arXiv subjects

Lei Jiang

Publications and source records attributed to Lei Jiang.

At least 37 records · Page 2Linked to original sources

arXiv2026

QAgent: An LLM-based Multi-Agent System for Autonomous OpenQASM programming

Programming quantum circuits at the OpenQASM level is essential for achieving hardware-aware optimization and reliable execution on noisy intermediate-scale quantum (NISQ) devices, yet it remains challenging due to the need for domain-specific planning, iterative code synthesis, and low-level calibration. In this paper, we present QAgent, the first autonomous multi-agent framework for end-to-end OpenQASM code generation. QAgent integrates schema-aware task planning, example- and tool-driven code synthesis, and hardware-aware calibration within a unified planning-synthesis-calibration workflow. The system leverages retrieval-augmented generation (RAG) to access structured kernel knowledge, examples, and backend constraints, and employs coordinated multi-agent reasoning with iterative execution feedback to ensure correctness. We evaluate QAgent on 12 representative quantum kernels and their compositions across five large language models (LLMs). Results show that QAgent improves Pass@1 accuracy by 47-70% on single-kernel tasks and achieves over 88% accuracy on multi-kernel workflows for large models, substantially outperforming existing baselines. Furthermore, under realistic hardware frequency drift, QAgent maintains near-unit execution fidelity through automated calibration, whereas SDK-based LLM methods suffer significant degradation. These results demonstrate that integrating planning, synthesis, and calibration is critical for reliable quantum program generation. The implementation of QAgent is open-sourced at https://github.com/fuzhenxiao/QAgent

cs.AI↗

arXiv2026

Surface code logical operations on a superconducting quantum processor

Fault-tolerant quantum computation requires logical operations that manipulate encoded information while preserving quantum error-correction protection. In planar surface-code architectures, code deformation and lattice surgery provide a local, measurement-based route to such operations. Here we experimentally realize key elements of patch-based surface-code logical processing on a 107-qubit superconducting quantum processor. We first implement a reusable primitive layer comprising merge and split, patch expansion and shrinkage, and deformations mediated by domain walls and twist defects. We then compose these primitives to realize logical state routing, the logical controlled-NOT gate, and the single-qubit Hadamard and phase gates, which together form a Clifford-generating set. All operations are implemented on distance-three rotated surface-code patches with multi-round syndrome extraction and neural-network decoding, without post-selection. Our results advance superconducting surface-code experiments from protected logical memory to active, patch-based fault-tolerant logical operations.

quant-ph↗

arXiv2026

GCN-DevLSTM: Path Development for Skeleton-Based Action Recognition

Skeleton-based action recognition (SAR) in videos is an important but challenging task in computer vision. The recent state-of-the-art (SOTA) models for SAR are primarily based on graph convolutional neural networks (GCNs), which are powerful in extracting the spatial information from skeleton data. However, their ability to capture temporal dynamics remains limited. To address this, we propose the G-Dev layer, which leverages path development-a principled and parsimonious representation for sequential data based on Lie group structures-to enhance temporal modeling. By integrating the G-Dev layer, the proposed DevLSTM module summarizes local temporal dynamics, reducing the time dimension while retaining high-frequency information. It can be conveniently applied to any temporal graph data, complementing existing advanced GCN-based models. Our empirical studies on the NTU-60, NTU-120 and Chalearn2013 datasets demonstrate that our proposed GCN-DevLSTM network consistently improves the strong GCN baseline models and achieves competitive performance. The code repository is publicly available at https://github.com/DeepIntoStreams/GCN-DevLSTM.

cs.CV↗

arXiv2026

SWAP: Symmetric Equivariant World-Model for Agile Robot Parkour

While latent world models enable the proactive predictions required for extreme parkour, their purely data-driven nature forces them to redundantly encode left-right symmetric interactions as independent patterns. This inflates the learning burden and hinders the capture of geometric regularities, restricting the latent space's efficiency for downstream policies. To address this, we propose SWAP, an end-to-end equivariant symmetric world model. This framework embeds symmetry directly into both the world model and the actor-critic networks. In real-world tests, the robot leaps across a 2.13 m gap and climbs a 1.63 m platform, breaking records for quadruped parkour. Furthermore, the framework exhibits robust geometric generalization to unseen mirrored terrains and exceptional zero-shot transferability across diverse outdoor environments. These results demonstrate that symmetry equivariance is an effective structural prior for pushing the physical boundaries of learned legged locomotion.

cs.RO↗

arXiv2026

When the Same Musical Knowledge Forgets Differently: A Clean Probe of Pathway-Dependent Forgetting

A model can learn that the piano piece Für Elise is calm and reflective by listening to the audio or by reading a text description, but does it matter which route that knowledge took when it is later at risk of being forgotten? Forgetting research in multimodal models measures what knowledge is lost under adaptation, yet has not asked whether acquisition route affects how easily that knowledge is forgotten. We call this untested premise the Pathway-Invariant Assumption. Music understanding enables a clean test because a music clip and a canonical text description can be aligned to the same perceptual content, allowing the same knowledge unit to enter a model through listening or reading while the target remains fixed. Across multiple architecturally distinct audio-language models, we observe a consistent asymmetry: text-pathway knowledge is forgotten more than matched audio-pathway knowledge under identical adaptation pressure. To attribute this effect to route rather than confounds, we introduce the Paired Pathway Controlled Protocol (PPCP), a three-phase design that establishes matched pathway baselines, activates both pathways under symmetric supervision on the same knowledge pool, and applies identical forgetting pressure to both pathways. The gap is stable across models and gain-controlled analyses, persists when contradictory overwrite is replaced by correct-label cross-domain learning, remains under single-modality pressure, and is not removed by lightweight replay. Two independent routing-depth controls confirm that the effect is not explained by architectural depth, pointing to input representation as the dominant factor. Under PPCP, our results demonstrate that forgetting is highly route-dependent, establishing acquisition route as a new analytical dimension for forgetting research and multimodal system design.

cs.SD↗

arXiv2026

Beyond WER: A Paired Acoustic Stress Test for Ambient Clinical Scribes

Ambient clinical scribes increasingly combine Automatic Speech Recognition with Large Language Models to automate documentation. However, traditional metrics like Word Error Rate mask systemic safety degradation. We present a paired acoustic stress test to isolate the causal impact of noise on clinical reasoning. For the same dialogues, we inject diverse noise types while keeping the downstream model configuration frozen. Crucially, we uncover a dangerous disconnect between signal fidelity and clinical safety. Stationary ambient noise increased the Word Error Rate by a negligible 0.71 percentage points yet nearly doubled the rate of unsafe outputs. Our analysis reveals that minor acoustic perturbations can invert clinical meaning without substantially inflating error rates. Furthermore, we demonstrate a lightweight mitigation strategy that mitigates safety degradation under noisy conditions without requiring model fine tuning.

cs.SD↗

arXiv2026

What Do LLMs Know About Alzheimer's Disease? Multi-loss Fine-Tuning and Probing for AD Detection

Reliable early detection of Alzheimer's disease (AD) is challenging, particularly due to the limited availability of labeled data. While large language models (LLMs) have shown strong transfer capabilities across do mains, adapting them to the AD domain through supervised fine-tuning remains largely unexplored. In this work, we empirically evaluate various model architectures across three heterogeneous transcript corpora (Pitt, CCC, ADRC) to investigate their effectiveness for text-based AD detection and analyze how task-relevant information is encoded within their internal representations. To the best of our knowledge, our fine-tuned BERT and T5 models establish a new state-of-the-art on the Pitt and CCC datasets, while achieving strong performance on ADRC. In parallel, the decoder-only Llama-1B achieves highly competitive results comparable to BERT and T5 across all three corpora, highlighting its effectiveness for AD detection. We further conduct a comprehensive evaluation of the Llama-1B backbone, analyzing cross-corpus transferability, optimal input chunk-size granularity, and the impact of clinical transcript markers. Also, we use linear probing to empirically show that fine-tuning shifts the representations of individual tokens, both linguistic markers and content words, in ways that reflect AD-related signal.

cs.CL↗

arXiv2026

HaluNet: Learning Hallucination Risk from Internal Signals in LLM Question Answering

Large language models (LLMs) achieve strong question answering (QA) performance but can produce fluent answers unsupported by available evidence. Existing hallucination detectors often rely on external verification, repeated sampling, or test-time judge calls, which can be costly for real-time QA. We propose \textbf{HaluNet}, a lightweight hallucination risk estimator that uses internal signals from one model generation. HaluNet jointly models token likelihood, predictive entropy, and hidden-state information, allowing probabilistic, distributional, and semantic evidence to inform an answer-level risk score. It is trained with LLM-as-a-Judge labels as scalable weak supervision and evaluated with independent human and multi-judge assessments. Experiments on SQuAD, TriviaQA, and Natural Questions show that HaluNet improves answer-level risk ranking across in-domain and out-of-domain settings. On a 300-example human evaluation, HaluNet achieves 0.874 AUROC and 0.869 AUPRC; its top 20\% highest-risk answers contain 96.5\% errors, yielding a 2.06$\times$ lift over the base error rate.

cs.CL↗

arXiv2026

ELAN4D: Embodiment-Centric 4D Supervision for Vision-Language-Action Models via Plug-and-Play Adaptation

Vision-Language-Action (VLA) models have shown promise for robotic manipulation, yet most existing policies operate reactively by directly regressing actions from current observations, without explicitly modeling future dynamics. This limits their ability to generalize under out-of-distribution perturbations. To address this issue, we propose ELAN4D, an embodiment-centric, 4D-aware training framework that enhances VLA policies with future robot keypoint tracks as predictive spatio-temporal supervision. Using only forward kinematics from proprioceptive states, we derive 3D displacement tracks of robot keypoints, such as joints and the end-effector, with negligible preprocess cost. These tracks provide metric and compact supervision without requiring external trackers or reconstruction. A plug-and-play auxiliary branch with a lightweight track decoder injects this 4D signal into the action expert while preserving the pretrained vision-language backbone through gradient isolation. The track decoder is discarded during inference, leaving the base policy interface unchanged. Extensive experiments on LIBERO, LIBERO-Plus, RoboTwin2.0 and real-world manipulation tasks demonstrate that ELAN4D consistently improves over strong VLA baselines, achieving the best overall performance and substantial gains under out-of-distribution perturbations, including camera, background, and layout shifts. These results highlight the effectiveness of embodiment-centric 4D supervision for building more robust and generalizable manipulation policies.

cs.RO↗

arXiv2026

Knowledge Dependency Estimation for Reliable Question Answering

Reliable question answering requires identifying not only whether an answer is correct, but also which available knowledge the prediction depends on. In realistic LLM-based QA, this knowledge may come from context, retrieval, decomposition, or intermediate reasoning, forming a noisy and redundant candidate space rather than a clean gold evidence set. We study \emph{knowledge dependency estimation}: estimating the sensitivity of a fixed black-box QA model to different candidate knowledge units. The challenge is to obtain fine-grained dependency scores without exhaustive test-time perturbation while modeling redundancy, substitutability, and complementarity. We propose \textbf{Knot}, a structured rank-aware knowledge dependency estimator. Knot learns from subset-level counterfactual supervision, models subset sensitivity through coverage over latent dependency factors, and derives rank-aware unit scores to identify influential candidates. Across multiple-choice and generative QA benchmarks, Knot outperforms all compared baselines in subset-sensitivity prediction and produces more faithful unit rankings than deployable baselines without extra QA-model calls; when used for practical risk screening, its dependency scores help flag error-prone QA predictions early.

cs.CL↗

arXiv2026

CarbonScaling: Extending Neural Scaling Laws for Carbon Footprint in Large Language Models

Large language models (LLMs) increasingly follow neural scaling laws that tie performance gains to rapidly expanding computational budgets, raising concerns about the sustainability of frontier-scale training. Existing carbon-estimation methods largely depend on regression over historical runs and fail to capture critical system-level factors, including hardware heterogeneity, distributed parallelism, communication overhead, and architectural sparsity. We present \textit{CarbonScaling}, a hardware-aware analytical framework for modeling the carbon scaling behavior of frontier LLM training. The framework integrates neural scaling laws, distributed training strategies, accelerator and interconnect modeling, and operational and embodied carbon accounting to estimate feasible hardware configurations and associated emissions. CarbonScaling jointly models tensor, pipeline, data, and expert parallelism while incorporating memory, bandwidth, utilization, and runtime constraints. Experimental validation demonstrates substantially higher fidelity than regression-based baselines and highlights the growing importance of embodied carbon at trillion-parameter scales. Source code: \url{https://github.com/UnchartedRLab/CarbonScaling}.

cs.CL↗

arXiv2026

Orbital AI Computing: Carbon Tradeoffs Across Satellite Scale

Low Earth Orbit (LEO) computing is emerging for low-latency, globally distributed AI services, enabled by advances in satellite constellations and reusable launch systems. However, its sustainability remains unclear. Prior work introduces ESpaS, a framework for estimating lifecycle carbon intensity, but models systems using generic datacenter configurations and does not capture modern AI hardware, where power, mass, and compute characteristics vary widely and launch emissions scale with system mass. In this work, we extend ESpaS with accelerator-aware modeling and evaluate two representative systems: a lightweight Jetson AGX Orin for small satellites and a high-performance DGX H100 enabled by large-payload launch platforms. We show that launch emissions act as a fixed carbon overhead: low-mass systems minimize absolute emissions, while high-performance systems amortize this cost more effectively, reducing carbon intensity. Consequently, the space-ground tradeoff is highly sensitive to hardware choice, highlighting the need for accelerator-aware baselines in orbital AI computing.

cs.DC↗

arXiv2026

Backdoor Threats in Variational Quantum Circuits: Taxonomy, Attacks, and Defenses

Variational quantum algorithms (VQAs) are a central paradigm for noisy intermediate-scale (NISQ) quantum computing, yet their reliance on predesigned and pretrained variational quantum circuits (VQCs) introduces critical security vulnerabilities, particularly backdoor attacks. These attacks embed hidden malicious behaviors that remain dormant under normal conditions but are activated by specific triggers, leading to adversarial outcomes such as incorrect predictions or manipulated objective values. This paper presents a survey of backdoor attacks in VQCs, covering data-poisoning, compiler-level, and quantum-native mechanisms. We formalize key terminology and threat models, and review existing attack strategies along with their empirical characteristics. We also analyze current detection and defense approaches, highlighting their limitations, especially against quantum-specific threats. By synthesizing recent advances, this survey outlines the evolving security landscape of VQCs and identifies key challenges and future directions for developing robust, quantum-aware defenses in hybrid quantum-classical systems.

quant-ph↗

arXiv2026

LLMSpace: Carbon Footprint Modeling for Large Language Model Inference on LEO Satellites

Large language models (LLMs) impose rapidly growing energy demands, creating an emerging energy and carbon crisis driven by large-scale inference. Solar-powered, AI-enabled low Earth orbit (LEO) satellites have been proposed to mitigate terrestrial electricity consumption, but their lifecycle carbon footprint remains poorly understood due to launch emissions, satellite manufacturing, and radiation-hardened hardware requirements. This paper presents \textit{LLMSpace}, the first carbon modeling framework for LLM inference on AI-enabled LEO satellites. LLMSpace jointly models operational and embodied carbon, peripheral subsystems, radiation-hardened accelerators and memories, and LLM-specific workload characteristics such as prefill-decode behavior and token generation. Using realistic satellite and GPU configurations, LLMSpace reveals key trade-offs among carbon footprint, inference latency, hardware design, and operational lifetime for sustainable space-based LLM inference. Source code: https://github.com/UnchartedRLab/LLMSpace.

cs.LG↗

arXiv2026

Reason Only When Needed: Efficient Generative Reward Modeling via Model-Internal Uncertainty

Recent advancements in the Generative Reward Model (GRM) have demonstrated its potential to enhance the reasoning abilities of LLMs through Chain-of-Thought (CoT) prompting. Despite these gains, existing implementations of GRM suffer from two critical limitations. First, CoT prompting is applied indiscriminately to all inputs regardless of their inherent complexity. This introduces unnecessary computational costs for tasks amenable to fast, direct inference. Second, existing approaches primarily rely on voting-based mechanisms to evaluate CoT outputs, which often lack granularity and precision in assessing reasoning quality. In this paper, we propose E-GRM, an efficient generative reward modeling framework grounded in model-internal uncertainty. E-GRM leverages the convergence behavior of parallel model generations to estimate uncertainty and selectively trigger CoT reasoning only when needed, without relying on handcrafted features or task-dependent signals. To improve reward fidelity, we introduce a lightweight discriminative scorer trained with a hybrid regression--ranking objective to provide fine-grained evaluation of reasoning paths. Experiments on multiple reasoning benchmarks show that E-GRM substantially reduces inference cost while consistently improving answer accuracy, demonstrating that model-internal uncertainty is an effective and general signal for efficient reasoning-aware reward modeling.

cs.CL↗

arXiv2026

Why Supervised Fine-Tuning Fails to Learn: A Systematic Study of Incomplete Learning in Large Language Models

Supervised Fine-Tuning (SFT) is the standard approach for adapting large language models (LLMs) to downstream tasks. However, we observe a persistent failure mode: even after convergence, models often fail to correctly reproduce a subset of their own supervised training data. We refer to this behavior as the Incomplete Learning Phenomenon(ILP). This paper presents the first systematic study of ILP in LLM fine-tuning. We formalize ILP as post-training failure to internalize supervised instances and demonstrate its prevalence across multiple model families, domains, and datasets. Through controlled analyses, we identify five recurrent sources of incomplete learning: (1) missing prerequisite knowledge in the pre-trained model, (2) conflicts between SFT supervision and pre-training knowledge, (3) internal inconsistencies within SFT data, (4) left-side forgetting during sequential fine-tuning, and (5) insufficient optimization for rare or complex patterns. We introduce a diagnostic-first framework that maps unlearned samples to these causes using observable training and inference signals, and study several targeted mitigation strategies as causal interventions. Experiments on Qwen, LLaMA, and OLMo2 show that incomplete learning is widespread and heterogeneous, and that improvements in aggregate metrics can mask persistent unlearned subsets. The findings highlight the need for fine-grained diagnosis of what supervised fine-tuning fails to learn, and why.

cs.CL↗

arXiv2026

Parameter Importance is Not Static: Evolving Parameter Isolation for Supervised Fine-Tuning

Supervised Fine-Tuning (SFT) of large language models often suffers from task interference and catastrophic forgetting. Recent approaches alleviate this issue by isolating task-critical parameters during training. However, these methods represent a static solution to a dynamic problem, assuming that parameter importance remains fixed once identified. In this work, we empirically demonstrate that parameter importance exhibits temporal drift over the course of training. To address this, we propose Evolving Parameter Isolation (EPI), a fine-tuning framework that adapts isolation decisions based on online estimates of parameter importance. Instead of freezing a fixed subset of parameters, EPI periodically updates isolation masks using gradient-based signals, enabling the model to protect emerging task-critical parameters while releasing outdated ones to recover plasticity. Experiments on diverse multi-task benchmarks demonstrate that EPI consistently reduces interference and forgetting compared to static isolation and standard fine-tuning, while improving overall generalization. Our analysis highlights the necessity of synchronizing isolation mechanisms with the evolving dynamics of learning diverse abilities.

cs.LG↗

arXiv2026

TARAC: Mitigating Hallucination in LVLMs via Temporal Attention Real-time Accumulative Connection

Large Vision-Language Models have demonstrated remarkable capabilities, yet they suffer from hallucinations that limit practical deployment. While various mitigation strategies exist, they often incur high computational overhead or require extensive retraining. In this paper, we address the issue of visual attention decay during generation, a key factor contributing to hallucinations. We propose Temporal Attention Real-time Accumulative Connection (TARAC), a novel training-free framework that dynamically accumulates and re-injects historical attention to sustain visual grounding. Inspired by cognitive reinforcement mechanisms, TARAC operates as a lightweight, plug-and-play module. Extensive experiments across diverse models (e.g., LLaVA, Qwen2-VL) and benchmarks demonstrate that TARAC significantly outperforms state-of-the-art methods. Remarkably, it achieves these gains with negligible inference overhead ($\sim$4\% TPOT increase), compared to the substantial costs of existing training-free baselines. Specifically, TARAC reduces hallucinated sentences by 25.2\% on CHAIR and improves Perception score by +10.65 on MME, validating its effectiveness and efficiency.

cs.CV↗