Academic Intelligence · Curated Daily

探索全球前沿学术脉络

AcademicHub 汇聚顶级期刊与预印本平台的实时文献。定制您的专属科研雷达,利用大语言模型自动生成交叉领域文献分析简报。

01.
medRxiv (Medicine) 2026-06-17

Cost-effectiveness of measles rapid diagnostic tests for replacing or expanding laboratory testing in Ethiopia

Background: In low- and middle-income countries, laboratory testing to rapidly detect measles outbreaks is limited by infrastructure availability and high costs. This study estimates the potential impact and cost-effectiveness of measles rapid diagnostic tests (RDTs) if implemented nationally in Ethiopia to either replace or expand current testing. Methods: An agent-based model to simulate measles outbreaks was calibrated to Ethiopian measles surveillance data. Modelled outbreak outcomes were aggregated over a 10-year period. Scenarios included using RDTs to (1) replace laboratory testing; (2) replace epidemiological linkage; and (3) increase case detection, in addition to replacing laboratory testing and epidemiological linkage. Testing and outbreak response costs (in 2025 US$) were obtained from Ethiopian Public Health Institute from a government perspective. Total costs and disability-adjusted life years (DALYs) for each scenario were compared to baseline. Results: All scenarios were cost saving compared to baseline. Replacing laboratory testing with RDTs saved US$4.2M (3.2M-4.9M) over 10-years, but due to very low testing rates the benefits of eliminating laboratory testing delays were offset by missed cases from the lower RDT sensitivity, leading to similar outbreak detection times and DALYs. Replacing epidemiological linkage with RDTs had similar DALYs but increased the cost savings to US$9.7M. Using RDTs to double case detection reduced outbreak detection time from 113 to 80 days, averted 17,000 DALYs, and saved US$4.3M. Conclusions: In Ethiopia, use of measles RDTs could be cost saving, and if used to expand testing could prevent measles infections through faster outbreak detection and response.

02.
arXiv (CS.CL) 2026-06-16

Rapid Poison: Practical Poisoning Attacks Against the Rapid Response Framework

The Rapid Response (RR) framework, deployed in production systems, including Anthropic's ASL-3 safeguards, continuously improves jailbreak-detection classifiers. When new jailbreaks emerge that bypass these classifiers, Rapid Response generates synthetic variants for training, helping the model generalize from the new attacks and quickly adapt. We reveal that prompt injection can infiltrate this pipeline to deliver poisoned samples into the classifier's training set, enabling two attack objectives: (I) targeted poisoning attacks that create false positives on harmless samples by categorizing them as a jailbreak, with a specific desired feature (e.g., certain formatting, subject, or keyword), (II) concept-based backdoor attacks that induce false negatives on jailbreak inputs, generalizing even to jailbreaks from attack strategies the defender explicitly trained against, when the backdoor trigger is present. Importantly, our threat model restricts adversaries to modifying only jailbreak samples (not benign data or labels), a constraint unexplored by prior work that makes the second objective particularly challenging. We address this with Omission Attack, which exploits a new phenomenon: when training on concept-absent unsafe samples, the classifier misassociates that concept's presence with the safe label. Both attacks cause substantial and in some cases near-complete label flipping at only a 1% poisoning rate, achieving up to 100% false positive rates and up to 96% false negative rates.

03.
arXiv (CS.LG) 2026-06-16

Towards Functional Correctness of Large Code Models with Selective Generation

arXiv:2505.13553v3 Announce Type: replace-cross Abstract: The hallucination of code generation models hinders their applicability to systems requiring higher safety standards. One critical bottleneck in addressing code hallucination is the difficulty of identifying the functional correctness of generated code, due to its unnatural form. We address this core bottleneck by automatically generating unit tests using dynamic code analysis tools, leveraging the executable nature of code. Accordingly, we propose a selective code generator that abstains from uncertain generations – based on the functional correctness evaluated by generated unit tests – to theoretically control the correctness among non-abstained answers, \ie the false discovery rate. Finally, we propose to use generated unit tests in evaluation as well as in learning for precise code evaluation, calling this paradigm FuzzEval. We demonstrate the efficacy of our method along with the controllability of code hallucination and reasonable selection efficiency.

04.
arXiv (CS.CL) 2026-06-18

As Easy as Rocket Science: Assessing the Ability of Large Language Models to Interpret Negation in Figurative Language

Figurative language and negation are two areas that challenge current language models, however, both are widely used throughout written and spoken language. Large language models (LLMs) are also widely used in everyday contexts where they cannot necessarily be tuned for a specific dataset. It is therefore essential to understand the ability of LLMs to correctly interpret text that includes both negation and figurative language. To investigate this, we develop a set of new annotations to an existing dataset of figurative language, and test a range of language models on the dataset. We find that the combination of negation and figurativeness can present a particular challenge, and that performance overall and across different negation types is particularly dependent on the prompt style used.

05.
arXiv (quant-ph) 2026-06-11

Permutation-Invariant N-body gates via Tavis-Cummings Hamiltonian

arXiv:2506.03453v3 Announce Type: replace Abstract: Global control provides a promising route to implementing multi-qubit gates without individual qubit addressing. This is especially appealing for permutation-invariant (PI) gates, whose symmetry is often broken when they are compiled into individually addressed one- and two-qubit gates. Important examples include SWAP, $\sqrt{iSWAP}$, and the n-qubit controlled-Z gate, which is equivalent, up to two single-qubit Hadamard gates, to the multi-qubit Toffoli gate. Motivated by this global-control perspective, we show that all PI unitaries on an arbitrary number of qubits can be realized using the Tavis-Cummings (TC) interaction, the multi-qubit version of the Jaynes-Cummings interaction, together with global uniform z and x fields. Here, the $n$ qubits are identically coupled to a single bosonic mode (oscillator), which is initialized in and returned to its vacuum state. A corollary is that all PI states, including GHZ and Dicke states, can be prepared using the same global control. For the case n=2 qubits, which is particularly important in quantum computing, we also find explicit pulse sequences for implementing all PI qubit unitaries that conserve angular momentum in the z direction, using only the TC interaction and global z fields. This includes controlled-Z, SWAP, and $\sqrt{iSWAP}$.

06.
arXiv (CS.AI) 2026-06-16

Training and Evaluating Diffusion Policies with Long Context Lengths

arXiv:2606.16447v1 Announce Type: cross Abstract: Imitation learning has enabled highly-dexterous robotic manipulation from RGB observations. Policies trained with these methods, however, typically condition robot actions on only a short history of observations. These policies cannot solve tasks that require memory and can get stuck repeatedly executing the same failing motions. In this work, we first benchmark policy performance as context length is incrementally increased from short to long, across a spectrum of tasks with varying local stability and memory requirements, and in multiple data regimes. To our knowledge, this is the first study to investigate context length in imitation learning at this level of detail. Our results challenge prior claims: naively scaling context length is not as brittle as advertised in literature. With an appropriate conditioning method and denoising backbone (UNet+Cross-Attention), single-task policies achieve high success rates on many tasks in the usual data regime even with naive scaling. Next, we propose a training algorithm to jointly train policies at multiple context lengths, further reducing the sample complexity of long-context learning. Finally, we apply our findings to re-evaluate some previously proposed solutions to long-context imitation learning.

07.
Nature (Science) 2026-06-10

Gen Z scepticism towards AI is a wake-up call — universities must take it seriously

作者:

The challenge for universities is not adopting artificial intelligence, but doing so in ways that the current generation of students can trust. The challenge for universities is not adopting artificial intelligence, but doing so in ways that the current generation of students can trust.

08.
arXiv (CS.AI) 2026-06-16

EChO-Agent: Evidence Chain Orchestration Agent for Audio Reasoning

arXiv:2606.15141v1 Announce Type: cross Abstract: While LALMs show promise on audio question answering, they fail to focus on question-relevant segments of audio and provide a clear, checkable reasoning process when dealing with complex audio reasoning. Reinforcement learning and tool-augmented prompting can help models better relate questions to audio but lack a reliable way to understand, integrate, and self-verify audio segments. To address this gap, we present EChO-Agent, a modular agent framework that reformulates complex audio QA as a planning, tool execution, evidence integration, and answer verification workflow. Experiments on MMAR benchmark show EChO-Agent improves both accuracy and rubric scores over baseline and ablation studies show evidence integration is the key factor.

09.
arXiv (CS.AI) 2026-06-11

GEAR-VLA: Learning Geometry-Aware Action Representations for Generalizable Robotic Manipulation

arXiv:2606.08530v2 Announce Type: replace-cross Abstract: Vision-Language-Action (VLA) models achieve strong benchmark performance but still struggle in real-world deployment with unseen objects, background shifts, and different robot embodiments. We argue that this stems from the lack of a unified geometry-aware manipulation representation, leaving existing VLAs vulnerable to low-level trajectory supervision, misaligned 3D features, and embodiment differences. To address this, we propose GEAR-VLA, a VLA framework for learning unified geometry-aware action representations for generalizable robotic manipulation. GEAR-VLA adopts coarse-to-fine action learning, where multi-source embodied pretraining equips the VLM with embodied reasoning and discrete action understanding before latent action tokens connect action semantics to a gradient-decoupled DiT continuous action expert. It further performs semantic-aligned 3D integration by aligning a trainable 3D spatial backbone with the VLA representation while freezing the original VLM-aligned visual pathway. To share this representation across robots, GEAR-VLA uses embodiment canonicalization, where embodiment-aware states and embodiment-invariant actions confine robot differences to the low-level interface. Extensive simulation and real-world experiments demonstrate strong generalization: GEAR-VLA achieves state-of-the-art performance on LIBERO, zero-shot LIBERO-Plus, and RoboTwin 2.0, reaches 85.9% success on AgileX and 81.0% on the pretraining-unseen LDT-01 embodiment, and obtains 90.1% success on a 6,360-trial universal grasping benchmark with 212 unseen objects. Code and models will be released at https://github.com/babynabeauty/GEAR-VLA.

10.
arXiv (CS.AI) 2026-06-19

Reward as An Agent for Embodied World Models

arXiv:2606.19990v1 Announce Type: new Abstract: While RL has become a promising tool for refining world models, existing methods largely rely on conservative rollouts near the training distribution, limiting exploration, behavioral diversity, and richer dynamic discovery. In this work, we challenge this conservative paradigm. We argue that the core limitation is not exploration itself, but the lack of reliable verification strategies to support broader exploration. Without reliable verification, expanded exploration becomes highly susceptible to reward hacking, where policies exploit imperfect rewards without achieving genuine improvement. To evaluate this motivation, we instantiate our method in embodied world models, where physical plausibility, and task completion provide a rigorous testbed for scalable RL under complex dynamics. On the verification side, we introduce Reward as an Agent, an agentic reward framework that actively evaluates generated behaviors to provide robust reward signals and mitigate reward hacking under distribution shifts. On the exploration side, we introduce Dynamic-Aware Rollout Diversification through DynDiff-GRPO, which explicitly expands action-space exploration to diversify trajectories, broaden state-action coverage, and encourage richer embodied behaviors beyond conservative rollout regimes. By unifying Reward as an Agent with DynDiff-GRPO, we enable RL on a more reliable reward foundation with substantially diversified sampling, effectively mitigating reward hacking while yielding significant accuracy gains across multiple open-source world models, thereby demonstrating that broader exploration can scale successfully when grounded in robust verification.

11.
arXiv (CS.LG) 2026-06-16

Beyond Accuracy: Measuring Bias Acknowledgment in Chain-of-Thought Reasoning for Responsible AI Evaluation

arXiv:2606.15127v1 Announce Type: new Abstract: Reasoning models are increasingly used in settings where the final answer is not the only object of review: educational tools may show students intermediate steps, decision-support systems may require human oversight, and audit workflows may inspect traces for misleading or biased input. In such settings, two responses can receive the same final-answer score while differing in whether the trace explicitly flags injected biasing content. Accuracy-only evaluation collapses these cases. We study this gap as a measurement blind spot for responsible evaluation and introduce a minimal trace-level diagnostic with two axes: susceptibility (whether the bias breaks a previously correct answer) and acknowledgment (whether the trace contains a rubric-defined surface reference to the injected content). Across thousands of biased GSM8K trials, GPT-4o and Claude Sonnet~4 have similar susceptibility rates ($1.3\%$ vs.\ $1.2\%$) but substantially different acknowledgment rates ($13.0\%$ vs.\ $75.0\%$) under the same rubric.

12.
arXiv (CS.LG) 2026-06-19

Structure-Oriented Randomized Neural Networks for Poisson-Nernst-Planck and Poisson-Nernst-Planck-Navier-Stokes Systems

arXiv:2606.19912v1 Announce Type: cross Abstract: We develop a structure-oriented randomized neural network framework, termed SO-RaNN, for the Poisson-Nernst-Planck (PNP) system and the Poisson-Nernst-Planck-Navier-Stokes (PNP-NS) system. The decoupled linearized subproblems are solved iteratively by randomized neural networks in a space-time framework. For the concentration variables, a pointwise cut-off is used to enforce positivity at the value level, and discrete mass-scaling factors are computed at selected correction instants and interpolated in time, so as to ensure exact mass matching at those instants and to promote approximate mass preservation between them. To introduce an auxiliary discrete dissipation mechanism, we further employ an SAV-type post-processing correction, which yields monotonicity of the SAV auxiliary variable under the ideal SAV update. For the PNP-NS system, a structure-preserving randomized neural network (SP-RaNN) is used for the velocity field, so that the velocity approximation satisfies the incompressibility constraint pointwise by construction. On the theoretical side, we derive residual-based estimates for the raw, uncorrected RaNN solvers of the linearized subproblems, formulate a conditional local-in-time convergence result for the raw outer Picard iteration of the PNP system, and analyze the value-level positivity correction together with the mass-correction and SAV post-processing steps. For the PNP-NS system, we establish an approximation result for the SP-RaNN space and provide a conditional error statement for the corresponding linearized Oseen-type problem. Numerical experiments demonstrate approximation accuracy in the source-driven manufactured tests and illustrate the intended value-level positivity correction, selected-time mass matching, computed free-energy curves based on the final gauge-fixed potential, and divergence-free approximation in benchmark tests.

13.
arXiv (CS.LG) 2026-06-12

Physics-Informed Neural Networks for Chemotherapy Pharmacokinetics: Benchmarking the Clinical Estimator and Exposing Parameter Identifiability

arXiv:2606.12658v1 Announce Type: new Abstract: Physics-Informed Neural Networks (PINNs) are an attractive tool for partial-observation problems in biology, where the governing dynamics are known but some compartments cannot be measured. Chemotherapy pharmacokinetics (PK) is a clean instance: drug concentration in plasma is routinely measured, but concentration in tissue – which determines tumour kill and off-target toxicity – is not. We benchmark a PINN against the standard clinical baseline (nonlinear least-squares on the analytical biexponential plasma solution, hereafter NLS) and a physics-agnostic neural baseline (a data-only MLP) on two PK problems. On the linear two-compartment problem, NLS is near-optimal; the PINN matches it to within a small constant factor while also producing the tissue curve in a single training pass, whereas the data-only MLP fails on tissue by roughly 10x. On a Michaelis-Menten extension (saturable elimination), the biexponential closed form no longer exists, so NLS is mis-specified and silently returns meaningless rate constants. The PINN instead exposes a deeper fact: the Michaelis-Menten two-compartment model is non-identifiable from plasma alone, and the PINN reports this honestly by converging to a basin with k12 -> 0. Adding two sparse tissue observations largely resolves identifiability: across five seeds the PINN recovers k21 to within 1% of truth and Vmax, Km to within one standard-deviation bar, while k12 moves in the correct direction (0.02 -> 0.82) but remains ~2 sigma below truth – a recovery the closed-form NLS estimator cannot attempt at all, because its biexponential ansatz describes only plasma. Our claim is not that PINNs beat NLS. It is that PINNs offer a uniform recipe that ties the textbook estimator on the textbook problem, exposes structural identifiability that the textbook estimator hides, and absorbs heterogeneous measurements within a single loss.

14.
medRxiv (Medicine) 2026-06-18

Consistency of sleep timing and duration are associated with more physical activity and favorable heart rate metrics in a naturalistic cohort

Background: Regularity of sleep patterns over time has increasingly gained traction as an important axis of sleep health. Since sleep habits are under some degree of behavioral control, understanding such patterns in naturalistic settings is particularly important. We quantified sleep variability and tested the hypothesis that regularity correlates with physical activity, resting heart rate (rHR), and heart rate variability (HRV). Methods: We analyzed real-world digital health data from over 81,000 participants (over 18 million nights) who provided informed consent to participate in the Apple Heart and Movement Study and elected to contribute sleep, activity, and heart rate data to the study. Variability was quantified using the standard deviation (SD) computed from total sleep time (TST), sleep start time (S-start), end time (S-end), and midpoint time (MP), as well as the Sleep Regularity Index (SRI). Results: The SD-based variability metrics correlated with one another (R values 0.74-0.92), and with the SRI metric (R values 0.62-0.64). More consistent sleep, by any metric, was associated with more activity and better rHR and HRV. The most consistent tertile for TST variability had higher median TST (6.9 vs 5.9 hours), more daily exercise (32.8 vs 20.4 minutes), lower rHR (62.4 vs 65.6 beats per minute), and higher HRV (40.6 vs 37.3), all p

15.
bioRxiv (Bioinfo) 2026-06-12

From Proteome Mining to Structural Validation: Phosphopyruvate Hydratase as a Structurally Tractable Drug Target in Kinetoplastid Parasites

Chagas disease, caused by Trypanosoma cruzi, demands novel therapeutic strategies that overcome the toxicity and limited efficacy of current treatments. To address this need, herein we report an integrative, target-centric strategy that combines parasite proteome mining, structural modeling, and experimental validation. Functional enrichment and druggability analyses identified phosphopyruvate hydratase (PPH) as a promising candidate due to its essential metabolic role and limited similarity to human homologs. Notably, proteome mining revealed the presence and conservation of PPH across kinetoplastid parasites, including Leishmania donovani, supporting its evaluation beyond T. cruzi. For the selected PPH sequences, AlphaFold-derived three-dimensional models underwent extensive molecular dynamics refinement, yielding stable conformational ensembles suitable for structure-based studies. Using this validated model, virtual screening of the Latin American Natural Products Database - LANaPDB - identified aptosimon as a top-ranked compound candidate. Molecular dynamics simulations further showed ligand-dependent binding behavior, suggesting alternative binding modes distinct from the canonical substrate configuration. In vitro assays demonstrated consistent antiparasitic activity against intracellular T. cruzi amastigotes (IC50 = 3.52 ug/mL) and Leishmania donovani promastigotes (IC50 = 13.06 ug/mL), supporting the biological relevance of the aptosimon-related lignan chemotype, hinokinin, across two kinetoplastid parasite models. Together, these results support PPH as a structurally tractable and biologically relevant candidate target, while identifying an aptosimon-related lignan chemotype, represented experimentally by hinokinin, as a cross-species antiparasitic scaffold that warrants further biochemical target-validation studies.

16.
bioRxiv (Bioinfo) 2026-06-13

ProtAff: Protein Binding Affinity Prediction via LoRA-Finetuned ESM-2

Predicting the binding affinity of protein–protein interactions remains a central challenge in computational biology. Structure prediction models such as AlphaFold3 (AF3) and Boltz-2 can produce high-quality docking poses, and their confidence scores indicate structure quality, but these same scores fail to rank binding affinity among confirmed binders. Here we present ProtAff, a sequence-only affinity prediction model built on ESM-2 (650M parameters) with low-rank adaptation (LoRA) fine-tuning and a cross-attention module. ProtAff is trained using a margin ranking loss on 362,567 affinity measurements spanning 20 heterogeneous data sources, and we removed all training samples whose target sequence exceeds 50% similarity to the test target EGFR. On the AdaptyvBio EGFR benchmark (N = 55), ProtAff achieves a Spearman correlation coefficient {rho} = 0.413, outperforming the best AF3 metric ({rho} = 0.054), the best Boltz-2 metric ({rho} = -0.046), and ML-based predictors MINT ({rho} = 0.242) and CrossAffinity ({rho} = 0.216). Applied to the AdaptyvBio Nipah virus binder design competition, a pipeline incorporating ProtAff for affinity ranking produced a design with KD = 0.132 nM (2 of 5 designs confirmed binding), a 2.8-fold improvement over the competition winner. On a cross-target discrimination benchmark of 91 VHH-antigen crystal structures, ProtAff underperforms structural methods for distinguishing cognate from non-cognate pairings, indicating that sequence-based affinity models are effective for within-target ranking but not for cross-target specificity.

17.
arXiv (CS.LG) 2026-06-16

OptEMA: Adaptive Exponential Moving Average for Stochastic Optimization with Zero-Noise Optimality

作者:

arXiv:2603.09923v4 Announce Type: replace Abstract: Exponential moving averages (EMAs) are a central component of widely used adaptive optimizers such as Adam. However, existing analyses of Adam-style methods often yield suboptimal guarantees in the zero-noise regime, rely on open-loop parameter schedules, or require prior knowledge of smoothness constants. Motivated by these limitations, we introduce OptEMA and analyze two complementary variants: OptEMA-M, which applies an adaptive, decreasing EMA coefficient to the first moment with a fixed second-moment decay, and OptEMA-V, which swaps these roles. At the heart of these variants is a Corrected AdaGrad-Norm coefficient schedule. This formulation renders OptEMA algorithmically closed-loop and Lipschitz-free, meaning its effective stepsizes are trajectory-dependent and require no parameterization via the Lipschitz constant. Under lower-boundedness, unbiasedness, bounded variance, average smoothness, and a bounded stochastic-gradient condition used to control the adaptive normalizers, we prove that both variants achieve the unified noise-adaptive rate $\tilde{\mathcal{O}} \left(T^{-1/2}+\sigma^{1/2}T^{-1/4}\right)$ for the averaged gradient norm. In the zero-noise regime, these bounds automatically reduce to the nearly optimal deterministic rate $\widetilde{\mathcal{O}}(T^{-1/2})$ without manual hyperparameter retuning.

18.
arXiv (CS.CV) 2026-06-11

Performance Analysis of YOLOv11 and YOLOv8 for Mixed Traffic Object Detection under Adverse Weather Conditions in Developing Countries

In modern vehicular systems, robust performance under harsh conditions has become a critical problem of autonomous driving. Our study delivers a comprehensive evaluation of the newest iteration of the YOLO series, which is YOLOv11 Nano architecture benchmarked against the widely adopted YOLOv8 Nano as a baseline on a custom fused dataset that combines the Indian Driving Dataset (IDD) [1] and Berkeley Deep Drive Dataset (BDD100K) [2]. We have analyzed the trade-offs among detection accuracy, inference speed, and computational efficiency in high-entropy scenarios involving dense mixed traffic, rain, and low-light conditions. Specifically, YOLOv11n achieves a mean Average Precision (mAP@50) of 46.6%, with a notable 3.2% improvement in Precision over the baseline, effectively reducing false positives in cluttered scenes. Furthermore, the proposed model exhibits enhanced energy efficiency, requiring 22% fewer FLOPs (6.3G vs. 8.1G) while maintaining real-time inference speed of 70.9 FPS on a Tesla T4 GPU, offering an optimal trade-off for safety-critical edge deployment.

19.
arXiv (CS.CL) 2026-06-16

Think-at-Hard: Selective Latent Iterations to Improve Reasoning Language Models

Improving the reasoning abilities of Large Language Models (LLMs), especially under parameter constraints, is crucial for real-world applications. Looped transformers address this by performing multiple latent iterations to refine each token beyond a single forward pass. However, we identify a latent overthinking phenomenon: most token predictions are already correct after the first pass, but are sometimes revised into errors in later iterations. We ask whether selectively skipping latent iterations can improve accuracy, and reveal significant potential with an oracle iteration policy that boosts performance by up to 7.3%. Motivated by this, we propose Think-at-Hard (TaH), a looped transformer optimized for selective iteration. TaH employs a lightweight neural decider to trigger latent iteration, only at tokens likely to be incorrect after the standard forward pass. During latent iterations, depth-aware Low-Rank Adaptation (LoRA) modules shift the objective from general next-token prediction to focused hard-token refinement. A duo-causal attention mechanism extends attention from the token sequence dimension to an additional iteration depth dimension, enabling cross-iteration information flow with full sequential parallelism. Experiments on nine benchmarks show consistent gains across math, QA, and coding tasks. With identical parameter counts, TaH outperforms always-iterate baselines by 3.8-4.4% while skipping iterations on 93% of tokens, and exceeds single-iteration Qwen3 baselines by 3.0-3.8%. When allowing

20.
arXiv (CS.CV) 2026-06-11

FronTalk: Benchmarking Front-End Development as Conversational Code Generation with Multi-Modal Feedback

We present FronTalk, a benchmark for front-end code generation that pioneers the study of a unique interaction dynamic: conversational code generation with multi-modal feedback. In front-end development, visual artifacts such as sketches, mockups and annotated creenshots are essential for conveying design intent, yet their role in multi-turn code generation remains largely unexplored. To address this gap, we focus on the front-end development task and curate FronTalk, a collection of 100 multi-turn dialogues derived from real-world websites across diverse domains such as news, finance, and art. Each turn features both a textual instruction and an equivalent visual instruction, each representing the same user intent. To comprehensively evaluate model performance, we propose a novel agent-based evaluation framework leveraging a web agent to simulate users and explore the website, and thus measuring both functional correctness and user experience. Evaluation of 20 models reveals two key challenges that are under-explored systematically in the literature: (1) a significant forgetting issue where models overwrite previously implemented features, resulting in task failures, and (2) a persistent challenge in interpreting visual feedback, especially for open-source vision-language models (VLMs). We propose a strong baseline to tackle the forgetting issue with AceCoder, a method that critiques the implementation of every past instruction using an autonomous web agent. This approach significantly reduces forgetting to nearly zero and improves the performance by up to 9.3% (56.0% to 65.3%). Overall, we aim to provide a solid foundation for future research in front-end development and the general interaction dynamics of multi-turn, multi-modal code generation. Code and data are released at https://github.com/shirley-wu/frontalk

21.
arXiv (CS.AI) 2026-06-19

Beyond Reasoning Gains: Mitigating General-Capability Forgetting in Large Reasoning Models

arXiv:2510.21978v2 Announce Type: replace-cross Abstract: Reinforcement learning with verifiable rewards (RLVR) has delivered impressive gains in mathematical and multimodal reasoning and has become a standard post-training paradigm for contemporary language and vision-language models. However, the RLVR recipe introduces a significant risk of capability regression, in which models forget foundational skills after prolonged training without employing regularization strategies. We empirically confirm this concern, observing that open-source reasoning models suffer performance degradation on core capabilities such as perception and faithfulness. While imposing regularization terms like KL divergence can help prevent deviation from the base model, these terms are computed on the current task and therefore do not guarantee preservation of broader knowledge. Meanwhile, commonly used experience replay across heterogeneous domains makes it nontrivial to decide how much training emphasis each objective should receive. To address this, we propose RECAP-a replay strategy with dynamic objective reweighting for general knowledge preservation. Our reweighting mechanism adapts online using short-horizon signals of convergence and instability, shifting the post-training focus away from saturated objectives and toward underperforming or volatile ones. Our method is end-to-end and readily applicable to existing RLVR pipelines without training additional models or heavy tuning. Extensive experiments on benchmarks using Qwen2.5-VL-3B and Qwen2.5-VL-7B demonstrate the effectiveness of our method, which not only preserves general capabilities but also improves reasoning by enabling more flexible trade-offs among in-task rewards.

22.
arXiv (CS.AI) 2026-06-16

Recurrent Reasoning on Symbolic Puzzles with Sequence Models

arXiv:2606.15686v1 Announce Type: new Abstract: Large language models often appear strong on symbolic and algorithmic tasks, yet this apparent strength can hide brittle behaviour when problems become longer, harder, or slightly out of distribution. A major limitation of current reasoning benchmarks is that many primarily test whether a model can produce a valid answer, while paying less attention to whether the solution is minimal, robust, and stable under controlled difficulty scaling. We introduce RecurrReason, a difficulty-controlled benchmark of four recurrent logic puzzles (Tower of Hanoi, River Crossing, Block World, and Checkers Jumping) with BFS-optimal trajectories and a single interpretable difficulty parameter $N \in \{1,\dots,10\}$, totalling 10{,}817 unique puzzles and 285{,}933 moves. We benchmark two Transformer families, an encoder-decoder model (T5-style) and a decoder-only model (GPT-2-style), under consistent data splits and evaluation criteria, training on $N{=}1$ to $7$ and evaluating on both held-out in-distribution instances and harder out-of-distribution instances at $N{=}8$ to $10$. Fine-tuned pre-trained T5 achieves 97.27\% validation and 81.00\% OOD accuracy on Block World; all models score 0.00\% on River Crossing under all conditions. Failure mode analysis reveals that architecture is a stronger determinant of success than scale. Pre-training transfers only to puzzles with locally structured transition functions. Our code and dataset will be open-sourced upon acceptance.

23.
arXiv (CS.AI) 2026-06-16

From Tokens to Regions: CUDA-Sensitive Instruction Tuning for GPU Kernel Generation

arXiv:2606.16231v1 Announce Type: cross Abstract: High-performance CUDA kernels are essential for scalable AI systems, while Large Language Models (LLMs) still struggle to generate correct kernels due to strict and implicit execution constraints. Existing LLM-based approaches either rely on costly agentic or reinforcement-learning (RL) pipelines, or adopt supervised fine-tuning (SFT) objectives that fail to explicitly model CUDA sensitivity, namely code tokens or regions tightly coupled with execution constraints. In this work, we investigate CUDA sensitivity from the perspective of token confidence patterns, showing that CUDA sensitivity appears at both token and region levels, where most CUDA-sensitive tokens are predicted with high confidence, while a smaller low-confidence subset forms regions corresponding to execution-critical structures. These findings suggest that effective CUDA kernel generation should both leverage high-confidence CUDA-sensitive tokens and preserve low-confidence CUDA-sensitive regions. Building on these insights, we propose \underline{CUDA-\underline{Se}nsitive Instruction \underline{T}uning (CuSeT)}, a low-cost post-training method within a simple SFT framework. CuSeT follows the principle of ``from tokens to regions'' by combining adaptive token-level masking with region-aware sample reweighting. Experiments show that CuSeT consistently improves functional correctness across multiple model families and scales, outperforming standard SFT and advanced SFT variants, while achieving competitive performance against frontier CUDA kernel generation models with substantially lower inference cost.

24.
arXiv (CS.CV) 2026-06-17

MOCHI: Motion Enhancement of Collaborative Human-object Interactions

Collaborative human-object interaction shows dynamic and complex movements that require mutual anticipation and continuous adjustment between participants and the shared object. Modeling such collaborative multi-human object interaction (MHOI) scenarios requires high-quality data acquisition as a foundational step; however, this is challenging due to the inherent complexity of MHOI where human-human and human-object interactions occur simultaneously. Such complexity leads to noisy MHOI captures characterized by several artifacts: contact misalignment between hands and objects, motion jitter and temporal inconsistencies in the captured sequences, and missing or incomplete finger-level articulation details. To address these challenges, we present MOCHI (MOtion Enhancement of Collaborative Human-object Interactions), a two-stage framework for enhancing noisy MHOI data. Our approach first generates physically plausible hand grasps through optimization from noisy body input, producing grasps that are both physically plausible and semantically consistent with the body pose, where these optimized grasps are extended into complete hand-object interaction sequences. Consequently, the full-body motion for all participants are refined through a diffusion-based noise optimization framework that uses single-person motion priors. During the optimization process, we introduce optimization objectives to encode human-object and human-human interaction information within these single-person priors. Experimental results demonstrate the effectiveness of our pipeline across diverse MHOI data, either acquired by existing capture methods or synthesized by generative models. We further show robustness of our system across varying numbers of participants and types of interactions, and demonstrate various applications including keyframe-based MHOI creation and data augmentation through varying object geometries.

25.
arXiv (CS.CV) 2026-06-16

Akasha 2: Hamiltonian State Space Duality and Visual-Language Joint Embedding Predictive Architectur

作者:

We present Akasha 2, a state-of-the-art multimodal architecture that integrates Hamiltonian State Space Duality (H-SSD) with Visual-Language Joint Embedding Predictive Architecture (VL-JEPA). The system leverages the Mamba-3 Selective State Space Model (SSM) augmented by a Sparse Mixture of Hamiltonian Experts (SMoE-HE) that enforces latent physical conservation laws through symplectic integration. For visual synthesis, we introduce Hamiltonian Flow Matching (HFM) and persistent 3D Gaussian Splatting (3DGS), enabling ultra-low latency (