Searcharxiv⌕ Search

arXiv subjects

Sheng Bi

Publications and source records attributed to Sheng Bi.

At least 19 recordsLinked to original sources

ARMOR: Adaptive Curriculum Meta-Learning for Noise-Robust RAG Reasoning

Retrieval-Augmented Generation (RAG) systems have demonstrated remarkable effectiveness in mitigating hallucinations by incorporating external knowledge. However, the retrieval process inevitably introduces noise, posing significant challenges to RAG robustness. Fundamentally, noise robustness is a composite competency encompassing noise identification, retrieval filtering, and robust reasoning. Unfortunately, existing methods often fail to capture this holistic nature, typically restricted to single-dimensional enhancements or hampered by noise-induced optimization interference. To address these limitations, we decompose RAG robustness into three sub-tasks: knowledge boundary perception, retrieval verification, and robust reasoning. Accordingly, we propose ARMOR, an Adaptive Curriculum Meta-Learning for Noise Robust RAG Reasoning. Inspired by the teaching process, ARMOR introduces an adaptive task scheduler as a teacher to curate training plans according to the absolute learning progress of RAG generators (Student). Distinct from GRPO, ARMOR utilizes inter-group advantage estimation during learning to effectively mitigate mode collapse caused by limited output spaces. Experiments on three complex reasoning datasets (2WikiMultiHopQA, MusiQue, and HotPotQA) indicate that ARMOR achieves synergistic improvements in retrieval noise robustness, significantly outperforming baselines in output quality.

cs.AI↗

Full-data accuracy with fewer labels for training and fine-tuning machine-learning force fields

Machine-learning force fields (MLFFs) are reliable only near their training distribution, making efficient construction of diverse training sets a major bottleneck for both train-from-scratch and foundation fine-tuning workflows. Active learning can reduce this cost, but standard model-committee uncertainty is impractical for foundation MLFFs because each committee member requires a separate fine-tuning run. We present an active-learning workflow based on last-layer-projection regression (LLPR), a forward-pass-cheap per-configuration uncertainty estimator. Across molecular, condensed-phase, and electrolyte systems, LLPR identifies compact, high-value training sets that recover full-data accuracy using only a small fraction of electronic-structure labels. In foundation-model fine-tuning, LLPR-selected configurations approach the practical fine-tuning ceiling with substantially fewer labels than random selection. In iterative electrolyte fine-tuning, LLPR detects unphysical local coordination before new labelling, calibrates uncertainty in force-error units, and supports retrospective convergence monitoring. The resulting models reproduce reference density and Li--O coordination structure, providing a scalable uncertainty-quantification strategy across MLFF training regimes.

physics.chem-ph↗

Kernel-Level Constant-Potential Molecular Dynamics in GROMACS

How charge rearranges within an electrode shapes the behavior of batteries, supercapacitors, and other electrochemical devices. Molecular simulations can capture this response by allowing the charge on each electrode atom to adjust to its surroundings, but repeatedly calculating these charges adds substantial computational cost. We present an open-source implementation of the constant-potential method in GROMACS that integrates these calculations into the program's optimized CPU and GPU routines. Comparisons with established codes validate the electrode charges, while charging, voltammetry, and impedance simulations demonstrate the response to changing electrical conditions. The implementation also supports electrodes that move or deform and models of finite metallic screening and electron spillover. In a pair of charged carbon nanotubes, allowing charge to move away from the facing surfaces reduces the repulsive force at contact by about half at an intermediate charge and keeps the pair bound where a uniform-charge model opens within the simulated time. These capabilities make the electrical response of electrodes accessible within GROMACS for studies of interfacial structure, charging dynamics, and electrode motion.

physics.chem-ph↗

Ai2-Kit: Streamlining AI-Accelerated Ab Initio Workflows for Complex Chemical Systems

Molecular simulations of complex chemical systems, such as catalysis, electrochemistry, and energy storage, often need to capture the interplay of effects such as electronic structure, finite-temperature fluctuations, and electric-field response. Such complexity is difficult to address with traditional ab initio calculations, which are limited by the time and length scales they can reach. AI-accelerated ab initio (AI2) methods use machine learning potentials trained on first-principles data to replace expensive electronic-structure calculations, extending ab initio accuracy to these regimes, but their routine application requires reliable workflows that connect first-principles calculations, model training, molecular dynamics, enhanced sampling, trajectory analysis, and HPC orchestration. Here we present ai2-kit, a software toolkit for developing accessible, reproducible, and extensible AI2 workflows. ai2-kit provides high-semantic-density command-line interfaces and Python APIs for structure and dataset conversion, batch task generation, active-learning screening, job orchestration, and workflow recovery. We demonstrate ai2-kit in four representative applications: active-learning-based machine learning potential construction, free-energy perturbation for redox and acid-base processes, electrochemical machine learning potentials for electrified interfaces, and spectroscopies from machine learning molecular dynamics. ai2-kit also provides AI-agent skills that help users adapt these use cases into customized workflows for their own chemical systems and computational software stacks. Together, ai2-kit helps turn AI2 methods from bespoke computational protocols into reusable and extensible workflows for complex chemical systems, from model construction to property prediction.

physics.chem-ph↗

Geometric obstruction to resolving the Hubble tension: orthogonality of scale and shape in distance measurements

We identify a geometric obstruction to resolving the Hubble tension by combining early-time sound-horizon reduction with late-time smooth dark energy. Within $Λ$CDM, the BAO--SN matter-density gap $ΔΩ_m = 0.037$ is exactly invariant under the sound-horizon rescaling $α\equiv r_s^{\rm mod}/r_s^{Λ{\rm CDM}}$, and late-time $w(z)$ deformations cannot eliminate this gap either: reconciling the two datasets requires \emph{opposite} deformations -- phantom ($w < -1$) for BAO, quintessence ($w > -1$) for SN at $z < 0.5$ -- an anti-alignment quantified by $\cosθ= -0.97$ in $w(z)$ space. A full MCMC analysis of DESI DR2 BAO, Planck plik\_lite, and Pantheon+ bears this out: the optimal $α^* = 0.992$ ($0.8\%$ $r_s$ reduction) brings the joint fit to $H_0 = 70.3 \pm 0.3\;\mathrm{km\,s^{-1}\,Mpc^{-1}}$, still $3.2σ$ below SH0ES, with the inter-dataset tension reduced but not removed. The obstruction reflects not a shortage of model freedom but an irreducible disagreement between probes. The deformation space $\{α, β_{\rm damp}, w(z)\}$ already spans $93\%$ of the $Ω_m$ response direction; nonetheless BAO and SN constrain $Ω_m$ through independent channels and disagree, while the residual $H_0$ deficit, anchored by the local distance ladder, resides in the absolute distance scale that $w(z)$ reshapes but cannot rescale.

astro-ph.CO↗

LPM 1.0: Video-based Character Performance Model

Performance, the externalization of intent, emotion, and personality through visual, vocal, and temporal behavior, is what makes a character alive. Learning such performance from video is a promising alternative to traditional 3D pipelines. However, existing video models struggle to jointly achieve high expressiveness, real-time inference, and long-horizon identity stability, a tension we call the performance trilemma. Conversation is the most comprehensive performance scenario, as characters simultaneously speak, listen, react, and emote while maintaining identity over time. To address this, we present LPM 1.0 (Large Performance Model), focusing on single-person full-duplex audio-visual conversational performance. Concretely, we build a multimodal human-centric dataset through strict filtering, speaking-listening audio-video pairing, performance understanding, and identity-aware multi-reference extraction; train a 17B-parameter Diffusion Transformer (Base LPM) for highly controllable, identity-consistent performance through multimodal conditioning; and distill it into a causal streaming generator (Online LPM) for low-latency, infinite-length interaction. At inference, given a character image with identity-aware references, LPM 1.0 generates listening videos from user audio and speaking videos from synthesized audio, with text prompts for motion control, all at real-time speed with identity-stable, infinite-length generation. LPM 1.0 thus serves as a visual engine for conversational agents, live streaming characters, and game NPCs. To systematically evaluate this setting, we propose LPM-Bench, the first benchmark for interactive character performance. LPM 1.0 achieves state-of-the-art results across all evaluated dimensions while maintaining real-time inference.

cs.CV↗

AD-CARE: A Guideline-grounded, Modality-agnostic LLM Agent for Real-world Alzheimer's Disease Diagnosis with Multi-cohort Assessment, Fairness Analysis, and Reader Study

Alzheimer's disease (AD) is a growing global health challenge as populations age, and timely, accurate diagnosis is essential to reduce individual and societal burden. However, real-world AD assessment is hampered by incomplete, heterogeneous multimodal data and variability across sites and patient demographics. Although large language models (LLMs) have shown promise in biomedicine, their use in AD has largely been confined to answering narrow, disease-specific questions rather than generating comprehensive diagnostic reports that support clinical decision-making. Here we expand LLM capabilities for clinical decision support by introducing AD-CARE, a modality-agnostic agent that performs guideline-grounded diagnostic assessment from incomplete, heterogeneous inputs without imputing missing modalities. By dynamically orchestrating specialized diagnostic tools and embedding clinical guidelines into LLM-driven reasoning, AD-CARE generates transparent, report-style outputs aligned with real-world clinical workflows. Across six cohorts comprising 10,303 cases, AD-CARE achieved 84.9% diagnostic accuracy, delivering 4.2%-13.7% relative improvements over baseline methods. Despite cohort-level differences, dataset-specific accuracies remain robust (80.4%-98.8%), and the agent consistently outperforms all baselines. AD-CARE reduced performance disparities across racial and age subgroups, decreasing the average dispersion of four metrics by 21%-68% and 28%-51%, respectively. In a controlled reader study, the agent improved neurologist and radiologist accuracy by 6%-11% and more than halved decision time. The framework yielded 2.29%-10.66% absolute gains over eight backbone LLMs and converges their performance. These results show that AD-CARE is a scalable, practically deployable framework that can be integrated into routine clinical workflows for multimodal decision support in AD.

cs.MA↗

Sequential Decision-Making under Uncertainty: A Robust MDPs review

Fueled by advances in both robust optimization theory and reinforcement learning (RL), robust Markov Decision Processes (RMDPs) have garnered increasing attention due to their powerful capability for sequential decision-making under uncertainty. In this paper, we provide a comprehensive overview of the theoretical foundations and recent developments in RMDPs, with a particular emphasis on ambiguity modeling. We examine the ``rectangular assumption", a key condition ensuring computational tractability in RMDPs but often resulting in overly conservative policies. Three widely used rectangular forms are summarized, and a novel proof is provided for the NP-hardness of non-rectangular RMDPs. We categorize RMDP formulation approaches into parametric, moment-based, and discrepancy-based models, analyzing the trade-offs associated with each representation. Beyond the traditional scope of RMDPs, we also explore recent efforts to relax rectangular assumptions and highlight emerging trends within the RMDP research community. These developments contribute to more practical and flexible modeling frameworks, complementing the classical RMDP results. Relaxing rectangular assumptions tailored to operations management is a promising area for future research, and there are also opportunities for further advances in developing fast algorithms and provably robust RL algorithms.

math.OC↗

Can LLMs Evaluate Complex Attribution in QA? Automatic Benchmarking using Knowledge Graphs

Attributed Question Answering (AQA) has attracted wide attention, but there are still several limitations in evaluating the attributions, including lacking fine-grained attribution categories, relying on manual annotations, and failing to compare attributions with only subtle differences. To bridge these gaps, we introduce Complex Attributed Question Answering (CAQA), a large-scale benchmark containing comprehensive attribution categories, automatically generated using Knowledge Graphs (KGs), and complex attribution scenarios. We have conducted extensive experiments to verify the effectiveness of CAQA, including the benchmarking of 25 automatic evaluators, their comparison with human evaluators, the testing of LLM evaluators fine-tuned by CAQA and so on. These experiments also lead to a series of important findings that can benefit the future research of AQA. All the codes and data are publicly accessible at https://github.com/HuuuNan/CAQA-Benchmark.

cs.CL↗

What Prevents Resolving the Hubble Tension through Late-Time Expansion Modifications?

We demonstrate that Type Ia supernovae (SNe Ia) observations impose the critical constraint for resolving the Hubble tension through late-time expansion modifications. Applying the Fisher-bias optimization framework to cosmic chronometers (CC), baryon acoustic oscillations (BAO) from DESI DR2, Planck CMB, and Pantheon+ data, we find that: (i) deformations in $H(z \lesssim 3)$ (via $w(z)$ reconstruction) can reconcile tensions between CC, Planck, DESI BAO, and SH0ES measurements while maintaining or improving fit quality ($Δχ^2 < 0$ relative to $Λ$CDM); (ii) In the neighborhood of Planck best-fit $Λ$CDM model, no cosmologically viable solutions targeting $H_0 \gtrsim 69$ satisfy SNe Ia constraints. MCMC validation confirms the maximum achievable $H_0 = 69.09\pm0.30$ ($χ^2_{\rm BF} \approx χ^2_{Λ\rm CDM}$) across all data combinations, indicating that the conflict between late-time $w(z)$ modifications and SNe Ia observations prevents complete resolution of the Hubble tension.

astro-ph.CO↗

From Superficial to Deep: Integrating External Knowledge for Follow-up Question Generation Using Knowledge Graph and LLM

In a conversational system, dynamically generating follow-up questions based on context can help users explore information and provide a better user experience. Humans are usually able to ask questions that involve some general life knowledge and demonstrate higher order cognitive skills. However, the questions generated by existing methods are often limited to shallow contextual questions that are uninspiring and have a large gap to the human level. In this paper, we propose a three-stage external knowledge-enhanced follow-up question generation method, which generates questions by identifying contextual topics, constructing a knowledge graph (KG) online, and finally combining these with a large language model to generate the final question. The model generates information-rich and exploratory follow-up questions by introducing external common sense knowledge and performing a knowledge fusion operation. Experiments show that compared to baseline models, our method generates questions that are more informative and closer to human questioning levels while maintaining contextual relevance.

cs.AI↗

OneEval: Benchmarking LLM Knowledge-intensive Reasoning over Diverse Knowledge Bases

Large Language Models (LLMs) have demonstrated substantial progress on reasoning tasks involving unstructured text, yet their capabilities significantly deteriorate when reasoning requires integrating structured external knowledge such as knowledge graphs, code snippets, or formal logic. This limitation is partly due to the absence of benchmarks capable of systematically evaluating LLM performance across diverse structured knowledge modalities. To address this gap, we introduce \textbf{\textsc{OneEval}}, a comprehensive benchmark explicitly designed to assess the knowledge-intensive reasoning capabilities of LLMs across four structured knowledge modalities, unstructured text, knowledge graphs, code, and formal logic, and five critical domains (general knowledge, government, science, law, and programming). \textsc{OneEval} comprises 4,019 carefully curated instances and includes a challenging subset, \textsc{OneEval}\textsubscript{Hard}, consisting of 1,285 particularly difficult cases. Through extensive evaluation of 18 state-of-the-art open-source and proprietary LLMs, we establish three core findings: a) \emph{persistent limitations in structured reasoning}, with even the strongest model achieving only 32.2\% accuracy on \textsc{OneEval}\textsubscript{Hard}; b) \emph{performance consistently declines as the structural complexity of the knowledge base increases}, with accuracy dropping sharply from 53\% (textual reasoning) to 25\% (formal logic); and c) \emph{diminishing returns from extended reasoning chains}, highlighting the critical need for models to adapt reasoning depth appropriately to task complexity. We release the \textsc{OneEval} datasets, evaluation scripts, and baseline results publicly, accompanied by a leaderboard to facilitate ongoing advancements in structured knowledge reasoning.

cs.CL↗

Magic Mushroom: A Customizable Benchmark for Fine-grained Analysis of Retrieval Noise Erosion in RAG Systems

Retrieval-Augmented Generation (RAG) systems enhance Large Language Models (LLMs) by incorporating external retrieved information, mitigating issues such as hallucination and outdated knowledge. However, RAG systems are highly sensitive to retrieval noise prevalent in real-world scenarios. Existing benchmarks fail to emulate the complex and heterogeneous noise distributions encountered in real-world retrieval environments, undermining reliable robustness assessment. In this paper, we define four categories of retrieval noise based on linguistic properties and noise characteristics, aiming to reflect the heterogeneity of noise in real-world scenarios. Building on this, we introduce Magic Mushroom, a benchmark for replicating "magic mushroom" noise: contexts that appear relevant on the surface but covertly mislead RAG systems. Magic Mushroom comprises 7,468 single-hop and 3,925 multi-hop question-answer pairs. More importantly, Magic Mushroom enables researchers to flexibly configure combinations of retrieval noise according to specific research objectives or application scenarios, allowing for highly controlled evaluation setups. We evaluate LLM generators of varying parameter scales and classic RAG denoising strategies under diverse noise distributions to investigate their performance dynamics during progressive noise encroachment. Our analysis reveals that both generators and denoising strategies have significant room for improvement and exhibit extreme sensitivity to noise distributions. Magic Mushroom emerges as a promising tool for evaluating and advancing noise-robust RAG systems, accelerating their widespread deployment in real-world applications. The Magic Mushroom benchmark is available at https://drive.google.com/file/d/1aP5kyPuk4L-L_uoI6T9UhxuTyt8oMqjT/view?usp=sharing.

cs.CL↗

Seaweed-7B: Cost-Effective Training of Video Generation Foundation Model

This technical report presents a cost-efficient strategy for training a video generation foundation model. We present a mid-sized research model with approximately 7 billion parameters (7B) called Seaweed-7B trained from scratch using 665,000 H100 GPU hours. Despite being trained with moderate computational resources, Seaweed-7B demonstrates highly competitive performance compared to contemporary video generation models of much larger size. Design choices are especially crucial in a resource-constrained setting. This technical report highlights the key design decisions that enhance the performance of the medium-sized diffusion model. Empirically, we make two observations: (1) Seaweed-7B achieves performance comparable to, or even surpasses, larger models trained on substantially greater GPU resources, and (2) our model, which exhibits strong generalization ability, can be effectively adapted across a wide range of downstream applications either by lightweight fine-tuning or continue training. See the project page at https://seaweed.video/

cs.CV↗

Single Image Unlearning: Efficient Machine Unlearning in Multimodal Large Language Models

Machine unlearning empowers individuals with the `right to be forgotten' by removing their private or sensitive information encoded in machine learning models. However, it remains uncertain whether MU can be effectively applied to Multimodal Large Language Models (MLLMs), particularly in scenarios of forgetting the leaked visual data of concepts. To overcome the challenge, we propose an efficient method, Single Image Unlearning (SIU), to unlearn the visual recognition of a concept by fine-tuning a single associated image for few steps. SIU consists of two key aspects: (i) Constructing Multifaceted fine-tuning data. We introduce four targets, based on which we construct fine-tuning data for the concepts to be forgotten; (ii) Jointly training loss. To synchronously forget the visual recognition of concepts and preserve the utility of MLLMs, we fine-tune MLLMs through a novel Dual Masked KL-divergence Loss combined with Cross Entropy loss. Alongside our method, we establish MMUBench, a new benchmark for MU in MLLMs and introduce a collection of metrics for its evaluation. Experimental results on MMUBench show that SIU completely surpasses the performance of existing methods. Furthermore, we surprisingly find that SIU can avoid invasive membership inference attacks and jailbreak attacks. To the best of our knowledge, we are the first to explore MU in MLLMs. We will release the code and benchmark in the near future.

cs.CV↗

HeGTa: Leveraging Heterogeneous Graph-enhanced Large Language Models for Few-shot Complex Table Understanding

Table understanding (TU) has achieved promising advancements, but it faces the challenges of the scarcity of manually labeled tables and the presence of complex table structures.To address these challenges, we propose HGT, a framework with a heterogeneous graph (HG)-enhanced large language model (LLM) to tackle few-shot TU tasks.It leverages the LLM by aligning the table semantics with the LLM's parametric knowledge through soft prompts and instruction turning and deals with complex tables by a multi-task pre-training scheme involving three novel multi-granularity self-supervised HG pre-training objectives.We empirically demonstrate the effectiveness of HGT, showing that it outperforms the SOTA for few-shot complex TU on several benchmarks.

cs.CL↗

PRIMO: Progressive Induction for Multi-hop Open Rule Generation

Open rule refer to the implication from premise atoms to hypothesis atoms, which captures various relations between instances in the real world. Injecting open rule knowledge into the machine helps to improve the performance of downstream tasks such as dialogue and relation extraction. Existing approaches focus on single-hop open rule generation, ignoring multi-hop scenarios, leading to logical inconsistencies between premise and hypothesis atoms, as well as semantic duplication of generated rule atoms. To address these issues, we propose a progressive multi-stage open rule generation method called PRIMO. We introduce ontology information during the rule generation stage to reduce ambiguity and improve rule accuracy. PRIMO constructs a multi-stage structure consisting of generation, extraction, and ranking modules to fully leverage the latent knowledge within the language model across multiple dimensions. Furthermore, we employ reinforcement learning from human feedback to further optimize model, enhancing the model's understanding of commonsense knowledge. Experiments show that compared to baseline models, PRIMO significantly improves rule quality and diversity while reducing the repetition rate of rule atoms.

cs.CL↗

Popularity Ratio Maximization: Surpassing Competitors through Influence Propagation

In this paper, we present an algorithmic study on how to surpass competitors in popularity by strategic promotions in social networks. We first propose a novel model, in which we integrate the Preferential Attachment (PA) model for popularity growth with the Independent Cascade (IC) model for influence propagation in social networks called PA-IC model. In PA-IC, a popular item and a novice item grab shares of popularity from the natural popularity growth via the PA model, while the novice item tries to gain extra popularity via influence cascade in a social network. The popularity ratio is defined as the ratio of the popularity measure between the novice item and the popular item. We formulate Popularity Ratio Maximization (PRM) as the problem of selecting seeds in multiple rounds to maximize the popularity ratio in the end. We analyze the popularity ratio and show that it is monotone but not submodular. To provide an effective solution, we devise a surrogate objective function and show that empirically it is very close to the original objective function while theoretically, it is monotone and submodular. We design two efficient algorithms, one for the overlapping influence and non-overlapping seeds (across rounds) setting and the other for the non-overlapping influence and overlapping seed setting, and further discuss how to deal with other models and problem variants. Our empirical evaluation further demonstrates that the proposed PRM-IMM method consistently achieves the best popularity promotion compared to other methods. Our theoretical and empirical analyses shed light on the interplay between influence maximization and preferential attachment in social networks.

cs.SI↗