Science & Mathematics

A coding agent that rewrites its own scaffold has to choose which proposed edit to expand before that edit has been evaluated, and the search then has to return one agent. Each run fits one linear readout on the parent's steering-split solve rate and eight coordinates of the proposed diff, from 96 scaffold edits labeled by four evaluations of a 16-instance fit split, and then freezes the coefficients. HumanEval+ reserves 48 instances as the steering split and Polyglot reserves 64. That split scores parents and keeps an archive of four nodes. Huxley-Gödel Machine selection draws a parent by Thompson sampling on the steering counts of nodes that have already been evaluated, and it expands the child that parent proposes. Every method then returns the earliest evaluated node whose steering-split solve count is maximal. The tables report the held-out score of that node on the other 100 HumanEval+ instances and the other 145 Polyglot instances, next to the root scaffold and the calibration edit with the highest fit-split label. Twelve runs of GPT-4o return a node at 87.0% on held-out HumanEval+, against 84.2% for the parent-rate ablation. On that split Huxley-Gödel Machine selection returns 85.1%, the root is at 81.2%, and the calibration-best edit is at 82.8%. On the same split Claude Sonnet 4.5 returns 86.6%, beside 86.5% for Huxley-Gödel Machine selection, 85.3% for the parent-rate ablation, and 85.1% for the calibration-best edit.

Individual patient data are the gold standard for secondary analysis of time-to-event evidence, but they are rarely released, and reconstruction from a published Kaplan-Meier figure could help. Existing methods all claim high fidelity, each judged by whichever agreements it happens to achieve, and each returns a single dataset although the map from figure to data is not bijective. We define fidelity explicitly: a dataset is admissible when it reproduces the risk table exactly, follows the digitized curve to within the image resolution, and respects the printed median, interval and event count. We propose StochIKM, a stochastic algorithm that digitizes the figure with a vision-capable language model and uses the high-fidelity definition as constraints, returning an admissible set of individual patient datasets. A solution is guaranteed to exist whenever the figure has been read correctly, so a failure indicts either the published figure or its digitization. In simulation and in 79 arms from published oncology trials, StochIKM reproduces the published risk tables and medians more closely than LLM-improved IPDfromKM, a high-impact alternative.

This paper addresses fine-grained prosody control for large-language-model (LLM) based text-to-speech (TTS) synthesis of Mandarin Chinese, motivated by multimedia content-creation scenarios---film dubbing, audiobook production, and interactive voice interfaces---in which prosodic adjustments must be placed on specific characters rather than applied uniformly to whole utterances. We present a data-driven framework in which fine-grained prosodic markers (pitch, duration, energy, and pause) are automatically extracted from raw audio at scale and embedded into the text stream as out-of-vocabulary Unicode symbols. The LLM of an existing TTS system is then fine-tuned to interpret these markers while its architecture remains unchanged, yielding character-level temporal control that complements existing utterance-level style control. An optional text-LLM front-end can translate natural-language instructions into marker sequences. Experiments on 2.15M Mandarin utterances (5,858 hours) with CosyVoice2 as the backbone demonstrate reliable control of character duration, pitch, pause duration, and global speaking rate, with perceptual quality essentially preserved. A cross-tokenizer analysis further indicates that the ASR-optimized tokenizer of the backbone attenuates prosodic variation relative to reconstruction-oriented alternatives---an observation we report as hypothesis-generating evidence for prosody-sensitive system design.

Machine learning systems are often developed under data-rich conditions but deployed when only incomplete information is available. A natural example arises in colour fundus photography (CFP), where both eyes are commonly imaged because they can provide complementary information for ocular disease assessment. Prior approaches that exploit bilateral information either fuse both eyes in a single model or use knowledge distillation to compress a dual-eye teacher into a lightweight student that still requires both images at test time. However, bilateral acquisition may not always be feasible. We thus propose a privileged learning framework in which a teacher model has access to bilateral CFP images during training and learns representations informed by information from both eyes, while the student is trained to transfer this knowledge from the bilateral teacher using only a single-eye image as input. At inference, the student operates exclusively on this single-eye input and does not require the contralateral image. Experimental results show consistent improvements in area under the receiver operating characteristic curve over conventional monocular training, with the largest gains observed for left-eye images for the detection of age-related macular degeneration (0.931 vs. 0.966) and diabetic retinopathy (0.789 vs. 0.836).%, while requiring only a single-eye image at inference. By enabling models to benefit from bilateral information during training while requiring only a single-eye image at inference, our approach may support more reliable deployment in resource-constrained clinical and screening environments.

Reward hacking is a major problem in current production machine learning systems, resulting in costly and dangerous failures. Here we study how to monitor for and mitigate reward hacking in realistic settings, including tool calls and long contexts. We apply a dynamic intervention to steering directions obtained by three different methods: difference-in-means vectors from labeled rollouts, gradient optimization-based vectors, and vectors extracted when models reference an anti-reward-hacking model spec. This dynamic intervention perturbs model representations less than uniform steering vector addition to the residual stream, allowing for more precise edits. We find that we are able to cheaply and significantly reduce reward hacking without correspondingly significant costs to model coherence or capability, and that steering vectors originally extracted from coding environments also generalize to reducing reward hacking in other domains. Our results pave the way towards latent space monitoring and control over realistic production systems.

Full fine-tuning has achieved remarkable success on reasoning-intensive tasks, but it often requires substantial GPU resource consumption. Parameter-efficient fine-tuning (PEFT) provides a promising alternative for reducing this cost, yet we observe a substantial performance gap on reasoning-intensive tasks. In this paper, we present three key observations showing that this gap is closely associated with norm-related under-adaptation, where PEFT produces markedly smaller weight-update norms than full fine-tuning. Based on these observations, we propose Norm-Guided Parameter-Efficient Fine-Tuning (NormTune), a simple yet effective strategy that adaptively strengthens insufficiently updated layers according to their weight-update norms. Extensive experiments show that NormTune consistently enhances multiple PEFT methods across challenging reasoning benchmarks. Specifically, when trained on S1K-1.1, integrating NormTune into rank-8 LoRA, Bone, and MoRA yields substantial average accuracy improvements of 24.24\%, 11.33\%, and 16.21\% , respectively, across AIME 2024, MATH 500, GPQA, and AIME 2025. Moreover, NormTune-enhanced rank-256 LoRA, Bone, and MoRA achieve performance comparable to S1 full fine-tuning, with respective average gains of 2.17\%, 0.93\%, and 3.00\% over this baseline.

A content-safe response can still promote priorities that conflict with its task: an assistant may demand compliance when asked to preserve autonomy. We study contextual value drift as a task-relative decoding problem with an explicit allowance for change. LaVGuard separates an authorized ten-dimensional profile from a frozen estimate of the values expressed in the visible prefix. A profile-relative density gate selects when to evaluate candidate continuations; their signed progress determines a logit tilt constrained by both changed-position and cumulative-KL budgets. We establish local surrogate improvement, pathwise budget feasibility, and a bound on response-distribution change. Profile conditioning raises boundary TPR at 5% FPR from .58 to .72. At the same .12 mean intervention ratio, the selection study reports risk-safety scores of .88 for LaVGuard and .72 for position-matched edits. The resulting interface makes task priorities actionable while keeping the extent of intervention explicit and bounded.

Continuous diffusion language models generate a whole sequence in parallel, never drawing a single word at any intermediate step; only the final step turns embeddings into text. Models such as ELF now report autoregressive-level quality at a fraction of the sampling steps of discrete-token diffusion models, and this design principle -- no step-wise discretization -- is what their own ablations credit for the gain. We show that this same principle carries a hidden cost: a collapse of the lexical long tail. Measured against held-out human text, ELF's output matches the human rate on common words almost exactly, yet its rarest words appear at only 0.30x the human rate, a gap that does not close from 105M to 652M parameters and that no sampler setting these models expose can reach. A same-size GPT-2, in contrast, spans zero to above the human rate by its decoding rule alone. We trace the difference to what continuous diffusion models lack: with no step that ever draws a discrete word, every step is instead driven by an average over the candidates still in play, and averaging a rare candidate together with the common word competing for its slot lands on the common word. An autoregressive model avoids this because its decoding rule samples one candidate rather than averaging over all of them. We propose GRAFT, a training-free method that grafts this missing choice back onto a frozen model: while the model is still undecided, GRAFT commits the rare words the average would otherwise discard, then regenerates the sequence from scratch with those commitments held in place. Across three ELF scales and on LangFlow, an independent continuous dLLM, GRAFT restores the rare-word share of generated text from 0.42-0.77x the human rate to 0.82-1.10x, a relative gain of 34-109%, and moves the entire word-frequency profile closer to human on every model tested, by 2.7x on ELF-B and 12x on LangFlow.

Multimodal click-through rate (CTR) prediction is central to search systems, where ranking depends on item content, query relevance, and user preference. Existing approaches incorporate content through dense embeddings or coarse discrete IDs, yet effectively integrating fine-grained heterogeneous content with ID-based predictors remains challenging. We propose FineSID, a progressive learning framework that unifies content tokenization, query--item semantic pretraining, and personalized post-training via a shared fine-grained discrete representation. Each stage optimizes an explicit objective while preserving token-level access to content. First, a pretrained multi-scale visual tokenizer and text tokenizers convert heterogeneous content into fine-grained Semantic IDs (SIDs), explicitly aligning them with the ID features. Second, full-parameter semantic pretraining under query--item relevance supervision establishes a robust matching prior, jointly optimizing the interactions of multimodal SIDs and conventional fields within a shared predictor. Third, this initialized predictor is directly fine-tuned with user click supervision to capture implicit personalized preferences. Extensive offline and online evaluations demonstrate substantial gains in both standard CTR prediction and core business metrics. Furthermore, FineSID improves within-product creative discrimination, demonstrating its semantic generalization beyond conventional ID memorization.

Frontier black-box Large Language Model (LLM) agents are proficient at long-horizon reasoning and precise action generation. However, when it comes to a specific agentic task, they have no internalized knowledge of how that particular environment unfolds. A world model offers a natural remedy, providing the agent with environment specific knowledge by foreseeing how the environment will respond to the action it is about to take. Prior world models for LLM agents mainly predict next observations at each step, which is costly and often uninformative because individual observations provide limited task-relevant guidance and not every step requires call for foresight. This raises a key question: what a world model should predict and when should it be invoked for agentic tasks? Our pilot study suggests that world model guidance is most useful when it anticipates pitfalls within a bounded multi-step future. Motivated by this, we introduce Pitfall Anticipation World Model (PAWM), which selectively intervenes when a potential pitfall arises and predicts the impending failure to steer the agent away from it. PAWM is trained in two stages: Supervised Fine-Tuning (SFT) on synthetic data labeled with agent pitfalls, followed by GRPO co-training with a black-box LLM to further optimize it in the target environment. PAWM improves Claude Sonnet 4.5 performance by 24.3% on ScienceWorld and 7.7% on WebArena-Lite, outperforming a range of baselines. These results demonstrate that PAWM provides an effective form of world model guidance for black-box LLM agents.

Engineering graduates in India enter a technology sector that is being reshaped by generative AI, yet the route into that sector runs through examinations and campus placement screens that largely exclude AI. Institutions in India have an approved intake of nearly 2.9 million seats in engineering and technology for 2026-2027, but HCI research on how students prepare for AI-mediated work has drawn mostly on Western and East Asian settings. We interviewed 23 students and recent graduates of Indian engineering and computing programmes about how they used generative AI across coursework, examinations and placements, and how they judged their own readiness for work. Using reflexive thematic analysis, we report five themes, covering AI-first workplaces and AI-free placement screens, how participants prepared for placement, how they used AI in coursework, the role of faculty authority and enforcement, and participants' concerns about their readiness for work. Drawing on signaling theory, we discuss what grades and screen scores certify once AI can help produce them, and we offer recommendations for employers, institutions and AI tool designers.

Synthesizing high-fidelity 7T MRI from routinely acquired 3T scans offers a practical pathway to enhanced neuroimaging without the prohibitive cost of ultra-high-field scanners. While existing deep learning approaches have achieved promising results, state-of-the-art methods typically rely on either large-scale architectures exceeding 45 million parameters or computationally intensive ensemble strategies, limiting their practicality in resource-constrained clinical environments. In this work, we propose a compact multi-domain attention framework, termed DGCM-SSFA, that achieves competitive synthesis quality with only 11M parameters. The proposed architecture integrates two complementary modules into a U-Net Transformer backbone. First, a Dual-Granularity Cross-Modulated Feature Modulator (DGCM) is embedded at multiple hierarchical levels to enhance local edge and texture details through Sobel-guided spatial attention and depthwise frequency-aware channel attention with cross-branch gating. Second, a Sparse Strip Frequency Attention (SSFA) module at the bottleneck captures global spatial context via strip-pooled attention with near-linear complexity, while simultaneously performing FFT-based spectral modulation in the frequency domain. These two modules operate synergistically with the backbone's native channel attention, forming a triple-granularity feature refinement pipeline. Extensive experiments on the UNC and BNU 3T-to-7T MRI datasets across T1w and T2w modalities demonstrate that DGCM-SSFA achieves strong structural fidelity, attaining an SSIM of 0.8431 on BNU T2w---approaching the best benchmark value of 0.8624---and 0.8100 on BNU T1w, while using substantially fewer parameters than Transformer-based alternatives such as ResViT (123.4M) and FS-RWKV (45.4M).

Traditional robotic navigation systems separate mapping, planning, and low-level control through cascaded modules, introducing computational overhead and accumulated latency. End-to-end reinforcement learning offers a compact unified policy that maps onboard observations directly to control actions. However, on inherently unstable platforms such as two-wheeled inverted-pendulum robots, navigation progress and obstacle avoidance can conflict with attitude stabilization in a strongly state-dependent manner. We propose a lightweight state-dependent objective coordination method for integrated navigation and control. A pitch-conditioned scalarization coefficient $β(s)\in[0,1]$ adjusts the relative contributions of balance stabilization and navigation objectives during training, without extra control hierarchies or auxiliary attitude-stabilization controllers, such as proportional-integral-derivative (PID) or linear-quadratic regulator (LQR) controllers. To support zero-shot deployment, simulation training models execution delay, action hold, asynchronous and noisy observations, and dynamics uncertainty. Gradient analysis reveals persistent directional interference and state-dependent variation in the objectives' relative scales. Comparisons with alternative reward-weighting, constraint, training, and control-structure strategies show that PPO+$β(s)$, where PPO denotes proximal policy optimization, achieves 98% and 95% success rates with small attitude excursions on the fixed obstacle map and the randomly generated high-obstacle-density map, respectively. Zero-shot hardware experiments further demonstrate bounded attitude behavior and consistent obstacle avoidance using only onboard observations. Training and evaluation code will be made publicly available upon acceptance.

Context: Reviewing a pull request requires verification evidence that addresses the proposed change. Coding agent benchmarks and studies of pull request outcomes do not directly assess this relationship. Objective: We test whether the V-Model's build and verify pairs can measure verification evidence in pull requests and support comparisons between agent-authored and human-authored changes. Method: VeeReview classifies four build and verify pairs using repository signals and a language-model judge. Three pairs use model judgements constrained by deterministic signals, and a rule over the diff and repository tree decides Module Design-Unit Test. We evaluated VeeReview against a 25-pull request reference set labelled by two raters and applied it to 1,454 matched pairs of agent and human pull requests from 741 repositories. Results: VeeReview assigned lower build-only rates to agent-authored pull requests on all four pairs, contrary to the preregistered hypothesis (mixed-effects odds ratios 0.52 to 0.86). Architecture-Integration was not significant. Standardization for linked issues, shipped tests and size reduced matched-rate differences by 34% to 60%, and the rule-decided pair's standardized interval included zero. The instrument did not meet its preregistered calibration criterion: agreement with adjudicated labels was 4% to 60% against a required 80%, and inter-rater agreement was weak (kappa = 0.154). Conclusions: The rule-decided pair shows that a process measure judged by a language model partly reflects how a pull request is presented rather than the verification work behind it. We report the cohort differences as observations made with this instrument, and we argue that such measures need a mechanically decided component and a validated human reference before they can support comparisons of verification practice.

MDS array codes achieve optimal storage efficiency in distributed storage systems, but their repair bandwidth depends critically on the sub-packetization level. While the theoretically optimal repair bandwidth (i.e., cut-set bound) of a high-rate exact-repair MDS array code is attainable with sufficiently large sub-packetization, the fundamental limits for fixed sub-packetization remain poorly understood. We study this problem for centralized linear exact repair of $h$ failures from $d$ helper nodes over a fixed finite field. We show that the rank profile of any linear repair scheme is constrained by two auxiliary mixed-alphabet codes. The first code is formed from the kernels of the helper projections and exists for all repair degrees; the second is formed from their row spaces and applies in the full-helper regime when $d\ge r$. These codes have complementary alphabet profiles, and their size and minimum-distance requirements are forced solely by the MDS property and repair correctness. This gives a unified route from classical coding-theoretic upper bounds to repair-bandwidth lower bounds. As a consequence, we obtain several explicit lower bounds via Singleton, Hamming, Plotkin, and Griesmer arguments, including known bounds as corollaries. We further formulate a Delsarte-type support-enumerator LP for mixed-alphabet codes and give an exact type-compressed implementation. For fixed sub-packetization \(\ell\), this yields a polynomial-time procedure for evaluating the resulting finite-parameter lower bound.