arXiv subjects

Fanxu Meng

Publications and source records attributed to Fanxu Meng.

At least 19 recordsLinked to original sources

arXiv2026

Linear Temporal Logic Translation via Human-Inspired Self-Constrained Reasoning for Robot Task Specification

Many robotic tasks are temporally extended and demand precise specifications of subgoals, constraints, and their temporal ordering. Yet human operators typically communicate such tasks in natural language, which is inherently ambiguous, underspecified, and context dependent. Translating human instructions into formal task specifications, such as Linear Temporal Logic (LTL), is therefore essential for verifiable and safe robotic execution. Existing LLM-based translators attempt to bridge this gap through open-ended reasoning or post-hoc constraint enforcement, but the former may violate domain constraints, whereas the latter can disrupt the reasoning needed for novel instructions. This paper proposes Self-Constrained Reasoning (SCR), a framework that mitigates this trade-off by internalizing structural knowledge into the model's decision-making process rather than imposing it as an external filter. By combining a structural constraint representation with a hierarchical decision-making formulation, SCR guides reasoning within a formally grounded space while preserving adaptability to unseen instructions. Experiments show that SCR improves both domain-constraint satisfaction and generalization, providing an effective and interpretable approach for translating human intent into verifiable specifications for robotic execution.

cs.RO↗

arXiv2026

Beyond KV Reconstruction: Functional Reconstruction for MLA Draft Models in Speculative Decoding

Multi-head latent attention (MLA) is increasingly important for long-context LLM inference because compact latent states replace the growing key-value (KV) cache and reduce decoding memory traffic. Yet most capable open checkpoints use multi-head or grouped-query attention (MHA/GQA), so conversion is needed to obtain MLA's cache efficiency without retraining from scratch. Speculative decoding offers complementary acceleration, but its speedup depends on agreement between draft proposals and target verification. We find that direct MHA/GQA-to-MLA conversion can sharply reduce this agreement: low-rank factorization and RoPE handling introduce attention-function errors that may be tolerable for standalone generation but substantially lower draft-token acceptance. We therefore formulate MLA draft construction as functional reconstruction rather than cache compression. Our end-to-end (E2E) method optimizes each converted MLA attention module to reproduce the post-output-projection response of its original MHA/GQA counterpart on calibration hidden states. This converter-agnostic post-conversion procedure preserves the converted cache and inference graph and requires neither verifier logits nor verifier supervision. We evaluate 192 model-converter-backend-method-task configurations spanning four Llama/Qwen draft-target pairs, TransMLA and MHA2MLA, HF and vLLM, and four 200-prompt tasks. With a 0.5-percentage-point reporting tolerance, Functional Reconstruction materially improves acceptance in 37 of 64 matched task cells, leaves 26 practically unchanged, and materially decreases one. Code and evaluation artifacts are available at https://github.com/swyhahaha/FunctionalMLA.

cs.LG↗

arXiv2026

GQLA: Group-Query Latent Attention for Hardware-Adaptive Large Language Model Decoding

Multi-head Latent Attention (MLA), the attention used in DeepSeek-V2/V3, jointly compresses keys and values into a low-rank latent and matches the H100 roofline almost perfectly. Its trained weights, however, expose only one decoding path - an absorbed MQA form - which ties efficient inference to H100-class compute-bandwidth ratios, forfeits tensor parallelism along the head axis, and yields no Multi-Token Prediction (MTP) gain on commodity inference GPUs such as the export-restricted H20. We propose Group-Query Latent Attention (GQLA), a minimal modification of MLA whose trained weights expose two algebraically equivalent decoding paths over the same parameters: an MQA-absorb path identical to MLA's, and a GQA path with a per-group expanded cache. The runtime picks the path that matches the target hardware - no retraining, no custom kernels - so a single set of GQLA weights pins the rooflines of both H100 (MQA-absorb, s_q=1) and H20 (GQA + MTP, s_q=2), while supporting up to 8-way zero-redundancy tensor parallelism on the GQA path. To avoid pretraining from scratch we extend TransMLA into TransGQLA, which converts a pretrained GQA checkpoint into a GQLA model; on LLaMA-3-8B it compresses the per-token KV cache to 28.125% of the GQA baseline on the MQA-absorb path while structurally preserving GQA-level traffic on the per-group path.

cs.LG↗

arXiv2026

MISA: Mixture of Indexer Sparse Attention for Long-Context LLM Inference

DeepSeek Sparse Attention (DSA) sets the state of the art for fine-grained inference-time sparse attention by introducing a learned token-wise indexer that scores every prefix token and selects the most relevant ones for the main attention. To remain expressive, the indexer uses many query heads (for example, 64 on DeepSeek-V3.2) that share the same selected token set; this multi-head design is precisely what makes the indexer the dominant cost on long contexts. We propose MISA (Mixture of Indexer Sparse Attention), a drop-in replacement for the DSA indexer that treats its indexer heads as a pool of mixture-of-experts. A lightweight router uses cheap block-level statistics to pick a query-dependent subset of only a few active heads, and only those heads run the heavy token-level scoring. This preserves the diversity of the original indexer pool while reducing the per-query cost from scoring every prefix token with every head to scoring it with only a handful of routed heads, plus a negligible router term computed on a small set of pooled keys. We further introduce a hierarchical variant of MISA that uses the routed pass to keep an enlarged candidate set and then re-ranks it with the original DSA indexer to recover the final selected tokens almost exactly. With only eight active heads and no additional training, MISA matches the dense DSA indexer on LongBench across DeepSeek-V3.2 and GLM-5 while running with eight and four times fewer indexer heads respectively, and outperforms HISA on average. It also preserves fully green Needle-in-a-Haystack heatmaps up to a 128K-token context and recovers more than 92% of the tokens selected by the DSA indexer per layer. Our TileLang kernel delivers roughly a 3.82 times speedup over DSA's original indexer kernel on a single NVIDIA H200 GPU.

cs.LG↗

arXiv2026

"From remembering to shaping": Narrating Shared Experiences by Co-Designing Cultural Heritage Artifacts in Collaborative VR

The ways people remember and recall places reveal an invisible aspect of cultural heritage (CH), reflecting how individuals and communities relate to these places. Heritage is communal, emerging through collaboratively constructed narratives rather than individual records. To probe how people may share collective memories, we designed an immersive two-person workflow for collaboratively co-designing 3D artifacts and environments in virtual heritage locations, using Generative AI (GenAI) to instantiate these intangible memories. Observations of the co-creation process revealed that participants merged prompts and model placements when negotiating different perspectives. They used spatial operations to compose scenes, and also to express personal and embodied experiences of CH. When GenAI failed to meet their needs, participants engaged in creative appropriation, re-purposing unsatisfactory generated objects as sources of design inspiration to further shared narratives. While GenAI may have a homogenizing effect on CH expression, this work shows how people may overcome limitations in immersive collaborative workflows.

cs.HC↗

arXiv2026

HISA: Efficient Hierarchical Indexing for Fine-Grained Sparse Attention

Token-level sparse attention mechanisms, exemplified by DeepSeek Sparse Attention (DSA), achieve fine-grained key selection by scoring every historical key for each query through a lightweight indexer, then computing attention only on the selected subset. While the downstream sparse attention itself scales favorably, the indexer must still scan the entire prefix for every query, introducing an per-layer bottleneck that grows prohibitively with context length. We propose HISA (Hierarchical Indexed Sparse Attention), a plug-and-play replacement for the indexer that rewrites the search path from a flat token scan into a two-stage hierarchical procedure: (1) a block-level coarse filtering stage that scores pooled block representations to discard irrelevant regions, followed by (2) a token-level refinement stage that applies the original indexer exclusively within the retained candidate blocks. HISA preserves the identical token-level top-sparse pattern consumed by the downstream Sparse MLA operator and requires no additional training. On kernel-level benchmarks, HISA achieves up to speedup at 64K context. On Needle-in-a-Haystack and LongBench, we directly replace the indexer in DeepSeek-V3.2 and GLM-5 with our HISA indexer, without any finetuning. HISA closely matches the original DSA in quality, while substantially outperforming block-sparse baselines.

cs.LG↗

arXiv2026

Breaking the Blocks: Continuous Low-Rank Decomposed Scaling for Unified LLM Quantization and Adaptation

Current quantization methods for LLMs predominantly rely on block-wise structures to maintain efficiency, often at the cost of representational flexibility. In this work, we demonstrate that element-wise quantization can be made as efficient as block-wise scaling while providing strictly superior expressive power by modeling the scaling manifold as continuous low-rank matrices ($S = BA$). We propose Low-Rank Decomposed Scaling (LoRDS), a unified framework that rethinks quantization granularity through this low-rank decomposition. By "breaking the blocks" of spatial constraints, LoRDS establishes a seamless efficiency lifecycle: it provides high-fidelity PTQ initialization refined via iterative optimization, enables joint QAT of weights and scaling factors, and facilitates high-rank multiplicative PEFT adaptation. Unlike additive PEFT approaches such as QLoRA, LoRDS enables high-rank weight updates within a low-rank budget while incurring no additional inference overhead. Supported by highly optimized Triton kernels, LoRDS consistently outperforms state-of-the-art baselines across various model families in both quantization and downstream fine-tuning tasks. Notably, on Llama3-8B, our method achieves up to a 27.0% accuracy improvement at 3 bits over NormalFloat quantization and delivers a 1.5x inference speedup on NVIDIA RTX 4090 while enhancing PEFT performance by 9.6% on downstream tasks over 4bit QLoRA, offering a robust and integrated solution for unified compression and adaptation of LLMs.

cs.LG↗

arXiv2026

BioPIE: A Biomedical Protocol Information Extraction Dataset for Experiment Understanding

Understanding biomedical experiments provides a foundation for downstream tasks, e.g., laboratory automation, and facilitates effective cross-disciplinary communication. Two challenges, High Information Density (HID) and Multi-Step Reasoning (MSR), pose unique difficulties for precise automatic experimental understanding. Extracting structured knowledge, e.g., Knowledge Graphs (KGs), is an effective approach to address the HID and MSR. However, existing biomedical datasets for structured knowledge Information Extraction (IE) are limited to a general or coarse-grained level, hindering fine-grained experimental understanding. To address this gap, we introduce Biomedical Protocol Information Extraction Dataset (BioPIE), a dataset providing procedure-centric KGs that captures entities, actions, and relations at a scale sufficient for reasoning across biomedical protocols. We evaluate both supervised and LLM-based IE methods on BioPIE to verify its effectiveness, and implement a biomedical question answering system to provide a quantitative illustration of BioPIE's effectiveness for downstream understanding tasks. The experimental results demonstrate improved understanding performance on both the HID and MSR question sets.

cs.AI↗

arXiv2025

Youtu-LLM: Unlocking the Native Agentic Potential for Lightweight Large Language Models

We introduce Youtu-LLM, a lightweight yet powerful language model that harmonizes high computational efficiency with native agentic intelligence. Unlike typical small models that rely on distillation, Youtu-LLM (1.96B) is pre-trained from scratch to systematically cultivate reasoning and planning capabilities. The key technical advancements are as follows: (1) Compact Architecture with Long-Context Support: Built on a dense Multi-Latent Attention (MLA) architecture with a novel STEM-oriented vocabulary, Youtu-LLM supports a 128k context window. This design enables robust long-context reasoning and state tracking within a minimal memory footprint, making it ideal for long-horizon agent and reasoning tasks. (2) Principled "Commonsense-STEM-Agent" Curriculum: We curated a massive corpus of approximately 11T tokens and implemented a multi-stage training strategy. By progressively shifting the pre-training data distribution from general commonsense to complex STEM and agentic tasks, we ensure the model acquires deep cognitive abilities rather than superficial alignment. (3) Scalable Agentic Mid-training: Specifically for the agentic mid-training, we employ diverse data construction schemes to synthesize rich and varied trajectories across math, coding, and tool-use domains. This high-quality data enables the model to internalize planning and reflection behaviors effectively. Extensive evaluations show that Youtu-LLM sets a new state-of-the-art for sub-2B LLMs. On general benchmarks, it achieves competitive performance against larger models, while on agent-specific tasks, it significantly surpasses existing SOTA baselines, demonstrating that lightweight models can possess strong intrinsic agentic capabilities.

cs.CL↗

arXiv2025

Shot and Architecture Adaptive Subspace Variational Quantum Eigensolver for Microwave Simulation

Quantum computing offers a promising paradigm for electromagnetic eigenmode analysis, enabling compact representations of complex field interactions and potential exponential speedup over classical numerical solvers. Recent efforts have applied variational quantum eigensolver (VQE) based methods to compute waveguide modes, demonstrating the feasibility of simulating TE and TM field distributions on noisy intermediate-scale quantum (NISQ) hardware. However, these studies typically employ manually designed, fixed-depth parameterized quantum circuits and uniform measurement-shot strategies, resulting in excessive quantum resource consumption, limited circuit expressivity, and reduced robustness under realistic noise conditions. To address these limitations, we propose an architecture and shot adaptive subspace variational quantum eigensolver for efficient microwave waveguide eigenmode simulation on NISQ devices. The proposed framework integrates a reinforcement learning (RL) based circuit design strategy and an adaptive shot allocation mechanism to jointly reduce quantum resource overhead. Specifically, the RL agent autonomously explores the quantum circuit space to generate hardware-efficient parameterized quantum circuits, while the adaptive measurement scheme allocates sampling resources according to Hamiltonian term weights. Numerical experiments on three- and five-qubit systems demonstrate that the proposed framework achieves accurate estimation of TE and TM mode eigenvalues, with a minimum absolute error down to $10^{-8}$ and reconstructed field distributions under noiseless conditions in excellent agreement with classical electromagnetic solutions.

quant-ph↗

arXiv2025

Quantum-Based Self-Attention Mechanism for Hardware-Aware Differentiable Quantum Architecture Search

The automated design of parameterized quantum circuits for variational algorithms in the NISQ era faces a fundamental limitation, as conventional differentiable architecture search relies on classical models that fail to adequately represent quantum gate interactions under hardware noise. We introduce the Quantum-Based Self-Attention for Differentiable Quantum Architecture Search (QBSA-DQAS), a meta-learning framework featuring quantum-based self-attention and hardware-aware multi-objective search. The framework employs a two-stage quantum self-attention module that computes contextual dependencies by mapping architectural parameters through parameterized quantum circuits, replacing classical similarity metrics with quantum-derived attention scores, then applies position-wise quantum transformations for feature enrichment. Architecture search is guided by a task-agnostic multi-objective function jointly optimizing noisy expressibility and Probability of Successful Trials (PST). A post-search optimization stage applies gate commutation, fusion, and elimination to reduce circuit complexity. Experimental validation demonstrates superior performance on VQE tasks and large-scale Wireless Sensor Networks. For VQE on H$_2$, QBSA-DQAS achieves 0.9 accuracy compared to 0.89 for standard DQAS. Post-search optimization reduces discovered circuit complexity by up to 44% in gate count and 47% in depth without accuracy degradation. The framework maintains robust performance across three molecules and five IBM quantum hardware noise models. For WSN routing, discovered circuits achieve 8.6% energy reduction versus QAOA and 40.7% versus classical greedy methods, establishing the effectiveness of quantum-native architecture search for NISQ applications.

quant-ph↗

arXiv2025

Family of two-parameter multipartite entanglement measures

Multipartite entanglement is regarded as a crucial physical resource in quantum network communication. However, due to the intrinsic complexity of quantum many-body systems, identifying a multipartite entanglement measure that is both efficiently computable and capable of accurately characterizing entanglement remains a challenging problem. To address these issues, we propose a family of two-parameter multipartite entanglement measures for mixed states, termed unified-entropy concentratable entanglements. Many well-known multipartite entanglement measures are recovered as special cases of this family of measures, such as the entanglement of formation and the concentratable entanglements introduced in [Phys. Rev. Lett. 127, 140501 (2021)]. We demonstrate that the unified-entropy concentratable entanglements constitutes a well-defined entanglement monotones, and establish several desirable properties it satisfies, such as subadditivity and continuity. We further investigate the ordering relations of unified-entropy concentratable entanglements and discuss how these quantities can be efficiently estimated on near-term quantum devices. As an application, we demonstrate that the unified-entropy concentratable entanglements can effectively distinguish between multi-qubit Greenberger-Horne-Zeilinger (GHZ) states and W states. The ordering relations of these entanglement measures are further validated using four-partite star quantum network states and four-qubit Dicke states. Moreover, we find that the unified-entropy concentratable entanglements exhibit greater sensitivity than the original concentratable entanglements in detecting certain four-partite star quantum network states.

quant-ph↗

arXiv2025

Law in Silico: Simulating Legal Society with LLM-Based Agents

Since real-world legal experiments are often costly or infeasible, simulating legal societies with Artificial Intelligence (AI) systems provides an effective alternative for verifying and developing legal theory, as well as supporting legal administration. Large Language Models (LLMs), with their world knowledge and role-playing capabilities, are strong candidates to serve as the foundation for legal society simulation. However, the application of LLMs to simulate legal systems remains underexplored. In this work, we introduce Law in Silico, an LLM-based agent framework for simulating legal scenarios with individual decision-making and institutional mechanisms of legislation, adjudication, and enforcement. Our experiments, which compare simulated crime rates with real-world data, demonstrate that LLM-based agents can largely reproduce macro-level crime trends and provide insights that align with real-world observations. At the same time, micro-level simulations reveal that a well-functioning, transparent, and adaptive legal system offers better protection of the rights of vulnerable individuals.

cs.AI↗

arXiv2025

Quantum Approximate Optimization Algorithm for Maximum Likelihood Detection in Massive MIMO

In the massive multiple-input and multiple-output (Massive MIMO) systems, the maximum likelihood (ML) detection problem is NP-hard and becoming classically intricate with the number of the transmitting antennas and the symbols increasing. The quantum approximate optimization algorithm (QAOA), a leading candidate algorithm running in the noisy intermediate-scale quantum (NISQ) devices, can show quantum advantage for approximately solving combinatorial optimization problems. In this paper, we propose the QAOA based the maximum likelihood detection solver of binary symbols. In proposed scheme, we first conduct a universal and compact analytical expression for the expectation value of the 1-level QAOA. Second, a bayesian optimization based parameters initialization is presented, which can speedup the convergence of the QAOA to a lower local minimum and improve the probability of measuring the exact solution. Compared to the state-of-the-art QAOA based ML detection algorithm, our scheme have the more universal and compact expectation value expression of the 1-level QAOA, and requires few quantum resources and has the higher probability to obtain the exact solution.

quant-ph↗

arXiv2025

TPLA: Tensor Parallel Latent Attention for Efficient Disaggregated Prefill and Decode Inference

Multi-Head Latent Attention (MLA), introduced in DeepSeek-V2, compresses key-value states into a low-rank latent vector, caching only this vector to reduce memory. In tensor parallelism (TP), however, attention heads are computed across multiple devices, and each device must load the full cache, eroding the advantage of MLA over Grouped Query Attention (GQA). We propose Tensor-Parallel Latent Attention (TPLA): a scheme that partitions both the latent representation and each head's input dimension across devices, performs attention independently per shard, and then combines results with an all-reduce. TPLA preserves the benefits of a compressed KV cache while unlocking TP efficiency. Unlike Grouped Latent Attention (GLA), every head in TPLA still leverages the full latent representation, maintaining stronger representational capacity. TPLA is drop-in compatible with models pre-trained using MLA: it supports MLA-style prefilling and enables efficient tensor-parallel decoding without retraining. Applying simple orthogonal transforms -- e.g., the Hadamard transform or PCA -- before TP slicing further mitigates cross-shard interference, yielding minimal accuracy degradation. By reducing the per-device KV cache for DeepSeek-V3 and Kimi-K2, we achieve 1.79x and 1.93x speedups, respectively, at a 32K-token context length while maintaining performance on commonsense and LongBench benchmarks. TPLA can be implemented with FlashAttention-3, enabling practical end-to-end acceleration.

cs.LG↗

arXiv2025

Hardware-Aware Quantum Kernel Design Based on Graph Neural Networks

Quantum kernels hold significant promise for achieving computational advantages in quantum machine learning (QML), yet their effectiveness critically depends on the design of expressive and hardware-compatible feature maps, a challenge that is particularly pronounced on Noisy Intermediate-Scale Quantum (NISQ) devices with limited qubits, gate errors, and restricted connectivity. In this work, we propose a hardware-aware framework for automated quantum kernel design that integrates quantum device characteristics with learning-based evaluation. Specifically, candidate quantum circuits explored within the hardware-aware circuit space are represented as directed acyclic graphs (DAGs) encoding hardware-specific information such as gate operations, qubit interactions, and noise properties, while a dual graph neural network (GNN) predictor is employed to estimate key surrogate metrics, including probability of successful trials (PST) and kernel-target alignment (KTA), enabling efficient and accurate assessment of circuit fidelity and kernel performance to facilitate the identification of task-specific quantum kernels. Furthermore, feature selection is incorporated to reduce input dimensionality and ensure compatibility with near-term devices. Extensive experiments on multiple benchmark datasets, including Credit Card (CC), MNIST-5, and FMNIST-4, demonstrate that our method consistently outperforms existing baselines in classification accuracy, effectively balancing hardware constraints and model expressivity under realistic noise conditions. These results highlight the potential of combining hardware-aware design with deep learning techniques to advance practical quantum kernel methods and facilitate their deployment on near-term quantum hardware.

quant-ph↗

arXiv2025

LoRASuite: Efficient LoRA Adaptation Across Large Language Model Upgrades

As Large Language Models (LLMs) are frequently updated, LoRA weights trained on earlier versions quickly become obsolete. The conventional practice of retraining LoRA weights from scratch on the latest model is costly, time-consuming, and environmentally detrimental, particularly as the diversity of LLMs and downstream tasks expands. This motivates a critical question: "How can we efficiently leverage existing LoRA weights to adapt to newer model versions?" To address this, we propose LoRASuite, a modular approach tailored specifically to various types of LLM updates. First, we compute a transfer matrix utilizing known parameters from both old and new LLMs. Next, we allocate corresponding layers and attention heads based on centered kernel alignment and cosine similarity metrics, respectively. A subsequent small-scale, skillful fine-tuning step ensures numerical stability. Experimental evaluations demonstrate that LoRASuite consistently surpasses small-scale vanilla LoRA methods. Notably, on backbone LLMs such as MiniCPM and Qwen, LoRASuite even exceeds the performance of full-scale LoRA retraining, with average improvements of +1.4 and +6.6 points on math tasks, respectively. Additionally, LoRASuite significantly reduces memory consumption by 5.5 GB and computational time by 78.23%.

cs.LG↗

arXiv2025

Hierarchically Encapsulated Representation for Protocol Design in Self-Driving Labs

Self-driving laboratories have begun to replace human experimenters in performing single experimental skills or predetermined experimental protocols. However, as the pace of idea iteration in scientific research has been intensified by Artificial Intelligence, the demand for rapid design of new protocols for new discoveries become evident. Efforts to automate protocol design have been initiated, but the capabilities of knowledge-based machine designers, such as Large Language Models, have not been fully elicited, probably for the absence of a systematic representation of experimental knowledge, as opposed to isolated, flatten pieces of information. To tackle this issue, we propose a multi-faceted, multi-scale representation, where instance actions, generalized operations, and product flow models are hierarchically encapsulated using Domain-Specific Languages. We further develop a data-driven algorithm based on non-parametric modeling that autonomously customizes these representations for specific domains. The proposed representation is equipped with various machine designers to manage protocol design tasks, including planning, modification, and adjustment. The results demonstrate that the proposed method could effectively complement Large Language Models in the protocol design process, serving as an auxiliary module in the realm of machine-assisted scientific exploration.

cs.AI↗