned experiments, supports physics and geoscience's goal of multi-modal data on the Earth's internal composition and structure. Based on geoscientific studies, this design is expected to achieve a 50--100-fold reduction in crustal background compared to similarly sized continental detectors, thereby enabling direct measurements of mantle geoneutrinos. The multiple stereoscopic projections enabled by the detector's unique mobility can map spatial variations in heat-producing elements within the mantle. Beyond discussing the design, we report on our collaboration's most recent hardware developments in the active prototyping of this detector. We briefly highlight the potential multiuse and interdisciplinary nature of this detector.
arXiv:2606.21535v2 Announce Type: replace
Abstract: High-dimensional vector spaces, particularly embedding spaces with dense semantic structure, are often interpreted primarily leveraging solely geometric relationships. In this work, we show that they can also be viewed as spectral energy networks induced by the topology of their underlying feature-space manifold with relevant improvements for downstream tasks. Building on this perspective, we introduce Graph Wiring, a general framework for exploiting feature-space spectral structure, together with Spectral Indexing, its task-specific instantiation for vector search. By coupling geometric similarity with spectral information, the proposed method improves Head-Tail coherence and semantic alignment relative to purely geometric retrieval methods. It further supports adaptive search behavior through tau-modulation, providing the flexibility increasingly required by modern Retrieval-Augmented Generation (RAG) pipelines. We present the complete algorithmic pipeline, establish its theoretical foundation through epiplexity, and evaluate the approach across benchmark and industrial settings using the open-source arrowspace library.
arXiv:2606.21587v2 Announce Type: replace
Abstract: Deep reinforcement learning is pivotal for closed-loop autonomous driving yet remains constrained by severe bottlenecks in sampling efficiency. Standard parallel sampling mitigates this but suffers from the straggler effect, where the premature termination of a single environment necessitates a synchronized batch re-initialization, leading to suboptimal sample utilization and prohibitive re-initialization latency. To address this, we propose FAST, a synchronous parallel framework tailored for closed-loop simulation. Specifically, FAST employs Dynamic Parallel Sampling Alignment (DPSA) to maintain vectorization synchronization by extending terminated episodes via virtual continuation, thereby decoupling the sampling loop from individual terminations. By dynamically triggering global truncation based on the termination rate of parallel clips, FAST effectively eliminates the bottleneck of premature resets without sacrificing data diversity. Furthermore, to strictly preserve theoretical consistency, we incorporate a Scaled Mask-Padding Optimization (SMPO) that leverages validity masking and adaptive loss normalization to nullify the bias from auxiliary padding data. Empirical evaluations demonstrate that FAST achieves at least a 1.78 times wall-clock speedup over the single-clip baseline while preserving statistical unbiasedness.
arXiv:2606.11324v2 Announce Type: replace
Abstract: We introduce Embodied-R1.5, a unified Embodied Foundation Model (EFM) that integrates comprehensive embodied reasoning capabilities, spanning embodied cognition, task planning, correction, and pointing, within a single architecture toward general physical intelligence. Leveraging three automated data construction pipelines to significantly expand the data coverage of critical capabilities, we build a large-scale data system of over 15B tokens, and design a multi-task balanced RL recipe to alleviate heterogeneous task conflicts. We further introduce a Planner-Grounder-Corrector (PGC) closed-loop framework that enables a single model to autonomously execute and self-correct over long-horizon tasks. With only 8B parameters, Embodied-R1.5 achieves SOTA on 16 out of 24 embodied VLM benchmarks, surpassing leading models like Gemini-Robotics-ER-1.5 and GPT-5.4. Benefiting from the internalized embodied capabilities, Embodied-R1.5 can be fine-tuned into a VLA with only a small amount of data, outperforming leading VLA models like $\pi_{0.5}$ across 4 popular manipulation benchmark suites. We further conduct extensive zero-shot real-robot experiments, validating performance in instruction following, affordance grounding, articulated object manipulation, and long-horizon complex tasks, demonstrating strong generalization to the physical world. We open-source model weights, datasets, training code, and EmbodiedEvalKit, an evaluation framework tailored for embodied tasks, to facilitate future research in EFMs.
arXiv:2606.24890v3 Announce Type: replace
Abstract: Can a small group of volunteers shape how AI systems discuss animal welfare, just by editing Wikipedia? We show that they can. Wikipedia appears in nearly every major language model training dataset and is weighted more heavily than web-crawled text. The Pro-Animal Wikipedians (PAW), a group of advocates who add sourced animal welfare content to relevant articles, have made 125 edits across 115 pages. Using gradient-based data attribution (Bergson; MAGIC), we traced how these edits influence language model behavior. TrackStar retrieval attribution on Llama 3.1 8B found that PAW-edited sections made up 68 percent of the highest-attributed documents for animal welfare queries (p < 0.0001) but only 52 percent for unrelated queries about the same companies (p = 0.53): the model links PAW content specifically to animal welfare topics, not to the entities in general. MAGIC counterfactual influence estimation on Llama-3.2-1B, run across five random training-order seeds, gave the same picture even more sharply: in every seed, the top-10 most influential documents on animal welfare queries were all PAW edits (10 of 10, 5 of 5 seeds), while on general queries the same top-10 sat at chance (4 to 6 of 10). Mean PAW influence exceeded mean control influence on animal welfare queries with p < 0.0001 in every seed, an effect 6 to 30 times larger than on general queries. Leave-subset-out validation gave Spearman rho = 1.00 for all 10 runs. When we fine-tuned separate models on PAW content versus control content, each model performed better specifically on the type of text it was trained on: the PAW-trained model cut perplexity on animal welfare text from 12.4 to 8.4, while the control-trained model cut perplexity on control text from 16.1 to 11.4. A small, coordinated Wikipedia editing campaign therefore measurably shapes how language models handle the topics those edits address.
arXiv:2606.24894v3 Announce Type: replace
Abstract: Large language models have shown strong fluency in scientific writing, yet the evaluation of related work generation (RWG) remains limited. Existing RWG evaluations largely inherit summarization-oriented metrics, using lexical or semantic similarity to reference sections as proxies for quality. However, related work writing is fundamentally a citation-level scholarly positioning task: it requires selecting, organizing, and framing prior work to clarify how a target paper relates to, differs from, and contributes beyond existing research.As a result, models may generate coherent and semantically-relevant text while exhibiting academically critical failures, such as inappropriate citation selection or misplaced references, that conventional metrics do not capture.To this end, we introduce \textbf{RWGBench}, a benchmark that evaluates RWG from the perspective of citation decision-making rather than text similarity. RWGBench is constructed from a large-scale collection of 40,108 computer science papers and a retrieval corpus of 1.09 million documents, with a carefully curated test set comprising 100 papers and their corresponding published related work sections.We propose a multi-dimensional evaluation framework that assesses citation selection, contextual appropriateness, organization, and discourse structure.Experiments reveal systematic limitations in current systems that are obscured by standard evaluations, while Oracle studies further disentangle retrieval-level and generation-level bottlenecks. Human evaluation further shows that our citation-centric metrics align substantially better with expert judgment than surface-level text metrics. RWGBench offers a citation-centric testbed for developing and evaluating related work generation systems that are better aligned with scholarly writing practices.
arXiv:2606.25059v2 Announce Type: replace
Abstract: Black-box LLMs (accessible only via API) are vulnerable to distillation attacks, in which an attacker queries the model and trains a student on its outputs. A recent line of work proposes output perturbation defenses that modify the teacher's output to reduce student performance while preserving utility for legitimate users. As a relatively new family of approaches, output perturbation defenses lack a shared threat model, making it difficult to compare them, reason about composing them with other attacks, or evaluate their robustness against realistic adversaries. This underspecification matters beyond technical evaluation: when defenses are deployed to protect intellectual property or justify regulatory compliance, an imprecise threat model can create a false sense of security. We propose a threat model framework that describes attackers along three dimensions: a query budget, a data budget, and an interface profile that captures how attackers interact with the API. Using antidistillation sampling as a case study, we show that whether the defense is considered effective depends on the assumed threat model. We argue that future work on distillation defenses, along with any governance or policy frameworks built around them, should explicitly specify and stress-test attacker capabilities along our three dimensions.
arXiv:2606.25680v2 Announce Type: replace
Abstract: Underwater vehicles operate from a fixed onboard energy budget that propulsion rapidly depletes, so a controller that completes its task while drawing less thruster power directly extends mission range and endurance. Reinforcement learning yields capable model-free controllers for station-keeping and trajectory tracking, but optimizing task accuracy alone drives the policy toward oscillatory, energy-wasting actuation. The established remedy subtracts an energy penalty from the reward, yet this sets the task-power trade-off through a single weight with no physical units: a target power level cannot be specified, the weight must be re-tuned for every vehicle and task, and a mismatched weight can even raise power. This paper instead formulates energy-efficient underwater control as a constrained Markov decision process in which average thruster power is subject to an explicit budget, solved with a PPO-Lagrangian algorithm. The power level is set by declaring a budget in physical units, and a single dual variable is updated online to meet it for each vehicle and task, without manual weight search. Across three vehicles and four tasks in the MarineGym simulator, the energy-constrained policy draws the least power in all twelve settings, reducing it by 14--65\% (up to 64.9\%) over a task-only baseline and below an energy-reward baseline everywhere, while remaining the smoothest in ten settings and preserving task accuracy except in one deliberately power-limited regime. Imposing energy as an explicit constraint thus offers a tuning-free route to energy-efficient underwater control that needs no per-vehicle, per-task weight search.
arXiv:2606.25847v3 Announce Type: replace
Abstract: This text is intended to offer a consistent presentation of the theory of collisions and stopping of charged particles in matter, limited to the range of intermediate kinetic energies where atomic aggregation effects are relatively unimportant and processes such as the creation of particle-antiparticle pairs are not likely to occur. The first three Chapters contain introductory material on the classical description of electromagnetic fields in matter, an overview of quantum wave equations for a particle in a central potential, and an account of elementary atomic-structure models. Chapters 4 and 5 are devoted to the classical and quantum theories of elastic collisions of charged particles with atoms. The theory of inelastic collisions and stopping is split into two parts: first, collisions with atoms are considered within the plane-wave Born approximation in Chapter 6, which includes a derivation of the Bethe stopping power formula; second, the theory of inelastic collisions in dense materials is based on the dielectric formalism, which is formulated for the electron gas, and extended to arbitrary materials by means of optical-data models in Chapter 7. Chapter 8 offers a detailed review of the theory of stopping, starting with the classical study by Bohr and ending with derivations of the Bloch and Barkas corrections to the stopping power. Chapter 9 deals with general aspects of transport theory, including derivations of energy-straggling distributions and multiple-scattering distributions, which are the basis for condensed simulation schemes of charged particle transport. Finally, Chapter 10 describes the Fortran programs elastic and sbethe, which implement the main theoretical models presented in the preceding Chapters and are distributed as ancillary information.
Helpfulness Hurts: Domain-Dependent Degradation of Mid-Trained Compassion Values Under Post-Training
arXiv:2606.26102v3 Announce Type: replace
Abstract: Standard post-training pipelines apply supervised fine-tuning (SFT) and reinforcement learning (RL) to make language models helpful, but these processes may inadvertently degrade values instilled during pre-training. We investigate whether the domain of post-training data differentially affects the retention of animal compassion values in a Llama 3.1 8B model mid-trained on compassion-oriented synthetic data, using both SFT (helpfulness via Dolly-15k vs. coding via Magicoder-110K) and GRPO (helpfulness via RLHFlow vs. coding via Magicoder), evaluated on the ANIMA 2.2 benchmark and MORU benchmark (Moral Reasoning Under Uncertainty). Helpfulness training significantly degrades animal compassion relative to coding training on ANIMA (SFT: 35.7% vs. 65.2%; GRPO: 18.7% vs. 32.0%), replicating across two independent helpfulness datasets and two training paradigms. On English MORU items, helpfulness training degrades general moral reasoning by 25.5 percentage points (46.4% vs. 71.9%), a striking gap that rivals the compassion effect in magnitude. However, this effect does not transfer cross-lingually: on the multilingual MORU benchmark, the domain effect disappears (SFT: 52.3% vs. 51.2%). In contrast, the animal compassion effect transfers consistently across languages, with Magicoder's ANIMA percentage-point gain over the base model 4.5 times larger on non-English items than English items. This divergence suggests that values instilled through mid-training are encoded more deeply and cross-lingually than reasoning improvements from domain-specific post-training. These results suggest that, for labs building on value-laden mid-training, coding-domain post-training may better preserve mid-trained values than helpfulness post-training without harming general reasoning capabilities.
arXiv:2606.26104v3 Announce Type: replace
Abstract: Animal-welfare advocates produce a lot of writing, and increasingly that writing trains the language models that millions of people then ask about animal welfare. Using vocabulary-matched stance-contrast probes on a held-out animal-welfare benchmark, we measure how each of ten linguistic features changes Llama-3.2-1B's preference for pro-animal-welfare reasoning when used as fine-tuning data. Eight of the ten features produce statistically significant shifts. Seven move the model toward stronger pro-animal-welfare reasoning: assertive certainty, explicit moral vocabulary, emotion words, evaluative claims, narrative structure, depicted harm severity, and immediate temporal framing. Two move it the other way: hedged language and concrete sensory description both dilute the pro-animal-welfare stance. First-person perspective has no statistically significant effect. The practical recommendation for anyone writing animal-welfare text that may end up in LLM training corpora: assert a position rather than describe a scene neutrally. The features that shift the model are the ones that make the writer's position explicit; the features that dilute it hold animal-welfare content but withhold stance.
arXiv:2606.26515v3 Announce Type: replace
Abstract: Test Time Training (TTT) is a mechanism in which a model adapts to an incoming test-sample by performing some self-supervised (SSL) task and updating its weights even during inference. This procedure does not require labels at test-time. This paper focuses on TTT for long-videos. A major concern with existing approaches is: 1) they perform TTT updates using a sliding window containing frames in the past, whose compute increases linearly with the size of window. This becomes computationally intractable when the videos are hours long. 2) TTT is performed even when temporally close frames look similar, thereby consuming a lot of compute.
We present the Frame Forgetting Network (FFN) that: 1) operates on only three frames within the sliding window, namely the frame that exits, the current frame and the frame after that. The model still manages to retain temporal context and work for hours long-videos; 2) mathematically define a surprise metric: how much new information the incoming frame contains with respect to the past seen frame. This facilitates determining how to modify the effective window size during TTT and constitutes the core mechanism of an adaptive windowing algorithm. Additionally, we curate a dataset EpicTours containing up to 3 hour long videos of walking city-tours, whereas earlier datasets on this problem were only 5 min long. We demonstrate FFNs empirical effectiveness on dense-segmentation, video classification tasks, generalization to depth-estimation, and multi-hour long videos.
arXiv:2606.26795v2 Announce Type: replace
Abstract: Video Diffusion Models (VDMs) is constrained by immense computational costs. While offline calibration-based acceleration suffers from calibration data dependency, prohibitive calibration duration, and susceptibility to distribution shifts, offline calibration-free methods eliminate these hurdles. However, since they rely on instantaneous zero-order approximations where the mapping between input and output differences varies in real-time, they are susceptible to observational noise and ignore the intrinsic momentum within the diffusion trajectory. In this paper, we propose NaviCache, a plug-and-play test-time self-calibration method re-conceptualizing feature evolution as an Inertial Navigation System (INS) problem. NaviCache bridges the fundamental domain gap and the non-stationary nature of diffusion by modeling the relative coupling between input and output variations. We introduce a dual-state estimation architecture that adaptively tracks the feature change ratio and its latent drift, initialized via a specialized Initial Alignment phase. By integrating a time-dependent noise schedule with an uncertainty-aware Measurement Update mechanism, NaviCache provides a theoretically grounded mechanism for error-bounded computation skipping. Extensive experiments on the HunyuanVideo, Wan, and Open-Sora series demonstrate that NaviCache exhibits more accurate error judgment for computation skipping and achieves outstanding comprehensive performance.
arXiv:2606.27537v5 Announce Type: replace
Abstract: Video generation models aspire to simulate dynamic environments, and several benchmarks now evaluate memory consistency across frames. However, most assess consistency only while the target remains in view, and the few that force objects out of view evaluate static scenes where nothing changes during occlusion. To bridge this gap, we introduce MemoBench, a diagnostic benchmark built around the disappear-and-reappear paradigm in dynamically changing environments: a target object undergoes a physical process, disappears from view, and must be correctly recovered in its updated state upon reappearance. We curate 360 ground-truth clips spanning synthetic and real-world scenes, and design an evaluation suite combining automated metrics with VQA-based assessment across four diagnostic pillars. Evaluation of eight state-of-the-art models reveals key insights and open challenges regarding memory consistency under the disappear-and-reappear paradigm.
arXiv:2606.27677v2 Announce Type: replace
Abstract: World-action models have shown promising robot-manipulation performance by jointly predicting future visual states and actions. However, existing methods mainly rely on short-term history and short-horizon future prediction, which is insufficient for long-horizon tasks whose correct execution depends on earlier observations and task progress. Such temporally dependent tasks require effective use of complementary temporal information, including recent local context, cross-stage historical events, immediate future dynamics, and global task progress. To address long-term forgetting and poor awareness of the global task state, we introduce DiM-WAM, a memory-augmented world-action model that integrates multi-scale historical context, local future dynamics, and global task progress. The memory extracts compact visual event information from real observations, updates multiple memory banks through independent similarity-based merging, and then reads the bank-identity- and time-embedded long-term context to condition video and action denoising. A progress-supervision objective further encourages memory tokens to encode not only completed historical events but also the current task stage and its implications for the remaining task. On RMBench, DiM-WAM raises average success from 28.4% with LingBot-VA to 69.8%, exceeding the explicit-memory Mem-0 baseline at 42.0%. On four real-world Franka tasks, it improves average stage success from 70.7% to 91.5% and full-task success from 52.5% to 80.0%. Project page: https://wangkai-casia.github.io/dim-wam.
arXiv:2607.10147v1 Announce Type: new
Abstract: Automated chest X-ray report generation has recently benefited from reinforcement learning (RL) and large language models. However, RL training often suffers from instability or limited exploration due to fixed Kullback-Leibler (KL) regularization and a static reference policy that accumulates KL pressure over time. We propose Response-Weighted and Validation-Anchored Policy Optimization (REVA-PO), a RL framework that stabilizes long-term training via Response-Weighted Regularization (RER) and Validation-Anchored Policy Reset (VAPR). RER dynamically adjusts per-response KL weights based on advantage and reference-policy entropy, relaxing constraints for high-quality responses while tightening them for low-quality ones. Complementarily, VAPR periodically synchronizes the reference and current policies to the best validation checkpoint, resetting accumulated regularization pressure to expand the viable exploration space. To ensure a robust starting point, we employ a three-stage pipeline consisting of warm-up training, classifier-guided supervised fine-tuning, and RL. Extensive evaluations on MIMIC-CXR and IU-Xray demonstrate that REVA-PO sets new state-of-the-art benchmarks in both linguistic quality and clinical accuracy. Notably, BLEU-4 improves by 5.1% on MIMIC-CXR and 3.6% on IU-Xray, while CheXpert F1 and RadGraph F1 scores increase by 4.5% and 12.8%, respectively, over prior leading methods. The code is publicly available at https://github.com/LiGuo12/REVA_PO/.
arXiv:2606.28533v2 Announce Type: replace
Abstract: Sequence learning has emerged as the promising paradigm in recommendation systems, surpassing traditional Deep Learning Recommendation Models (DLRM) by capturing the temporal nuances of user behavior. However, current state-of-the-art architectures operate under a limiting analogy: they treat user history as a monolithic chronological sequence like a sentence in a Large Language Model (LLM). We observe a fundamental divergence between natural language and recommendation data: unlike the linear, logical flow of text, user history is inherently multi-faceted. A user's journey is a fragmented reflection of diverse interests, resulting in much weaker coherence between items than is found in LLM training data. This lack of structural unity leads to context pollution. In single-sequence modeling, unrelated behaviors compete for the same attention budget. This "noisy" signal dilutes the model's focus, effectively capping its ability to discern high-intent patterns from background activity. To address this, we propose Constructive Multi-Sequence Learning (CMSL), a paradigm shift from passive sequence ingestion to active "context engineering" that constructs multiple coherent sequences in latent space. CMSL leverages a learnable Sequence Construction Module to disentangle user history into "pure" thematic strands, followed by a linear attention mechanism to efficiently model these strands at scale. CMSL has been deployed across ranking and retrieval tasks and across four major surfaces at Meta.
arXiv:2606.28622v2 Announce Type: replace
Abstract: Differentiable rendering has emerged as a powerful approach for 3D reconstruction and novel view synthesis. State-of-the-art differentiable rendering methods combine a variety of custom representations of 3D geometry and appearance with specialized renderers. However, most downstream tasks in computer graphics rely on 3D meshes. While prior work has attempted differentiable rendering with mesh representations, these approaches are limited to object-centric scenes and fail to reconstruct large-scale, unbounded scenes. In this work, we introduce Meshtryoshka, a novel mesh differentiable rendering framework that combines an off-the-shelf triangle rasterizer with a 3D representation that consists of nested mesh shells which resemble a matryoshka doll. In every forward pass, the mesh shells are extracted anew from a 3D signed distance function via iso-surface extraction, and the opacities for each vertex are computed as a function of signed distance. Each mesh shell is then rasterized independently, and the final image is created via alpha compositing. Crucially, mesh vertex positions are updated only indirectly via gradients that flow through the opacity values into the signed distance function, and hence, our method is compatible with off-the-shelf mesh renderers that need not be differentiable with respect to vertex positions. On object-centric scenes, our method performs competitively with surface-based differentiable rendering techniques. Our differentiable mesh rendering method scales to unbounded, real-world 3D scenes, where it yields high-quality novel view synthesis results approaching those of state-of-the-art, non-mesh methods. Our method suggests that it may be possible to solve the differentiable rendering problem without relying on specialized renderers, only using conventional tools from the computer graphics toolbox.
arXiv:2606.28748v3 Announce Type: replace
Abstract: Temporal boundaries in time-modulated media provide a powerful route toward wave manipulation beyond conventional spatial boundaries. Here, we investigate nonlinear temporal boundaries generated by interaction quenches in a synthetic lattice with exactly solvable interacting dynamics. Unlike conventional temporal boundaries arising from abrupt changes of single-particle dispersion, the present system realizes a self-induced temporal medium in which the propagating wave packet dynamically determines its own effective dispersion and transport properties. By solving the nonlinear Schr\"odinger dynamics analytically, we show that the interaction generates an emergent wave-packet-dependent band structure and a state-dependent temporal refractive response while preserving fully controllable evolution. Based on this framework, we establish a nonlinear temporal-scattering picture and uncover phenomena including amplitude-dependent temporal reflection/refraction and nonlinear temporal birefringence. Furthermore, we demonstrate that gradient-induced Bloch oscillations suppress wave-packet diffusion and enable coherent periodic transport with exact state reconstruction. Our results extend temporal reflection and refraction from dispersion-quenched linear systems to interaction-quenched nonlinear media and provide a tractable framework for nonlinear wave manipulation in synthetic lattices.
arXiv:2606.28925v2 Announce Type: replace
Abstract: Tool and agent routing from natural-language prompts is naturally a set-valued prediction problem: a single query may require multiple agents, while over-selection increases execution cost. The benchmark introduced here is derived from WildChat and contains 3,000 prompts over a fixed 12-agent catalog, with AI-assisted heuristic labels under a fixed schema and controlled rebalancing for multi-label evaluation. The evaluation protocol combines set-level metrics (Precision, Recall, F1, Jaccard, and Exact Match), latency, an execution-oriented capability-coverage simulation, and a constrained weighted-routing setting based on ordinal agent-cost tiers. Compared methods include nearest-neighbor matching, linear multilabel classification, dependency-aware baselines, a fine-tuned encoder, deterministic weighted post-scoring via Weighted Agent Routing (WAR), and a zero-shot LLM baseline. Results show that supervised routers substantially outperform nearest-neighbor and zero-shot LLM routing. The fine-tuned encoder achieves the strongest unconstrained set accuracy, while the linear multilabel model provides the strongest practical baseline. In the constrained setting, the weighted routing layer improves utility when applied on top of strong supervised scorers, with the largest gain observed for Encoder+WAR. Overall, the benchmark and evaluation protocol support reproducible study of accuracy-cost trade-offs in fixed-catalog multi-agent routing.
arXiv:2606.29151v3 Announce Type: replace
Abstract: Semantic query processing engines (SQPEs) extend relational query processing with semantic operators that are executed via model inference over unstructured data. Optimizing such queries is inherently multi-objective: model inference dominates latency and monetary cost, and outputs are stochastic and backend-dependent, so quality must be optimized alongside efficiency. Existing SQPE optimizers do not expose each semantic operator instance's intermediate task outputs as a relational optimization object, leaving optimization unable to filter, reorder, route, threshold, or jointly tune them. We present CADENZA, which compiles each semantic operator instance--a template bound to a natural-language intent--into an intent-specific plan space of typed task DAGs and selects an executable plan under user-specified quality-latency-cost trade-offs. CADENZA introduces task-extended relational algebra (TxRA), a conservative extension of relational algebra with task-specific operators. The logical planner synthesizes seed TxRA plans, applies structural rewrites whose safety conditions are checked from operator dependencies, and enumerates semantics-guided alternatives from alternative-generation templates. The physical planner compiles each task-specific operator into a router over heterogeneous backends and jointly tunes routing cutpoints, backend parameters, and relational thresholds with Bayesian optimization. On SemBench, CADENZA improves the scenario-level averages of quality, latency, and cost by up to +0.49, 165.7x, and 310.3x, respectively, relative to state-of-the-art.