arXiv subjects

Lu Yin

Publications and source records attributed to Lu Yin.

At least 19 recordsLinked to original sources

arXiv2026

Bernoulli On-Policy Distillation: Don't Ignore High-Confidence Student Tokens

On-policy distillation (OPD) trains a student language model on its own trajectories using token-level supervision from a stronger teacher. We show that standard sampled reverse-KL OPD systematically suppresses updates to high-confidence tokens, such that their weak gradient contribution does not necessarily imply weak or uninformative teacher supervision. We introduce Bernoulli On-Policy Distillation (Bernoulli OPD), which directly matches the teacher and student probabilities for that token. This yields a bounded probability space correction that preserves teacher-student disagreement while avoiding the additional confidence-dependent suppression induced by vanilla OPD. Fixed-rollout analyses show that Bernoulli OPD substantially restores optimization signal to high-confidence tokens while remaining sensitive to teacher-student disagreement. We further analyze the resulting model behavior through confidence distributions, answer calibration, policy movement, and training dynamics, showing that Bernoulli OPD's gains cannot be explained simply by longer reasoning, larger optimization steps, or more uniform imitation of the teacher. Code is available at https://github.com/duterscmy/Bernoulli-OPD.

cs.LG↗

arXiv2026

Scaling to Tens of Thousands of Test-Time Iterations with Loop-Native Attention Residuals

In this paper, we argue that Looped Transformers need their own residual connections to prevent performance degradation as the number of iterations grows. We first introduce an interesting phenomenon we observed in Looped Transformers, called latent overthinking: noisy state updates overwrite correct intermediate deductions and even undo completed solutions. This leaves subsequent iterations to recover lost information from an already degraded representation: once an error arises in an earlier loop, often as a result of long-range propagation through the recurrence, later loops find it difficult to correct. In this paper, we introduce InfiLoop, a loop-native residual connection that learns which past computations to retain and how much to accept from each new update. InfiLoop combines content-based weighting with learned temporal decay to maintain a running summary of recurrent states. An exact streaming recurrence keeps its persistent aggregation memory constant as the loop count grows. The resulting adaptive update suppresses unreliable proposals and preserves useful intermediate states. Across extensive reasoning tasks, a 7M-parameter InfiLoop model outperforms existing recursive architectures, reaching 97.9% exact accuracy on Sudoku-Extreme, and 13.6% pass@2 on ARC-AGI-2. Notably, on Sudoku-Extreme, InfiLoop continues to improve with test-time looping beyond 20,000 effective steps, showing that added depth translates directly into stronger reasoning. Our code is available at https://github.com/pixeli99/InfiLoop.

cs.AI↗

arXiv2026

DataSense-Bench: The First Step Toward an AI Scientist

As claims about recursive self-improvement (RSI) and artificial general intelligence (AGI) proliferate, we ask a simple question: do frontier AI models have a sense of data, i.e., can they reliably select the right data for training? We introduce DataSense-Bench to study this capability through the fundamental problem of data selection and performance forecasting in machine learning. We ask AI agents to select and rank candidate training subsets that can be used to fine-tune a small LLM model. Agents are allowed to inspect the data, write and execute analysis code, and run model forward passes, but can not train the model or access the actual evaluation tasks. We then fine-tune the base model on each selected subset and evaluate its post-training performance under a standardized protocol. We instantiate the benchmark in terminal problem solving and tool use, selecting trajectories from OpenThoughts-Agent and EnvScaler and evaluating on TBLite and BFCL, respectively. We then evaluate the agents along two complementary dimensions: the post-training performance of the top-ranked subset, reflecting the ability to identify high-value training data, and ranking accuracy, reflecting the ability to predict the relative performance of the selected subsets. In our experiments, selection gains over random selection are limited; agents do not reliably rank their selected groups, and ranking ability does not hold consistently across tasks: Astra identifies the best group in all three tool-use runs but in only one of three terminal runs. Analysis of execution traces on both tasks shows that agents often use similar data signals while interpreting their training value differently.

cs.LG↗

arXiv2026

Distinguishing the origin of cosmic birefringence: dark energy, dark matter, and neutrino asymmetry

Cosmic birefringence, the isotropic rotation of the plane of linear polarization of photons from sources at cosmological distances, can be generated by different parity-violating mechanisms. While the net rotation angle of a single polarized source carries little information about its physical origin, sources at various redshifts $z$ can be used to distinguish between different origins. Parity-sensitive correlations of the $E$- and $B$-mode polarization fields of the cosmic microwave background measured on large and small angular scales probe sources at $z\simeq 10$ and $1100$, respectively. Thus, the detailed shape of the $EB$ power spectrum is sensitive to the redshift evolution of the birefringence source. In this paper, we calculate the $EB$ power spectrum due to a neutrino-asymmetry current, and compare it to axion dark energy and dark matter using a common Chern--Simons framework. We find that the $EB$ power spectrum due to a neutrino asymmetry is suppressed at low multipoles, $\ell\lesssim 20$, similar to the dark matter case. This can be clearly distinguished from the dark energy case, which has no such suppression. Large-scale polarization measurements, including \textit{LiteBIRD}, can therefore distinguish late-time dark energy birefringence from that generated at higher redshifts by axion dark matter or neutrino asymmetry. We also find that the evolution of the neutrino density during recombination alters the shape of the $EB$ power spectrum at high multipoles. This signature can be used to distinguish between the neutrino and axion dark matter interpretations, using ground-based experiments such as the Simons Observatory.

astro-ph.CO↗

arXiv2026

Looping Beyond Twice: A Scalable Recipe for Looped Mixture-of-Experts

Looped Transformers introduce recurrent depth as a new scaling axis for LLMs: by repeatedly applying shared Transformer blocks, they increase effective depth without increasing parameter count. However, the benefits of looping remain unclear for large MoE LLMs under FLOPs-matched comparisons. The main reason is that the gains from additional iterations diminish quickly and can even turn into degradation, so the extra FLOPs spent on looping yield little substantial improvement. Consequently, prior work typically settles on two loops. We identify two main obstacles to scaling looped MoE. First, looping inherits and amplifies the curse of depth: hidden-state variance grows with each iteration as residual updates accumulate, which destabilizes deep recurrence and causes representations to drift. Second, looped MoE suffers from expert selection collapse: routers repeatedly select the same experts across loops, so extra iterations add computation without adding computational diversity. Guided by this diagnosis, we propose LOOM, built on a single principle: each loop should contribute new computation while keeping the recurrent state stable. LOOM stabilizes recurrence by scaling residual updates to bound variance growth and re-injecting the input embedding at every loop, and diversifies it through per-loop routers that engage different experts and a Looping Residual that carries earlier outputs forward. Experiments across 100M-1.7B models show stable scaling to 9-12 loops. Under near-iso-FLOP, the 700M model performs best at 5 loops, reducing perplexity from 18.36 to 16.54 and improving average zero-shot accuracy from 38.84% to 39.53% over the non-looped baseline. Without FLOP matching, the 1.7B model trained on 60B tokens peaks at 9 loops, reducing perplexity from 9.62 to 7.77 and improving average zero-shot accuracy from 42.4% to 47.7%. Code is available https://github.com/hed-ucas/LOOM.

cs.LG↗

arXiv2026

SE-ADD: Self-Evolving Audio Deepfake Detection with Mistake-Driven Supervision

Audio deepfake detection (ADD) must remain effective when new spoofing attacks emerge after deployment. Emerging audio language model (ALM)-based ADD methods are built on predefined supervision from ground-truth labels or verified forensic rationales. However, this paradigm overlooks an ALM's own mistakes, which indicate where targeted supervision is most needed. To this end, we first introduce evolving spoofing environments for ALM-based ADD, where a new attack becomes dominant while previously observed attacks persist. Motivated by the above learning-from-mistakes perspective, we further propose SE-ADD, a self-evolving framework that iteratively adapts an ALM via low-rank adaptation (LoRA) using mistake-driven supervision built from its verdicts and self-generated forensic cues. All training samples receive direct authenticity supervision, while misclassified ones receive additional cue-augmented supervision. As verdicts and cues are regenerated by the updated ALM, the resulting supervision evolves accordingly. Experiments on two ALMs demonstrate the effectiveness of SE-ADD in generalizing to unseen attacks, reducing the equal error rate (EER) from $36.72\%$ to $7.52\%$ for Qwen2-Audio and from $19.93\%$ to $3.97\%$ for MOSS-Audio.

cs.SD↗

arXiv2026

SEAR: Spoofing Evidence-Grounded Audio Reasoning Benchmark for Audio Language Models

Audio language models (ALMs) are increasingly used for audio deepfake detection (ADD), yet existing benchmarks assess their verdicts or rationale plausibility without verifying the underlying acoustic evidence. To address this issue, we first introduce spoofing evidence-grounded audio reasoning (SEAR), a four-task AQA benchmark to evaluate ALM-based ADD through acoustic evidence identification and quantification, deepfake detection, and forensic rationale generation. We further propose a bona-fide-based acoustic evidence agent (BAEA), which equips a frozen ALM with controlled acoustic tools under \textsc{fixed} or \textsc{adaptive} evidence-acquisition policies. Experiments with six ALMs reveal a clear gap between plausible rationales and verifiable acoustic evidence reasoning, while BAEA-\textsc{Fixed} improves final verdicts and forensic rationales on both evaluation partitions. Controlled interventions further show that misleading evidence degrades both detection and grounding performance.

cs.SD↗

arXiv2026

DiaLLM: An Investigation into the Robustness-Generation Gap in English Dialect Adaptation

Large language models increasingly understand dialectal English, yet still produce only standard, US-leaning English, leaving dialectal generation, the harder half of the problem, largely unaddressed. We introduce DiaLLM, which continually pretrains three open-weight language model families on the International Corpus of English and applies implicit and explicit post-training paradigms, each combined with three model alignment strategies, giving the first controlled comparison of these components across Australian, Indian, and Northern British English. Our results reveal a robustness-generation gap: benchmarks are shaped by continual pretraining and SFT, while alignment visibly reshapes generation in ways benchmarks do not capture. Explicit variety-targeted adaptation produces output reliably recognised as dialectal and judged more dialectal than broad alignment, yet where human judgement was directly assessed, the method that most aggressively optimises the dialectal reward is not the one judged most dialectal. Independent linguistic analysis corroborates this reward-quality gap, most clearly on two of the three families. No single alignment method dominates, and closing the gap will require richer reward designs and continued investment in dialectal resources. We release all code, checkpoints, and preference datasets.

cs.CL↗

arXiv2026

Resolving the Hubble Tension in the Early Dark Energy Framework with JWST and DESI Data

In the JWST and DESI era, the JWST high-redshift galaxy observations and DESI baryon acoustic oscillation (BAO) measurements severely challenge the standard $Λ$CDM model, while the $H_0$ tension becomes increasingly prominent. In this work, we investigate the capability of the early dark energy (EDE) model to alleviate the $H_0$ tension utilizing cosmic microwave background data from Planck, ACT, and SPT, BAO data from DESI, and ultraviolet luminosity function observations from the JWST. Within the canonical axion EDE framework, the CMB+DESI+JWST data significantly increase the $H_0$ value to $71.58\pm1.05\,\mathrm{km\,s^{-1}\,Mpc^{-1}}$, alleviating the $H_0$ tension to the $1.0σ$ level. Simultaneously, this model improves the fit to the JWST data and exhibits statistical performance significantly better than the $Λ$CDM model, with $Δχ^2_{\mathrm{tot}} = -18.26$ and $Δ\mathrm{DIC} = -11.89$. Our results highlight the complementary advantages of JWST high-redshift galaxy data alongside early- and late-time observations in testing EDE and alleviating the $H_0$ tension.

astro-ph.CO↗

arXiv2026

Grounded Checklist Partial Credit for Agent Skill Trajectories

Language-model agents increasingly tackle long-horizon tasks in interactive environments, yet their evaluation commonly relies on task-level success rates by reducing an entire execution trajectory to whether the task passes an official verifier. This binary score hides partial progress and is particularly limited for procedural agent skill evaluations, since a skill can alter execution without changing the final outcome. While checklists provide finer-grained evaluation by scoring individual task requirements, costly manual authoring and unreliable automatic generation make trustworthy evaluation difficult to scale. To address these challenges, we introduce Grounded Checklist Partial Credit (GCPC), a human-governed and LLM-instantiated partial-credit evaluation of agent trajectories. Humans define reusable rules once, from which an LLM instantiates a task-specific checklist grounded in the task instruction and official verifier. To keep judgment tied to evidence, a judge scores each item from execution log evidence alone and abstains when evidence is missing. A separate scripted step then applies the official verifier outcome to the score. Across a 4,455-trajectory, deduplicated SkillsBench evaluation population, GCPC better discriminates official PASS and FAIL outcomes than holistic judging on the shared subset (AUC 0.689 vs. 0.619). Human evaluation on 96 trajectories from 12 tasks shows that GCPC aligns more closely with human assessments of progress. Applied to 1,946 matched with/without-skill pairs, GCPC exposes the effects hidden by pass@1: among 879 pairs whose binary outcome does not change, 20.9% improve by more than 0.10 while 18.7% regress by the same margin. The GCPC pipeline also transfers to Terminal-Bench and SWE-bench, demonstrating applicability beyond skill-conditioned evaluation.

cs.SE↗

arXiv2026

The Curse of Depth in Large Language Models

In this paper, we introduce the Curse of Depth, a concept that highlights, explains, and addresses the recent observation in modern Large Language Models (LLMs) where nearly half of the layers are less effective than expected. We first confirm the wide existence of this phenomenon across the most popular families of LLMs such as Llama, Mistral, DeepSeek, and Qwen. Our analysis, theoretically and empirically, identifies that the underlying reason for the ineffectiveness of deep layers in LLMs is the widespread usage of Pre-Layer Normalization (Pre-LN). While Pre-LN stabilizes the training of Transformer LLMs, its output variance exponentially grows with the model depth, which undesirably causes the derivative of the deep Transformer blocks to be an identity matrix, and therefore barely contributes to the training. To resolve this training pitfall, we propose LayerNorm Scaling (LNS), which scales the variance of output of the layer normalization inversely by the square root of its depth. This simple modification mitigates the output variance explosion of deeper Transformer layers, improving their contribution. Across a wide range of model sizes (130M to 7B), our experiments show that LNS consistently outperforms previous normalization and scaling techniques in enhancing LLM pre-training performance. Moreover, this improvement seamlessly carries over to supervised fine-tuning. All these gains can be attributed to the fact that LayerNorm Scaling enables deeper layers to contribute more effectively during training. Our code is available at \href{https://github.com/lmsdss/LayerNorm-Scaling}{LayerNorm-Scaling}.

cs.LG↗

arXiv2026

Joint constraints on cosmic birefringence and early dark energy from ACT, Planck, DESI, and PantheonPlus

With the increasing number of high-precision astronomical observations, physical quantities that were previously inaccessible to accurate calculations, such as cosmic birefringence, have once again become a focal point of interest. Such phenomena induce a nonvanishing cross-correlation between the $E$- and $B$-mode polarizations of the cosmic microwave background (CMB), thereby providing a direct observational signature of parity violation. The Chern-Simons coupling between the scalar field in early dark energy (EDE) models and CMB photons is regarded as a plausible mechanism for generating cosmic birefringence. Recent data from the Atacama Cosmology Telescope (ACT) deliver $EB$ measurements at higher multipole moments than those previously achieved by {Planck}, while DESI and PantheonPlus datasets provide new and stringent constraints on the late-time expansion history. Using a joint analysis of {Planck}, DESI DR1, Pantheon+, and ACT data, we perform a full-parameter constraint on the cosmic birefringence effects induced by the EDE-CMB photon coupling. Our results favor a higher Hubble constant, $H_0 = 76.9^{+2.9}_{-2.5}\,\rm km\,s^{-1}\,Mpc^{-1}$, and a relatively large EDE fraction, $f_{\mathrm{EDE}} = 0.232^{+0.074}_{-0.047}$. By comparing the cosmological evolution of this model across different data combinations, we find that the ACT-$EB$ data combined with {Planck} + DESI + PantheonPlus provide good constraints to both early- and late-Universe observations.

astro-ph.CO↗

arXiv2026

When Fewer Layers Break More Chains: Layer Pruning Harms Test-Time Scaling in LLMs

Layer pruning has emerged as a widely adopted technique for improving the efficiency of large language models (LLMs). Although existing methods demonstrate strong performance retention on general knowledge tasks, their effect on long-chain reasoning, a more brittle yet crucial capability, remains largely unexplored. In this work, we study the impact of layer pruning on long-chain reasoning through the lens of test-time scaling, a key mechanism in modern LLMs that enables strong reasoning capacity by allocating more computation at inference time. With extensive experiments, we demonstrate that pruning even one or two layers can severely impair test-time scaling, with performance collapsing drastically on long reasoning benchmarks even when performance on knowledge-intensive and shallow reasoning tasks remains stable. Furthermore, we find that standard supervised fine-tuning remedies fail to recover test-time scaling once it has deteriorated. Through in-depth analyses, we identify the mechanisms underlying this fragility of test-time scaling and highlight the fundamental risks of applying layer pruning to reasoning-intensive LLMs. These findings call for a rethinking of layer pruning strategies and provide insights for developing methods that preserve the robustness of reasoning. We open-source the codebase in \href{https://github.com/keyu-wang-2002/Layer-Pruning-Harms-Inference-Scaling}{https://github.com/keyu-wang-2002/Layer-Pruning-Harms-Inference-Scaling}.

cs.LG↗

arXiv2026

AI vs Human Expert Reasoning: Assessing Agreements in Building Typology Predictions based on Street View Imagery

This research investigates the potential of Vision-Language Models (VLMs) to infer building typologies: Construction, Current Use, and Storeys from Google Street View (GSV) images. Predictions generated by VLMs are compared with inference by human experts (civil engineers and architects) as a source of manually labelled ground-truth data. We evaluate several state-of-the-art VLMs, including GPT-4o, Claude 3.5 Sonnet, and Gemini 2.0 Flash. By applying different scaling strategies and prompting techniques, we found that Chain-of-Thought prompts provide an overall more stable model performance. We also investigate the reasoning behind VLMs' building-typology predictions by examining the probabilities of keywords appearing in AI explanations. This enabled us to analyse patterns in these reasonings and identify key themes driving both agreements and disagreements between VLM and expert labels. We find that AI tends to focus on visual indicators, whereas human experts place greater emphasis on broader contextual cues and domain knowledge, in addition to visual cues. Overall, VLM can approximate experts' capability in building-typology classification at scale, with an average accuracy of approximately 70%. The study demonstrates the VLM's potential for AI automation in tasks that require pattern recognition and object identification in an urban context. AI have the potential to serve as complementary and collaborative tools for urban analysis, leveraging their strengths in understanding visual patterns. This study contributes to the exploration of the efficiency and scalability of AI visual prediction and provides insights into the reasoning processes that could support automation processes in urban analysis and prediction.

cs.AI↗

arXiv2026

CPR: Chained Perceptual Refinement for Coarse-to-Fine Medical Image Classification

High resolution medical images contain fine grained, spatially sparse cues that are critical for diagnosis, yet preserving full resolution incurs substantial computational and memory costs. Most deep models process images uniformly, leading to redundant computation or loss of diagnostic detail under downsampling. We propose Chained Perceptual Refinement, CPR, a coarse to fine framework that formulates medical image analysis as a sequential global to local decision process. Starting from a low resolution global view, CPR dynamically predicts the location and spatial extent of refinement regions, extracts high resolution evidence from the original image, and incrementally integrates it with global context. By keeping the backbone input size fixed while contracting the perceptual field, CPR preserves diagnostic fidelity with constant peak GPU memory. Extensive experiments on five medical imaging datasets and multiple backbone architectures demonstrate that CPR consistently outperforms both fixed resolution and multi scale state of the art baselines, achieving improvements of up to 2.27 percentage points over the second best method. It also achieves up to a 19.6 fold reduction in GFLOPs at matched accuracy, establishing a superior accuracy and efficiency trade off for high resolution medical image analysis. The code is available on GitHub.

cs.CV↗

arXiv2026

TSegAgent: Zero-Shot Tooth Segmentation via Geometry-Aware Vision-Language Agents

Automatic tooth segmentation and identification from intra-oral scanned 3D models are fundamental problems in digital dentistry, yet most existing approaches rely on task-specific 3D neural networks trained with densely annotated datasets, resulting in high annotation cost and limited generalization to scans from unseen sources. Thus, we propose TSegAgent, which addresses these challenges by reformulating dental analysis as a zero-shot geometric reasoning problem rather than a purely data-driven recognition task. The key idea is to combine the representational capacity of general-purpose foundation models with explicit geometric inductive biases derived from dental anatomy. Instead of learning dental-specific features, the proposed framework leverages multi-view visual abstraction and geometry-grounded reasoning to infer tooth instances and identities without task-specific training. By explicitly encoding structural constraints such as dental arch organization and volumetric relationships, the method reduces uncertainty in ambiguous cases and mitigates overfitting to particular shape distributions. Experimental results demonstrate that this reasoning-oriented formulation enables accurate and reliable tooth segmentation and identification with low computational and annotation cost, while exhibiting strong generalization across diverse and previously unseen dental scans.

cs.CV↗

arXiv2026

Distinguishing Imitation Error from Intrinsic Motion Learning Difficulty

Physics-based motion imitation is central to humanoid control, yet current evaluation metrics (e.g., MPJPE) only quantify imitation outcomes, not their underlying causes. This conflation obscures a critical diagnostic question: when imitation error occurs, does it stem from policy limitations or the intrinsic learning difficulty of the target motion? To resolve this ambiguity, we propose the Torque Variation Score (TVS), a physics-grounded metric that quantifies the inherent learning difficulty of a motion independently of any policy's performance. TVS measures the magnitude of torque variation required to correct small pose perturbations, directly capturing how dynamical properties shape the reinforcement learning landscape. We establish that high-TV motions induce flat reward landscapes and vanishing policy gradients, explaining persistent imitation failures. Extensive experiments with state-of-the-art methods (UHC, PHC+) confirm TVS strongly correlates with imitation error and enables principled error attribution: high error on low-TV motions indicates policy deficiency, while high error on high-TV motions reflects fundamental learning constraints. Beyond error diagnosis, TVS facilitates three practical applications: Maximum Imitable Difficulty (MID) for policy capability assessment, Difficulty-Stratified Joint Error (DSJE) for granular performance profiling, and Flawed Motion Detection for identifying segments with abnormally high learning difficulty to support mocap data curation and quality control. TVS provides a rigorous lens to distinguish policy-induced errors from motion-inherent challenges and enhances motion dataset reliability.

cs.GR↗

arXiv2026

Leave it to the Specialist: Repair Sparse LLMs with Sparse Fine-Tuning via Sparsity Evolution

Sparse large language models (LLMs) offer an attractive direction toward efficient deployment, but adapting them to downstream tasks remains challenging. The central difficulty is to enable effective task adaptation without sacrificing the efficiency advantages of sparsity. Existing fine-tuning methods are not well-suited to this setting, as they either introduce additional dense parameters or assume a fixed sparse topology, limiting their compatibility with sparse LLMs. In this paper, we propose Sparsity Evolution Fine-Tuning (SEFT), a fine-tuning framework designed specifically for sparse LLMs. SEFT allows sparse structure to evolve during fine-tuning by periodically reallocating sparse task-specific updates and reactivating previously pruned weights when beneficial. At the same time, SEFT preserves the efficiency advantages of sparsity through topology adaptation based on parameter importance. Experiments on LLaMA, DeepSeek, and Mistral models across multiple benchmarks show that SEFT delivers stronger performance while offering superior memory and time efficiency compared to existing baselines. Our code is publicly available at: https://github.com/QiaoXiao7282/SEFT.

cs.AI↗