arXiv:2606.19938v3 Announce Type: replace Abstract: We propose triangular consistency as a first-principled constraint for optical flow, which is agnostic to network architecture, supervision type, and dataset, and applies to both image-pair and multi-frame settings. This simple but powerful constraint is to compose two flows to induce a third flow and enforce consistency among the three. The composed flows may arise from (i) image pairs, yielding cycle consistency; (ii) multiple video frames, producing longer-range motion through temporal chaining; or (iii) image pairs combined with controlled synthetic transformations, which becomes data augmentation. This triangular consistency introduces negligible computational overhead and requires no additional annotations. Since it is derived directly from the geometry of optical flow, it does not rely on model-specific assumptions and serves as a ``universal'' plug-and-play component for optical flow training. Experiments show consistent improvement across supervised, unsupervised, and transfer learning settings.
Science Journals
arXiv:2606.20280v2 Announce Type: replace Abstract: Leveraging Multimodal Large Language Models (MLLMs) via contrastive learning has become a mainstream paradigm for improving the performance of Universal Multimodal Retrieval (UMR). However, previous works have ignored the grain blindness when adapting the contrastive paradigm into retrieval tasks. Grain blindness refers to the tendency of the model to overlook grain-level information contained in the query, which is crucial for effectively handling complex queries. This stems from contrastive learning treating samples as a binary classification (positive/negative), while ignoring the different information carried by each negative sample. To address this, we argue that negatives should be treated differently according to their similarity to the positive sample, enabling the model to learn distinct grain information from each negative. In this paper, we introduce a simple but effective framework, called ELVA, a novel rule-based RL framework that mitigates grain blindness through ranking-driven MLLMs. 1) Instead of relying on reward models, we extend Reinforcement Learning with Verifiable Rewards (RLVR) to retrieval tasks, allowing the model to explore new ranking behaviors without explicit ranking labels. 2) By utilizing rule-based rewards, our approach jointly optimizes the ranking of negative samples while enlarging the similarity gap between positive and negative. To more precisely measure grain blindness, we further introduce MRBench, a new benchmark specifically designed for multi-grain query scenarios. ELVA achieves state-of-the-art results across standard retrieval benchmarks, and its notable 13.1% improvement on MRBench further demonstrates its effectiveness in alleviating grain blindness.
arXiv:2606.20408v3 Announce Type: replace Abstract: Large language model (LLM) agents are increasingly proposed as supervisory components for safety-critical systems, yet their robustness under sustained, adaptive adversarial pressure remains poorly characterized. We present NRT-Bench, a benchmark for multi-turn red-teaming of LLM agents acting as operators of a safety-critical system, instantiated in a simulated nuclear power plant control room. A five-role operator team, each backed by a configurable LLM, runs a plant governed by six critical safety functions (CSFs), while adversaries inject messages over four channels in bounded multi-turn sessions with per-turn feedback. Harm is an objective signal rather than LLM-judged text: a run terminates the moment any CSF is lost, attributed to the causing message. Evaluating four frontier operator models under a fixed-attack paired-replay protocol, we find that adaptive multi-turn attacks reliably push the operator team past a safety limit: across the four models, between 8.7% and 12.1% of attack sessions end with the plant losing a critical safety function. Although the four models look almost equally robust by this aggregate rate, their failures barely overlap: of $149$ sessions, none defeat all four models while a third defeat at least one, so vulnerabilities are nearly disjoint across models rather than nested. The effect of added defences is strongly model-dependent: the same guardrail stack or safety-advisor agent that lowers attack success for one model can raise it for another. We release the simulation venue, attack dataset, and replay tooling for reproducible safety evaluation of LLM agents.
arXiv:2606.20659v2 Announce Type: replace Abstract: Agent skills encode reusable procedural knowledge for large language model (LLM) agents, and existing benchmarks show that such skills can improve task-level performance. However, a task outcome does not reveal which parts of a reusable skill were exercised, nor whether the agent followed the relevant skill instructions when those parts were exercised. This gap makes it unclear whether a skill has been adequately tested, or whether observed task failures provide actionable evidence for improving agent skill effectiveness. To fill this gap, we introduce skill coverage, a trajectory-based test-adequacy metric for reusable agent skills. Our framework extracts skill behavior constraints from each skill, translating natural-language skill instructions into semi-structured constraints that specify the expected agent behavior under particular conditions. It then determines whether each constraint is covered by an agent trajectory and, for covered constraints, assigns a Pass or Fail verdict according to the agent behavior. We apply this framework to SkillsBench. The results show that agent trajectories on the benchmark leaderboard cover only 38.66 to 45.51% of the extracted skill behavior constraints on average. We then use Fail verdicts to strengthen the corresponding skill content only by emphasizing the original instructions that the agent failed to follow, and run the same tasks with the strengthened skills. This emphasis yields an average 16.0% recovery rate of the failed tasks across the five agent-model rows. These results show that skill coverage is both a test-adequacy metric and a fine-grained signal for observing skill-use behavior. In failed tasks, failed constraint labels provide actionable evidence for improving agent skill effectiveness. A project website accompanies the paper.
arXiv:2607.04941v1 Announce Type: new Abstract: Full-duplex spoken dialogue models are trained on conversational speech in which each speaker is represented as a separate stream, but existing large-scale public speech corpora are mostly monaural, making them unsuited for SDLM training. We present DuplexChat, an open-source corpus for full-duplex spoken dialogue models, and DuplexChat-Pipe, a pipeline for constructing speaker-separated full-duplex dialogue speech from public podcast feeds. DuplexChat-Pipe filters language-specific podcast feeds, retrieves and cleans episode audio, extracts diarization-guided two-speaker dialogue clips, and applies speech separation and restoration to produce one channel per speaker. Running this pipeline yields a speaker-separated spoken dialogue corpus covering 282,634 hours of English and 132,723 hours of Japanese. Analysis results on DuplexChat show that it contains turn-taking dynamics present in human dialogues.
arXiv:2606.20742v2 Announce Type: replace Abstract: UAV-based pavement inspection can reduce the cost and risk of road-surface monitoring, but real-world deployment remains difficult when traffic, pedestrians, and temporary occlusions affect defect visibility. This paper presents a Unity-based digital twin framework for traffic-aware UAV pavement monitoring in open-traffic conditions. The proposed environment integrates procedurally generated road defects, dynamic traffic agents, autonomous UAV navigation, and a multitask YOLOv8n perception module for detecting road defects, pedestrians, and vehicles while classifying road-defect subtypes. After synthetic-domain fine-tuning, the perception model achieved 0.959 mAP@0.5 and 0.940 macro F1-score on a held-out synthetic test set generated from the simulator. The digital twin was then used to evaluate hover-and-recheck, micro-repositioning, and skip-and-revisit recovery strategies across different traffic densities and flight altitudes. Results show that flight altitude strongly affects inspection coverage, while recovery strategies introduce different trade-offs between coverage, mission duration, energy consumption, and revisit behaviour. These findings demonstrate that digital twins can support the development and evaluation of traffic-aware UAV inspection strategies before real-world deployment. The full implementation and trained models are available at https://github.com/EdwinTSalcedo/RDMO-DigitalTwin.
arXiv:2606.20971v2 Announce Type: replace Abstract: We introduce UNITY, a Universal-to-Specialized adapter for efficient and scalable composite conditioning in diffusion based image generation. Unlike prior methods that train separate adapters for each conditioning modality, UNITY jointly learns shared semantics across multiple conditioning types and subsequently specializes without modifying the underlying architecture. The proposed two stage training paradigm consists of a Universal Stage that captures cross modal representations across all conditioning modalities using half of the total training steps, followed by a Specialization Stage that refines modality specific features using the remaining training budget. At the core of UNITY are the Morphable Attention Flow (MAF) Network and Morph Wrapper modules, which enable channel aware and spatially adaptive feature alignment through learnable flow fields and attention based fusion. This constant complexity formulation supports flexible operation under both single and composite conditioning settings while significantly reducing inference latency and memory consumption. Extensive experiments across multiple datasets demonstrate that UNITY achieves state of the art image fidelity while maintaining superior memory efficiency. Code: https://github.com/arya-domain/UNITY
arXiv:2606.21883v2 Announce Type: replace Abstract: In this paper, we propose a modified projected Walk on Spheres method (MPWoS) for screened Poisson equations on embedded manifolds. The method employs local extensions together with the Green representation in local Euclidean balls, coupled with a closest-point projection that maps the boundary samples back to the manifold. This formulation yields a meshfree and highly parallelizable stochastic recursion in the ambient Euclidean space, rather than a direct discretization of the Laplace-Beltrami operator on the manifold. The proposed approach can be viewed as a high-dimensional extension and modification of the projected Walk on Spheres method introduced for surface PDEs in [Sugimoto et al., SIGGRAPH Asia 2024 Conference Papers, pp. 1-10], with three main distinctions: a compensation term that corrects the discrepancy between the ambient Laplacian applied to the closest-point extension and the intrinsic Laplace-Beltrami operator on the manifold, an adaptive radius strategy determined by local geometric and boundary information, and a rigorous error analysis for the proposed algorithm. Under assumptions on the geometric projection and the prescribed compensation accuracy, we establish mean-square error estimates for the proposed Monte Carlo method in both the boundary and closed-manifold settings. Extensive numerical examples on parametrized, implicit, high-dimensional (up to 1000 dimensions), and point-cloud manifolds are presented to illustrate the convergence and efficiency of the proposed method across different geometries.
arXiv:2606.21942v2 Announce Type: replace Abstract: Spanwise wall oscillation can enhance convective heat transfer disproportionately to its drag penalty, a departure from the Reynolds analogy termed dissimilar heat transfer (DHT). The companion study of Gu'erin et al. (2026) established that an optimised quasi-plateau waveform attains an analogy factor $\overline{A}n \approx 1.09$ at $Pr = 1$ and attributed this preferential thermal enhancement to the absence of a pressure-strain redistribution channel in the temperature variance equation, but the mechanism had not been quantitatively verified. The present study addresses this gap through phase-resolved variance transport budget analysis from direct numerical simulation of turbulent channel flow at $Re\tau = 200$, $Pr = 1$. Two complementary pressure-mediated mechanisms are identified. At the Stokes-strain reversal, the pressure-strain redistribution $\Pi_{uu}$ imposes a pronounced drain on the streamwise velocity variance with no counterpart in the temperature variance equation: the divergence-free constraint redistributes momentum variance among velocity components but has no scalar analogue. During the quasi-steady plateau phases, the pressure-temperature-gradient correlation $\Pi_{v\theta}$ preferentially enhances the wall-normal scalar flux relative to the momentum flux. The concentration of both mechanisms within the reversal and plateau phases, rather than at the Stokes-layer penetration maxima, identifies the duration of the quasi-steady phases as the controlling parameter for DHT enhancement, resolving the paradox whereby increased penetration depth does not produce increased dissimilarity.
arXiv:2501.00048v2 Announce Type: replace Abstract: Every year in the United States, 800,000 individuals suffer a stroke - one person every 40 seconds, with a death occurring every four minutes. While individual factors vary, certain predictors are more prevalent in determining stroke risk. As strokes are the second leading cause of death and disability worldwide, predicting stroke likelihood based on lifestyle factors is crucial. Showing individuals their stroke risk could motivate lifestyle changes, and machine learning offers solutions to this prediction challenge. Neural networks excel at predicting outcomes based on training features like lifestyle factors, however, they're not the only option. Logistic regression models can also effectively compute the likelihood of binary outcomes based on independent variables, making them well-suited for stroke prediction. This analysis will compare both neural networks (dense and convolutional) and logistic regression models for stroke prediction, examining their pros, cons, and differences to develop the most effective predictor that minimizes false negatives.
arXiv:2501.06369v3 Announce Type: replace Abstract: We consider the Stokes-Darcy coupled problem, which models the interaction between free-flow and porous medium flow. By enforcing the normal flux continuity interface condition directly within the finite-element spaces, we establish unified well-posedness results for the coupled system under various boundary condition scenarios. Using the operator preconditioning framework, we develop a parameter-robust preconditioner that avoids the use of fractional operators. Numerical experiments employing both $H(\operatorname{div})$-conforming and nonconforming finite-element methods are presented to confirm the theoretical findings and demonstrate the robustness of the proposed block preconditioners with respect to the physical parameters and mesh size.
arXiv:2501.10711v5 Announce Type: replace Abstract: Code-related benchmarks play a critical role in evaluating large language models (LLMs), yet their quality fundamentally shapes how the community interprets model capabilities. In the past few years, awareness of benchmark quality has grown. Yet, after a decade-scale (2014-2025) survey over 672 code benchmarks, we observed a lag between growing awareness and actual practice. For example, in 2025 alone, the number of benchmarks that ignore code coverage when providing test cases nearly matches the total count accumulated across the previous ten years. In response, we take a clear position: Code benchmarks must prioritize rigor in benchmark construction, reliability in evaluation, and reproducibility in release. To operationalize this position, we introduce a code benchmark guideline HOW2BENCH with 55 checklists. Finally, our further human study also exposed that the current issues not only stem from the significant effort required, but also from a lack of awareness regarding their importance.
arXiv:2606.22005v3 Announce Type: replace Abstract: We present InstructFX2FX, a system for sequential audio effect refinement through multi-turn natural-language instructions. Existing text-to-effect systems are largely single-shot, mapping one textual descriptor to one preset. Real audio engineering is instead sequential: engineers refine an existing effect chain through successive instructions. This poses a stateful problem that single-shot systems do not address: given the current effect parameters state and a new instruction, update the sound while preserving what earlier instructions already achieved. InstructFX2FX addresses this with a hybrid architecture that divides labor between a language model and CLAP-guided optimization. The LLM serves as a high-level planner that selects effects and proposes the initial parameter state, motivated by recent evidence that LLMs can outperform CLAP-based optimization for single-turn text-to-effect mapping; CLAP-guided optimization then refines the existing parameter state, providing a more stable and robust refinement mechanism than LLM reprompting. In the demo, attendees drive a dry recording through successive natural-language instructions: after each turn, they choose how strongly the effect is applied, then issue the next instruction based on what still differs from the sound they intend. In a preliminary evaluation on SocialFX-derived descriptor pairs, CLAP-guided refinement achieves lower DSP-feature MMD than an LLM+LLM initialize-then-reprompt baseline on 9 of 10 pairs. Trajectory analysis further shows that, for differentiable effects, optimization tends to gradually move the audio toward the new target while retaining the effects of the previous instruction, highlighting the potential for gradual refinement.
arXiv:2606.22047v2 Announce Type: replace Abstract: Guided acoustic wave Brillouin scattering (GAWBS) is a major obstacle in fiber-based quantum and high-speed classical communication systems as well as in interferometry. The transverse phonons driving it modulate the light field in the fiber core, adding thermal noise to the signal. To this day, there is no known method to eliminate GAWBS from the fiber or to compensate its effects completely. In this letter, we present twisted photonic crystal fibers (t-PCF) as the first-ever fiber system allowing a complete removal of mixed torsional radial GAWBS in a Stokes basis. The torsional radial modes modulate the fiber asymmetrically in the transverse direction, resulting in linear birefringence. While pure phase modulation is added as common noise in the guided fiber modes and can be easily removed through self-referencing, linear birefringence induces polarization modulation, which cannot be counteracted. In t-PCFs, the transverse symmetry of the geometry translates to multiple symmetries in the acoustic and optical domains in the circular basis. This enables equal phase accumulation in certain orientations in the two optical modes. Through experiments and theory, we show that the GAWBS-induced phase can be compensated down to the quantum-noise limit by self-referencing in a symmetric interferometer with Stokes detection.
arXiv:2606.22094v2 Announce Type: replace Abstract: Accurate yaw estimation is a bottleneck in cross-view localization between ground view and Bird's Eye View (BEV). Existing methods couple yaw with translation and rely on height or projection assumptions that degrade under large yaw ambiguity. We disentangle yaw from location accuracy and introduce LAYS, a radially invariant line-consensus voting method. By exploiting the radial invariance of our formulation, we achieve sub-degree yaw precision via 3D voting over all candidate poses, while eliminating the need for accurate location. Our key observation is that a ground-image column matched to BEV pixels induces the same yaw across all camera positions along the radial direction of the pixels. LAYS matches BEV pixels to ground columns using feature similarity and accumulates the induced yaw votes into discrete 3D bins, where correct correspondences along the radial line concentrate into a sharp peak for the correct yaw. Experiments on Mapillary, Ford, KITTI, and VIGOR show significant gains under unknown yaw, particularly for normal FoV with unknown yaw (+28$\sim$45\%p), and using LAYS as a yaw prior improves downstream 3-DoF localization.
arXiv:2510.10355v2 Announce Type: replace Abstract: The semi-implicit (partly decoupled, also called staggered or fraction-step) time discretization is applied to compressible nonlinear dynamical models of viscoelastic solids in the Eulerian description, i.e.\ in the actual deforming configuration, formulated fully in terms of rates. The Kelvin-Voigt rheology and also, in the deviatoric part, the Jeffreys rheology are considered. The numerical stability and, considering the Stokes-type viscosity multipolar of the 2nd-grade, also convergence towards weak solutions are proved in three-dimensional situations, exploiting the convexity of the kinetic energy when written in terms of linear momentum. No (poly)convexity of the stored energy is required and some enhancements (specifically towards damage and diffusion models) are briefly outlined, too.
arXiv:2606.22574v2 Announce Type: replace Abstract: While synthetic data generation resolves the manual labeling bottleneck in computer vision, minimizing the syn-to-real domain gap requires optimizing rendering variables. This paper presents a systematic study analyzing the impact of lighting configurations and background complexity on object detection performance. We introduce SmartSDG, an automated, reproducible pipeline built on NVIDIA Isaac Sim using Physically-Based Shading (PBS), alongside ILLUM\_INTRUCK, a new multi-object industrial benchmark dataset. Through 18 controlled experiments utilizing a state-of-the-art YOLOv12 framework, we demonstrate that complex, indirect lighting configurations paired with domain-relevant background variability significantly increase visual cue richness. Our quantitative findings show that avoiding direct specular peaks preserves crucial surface textures, mitigates the domain gap, reduces false positives, and accelerates model convergence compared to using conventional direct-light synthetic data. Ultimately, we provide actionable virtual scene design guidelines to maximize object detection robustness in industrial automation.
arXiv:2606.24107v2 Announce Type: replace Abstract: Short dramas, with their rapid shot rhythms, dialogue-driven focus shifts, and demanding cinematographic grounding, pose challenges that prompt-level or text-only video generation pipelines struggle to meet. We study plot-to-short-drama generation, where a global plot and local context are transformed into visually grounded multi-shot videos. We propose DramaDirector, a geometry-grounded framework that lets the planner borrow cinematographic geometry from a gallery of real short-drama shots indexed by depth and pose. DramaDirector decouples each shot into static visual and dynamic narrative conditions, trains the planner with schema-constrained SFT and GRPO under a learned text-visual alignment reward, and retrieves depth-pose references to guide first-frame generation and image-to-video synthesis. We also introduce DramaBoard, a benchmark built from 35 live-action dramas, 2.8K episodes, and 81K shots, with structured storyboards and multi-dimensional evaluation protocols. Experiments show that DramaDirector improves over representative multi-agent and video generation baselines on faithfulness, consistency, and controllability. Our code is released at: https://github.com/iLearn-Lab/DramaDirector
arXiv:2606.24155v4 Announce Type: replace Abstract: Existing medical AI benchmarks lack process visibility, atomic skill evaluation, and integrated hallucination detection. We introduce MedBench v5, a redesigned benchmark for clinical multimodal models (language, vision-language, and agent systems) that moves from static QA to dynamic, process-oriented evaluation. MedBench v5 features: (1) a dual-dimensional framework combining Clinical Cognitive Responsiveness (13 sub-dimensions) and Medical Atomic Skills (4 agent environments), covering 63 tasks; (2) three switchable information-flow stressors (omission, contradiction, evidence delay) for factorized degradation analysis; (3) a dynamic process audit protocol with five reasoning nodes that produces model-specific failure fingerprints; (4) hallucination propagation monitoring across initiation, propagation, anchoring, and contradiction interaction-capturing silent hallucination. Experiments on frontier models show that strong overall task performance does not guarantee process stability: stressors mainly disrupt contradiction detection, diagnosis updating, hallucination propagation, and contradiction-based self-correction, while final evidence grounding can remain superficially stable. MedBench v5 provides a unified infrastructure for capability profiling, controllable stress testing, process auditing, and hallucination trajectory analysis in clinical AI evaluation.
arXiv:2607.04974v1 Announce Type: new Abstract: Multi-Modal Agents (M-agents), empowered by Large Language Models (LLMs), excel in various complex, open-world scenarios such as autonomous driving and robotics. However, their unique requirements to interact with dynamic and diverse multi-modal environments introduce novel implementation challenges beyond those faced by traditional agents. Outdated perception, untrustworthy planning and inapplicable execution could cause traffic accident and financial loss. Despite growing study on agent issues, there has not been a systematic study focusing on M-agent-specific implementation bugs. To address this gap, we conducted the first systematic study of implementation bugs in M-agents. We collected 34 representative M-agents from diverse sources and, through meticulous filtering,identified 158 M-agent-specific bugs from 1,268 issue reports. Using a top-down strategy, we developed a comprehensive taxonomy that classifies bugs by global symptoms, functionality component-level symptoms, and root causes. We then implemented MATester, an automatic proof-of-concept bug identifier by analyzing runtime inter-component outputs. When applied to 12 extra M-agents, MATester successfully covered 61.4% of known open issues and discovered 31 additional bugs, demonstrating the practical usefulness of our study. Our work provides a comprehensive reference and guideline for classification, prevention and fix of M-agent bugs.
arXiv:2606.24530v2 Announce Type: replace Abstract: We introduce NatureBench, a cross-discipline benchmark of 90 tasks distilled from peer-reviewed Nature-family publications, designed to evaluate whether AI coding agents can move beyond reproduction toward discovery on real scientific problems. NatureBench is built on NatureGym, an automated pipeline that constructs a standardized, per-task containerized environment from a source paper, addressing the environment-fragmentation problem that has limited the credibility of prior agent-on-research benchmarks. Evaluating ten frontier agent configurations under a strict web-search-disabled protocol, we find that the strongest model surpasses SOTA on only 17.8% of tasks under the g>0.1 criterion. Analysis of method pathways reveals that agents succeed primarily through methodological translation, converting scientific tasks into familiar supervised prediction problems, rather than through genuine scientific invention. Failures are dominated by wrong method choice and insufficient compute budget, not by task misunderstanding. We release the benchmark, the NatureGym pipeline, and a public leaderboard with maintainer-side reproduction. Code: https://github.com/FrontisAI/NatureBench
arXiv:2606.24633v2 Announce Type: replace Abstract: Human demonstrations for robot imitation learning often contain mistakes and corrective behaviors, such as imprecise grasps, object misalignment, unstable contact, and repeated attempts. While these segments are commonly treated as noisy or suboptimal data, they provide valuable evidence about when execution deviates from a desirable path and how task feasibility can be restored. However, existing reward and value models often rely on monotonic progress assumptions, which capture coarse task advancement but may overlook local execution errors and corrective behaviors in imperfect demonstrations. In this work, we propose ReTVL (ReTry-Supervised Value Learning), a framework for learning mistake-sensitive value functions from mixed-quality robot demonstrations by leveraging retry events as sparse supervision. ReTVL captures the local degradation-and-recovery structure around mistakes by combining global progress calibration with local pairwise preference learning induced by sparsely annotated retry keypoints. The learned value model is then used to reweight demonstration chunks for downstream behavior cloning, reducing the influence of harmful execution errors while preserving useful corrective behaviors. Experiments on real-robot manipulation tasks show that ReTVL produces more fine-grained value estimates than progress-based baselines and improves imitation learning from imperfect demonstrations.
arXiv:2606.24722v2 Announce Type: replace Abstract: Frontier AI training is increasingly shaped by access to dense, centrally controlled accelerator clusters. This creates a structural advantage for hyperscalers and large centralized laboratories, and makes open or independent AI efforts depend on scarce capital, privileged infrastructure, and data-center geography. We present Spheroid BlockTrain, a decentralized training protocol in which a model is partitioned into independently trainable blocks, each optimized on a local objective derived from the same global target and composed at inference into one model. On byte-level WikiText, BlockTrain reaches cross entropy 1.359 (perplexity 3.89), within about 0.04 CE of a same-setup end-to-end Transformer reference, while each active worker trains only one block and avoids full-model optimizer state. A shared six-worker block training run reaches CE 1.385 by averaging same-block updates into one assembled model. HTTP/TCP transport experiments move real serialized checkpoints and updates, including a public-IP three-host run that improves CE from 5.580 to 1.811 while moving 15.22 GB. For inference, the current BlockTrain path uses one block-stack traversal per full output and serves over direct TCP across three public-network GPU hosts up to a 75.80B-parameter logical fp16 shape, outperforming a matched plain-autoregressive TCP pipeline baseline because it emits a full sequence per WAN pipeline traversal rather than one token per traversal.
arXiv:2606.24780v2 Announce Type: replace Abstract: Progress in deep learning is, at scale, more a matter of systems engineering than of modelling: the behaviour of a model in training (its throughput, its memory footprint, and the numerical fidelity of the result) is determined less by the architecture itself than by how that architecture is expressed on the hardware. To achieve absolute control over this hardware expression while abstracting away systems complexity to make modelling seamless and eliminating the need for repetitive orchestration logic, BluTrain was architected from first principles as a robust, lightweight, and architecture-general training framework in standard C++ and the core CUDA programming model. Every layer is implemented natively: a typed tensor module with reverse-mode autograd, a linear-algebra library, a caching allocator, a multi-mode distributed-execution module, and an MLIR-based deep-learning compiler. In formal evaluations training a 124M-parameter GPT-2 baseline in FP32 on an 8-GPU 6000 Ada system, BluTrain outperforms industry-standard baselines in both throughput (sustaining an average of 407K tokens/s versus PyTorch's 395K tokens/s) and memory efficiency (achieving up to a 22% footprint reduction), while strictly preserving numerical fidelity and converging to a marginally lower final validation loss. With every layer explicitly open to native tuning, the performance ceiling is the framework's own to raise.
arXiv:2607.05164v1 Announce Type: new Abstract: The Traveling Thief Problem (TTP) combines the Traveling Salesperson Problem with the Knapsack Problem. In this problem, a finite metric space is given, and at each location an item with some profit and weight is placed. An agent seeks to collect a subset of the items. To do so, the agent must decide which items to collect and to determine a cyclic tour visiting the corresponding locations. While collecting an item yields its profit as a reward, the agent's speed decreases as more weight is picked up. The problem involves two competing objectives: maximizing the total profit of the collected items and minimizing the travel time of the tour. While many heuristics and exact algorithms (with a non-polynomial running time) have been developed, no approximation algorithms are known for any variant of the TTP. We aim at computing an $(\alpha_1,\alpha_2)$-approximate Pareto set that, for every solution, contains another solution collecting at least a $\frac{1}{\alpha_1}$ fraction of its profit while requiring at most $\alpha_2$ times its travel time. Our main result is an algorithm that calculates a $(9 + \epsilon,9 + \epsilon)$-approximate Pareto set in polynomial time. We also consider the setting in which the set of items to be collected is given in advance, so that the agent only has to compute a tour through the corresponding locations that minimizes the total travel time. This is the so-called Weighted TSP. For this setting, we present a $(2e + \epsilon)$-approximation algorithm.