arXiv subjects

Jing Bai

Publications and source records attributed to Jing Bai.

At least 19 recordsLinked to original sources

arXiv2026

MakoXC: Rearchitecting DFT Exchange-Correlation with Matrix-Aligned and Knowledge-Organized Sparsity

Density Functional Theory (DFT) is indispensable for materials science and drug discovery, yet the exchange--correlation (XC) evaluation remains a major bottleneck due to its cubic scaling. Although linear-scaling methods exploit electronic nearsightedness to reduce asymptotic complexity, they produce irregular sparse workloads that hide implicit sparsity and prevent efficient use of modern AI accelerators. We present MakoXC, a modular matrix-aligned XC evaluation engine that rearchitects nearsightedness-induced sparsity into regular, accelerator-friendly computations. MakoXC co-designs three key techniques: (1) Matrix-Aligned Cells reorganize nearsightedness-induced interactions into dense, accelerator-aligned data clusters; (2) Sparsity-Guided Activation translates deeper implicit sparsity into numerically correct structured execution for practical linear scaling; and (3) Kernel-Fused Pipeline consolidates fragmented workloads into a unified, compute-intensive execution path that fully unleashes accelerator throughput. Extensive evaluations show that MakoXC achieves average speedups of 67.8$\times$ speedup over standard XC evaluation and 4.7$\times$ over state-of-the-art linear-scaling methods. When integrated into a production-grade commercial DFT package, MakoXC scales XC evaluation to ubiquitin (1,231 atoms, def2-SVP) on 64 GPUs, enabling the end-to-end DFT calculation to complete in under five minutes. By restructuring XC evaluation into a unified, structured computation, MakoXC demonstrates how scientific workloads can achieve genuine low complexity while maximizing parallel efficiency on AI accelerators.

cs.DC↗

arXiv2026

LatentRevise: Learning from Zero-Hit Reasoning

Reinforcement learning with verifiable rewards (RLVR) is bottlenecked by hard prompts on which correct trajectories have low probability, so sampling misses them within a practical budget and leaves the policy update with little useful signal. We frame such zero-hit prompts as RLVR's sampling frontier, where new reasoning behavior is most valuable yet least likely to be sampled. Importantly, failed rollouts can be informative: they expose where the model's reasoning went wrong. We introduce LatentRevise, a first-order latent revision method that recovers training signal for this zero-hit regime. Given a failed rollout and the gold answer as an anchor, LatentRevise optimizes the input embeddings of its reasoning prefix under two complementary gradients, moving the prefix away from the failed continuation and toward the gold answer. The optimization is constrained to the convex hull of the model's vocabulary embeddings, so each update moves the latent toward a real token embedding rather than an arbitrary feature direction. We find that continuations from the revised prefix lengthen, exhibit self-reflection, and reach correct answers missed by the original rollouts. Used as training data, these trajectories improve SFT and RLVR on math benchmarks over standard baselines.

cs.CL↗

arXiv2026

Availability-Aware and Efficiency-Driven AI Service Chain Provisioning in Multi-Domain Edge Intelligence Cloud

In a multi-domain edge intelligence cloud (MDEIC) managed by multiple network operators, AI services are delivered by chains of virtual network functions (VNFs) executed in sequence, called AI service chains (AISCs). Therefore, achieving an efficient and economical AISC provisioning approach is essential. However, the interaction between the environmental characteristics (heterogeneity, resource constraints and limited information visibility) of MDEIC and the time-dependence of AISCs, introduces various challenges to AISC provisioning in MDEIC. In this paper, we first formulate the AISC provisioning problem as a partially observable stochastic game (POSG). Then, we propose a graph-and-time-based multi-agent AISC provisioning (GT-MAAISCP) approach to achieve the collaborative optimization of AISC provisioning cost, delay and availability. Specifically, each agent uses the graph-time dueling network (GTDN) architecture to extract network topology information and temporal relationships. Finally, the experimental results demonstrate that the proposed approach outperforms benchmark approaches in MDEIC and also illustrate its performance under varying network topologies and different numbers of local EICs (LEICs).

cs.NI↗

arXiv2026

AISC deployment in dynamic UAV-assisted MEC network: a reinforcement learning method based on heterogeneous graph attention neural network

Unmanned aerial vehicles-assisted mobile edge computing (UMEC) can execute compute-intensive and latency-critical artificial intelligence (AI) services, which can be provided by multiple UAVs collaborating in the air to perform inference tasks. Completing an AI service requires multiple inferences, each of which is implemented by an AI service chain consisting of multiple virtual network functions (VNFs). The application of AISC relies on an efficient AISC deployment strategy to determine which UAV to deploy VNF on. However, the UMEC network topology is highly dynamic due to the high-speed movement of UAVs or their departure/arrival, which makes the AISC deployment in the UMEC network challenging. In addition, the intricate relationships between UMEC environment and AISC, as well as between individual VNFs in an AISC, can also affect the effectiveness of AISC deployment strategy. Moreover, under the constraints of energy consumption and load balancing, it is also difficult to optimize the AISC strategy to minimize AISC completion time for enhancing the quality of AI service. To address the above challenges, this paper proposes a double deep attention Q-network based on heterogeneous graph neural networks, which incorporates heterogeneous graph to capture diverse relationships in UMEC and utilizes attention mechanisms to adaptively focus on critical nodes and links for intelligent AISC deployment. The experimental results demonstrate that the proposed algorithm performs excellently in AISC completion time, AISC completion rate, load balancing and energy consumption.

cs.NI↗

arXiv2026

Step 3.5 Flash: Open Frontier-Level Intelligence with 11B Active Parameters

We introduce Step 3.5 Flash, a sparse Mixture-of-Experts (MoE) model that bridges frontier-level agentic intelligence and computational efficiency. We focus on what matters most when building agents: sharp reasoning and fast, reliable execution. Step 3.5 Flash pairs a 196B-parameter foundation with 11B active parameters for efficient inference. It is optimized with interleaved 3:1 sliding-window/full attention and Multi-Token Prediction (MTP-3) to reduce the latency and cost of multi-round agentic interactions. To reach frontier-level intelligence, we design a scalable reinforcement learning framework that combines verifiable signals with preference feedback, while remaining stable under large-scale off-policy training, enabling consistent self-improvement across mathematics, code, and tool use. Step 3.5 Flash demonstrates strong performance across agent, coding, and math tasks, achieving 85.4% on IMO-AnswerBench, 86.4% on LiveCodeBench-v6 (2024.08-2025.05), 88.2% on tau2-Bench, 69.0% on BrowseComp (with context management), and 51.0% on Terminal-Bench 2.0, comparable to frontier models such as GPT-5.2 xHigh and Gemini 3.0 Pro. By redefining the efficiency frontier, Step 3.5 Flash provides a high-density foundation for deploying sophisticated agents in real-world industrial environments.

cs.CL↗

arXiv2025

Step-DeepResearch Technical Report

As LLMs shift toward autonomous agents, Deep Research has emerged as a pivotal metric. However, existing academic benchmarks like BrowseComp often fail to meet real-world demands for open-ended research, which requires robust skills in intent recognition, long-horizon decision-making, and cross-source verification. To address this, we introduce Step-DeepResearch, a cost-effective, end-to-end agent. We propose a Data Synthesis Strategy Based on Atomic Capabilities to reinforce planning and report writing, combined with a progressive training path from agentic mid-training to SFT and RL. Enhanced by a Checklist-style Judger, this approach significantly improves robustness. Furthermore, to bridge the evaluation gap in the Chinese domain, we establish ADR-Bench for realistic deep research scenarios. Experimental results show that Step-DeepResearch (32B) scores 61.4% on Scale AI Research Rubrics. On ADR-Bench, it significantly outperforms comparable models and rivals SOTA closed-source models like OpenAI and Gemini DeepResearch. These findings prove that refined training enables medium-sized models to achieve expert-level capabilities at industry-leading cost-efficiency.

cs.CL↗

arXiv2025

EEPO: Exploration-Enhanced Policy Optimization via Sample-Then-Forget

Balancing exploration and exploitation remains a central challenge in reinforcement learning with verifiable rewards (RLVR) for large language models (LLMs). Current RLVR methods often overemphasize exploitation, leading to entropy collapse, diminished exploratory capacity, and ultimately limited performance gains. Although techniques that increase policy stochasticity can promote exploration, they frequently fail to escape dominant behavioral modes. This creates a self-reinforcing loop -- repeatedly sampling and rewarding dominant modes -- that further erodes exploration. We introduce Exploration-Enhanced Policy Optimization (EEPO), a framework that promotes exploration via two-stage rollouts with adaptive unlearning. In the first stage, the model generates half of the trajectories; it then undergoes a lightweight unlearning step to temporarily suppress these sampled responses, forcing the second stage to explore different regions of the output space. This sample-then-forget mechanism disrupts the self-reinforcing loop and promotes wider exploration during rollouts. Across five reasoning benchmarks, EEPO outperforms GRPO, achieving average relative gains of 24.3% on Qwen2.5-3B, 33.0% on Llama3.2-3B-Instruct, and 10.4% on Qwen3-8B-Base.

cs.CL↗

arXiv2025

Beyond Two-Stage Training: Cooperative SFT and RL for LLM Reasoning

Supervised fine-tuning (SFT) and reinforcement learning with verifiable rewards (RLVR) are two widely used post-training paradigms for improving the reasoning ability of large language models (LLMs). Recent methods attempt to integrate SFT and RLVR in a single stage by reweighting or scheduling their objectives. However, such coupling can be counterproductive because supervised updates are not uniformly beneficial for reward optimization. To address this, we propose BRIDGE, a scalable framework in which SFT learns to supervise RL by selectively transferring knowledge that improves reward optimization. Specifically, BRIDGE alternates two updates at each meta-training step: a base-model update that fuses the SFT and RL gradients, and an update to a lightweight low-rank adapter (LoRA) that coordinates the two objectives by maximizing a cooperative-gain signal, defined as the reward of joint SFT-RL training over an RL-only baseline. Across five mathematical reasoning benchmarks, BRIDGE consistently outperforms two-stage cold start, naive mixing, and representative single-stage integration baselines, yielding over three points average absolute improvement and more stable training dynamics. We further show that BRIDGE extends to logical reasoning and generalizes out-of-distribution to code and science without additional training, while staying robust under noisy rewards.

cs.CL↗

arXiv2025

Probing nontrivial fusion of Majorana zero modes via near-adiabatic coupling

We propose and simulate a near-adiabatically coupling probing scheme for nontrivial fusion of a pair of Majorara zero modes (MZMs). The scheme can avoid the complexity of oscillating charge occupation in the probing quantum dot, making thus practical measurements more feasible. We also show how to extract the information of nonadiabatic transition and fermion parity violation caused during moving the MZMs together to fuse, from the initial states prepared with definite fermion parity. All the simulations, including the effective coupling between the fusing MZMs, and their coupling to the probing quantum dot, are based on the lattice model of a Rashba quantum wire in proximity contact with an s-wave superconductor, under the modulation of mini-gate voltage control.

cond-mat.mes-hall↗

arXiv2025

ReVeal: Self-Evolving Code Agents via Reliable Self-Verification

Reinforcement learning with verifiable rewards (RLVR) has advanced the reasoning capabilities of large language models. However, existing methods rely solely on outcome rewards, without explicitly optimizing verification or leveraging reliable signals from realistic environments, leading to unreliable self-verification and limited test-time scaling. To address this, we widen the verification-generation asymmetry by explicitly optimizing self-verification, making it a reliable driver of deeper test-time scaling. We introduce ReVeal, a multi-turn reinforcement learning framework that evolves code generation through self-verification and tool-based evaluation. ReVeal structures long-horizon reasoning as iterative generation-verification turns and incorporates TAPO for turn-level credit assignment, fostering the co-evolution of code and test generation. At inference, this strengthened self-verification enables the model to use self-constructed tests and tool feedback to continuously evolve code for 20+ turns on LiveCodeBench despite training on only three. It also significantly improves Pass@k, indicating stronger exploration that expands the reasoning boundaries of the base model. These findings highlight the promise of ReVeal as a scalable paradigm for RL training and test-time scaling, paving the way for more robust and autonomous AI agents.

cs.SE↗

arXiv2025

Vulnerability-Aware Alignment: Mitigating Uneven Forgetting in Harmful Fine-Tuning

Harmful fine-tuning (HFT), performed directly on open-source LLMs or through Fine-tuning-as-a-Service, breaks safety alignment and poses significant threats. Existing methods aim to mitigate HFT risks by learning robust representation on alignment data or making harmful data unlearnable, but they treat each data sample equally, leaving data vulnerability patterns understudied. In this work, we reveal that certain subsets of alignment data are consistently more prone to forgetting during HFT across different fine-tuning tasks. Inspired by these findings, we propose Vulnerability-Aware Alignment (VAA), which estimates data vulnerability, partitions data into "vulnerable" and "invulnerable" groups, and encourages balanced learning using a group distributionally robust optimization (Group DRO) framework. Specifically, VAA learns an adversarial sampler that samples examples from the currently underperforming group and then applies group-dependent adversarial perturbations to the data during training, aiming to encourage a balanced learning process across groups. Experiments across four fine-tuning tasks demonstrate that VAA significantly reduces harmful scores while preserving downstream task performance, outperforming state-of-the-art baselines.

cs.LG↗

arXiv2025

Heliostat Optical Error Inspection with Polarimetric Imaging Drone

On a Concentrated Solar Power (CSP) field, optical errors have significant impacts on the collection efficiency of heliostats. Fast, cost-effective, labor-efficient, and non-intrusive autonomous field inspection remains a challenge. Approaches using imaging drone, i.e., Unmanned Aerial Vehicle (UAV) system integrated with high resolution visible imaging sensors, have been developed to address these challenges; however, these approaches are often limited by insufficient imaging contrast. Here we report a polarimetry-based method with a polarization imaging system integrated on UAV to enhance imaging contrast for in-situ detection of heliostat mirrors without interrupting field operation. We developed an optical model for skylight polarization pattern to simulate the polarization images of heliostat mirrors and obtained optimized waypoints for polarimetric imaging drone flight path to capture images with enhanced contrast. The polarimetric imaging-based method improved the success rate of edge detections in scenarios which were challenging for mirror edge detection with conventional imaging sensors. We have performed field tests to achieve significantly enhanced heliostat edge detection success rate and investigate the feasibility of integrating polarimetric imaging method with existing imaging-based heliostat inspection methods, i.e., Polarimetric Imaging Heliostat Inspection Method (PIHIM). Our preliminary field test results suggest that the PIHIM hold the promise to enable sufficient imaging contrast for real-time autonomous imaging and detection of heliostat field, thus suitable for non-interruptive fast CSP field inspection during its operation.

physics.app-ph↗

arXiv2025

ADNP-15: An Open-Source Histopathological Dataset for Neuritic Plaque Segmentation in Human Brain Whole Slide Images with Frequency Domain Image Enhancement for Stain Normalization

Alzheimer's Disease (AD) is a neurodegenerative disorder characterized by amyloid-beta plaques and tau neurofibrillary tangles, which serve as key histopathological features. The identification and segmentation of these lesions are crucial for understanding AD progression but remain challenging due to the lack of large-scale annotated datasets and the impact of staining variations on automated image analysis. Deep learning has emerged as a powerful tool for pathology image segmentation; however, model performance is significantly influenced by variations in staining characteristics, necessitating effective stain normalization and enhancement techniques. In this study, we address these challenges by introducing an open-source dataset (ADNP-15) of neuritic plaques (i.e., amyloid deposits combined with a crown of dystrophic tau-positive neurites) in human brain whole slide images. We establish a comprehensive benchmark by evaluating five widely adopted deep learning models across four stain normalization techniques, providing deeper insights into their influence on neuritic plaque segmentation. Additionally, we propose a novel image enhancement method that improves segmentation accuracy, particularly in complex tissue structures, by enhancing structural details and mitigating staining inconsistencies. Our experimental results demonstrate that this enhancement strategy significantly boosts model generalization and segmentation accuracy. All datasets and code are open-source, ensuring transparency and reproducibility while enabling further advancements in the field.

eess.IV↗

arXiv2025

Map Feature Perception Metric for Map Generation Quality Assessment and Loss Optimization

In intelligent cartographic generation tasks empowered by generative models, the authenticity of synthesized maps constitutes a critical determinant. Concurrently, the selection of appropriate evaluation metrics to quantify map authenticity emerges as a pivotal research challenge. Current methodologies predominantly adopt computer vision-based image assessment metrics to compute discrepancies between generated and reference maps. However, conventional visual similarity metrics-including L1, L2, SSIM, and FID-primarily operate at pixel-level comparisons, inadequately capturing cartographic global features and spatial correlations, consequently inducing semantic-structural artifacts in generated outputs. This study introduces a novel Map Feature Perception Metric designed to evaluate global characteristics and spatial congruence between synthesized and target maps. Diverging from pixel-wise metrics, our approach extracts elemental-level deep features that comprehensively encode cartographic structural integrity and topological relationships. Experimental validation demonstrates MFP's superior capability in evaluating cartographic semantic features, with classification-enhanced implementations outperforming conventional loss functions across diverse generative frameworks. When employed as optimization objectives, our metric achieves performance gains ranging from 2% to 50% across multiple benchmarks compared to traditional L1, L2, and SSIM baselines. This investigation concludes that explicit consideration of cartographic global attributes and spatial coherence substantially enhances generative model optimization, thereby significantly improving the geographical plausibility of synthesized maps.

cs.CV↗

arXiv2025

We Need to Effectively Integrate Computing Skills Across Discipline Curricula

Computing is increasingly central to innovation across a wide range of disciplinary and interdisciplinary problem domains. Students across noncomputing disciplines need to apply sophisticated computational skills and methods to fields as diverse as biology, linguistics, and art. Furthermore, computing plays a critical role in "momentous geopolitical events", such as elections in several countries including the US, and is changing how people "work, collaborate, communicate, shop, eat, travel, get news and entertainment, and quite simply live". Traditional computing courses, however, fail to equip non-computing discipline students with the necessary computing skills - if they can even get into classes packed with CS majors. A pressing question facing academics today is: How do we effectively integrate computing skills that are useful for the discipline into discipline curricula? We advocate an approach where courses in discipline X include the computing relevant to the learning outcomes of that course, as used by practitioners in X. We refer to the computing skills relevant to a course in discipline X as an "ounce of computing skills", to highlight our belief regarding the amount of computing to be integrated in that course. In this article, we outline our insights regarding the development of an ounce of computing skills for a discipline course, and the evaluation of the developed ounce. The key takeaways are that the goal has to be to advance students in their disciplines, and only the disciplinary experts can tell us how computing is used in that discipline. Computer scientists know how to teach computing, but the classes can't be about CS values. The disciplinary values are paramount.

cs.CY↗

arXiv2025

Bridging Scales in Map Generation: A scale-aware cascaded generative mapping framework for seamless and consistent multi-scale cartographic representation

Multi-scale tile maps are essential for geographic information services, serving as fundamental outcomes of surveying and cartographic workflows. While existing image generation networks can produce map-like outputs from remote sensing imagery, their emphasis on replicating texture rather than preserving geospatial features limits cartographic validity. Current approaches face two fundamental challenges: inadequate integration of cartographic generalization principles with dynamic multi-scale generation and spatial discontinuities arising from tile-wise generation. To address these limitations, we propose a scale-aware cartographic generation framework (SCGM) that leverages conditional guided diffusion and a multi-scale cascade architecture. The framework introduces three key innovations: a scale modality encoding mechanism to formalize map generalization relationships, a scale-driven conditional encoder for robust feature fusion, and a cascade reference mechanism ensuring cross-scale visual consistency. By hierarchically constraining large-scale map synthesis with small-scale structural priors, SCGM effectively mitigates edge artifacts while maintaining geographic fidelity. Comprehensive evaluations on cartographic benchmarks confirm the framework's ability to generate seamless multi-scale tile maps with enhanced spatial coherence and generalization-aware representation, demonstrating significant potential for emergency mapping and automated cartography applications.

eess.IV↗

arXiv2024

BSM: Small but Powerful Biological Sequence Model for Genes and Proteins

Modeling biological sequences such as DNA, RNA, and proteins is crucial for understanding complex processes like gene regulation and protein synthesis. However, most current models either focus on a single type or treat multiple types of data separately, limiting their ability to capture cross-modal relationships. We propose that by learning the relationships between these modalities, the model can enhance its understanding of each type. To address this, we introduce BSM, a small but powerful mixed-modal biological sequence foundation model, trained on three types of data: RefSeq, Gene Related Sequences, and interleaved biological sequences from the web. These datasets capture the genetic flow, gene-protein relationships, and the natural co-occurrence of diverse biological data, respectively. By training on mixed-modal data, BSM significantly enhances learning efficiency and cross-modal representation, outperforming models trained solely on unimodal data. With only 110M parameters, BSM achieves performance comparable to much larger models across both single-modal and mixed-modal tasks, and uniquely demonstrates in-context learning capability for mixed-modal tasks, which is absent in existing models. Further scaling to 270M parameters demonstrates even greater performance gains, highlighting the potential of BSM as a significant advancement in multimodal biological sequence modeling.

q-bio.GN↗

arXiv2024

CPL: Critical Plan Step Learning Boosts LLM Generalization in Reasoning Tasks

Post-training, particularly reinforcement learning (RL) using self-play-generated data, has become a new learning paradigm for large language models (LLMs). However, scaling RL to develop a general reasoner remains a research challenge, as existing methods focus on task-specific reasoning without adequately addressing generalization across a broader range of tasks. Moreover, unlike traditional RL with limited action space, LLMs operate in an infinite space, making it crucial to search for valuable and diverse strategies to solve problems effectively. To address this, we propose searching within the action space on high-level abstract plans to enhance model generalization and introduce Critical Plan Step Learning (CPL), comprising: 1) searching on plan, using Monte Carlo Tree Search (MCTS) to explore diverse plan steps in multi-step reasoning tasks, and 2) learning critical plan steps through Step-level Advantage Preference Optimization (Step-APO), which integrates advantage estimates for step preference obtained via MCTS into Direct Preference Optimization (DPO). This combination helps the model effectively learn critical plan steps, enhancing both reasoning capabilities and generalization. Experimental results demonstrate that our method, trained exclusively on GSM8K and MATH, not only significantly improves performance on GSM8K (+10.5%) and MATH (+6.5%), but also enhances out-of-domain reasoning benchmarks, such as HumanEval (+12.2%), GPQA (+8.6%), ARC-C (+4.0%), MMLU-STEM (+2.2%), and BBH (+1.8%).

cs.AI↗