arXiv:2412.20750v4 Announce Type: replace Abstract: Large-scale Vision-Language Models (VLMs) have achieved notable progress in aligning visual inputs with text. However, their ability to deeply understand the unique physical properties of non-RGB vision sensor images remains limited. In this paper, we revisit and analyze these limitations and introduce a novel, cost-efficient paradigm that significantly advances sensor image understanding-without requiring extensive training data or any modifications to the existing VLM architectures. Specifically, we propose Sensor-Aware Attributes Fine-Tuning (SAFT) with the Diverse Negative Attributes (DNA) optimization, which leverages minimal sensor-specific data to enable robust learning of non-RGB characteristics and overcome RGB-centric biases inherent in current VLMs. In addition, we present VS-TDX-the first comprehensive, public benchmark designed to rigorously evaluate VLMs' sensor-specific understanding across diverse and realistic scenarios. Through extensive experiments on VLMs and various sensor modalities, we validate that our method consistently delivers superior performance and generalization under resource-constrained and architecture-invariant settings. Our approach provides a practical advance towards scalable deployment of VLMs in increasingly sensor-diverse real-world environments.
Science Journals
arXiv:2501.13515v2 Announce Type: replace Abstract: We present an adaptation of the so-called structural method \cite{CMM23} for Hamiltonian systems, and redesign the method for this specific context, which involves two coupled differential systems. Structural schemes decompose the problem into two sets of equations: the physical equations, which describe the local dynamics of the system, and the structural equations, which only involve the discretization on a very compact stencil. They have desirable properties, such as unconditional stability or high-order accuracy. We first give a general description of the scheme for the scalar case (which corresponds to e.g. spring-mass interactions or pendulum motion), before extending the technique to the vector case (treating e.g. the $n$-body system). The scheme is also written in the case of a non-separable system (e.g. a charged particle in an electromagnetic field). We give numerical evidence of the method's efficiency, its capacity to preserve invariant quantities such as the total energy, and draw comparisons with the traditional symplectic methods.
arXiv:2503.02368v4 Announce Type: replace Abstract: While guided decoding, especially value-guided methods, has emerged as a cost-effective alternative for controlling language model outputs without re-training models, its effectiveness is limited by the accuracy of the value function. We identify that this inaccuracy stems from a core distributional gap: existing methods train static value functions on trajectories sampled exclusively from the base policy, which inherently confines their training to a narrow and suboptimal view of the potential output space. We propose Iterative Value Refinement, a evolutionary framework designed to narrow this gap. It employs Value Exploration to provide a more comprehensive and robust training signal, complemented by Iterative Self-Refinement, which uses the improved value function from one iteration to guide the generation of higher-quality data for the next. Extensive experiments on text summarization, multi-turn dialogue, and instruction following demonstrate the effectiveness of our framework in aligning language models. Our approach not only achieves alignment but also significantly reduces computational costs by leveraging principled value function optimization for efficient and effective control.
arXiv:2607.04674v1 Announce Type: new Abstract: Recovering dynamic environment maps from a single in-the-wild video is crucial for photorealistic rendering, yet remains a challenge. Recent video generation models can produce photorealistic scenes with complex lighting, possessing an inherent understanding of lighting. In this paper, we introduce V-LITE (Video generation models are inherent lighting estimators), a framework that unlocks this internal knowledge by reframing lighting estimation as a guided video inpainting task. Inspired by VFX industry practices, we insert a synthetic chrome ball into the scene to compel the model to generate physically plausible reflections from the surrounding spatio-temporal context. To bridge the gap from LDR-native models to the HDR domain, we design an HDR-aware VAE and employ an efficient LoRA-based fine-tuning strategy. We then construct a mixed dataset comprising high-fidelity HDR images to provide realistic HDR priors, and in-the-wild HDR videos to provide dynamic spatio-temporal context. Extensive experiments demonstrate that V-LITE produces temporally coherent HDR environment maps, revealing that modern video diffusion models are not merely synthesizers but also powerful, inherently capable estimators of physical scene lighting.
arXiv:2607.04675v1 Announce Type: new Abstract: This paper presents the IEEE International Conference on Multimedia and Expo (ICME) 2026 Grand Challenge on Cross-Scenario Defect Detection and Fine-Grained Severity Grading for High-Precision Manufacturing. The challenge is motivated by two key limitations of existing industrial defect inspection systems: (1) current deep learning-based methods often suffer significant performance degradation when deployed in unseen production scenarios, and (2) most benchmarks neglect severity-aware assessment, which is critical for risk control and yield optimization. To address these limitations, we design two complementary tracks: Track 1 (Cross-Scenario Defect Detection) targets accurate defect detection, localization, and classification across diverse unseen production environments; Track 2 (Fine-Grained Severity Grading) requires assigning each detected defect an industry-standard severity level, including Acceptable, Marginal NG, NG, and Gross NG. We construct a large-scale industrial dataset of high-resolution microscopic images spanning seven representative defect categories, comprising over 3,800 images with pixel-level instance annotations for Track 1 and over 2,600 images with severity-grade labels for Track 2. The challenge attracted 86 registered participants with 130 submissions; during the final testing phase, 21 teams submitted results and 12 teams provided models with technical reports. The resulting benchmark, together with the diverse and effective solutions contributed by participating teams, sets a new standard for industrial defect analysis research.
arXiv:2607.05040v1 Announce Type: new Abstract: Introduced by Lee, Ko, and Shin (VLDB 2020), a hypergraph motif is a connected subhypergraph consisting of three hyperedges whose intersections satisfy a prescribed pattern. Such patterns are represented by Venn diagrams $\mathcal{V}\in\{0,1\}^7$, indicating which of the seven regions determined by three sets must be empty or non-empty. Lee et al. designed and implemented exact and approximate algorithms for counting, in a hypergraph $G$, the motifs specified by $\mathcal{V}$; their algorithms run in worst-case cubic time in the number of hyperedges of $G$. This cubic worst case can occur even for hypergraphs of bounded rank, and already for $2$-uniform hypergraphs, that is, for simple graphs. In this work, we give a complete fine-grained picture of the parameterised complexity of exact hypergraph motif counting with respect to the rank of the input hypergraph. We use $\tilde{O}$ to hide polylogarithmic factors in the input size. First, we show that every Venn diagram $\mathcal{V}$ admits an exact counting algorithm running in FPT-near-quadratic time, \[ f(\mathsf{rank}(G))\cdot \tilde{O}(|E(G)|^2), \] for some computable function $f$. Second, we precisely characterise when this can be improved to FPT-near-linear time. We prove that such an algorithm exists exactly for the degenerate Venn diagrams, namely those that force one of the three hyperedges to be fully contained in another. For all non-degenerate Venn diagrams, we show that no FPT-near-linear-time algorithm exists unless either the Triangle Hypothesis or the Hyperclique Hypothesis fails. Exact hypergraph motif counting is thus always fixed-parameter near-quadratic in the rank, and the degenerate Venn diagrams are precisely the cases admitting fixed-parameter near-linear time.
arXiv:2505.05361v2 Announce Type: replace Abstract: In this paper, we focus on the numerical analysis of quantitative photoacoustic tomography. Our goal is to reconstruct the optical coefficients, i.e., the diffusion and absorption coefficients, using multiple internal observational data. The foundation of our numerical algorithm lies in solving an inverse diffusivity problem and a direct problem associated with elliptic equations. The stability of the inverse problem depends critically on a non-zero condition in the internal observations, a condition that can be met using randomly chosen boundary excitation data. Utilizing these randomly generated boundary data, we implement an output least squares formulation combined with finite element discretization to solve the inverse problem. In this scenario, we provide a rigorous error estimate in $L^2(\Omega)$ norm for the numerical reconstruction using a weighted energy estimate, inspired by the analysis of a newly proposed conditional stability result. The resulting error estimate serves as a valuable guide for selecting appropriate regularization parameters and discretization mesh sizes according to the noise levels present in the data. Several numerical experiments are presented to support our theoretical results and illustrate the effectiveness of our numerical scheme.
arXiv:2505.15379v2 Announce Type: replace Abstract: We present the P$^3$ dataset, a large-scale multimodal benchmark for building vectorization, constructed from aerial LiDAR point clouds, high-resolution aerial imagery, and vectorized 2D building outlines, collected across three continents. The dataset contains over 10 billion LiDAR points with decimeter-level accuracy and RGB images at a ground sampling distance of 25 centimeter. While many existing datasets primarily focus on the image modality, P$^3$ offers a complementary perspective by also incorporating dense 3D information. We demonstrate that LiDAR point clouds serve as a robust modality for predicting building polygons, both in hybrid and end-to-end learning frameworks. Moreover, fusing aerial LiDAR and imagery further improves accuracy and geometric quality of predicted polygons. The P$^3$ dataset is publicly available, along with code and pretrained weights of three state-of-the-art models for building polygon prediction at https://github.com/raphaelsulzer/PixelsPointsPolygons .
arXiv:2607.04879v1 Announce Type: new Abstract: Although multi-source fusion positioning systems have achieved significant progress, accurate and reliable heading estimation remains a critical challenge due to the lack of gravitational constraints and the inherent weak observability of heading in complex environments. Most existing methodologies are specifically tailored for the startup phase, relying on a singular initial alignment to establish the heading reference. Consequently, these approaches lack the adaptability required to refine heading estimates dynamically, which renders the system highly vulnerable to accumulated drift and observation noise during prolonged navigation or immediately following GNSS signal outages. To address these limitations, this paper proposes WinTA-GIL, a novel heading refinement framework that integrates information from Global Navigation Satellite System (GNSS), Inertial Measurement Unit (IMU), and Light Detection and Ranging (LiDAR) through a temporal window-based optimization strategy. Unlike conventional alignment methods restricted to the startup phase, WinTA-GIL leverages high-precision local trajectories from LiDAR-Inertial Odometry (LIO) to register against filtered GNSS observations. This approach transforms heading estimation into a repeatable, trajectory-based consistency optimization problem. In particular, an adaptive re-estimation mechanism based on state discrimination is incorporated to trigger heading corrections whenever necessary, thereby effectively suppressing the inertial drift accumulated during challenging conditions. Extensive experiments on both open-source and self-collected datasets demonstrate that WinTA-GIL significantly outperforms state-of-the-art approaches in both estimation accuracy and system robustness.
arXiv:2506.09740v2 Announce Type: replace Abstract: Diffusion models excel at image generation. Recent studies have shown that these models not only generate high-quality images but also encode text-image alignment information through attention maps or loss functions. This information is valuable for various downstream tasks, including segmentation, text-guided image editing, and compositional image generation. However, current methods heavily rely on the assumption of perfect text-image alignment in diffusion models, which is not the case. In this paper, we propose using zero-shot referring image segmentation as a proxy task to evaluate the pixel-level image and class-level text alignment of popular diffusion models. We conduct an in-depth analysis of pixel-text misalignment in diffusion models from the perspective of training data bias. We find that misalignment occurs in images with small-sized, occluded, or rare object classes. Therefore, we propose ELBO-T2IAlign--a simple yet effective method to calibrate pixel-text alignment in diffusion models based on the evidence lower bound (ELBO) of likelihood. ELBO-T2IAlign is training-free and generic: it requires no additional annotations, model retraining, or architectural modifications, and it can be directly applied to different diffusion backbones. Extensive experiments on zero-shot referring image segmentation, text-guided image editing, and compositional image generation verify that the proposed calibration improves pixel-text alignment across complementary downstream tasks.
arXiv:2506.13770v2 Announce Type: replace Abstract: We introduce Color Disentangled Style Transfer (CDST), a novel and efficient two-stream style transfer training paradigm which completely isolates color from style and forces the style stream to be color-blinded. With one same model, CDST unlocks universal style transfer capabilities in a tuning-free manner during inference. Especially, the characteristics-preserved style transfer with style and content references is solved in the tuning-free way for the first time. CDST significantly improves the style similarity by multi-feature image embeddings compression and preserves strong editing capability via our new CDST style definition inspired by Diffusion UNet disentanglement law. By conducting thorough qualitative and quantitative experiments and human evaluations, we demonstrate that CDST achieves state-of-the-art results on various style transfer tasks.
arXiv:2507.04221v3 Announce Type: replace Abstract: We introduce Context Tuning, a simple and effective method to significantly enhance few-shot adaptation of large language models (LLMs) without weight updates. In-Context Learning (ICL) forms a memory representation of the demonstrations in a single forward pass but cannot refine it when insufficient. Prompt-based methods offer lightweight adaptation by optimizing a trainable prompt or prefix but initialize it independently of the demonstrations. In contrast, Context Tuning leverages the model's inherent ICL ability to initialize a trainable memory representation from demonstrations, then refines it through gradient-based optimization. Extensive evaluations on benchmarks such as CrossFit, UnifiedQA, MMLU, BIG-Bench Hard, and ARC demonstrate that Context Tuning outperforms both ICL and traditional prompt-based adaptation methods while achieving competitive accuracy with Test-Time Training at significantly higher training efficiency.
arXiv:2607.05045v1 Announce Type: new Abstract: Modern machine-learning weather prediction (MLWP) has largely inherited the initial-value-problem (IVP) framing of numerical weather prediction (NWP). This inheritance leads to a dominant paradigm of learned autoregressive time-stepping and constrains how the learning problem is defined and architectures are favored. In this study we make the inheritance explicit, contrast two philosophical traditions: "scientific surrogate modeling," where machine learning (ML) is embedded within a physical system and must respect its structural constraints, and "free-form data-driven modeling," where atmospheric fields are treated as spatiotemporal sequences and models learn latent dynamics without explicit physical constraints. By reviewing the governing primitive equations, surveying recent literature, and analyzing concrete physical examples, we map each modeling paradigm to either a state-conditioned or evolution operator formulation. We conclude that principled model selection requires explicitly aligning architecture and training objectives with either the physical system structure or the statistical structure of the data.
arXiv:2607.05298v1 Announce Type: cross Abstract: Grain-refined metals typically exhibit high strength, yet their engineering applications are often constrained by grain coarsening under thermo-mechanical loading. Recent experiments have revealed abnormal grain growth (AGG) in ultrafine-grained Ni thin films subjected to cyclic loading at room temperature. Unlike conventional AGG, which generally requires significant plastic deformation or high temperatures, this phenomenon occurs within the regime of macroscopic elastic deformation. This AGG is characterized by the preferential growth of grains with an in-plane <100> orientation aligned with the loading direction. Here, we investigate the underlying physical mechanisms by combining phase-field simulations with micromechanical analysis. The results indicate that elastic energy reduction provides a thermodynamically plausible driving force for this orientation-selective grain growth. Phase-field simulations reveal the evolution kinetics of AGG and confirm that local grain geometry and stress states play critical roles in determining the grain growth pathway. By applying this framework to systems with varying elastic anisotropy, we establish a general approach for investigating elastically driven AGG in polycrystalline materials.
arXiv:2607.05321v1 Announce Type: cross Abstract: Molecular passivation of surface defects is key to improving the optoelectronic performance of hybrid halide perovskite materials, but the underlying atomistic mechanisms are incompletely understood. While machine-learned interatomic potentials are now widely used to simulate complex molecular and crystalline systems, their application to experimentally-realistic scenarios - such as molecules coordinating to perovskite surfaces - is still far from trivial. Here, we describe a multistep training pipeline, resembling continuous fine-tuning used for large language models, to underpin atomistic modelling and computational experiments in this domain. Our protocol involves two components: (i) a large, curated, and open dataset of diverse metal and hybrid halide perovskite structures ('hyP-26'); and (ii) a small, specialised dataset for an amino-silane molecule passivating the surface, providing highly specific information for fine-tuning. We apply this approach to explore collective behaviour at a mixed-composition halide perovskite surface passivated with a varying coverage of amino-silane molecules, revealing an evolution of interactions with increasing molecular surface coverage.
arXiv:2508.02296v3 Announce Type: replace Abstract: Retrieval-Augmented Generation (RAG) systems are increasingly deployed in high-stakes domains, where safety depends not only on how a system answers, but also on whether a query should be answered given a knowledge base (KB). Out-of-domain (OOD) queries can cause dense retrieval to surface weakly related context and lead the generator to produce fluent but unjustified responses. We study lightweight, KB-aligned OOD detection as an always-on gate for RAG systems. Our approach applies PCA to KB embeddings and scores queries in a compact subspace selected either by explained-variance retention (EVR) or by a separability-driven t-test ranking. We evaluate geometric semantic-search rules and lightweight classifiers across 16 domains, including high-stakes COVID-19 and Substance Use KBs, and stress-test robustness using both LLM-generated attacks and an in-the-wild 4chan attack. We find that low-dimensional detectors achieve competitive OOD performance while being faster, cheaper, and more interpretable than prompted LLM-based judges. Finally, human and LLM-based evaluations show that OOD queries primarily degrade the relevance of RAG outputs, showing the need for efficient external OOD detection to maintain safe, in-scope behavior.
arXiv:2607.05098v1 Announce Type: new Abstract: When sensitive attributes are continuous and high-dimensional $-$ demographic score vectors, posteriors over attributes, age or income profiles $-$ enforcing full statistical independence is often too restrictive, and existing relaxations rely on indirect dependence penalties or adversarial schemes that do not directly target the fairness-accuracy trade-off. We instead consider mean demographic parity through DPVar, the variance of the conditional-mean prediction given the sensitive attribute, and show that optimizing it yields a functional bilevel problem. We propose two algorithms for this problem: FBO, which uses a closed-form adjoint we derive for the squared-loss case to obtain an exact hypergradient, and ITD, which differentiates through unrolled inner steps and extends beyond squared loss. On synthetic data and a new semi-synthetic benchmark built from 60 tabular regression datasets, both methods achieve the lowest or near-lowest aggregate fairness-accuracy regret, and consistently match or outperform strong HSIC, adversarial, linear-dependence, and generalized-DP baselines.
arXiv:2607.05324v1 Announce Type: cross Abstract: We present the first constant-amortized-time algorithms for generating all length-$n$ necklaces and Lyndon words over a $k$-letter alphabet in colexicographic order, for arbitrary $k\geq 2$. Our approach introduces a novel class of words called \emph{quasinecklaces}, which serve as an easily generated superset of necklaces through which all necklaces can be efficiently identified. We derive a formula for the number $Q_k(n)$ of length-$n$ quasinecklaces and show that $Q_k(n)$ is proportional to the number of length-$n$ necklaces, which is the key property needed to achieve constant amortized time. We also apply our results to efficiently generate a well-known de Bruijn sequence and efficiently generate necklaces and Lyndon words subject to a weight constraint.
arXiv:2508.02932v2 Announce Type: replace Abstract: Low-Rank Adaptation (LoRA) has gained popularity as a fine-tuning approach for Large Language Models (LLMs) due to its low resource requirements and good performance. While numerous studies have investigated ways to improve LoRA serving efficiency by serving multiple LoRAs concurrently, existing methods assume that a wide range of LoRA adapters are available for serving. In our work, we conduct extensive empirical studies to show that current LoRA training paradigms do not efficiently utilize hardware resources and incur high overhead to obtain a performant LoRA adapter. Leveraging these insights, we propose PLoRA, which automatically orchestrates concurrent LoRA fine-tuning jobs under given hardware and model constraints and develops performant kernels to improve training efficiency. Across a range of LLMs and LoRA configurations, PLoRA improves training throughput by up to 12.8x and reduces the overall fine-tuning makespan by up to 7.52x compared to existing approaches.
arXiv:2508.04796v3 Announce Type: replace Abstract: Tokenization is the first -- and often least scrutinized -- step of most NLP pipelines. Standard algorithms for learning tokenizers rely on frequency-based objectives, which favor languages dominant in the training data and consequently leave lower-resource languages with tokenizations that are disproportionately longer, morphologically implausible, or even riddled with $ $ placeholders. This phenomenon ultimately amplifies computational and financial inequalities between users from different language backgrounds. To remedy this, we introduce Parity-aware Byte Pair Encoding (BPE), a variant of the widely-used BPE algorithm. At every merge step, Parity-aware BPE applies a fair-max rule that maximizes the compression gain of the currently worst-compressed language, trading a small amount of global compression for cross-lingual parity. We find empirically that Parity-aware BPE reduces tokenization inequality -- operationalized by the Gini coefficient of per-language token costs -- by up to 89% relative to Classical BPE. This comes with negligible impact on global compression rate and no evidence of systematic degradation in downstream LM performance.
arXiv:2607.04757v1 Announce Type: cross Abstract: The InP high-electron-mobility transistor (HEMT) is indispensable for low-noise amplifiers (LNAs) in radio astronomy and quantum computing. The composition of the $\mathrm{In_{x}Ga_{1-x}As}$ channel in InP HEMT is known to influence the LNA noise performance. However, the various physical mechanisms responsible for noise generation are not fully characterized and understood. Here, we investigate the $\mathrm{In_{x}Ga_{1-x}As}$ channel noise from 4 K to 300 K for 100-nm gate-length InP HEMTs with channel indium content of 53\%, 60\% and 70\%. Channel noise was quantified by extracting the equivalent drain noise temperature $\mathit{T}_{d}$ using both on-wafer and LNA-based measurements, covering 40-300 K and 4-40 K, respectively. The 60\% indium channel InP HEMT exhibited the lowest channel noise across the full temperature range. The $\mathit{T}_{d}$ extracted from on-wafer characterization was found to obey a parabolic temperature dependence which predicted the $\mathit{T}_{d}$ at 4 K for all InP HEMTs in good agreement with LNA-based measurements. By expressing the channel noise as the sum of one thermal and one excess noise term, it was found that the former increased linearly with ambient temperature and dominated at 300 K. The channel noise at 4 K was determined by the excess noise term and exhibited a non-monotonic dependence on the channel indium content in the InP HEMT. The results suggest that the excess noise in the InP HEMT originates not only from temperature-independent shot noise but also from impact ionization and real-space transfer noise.
arXiv:2607.04745v1 Announce Type: new Abstract: Modern thermal visual place recognition (TIR-VPR) frontends based on foundation models achieve remarkable closed-set retrieval but suffer from an overconfident forced-matching failure mode. Under out-of-distribution (OOD) or unmapped conditions, they generate highly plausible yet false loop candidates without a drop in similarity scores. While classical multi-hypothesis tracking (MHT) backends can mitigate these ambiguities by maintaining divergent trajectory beliefs, their exponential computational overhead violates real-time robotic constraints. To bridge this gap, we present Trajectory-Anchor Optimization (TAO). To counter the combinatorial challenge of evaluating parallel hypotheses (e.g., K=100), TAO compresses multi-view temporal verification into a batched SE(2) Procrustes alignment problem. By leveraging tensor-level vectorization and single-invocation batched SVD, this formulation bypasses the dynamic tree expansion of MHT, guaranteeing a strictly bounded per-frame execution loop of O(KN). Under a strict zero-leakage evaluation protocol, we show that while a passive geometric backend cannot mathematically separate metric localization errors from coherent hallucinations at a micro-scale (<5m) due to local visual ambiguities, TAO serves as an efficient fail-safe filter at a macro-scale. Within a 5m radius, hallucinations often possess a locally consistent geometry that deceives rigid alignment. However, beyond this threshold, the K=100 disparate hypotheses disperse spatially across the global map. This dispersion breaks the rigid temporal co-visibility constraint within the sliding window (N=20), causing the joint optimization residual to escalate sharply. Consequently, TAO establishes a distinct macroscopic convergence basin (10m) where multi-view geometric consistency reliably isolates catastrophic topological breaks and suppresses critical false acceptances.
Hydro-mechanical Model for Slope Stability Assessment: A polygonal stabilization-free discretization
arXiv:2607.04778v1 Announce Type: new Abstract: Rainfall-induced landslides are governed by the interaction between subsurface water flow and soil mechanics, requiring robust numerical methods for the simulation of variably saturated porous media. In this work, we consider a semi-coupled hydro-mechanical model based on Richards' equation and linear elasticity and propose a numerical framework based on a stabilization-free Virtual Element Method for its spatial discretization. The proposed approach naturally accommodates general polygonal meshes while avoiding problem-dependent stabilization terms, whose design may become challenging when heterogeneous and strongly non-linear coefficients are involved. The approach is combined with a mass-lumping technique to improve stability in the treatment of the storage term and with Nitsche's method to weakly impose seepage-face and infiltration boundary conditions, allowing for the automatic switching between Neumann and Dirichlet conditions. Time integration is performed using the backward Euler scheme, while non-linearities are handled through a Picard iteration. Numerical experiments demonstrate the stability and robustness of the proposed methodology and show its effectiveness in simulating rainfall infiltration and evaluating slope stability through the Local Factor of Safety.
arXiv:2607.04782v1 Announce Type: new Abstract: Many real-world games do not admit a fixed, compact rule set: instead, their dynamics are defined by interactions among a large and often evolving collection of game pieces, making general-purpose policy learning impractical. Magic: the Gathering (MTG) exemplifies this setting, where the cards themselves define and alter gameplay rules, strategic constraints, and long-term outcomes, while the pool of available cards is ever-changing. We study Draft, a constrained deck-building format of MTG in which eight players make 39-45 sequential selections from semi-random packs to construct a 40-card deck under partial information. By isolating the card selection process from gameplay, Draft provides a tractable yet non-trivial setting for studying decision-making driven by combinatorial card synergies. We propose an encoder-based model that produces set-contextualized card embeddings to encode the draft decision sequence, with a consistent improvement over linear baselines on large-scale real-world data, establishing a first learned benchmark for outcome prediction in MTG Draft. Our code is available at github.com/akulen/MtGDraftEncoder.
arXiv:2508.16771v3 Announce Type: replace Abstract: Code Language Models (CodeLLMs) learn token importance from data correlations, whereas human developers attend selectively to semantically salient code. We present EyeMulator, a model-agnostic method that injects human visual-attention priors into CodeLLM fine-tuning without architectural changes. EyeMulator distills eye-tracking data into semantic salience and gaze-transition priors, then uses them to reweight token-level training losses. Across six backbones, two data regimes, and three CodeXGLUE tasks, the reported configurations yield positive matched-metric deltas in all 36 model-task-setting cells. Effects are largest for structure-preserving completion and translation, while summarization shows smaller but positive METEOR deltas. Session-mode and component-ablation analyses further show that reading, writing, semantic, and transition-derived priors provide complementary signal. Human-attention artifacts are available at https://zenodo.org/records/17205682.