arXiv subjects

Xiangyang Liu

Publications and source records attributed to Xiangyang Liu.

At least 19 recordsLinked to original sources

arXiv2026

FARS: A Fully Automated Research System Deployed at Scale

Recent automated research systems show that language-model agents can generate hypotheses, run experiments, and write complete manuscripts, but most evidence still comes from selected examples, human-framed topics, or a few pre-defined research tasks. We present FARS (Fully Automated Research System), a fully automated AI-for-AI research system designed to operate across research topics at scale. FARS autonomously generates and advances projects through ideation, planning, experimentation, and writing, using stage-specific agents coordinated through a shared workspace that records proposals, code, logs, results, and manuscripts. In its first public deployment, FARS produced 166 complete research papers spanning 67 fine-grained AI/ML topics while preserving intermediate artifacts as an auditable corpus rather than a curated set of successes. We evaluate this corpus with 282 structured reviews from volunteer reviewers covering 140 papers, including overall ratings, sub-scores, integrity checks, and LLM-use disclosure. The reviews indicate that FARS can produce review-worthy and occasionally strong AI/ML research artifacts in a large-scale public deployment, while also exposing recurring failure modes in narrow experimental scope, methodological limitations, and integrity issues.

cs.AI↗

arXiv2026

Output-Space Allocation Costs for Calibration-Guided LLM Compression: An Empirical Study

Training-free compression methods for large language models (LLMs) often use calibration data to guide compression decisions. ROCKET, a recent method combining sparse-dictionary factorization with multi-choice knapsack problem (MCKP) allocation, derives its per-layer factorization from an output reconstruction objective but uses weight-space Frobenius error as the MCKP allocation cost. We investigate whether aligning the allocation cost with the output-space objective improves compressed model fidelity. On Qwen3-8B at 50\% compression, our ROCKET-ActCost achieves +0.8 percentage points higher average accuracy across 8 zero-shot benchmarks (53.1\% vs 52.3\%), but increases WikiText perplexity by 16\% (61.46 vs 52.98). This accuracy-perplexity tradeoff reveals that different allocation objectives favor different downstream metrics. The high correlation ($>$0.99) between weight-space and output-space errors limits allocation divergence, explaining the modest effect size. On Llama-3.2-1B at 20\% compression, the two methods produce near-identical results (53.3\% vs 53.5\% accuracy, 14.45 vs 14.66 PPL), suggesting that the effect of the cost function is minor at lower compression ratios.

cs.CL↗

arXiv2026

NLL-Guided Full-Attention Layer Selection for Training-Free Sliding-Window Adaptation

Hybrid attention models that mix full and sliding-window attention across layers offer a promising approach to efficient long-context inference, but the critical question of \emph{which layers} should retain full attention remains unsolved. Existing methods use either fixed periodic patterns or attention-based heuristics that may not capture what matters for downstream accuracy. We propose NLL-guided layer selection, a training-free method that directly measures each layer's importance by computing the negative log-likelihood degradation on answer tokens when that layer uses sliding-window instead of full attention. On LongMemEval with Qwen3-4B, our method achieves 64.6\% accuracy using only 1/4 full-attention layers, matching the 1/2-FA periodic baseline (65.0\%) while halving the computational budget. NLL-guided selection outperforms the SWAA-reported periodic 1/4-FA baseline by 10.4 percentage points and a matched LightTransfer-style baseline by 26.4 percentage points. De-confounding analysis shows the signal is consistent with long-range attention needs rather than generic layer sensitivity. The method requires only $\sim$15 minutes of one-time calibration, advancing the efficiency-accuracy Pareto frontier for long-context LLM deployment.

cs.CL↗

arXiv2026

Position Bias Correction is Insufficient for One-Pass Attention Sorting

Long-context language models suffer from position bias, where information in middle positions is underutilized. Attention Sorting addresses this by iteratively reordering documents based on attention patterns, but its multiple sort-and-generate cycles increase deployment cost. We hypothesize that position bias is the primary bottleneck and propose Debiased One-Pass Attention Sorting, which estimates a per-prompt position-bias curve from the low-attention majority of documents and uses it to correct raw attention scores (via subtraction or division) to enable single-pass sorting. Our experiments on two models refute this hypothesis in the tested setting: on LLaMA-2-7B-32K-Instruct, debiasing produces identical results to uncalibrated single-pass sorting (94.83\% containment accuracy), while on YaRN-Llama-2-7b-64k, debiasing improves accuracy by 8.67 percentage points but remains 14.84pp behind iterative sorting, closing only 37\% of the gap. These results suggest that position-bias correction is insufficient to match iterative sorting, and that repeated reordering provides additional benefits beyond bias correction.

cs.CL↗

arXiv2026

High-pressure phase stability and superconductivity in La-Zr-H hydrides

Hydrogen-rich ternary hydrides are promising candidates for high-Tc superconductivity at megabar pressures, yet their chemical space is vast and largely unexplored. Combining evolutionary structure searches with first-principles calculations, we comprehensively investigate the La-Zr-H ternary system in the 150-300 GPa pressure range. Zero-point energy-corrected convex hull analysis identifies multiple stable superconducting phases, including R3m-Zr2H17 at 300 GPa and P6/mmm-LaZr2H24 at 200 GPa, both of which are thermodynamically and dynamically stable and exhibit strong electron-phonon coupling. Solution of the Eliashberg equations predicts high superconducting transition temperatures of Tc = 209 K for R3m-Zr2H17 at 300 GPa and Tc = 202 K for P6/mmm-LaZr2H24 at 200 GPa. In addition to these stable phases, we identify a high-symmetry metastable compound, P6m2-LaZrH18, which lies just 0.027 eV/atom above the convex hull yet remains dynamically stable and exhibits a high predicted Tc of 206 K at 300 GPa. We find that, across all phases, the elevated Tc correlates with the high-symmetry structure with dense hydrogen cages, favorable electron counts per hydrogen, and a large hydrogen-derived density of states at the Fermi level. Finally, a random- forest machine learning model, trained on diverse hydrides superconductivity data, reproduces these structure-property trends across predicted structures, enabling to identify potential hydrides with high predicted Tc for targeted follow-up calculations and future high-pressure experiments.

cond-mat.mtrl-sci↗

arXiv2025

Nonlinear optical response in kagome lattice with inversion symmetry breaking

The kagome lattice is a fundamental model structure in condensed matter physics and materials science featuring symmetry-protected flat bands, saddle points, and Dirac points. This structure has emerged as an ideal platform for exploring various quantum physics. By combining effective model analysis and first-principles calculations, we propose that the synergy among inversion symmetry breaking, flat bands, and saddle point-related van Hove singularities within the kagome lattice holds significant potential for generating strong second-order nonlinear optical response. This property provides an inspiring insight into the practical application of the kagome-like materials, which is helpful for a comprehensive understanding of kagome lattice-related physics. Moreover, this work offers an alternative approach for designing materials with strong a second-order nonlinear optical response.

physics.optics↗

arXiv2025

Making Large Language Models Better Reasoners with Orchestrated Streaming Experiences

Large language models (LLMs) can perform complex reasoning by generating intermediate thoughts under zero-shot or few-shot settings. However, zero-shot prompting always encounters low performance, and the superior performance of few-shot prompting hinges on the manual-crafted demonstrations. In this paper, we present RoSE (Reasoning with Orchestrated Streaming Experiences), a general framework for solving reasoning tasks that can self-improve without complex external efforts. To enable RoSE, we describe an architecture that extends an LLM to store all answered questions and their thoughts in a streaming experience pool then orchestrates helpful questions from the pool to assist in answering new questions. To set up a question-aware orchestration mechanism, RoSE first calculates the similarity of each question in the pool with a new test question. Since the solution to each answered question is not always correct, RoSE will sort the questions according to their similarity with the new question, and then uniformly divide them into multiple buckets. It finally extracts one question from each bucket to make these extracted questions more diverse. To make these extracted questions help RoSE answer new questions as much as possible, we introduce two other attributes of uncertainty and complexity for each question. RoSE will preferentially select the questions with low uncertainty and high complexity from each bucket. We evaluate the versatility of RoSE in various reasoning tasks, LLMs, and CoT methods.

cs.CL↗

arXiv2024

Magnetism and weak electronic correlations in Kagome metal ScV$_6$Sn$_6$

As one class of typical quantum materials, Kagome metals in $A$V$_3$Sb$_5$($A$ = K, Rb, Cs) have attracted extensive attentions due to their interesting physical properties and different quantum phases of charge density wave (CDW), superconductivity and nontrivial topology. Recently, a new CDW phase in ScV$_6$Sn$_6$ was experimentally observed and inspired a wide study of the mechanism of driving force. To have a clear understanding of the correlation effect in the CDW phase in ScV$_6$Sn$_6$, we performed a systematic density functional theory plus dynamical mean field theory (DFT + DMFT) calculations. The resulting static local spin susceptibility is nearly independent of temperature, indicating the absence of local moment on atom V, in full agreement with experimental measurements. The mass enhancements of quasiparticles and bandwidth renormalizations near the Fermi level show a weak correlation strength in ScV$_6$Sn$_6$. In addition, the comparable mass enhancements of quasiparticles in ScV$_6$Sn$_6$ with CDW order and YV$_6$Sn$_6$ without CDW phase suggests that electronic correlations corresponding to Fermi surface nesting do not play the dominant role in the formation of CDW order in ScV$_6$Sn$_6$.

cond-mat.str-el↗

arXiv2024

DetectiveQA: Evaluating Long-Context Reasoning on Detective Novels

Recently, significant efforts have been devoted to enhancing the long-context capabilities of Large Language Models (LLMs), particularly in long-context reasoning. To facilitate this research, we propose \textbf{DetectiveQA}, a dataset specifically designed for narrative reasoning within long contexts. We leverage detective novels, averaging over 100k tokens, to create a dataset containing 1200 human-annotated questions in both Chinese and English, each paired with corresponding reference reasoning steps. Furthermore, we introduce a step-wise reasoning metric, which enhances the evaluation of LLMs' reasoning processes. We validate our approach and evaluate the mainstream LLMs, including GPT-4, Claude, and LLaMA, revealing persistent long-context reasoning challenges and demonstrating their evidence-retrieval challenges. Our findings offer valuable insights into the study of long-context reasoning and lay the base for more rigorous evaluations.

cs.CL↗

arXiv2024

Molecular Beam Epitaxy Growth and Doping Modulation of Topological Semimetal NiTe$_2$

In this study, high-quality thin films of the topological semimetal phase NiTe$_2$ were prepared using molecular beam epitaxy (MBE) technique, confirmed through X-ray diffraction with pronounced Laue oscillations. Electrical transport experiments reveal that thick films have properties similar to bulk materials. By employing co-deposition, we introduced either magnetic or non-magnetic elements during the growth of thinner films, significantly altering their electrical properties. Notably, magnetic element Cr induces long-range ferromagnetic ordering, leading to the observation of significant anomalous Hall effect in NiTe2 thin films. The Hall conductivity remains nearly constant well below the Curie temperature, indicating the correlation with the intrinsic topological nature of the band structure. Theoretical first principles band calculations support the generation of the Weyl semimetal state in the material through magnetic doping. These findings pave the way for exploring more magnetic Weyl semimetal materials and related low-dimensional quantum devices based on topological semimetal materials.

cond-mat.mtrl-sci↗

arXiv2024

Prediction of dual quantum spin Hall insulator in NbIrTe$_4$ monolayer

Dual quantum spin Hall insulator (QSHI) is a newly discovered topological state in the 2D material TaIrTe$_4$, exhibiting both a traditional $Z_2$ band gap at charge neutrality point and a van Hove singularity (VHS) induced correlated $Z_2$ band gap with weak doping. Inspired by the recent progress in theoretical understanding and experimental measurements, we predicted a promising dual QSHI in the counterpart material of the NbIrTe4 monolayer by first-principles calculations. In addition to the well-known band inversion at the charge neutrality point, two new band inversions were found after CDW phase transition when the chemical potential is near the VHS, one direct and one indirect $Z_2$ band gap. The VHS-induced non-trivial band gap is around 10 meV, much larger than that from TaIrTe$_4$. Furthermore, since the new generated band gap is mainly dominated by the $4d$ orbitals of Nb, electronic correlation effects should be relatively stronger in NbIrTe$_4$ as compared to TaIrTe$_4$. Therefore, the dual QSHI state in the NbIrTe$_4$ monolayer is expected to be a good platform for investigating the interplay between topology and correlation effects.

cond-mat.str-el↗

arXiv2024

Flames: Benchmarking Value Alignment of LLMs in Chinese

The widespread adoption of large language models (LLMs) across various regions underscores the urgent need to evaluate their alignment with human values. Current benchmarks, however, fall short of effectively uncovering safety vulnerabilities in LLMs. Despite numerous models achieving high scores and 'topping the chart' in these evaluations, there is still a significant gap in LLMs' deeper alignment with human values and achieving genuine harmlessness. To this end, this paper proposes a value alignment benchmark named Flames, which encompasses both common harmlessness principles and a unique morality dimension that integrates specific Chinese values such as harmony. Accordingly, we carefully design adversarial prompts that incorporate complex scenarios and jailbreaking methods, mostly with implicit malice. By prompting 17 mainstream LLMs, we obtain model responses and rigorously annotate them for detailed evaluation. Our findings indicate that all the evaluated LLMs demonstrate relatively poor performance on Flames, particularly in the safety and fairness dimensions. We also develop a lightweight specified scorer capable of scoring LLMs across multiple dimensions to efficiently evaluate new models on the benchmark. The complexity of Flames has far exceeded existing benchmarks, setting a new challenge for contemporary LLMs and highlighting the need for further alignment of LLMs. Our benchmark is publicly available at https://github.com/AIFlames/Flames.

cs.CL↗

arXiv2024

Switchable quantized signal between longitudinal conductance and Hall conductance in dual quantum spin Hall insulator TaIrTe$_4$

Topological insulating states in two-dimensional (2D) materials are ideal systems to study different types of quantized response signals due to their in gap metallic states. Very recently, the quantum spin Hall (QSH) effect was discovered in monolayer $\text{TaIrTe}_4$ via the observation of quantized longitudinal conductance that rarely exists in other 2D topological insulators. The non-trivial $Z_2$ topological charges can exist at both charge neutrality point and the van Hove singularity point with correlation effect induced band gap. Based on this model 2D material, we studied the switch of quantized signals between longitudinal conductance and transversal Hall conductance via tuning external magnetic field. In $Z_2$ topological phase of monolayer $\text{TaIrTe}_4$, the zero Chern number can be understood as 1-1=0 from the double band inversion from spin-up and spin-down channels. After applying a magnetic field perpendicular to the plane, the Zeeman split changes the band order for one branch of the band inversion from spin-up and spin-down channels, along with a sign charge of the Berry phase. Then the net Chern number of 1-1=0 is tuned to 1+1=2 or -1-1=-2 depending on the orientation of the magnetic field. The quantized signal not only provides another effective method for the verification of topological state in monolayer $\text{TaIrTe}_4$, but also offers a strategy for the utilization of the new quantum topological states based on switchable quantized responses.

cond-mat.mes-hall↗

arXiv2024

Can AI Assistants Know What They Don't Know?

Recently, AI assistants based on large language models (LLMs) show surprising performance in many tasks, such as dialogue, solving math problems, writing code, and using tools. Although LLMs possess intensive world knowledge, they still make factual errors when facing some knowledge intensive tasks, like open-domain question answering. These untruthful responses from the AI assistant may cause significant risks in practical applications. We believe that an AI assistant's refusal to answer questions it does not know is a crucial method for reducing hallucinations and making the assistant truthful. Therefore, in this paper, we ask the question "Can AI assistants know what they don't know and express them through natural language?" To answer this question, we construct a model-specific "I don't know" (Idk) dataset for an assistant, which contains its known and unknown questions, based on existing open-domain question answering datasets. Then we align the assistant with its corresponding Idk dataset and observe whether it can refuse to answer its unknown questions after alignment. Experimental results show that after alignment with Idk datasets, the assistant can refuse to answer most its unknown questions. For questions they attempt to answer, the accuracy is significantly higher than before the alignment.

cs.CL↗

arXiv2023

Evaluating Hallucinations in Chinese Large Language Models

In this paper, we establish a benchmark named HalluQA (Chinese Hallucination Question-Answering) to measure the hallucination phenomenon in Chinese large language models. HalluQA contains 450 meticulously designed adversarial questions, spanning multiple domains, and takes into account Chinese historical culture, customs, and social phenomena. During the construction of HalluQA, we consider two types of hallucinations: imitative falsehoods and factual errors, and we construct adversarial samples based on GLM-130B and ChatGPT. For evaluation, we design an automated evaluation method using GPT-4 to judge whether a model output is hallucinated. We conduct extensive experiments on 24 large language models, including ERNIE-Bot, Baichuan2, ChatGLM, Qwen, SparkDesk and etc. Out of the 24 models, 18 achieved non-hallucination rates lower than 50%. This indicates that HalluQA is highly challenging. We analyze the primary types of hallucinations in different types of models and their causes. Additionally, we discuss which types of hallucinations should be prioritized for different types of models.

cs.CL↗

arXiv2023

The Rise and Potential of Large Language Model Based Agents: A Survey

For a long time, humanity has pursued artificial intelligence (AI) equivalent to or surpassing the human level, with AI agents considered a promising vehicle for this pursuit. AI agents are artificial entities that sense their environment, make decisions, and take actions. Many efforts have been made to develop intelligent agents, but they mainly focus on advancement in algorithms or training strategies to enhance specific capabilities or performance on particular tasks. Actually, what the community lacks is a general and powerful model to serve as a starting point for designing AI agents that can adapt to diverse scenarios. Due to the versatile capabilities they demonstrate, large language models (LLMs) are regarded as potential sparks for Artificial General Intelligence (AGI), offering hope for building general AI agents. Many researchers have leveraged LLMs as the foundation to build AI agents and have achieved significant progress. In this paper, we perform a comprehensive survey on LLM-based agents. We start by tracing the concept of agents from its philosophical origins to its development in AI, and explain why LLMs are suitable foundations for agents. Building upon this, we present a general framework for LLM-based agents, comprising three main components: brain, perception, and action, and the framework can be tailored for different applications. Subsequently, we explore the extensive applications of LLM-based agents in three aspects: single-agent scenarios, multi-agent scenarios, and human-agent cooperation. Following this, we delve into agent societies, exploring the behavior and personality of LLM-based agents, the social phenomena that emerge from an agent society, and the insights they offer for human society. Finally, we discuss several key topics and open problems within the field. A repository for the related papers at https://github.com/WooooDyy/LLM-Agent-Paper-List.

cs.AI↗

arXiv2023

Low-Parameter Federated Learning with Large Language Models

We study few-shot Natural Language Understanding (NLU) tasks with Large Language Models (LLMs) in federated learning (FL) scenarios. It is a challenging task due to limited labeled data and communication capacities in FL, especially with mobile devices. Recent studies show LLMs can be prompted to perform few-shot NLU tasks like sentiment analysis and arithmetic reasoning. However, the huge sizes of LLMs result in high computation and communication costs, making classical FL schemes impractical. To address these challenges, we propose Low-Parameter Federated Learning (LP-FL). LP-FL combines few-shot prompt learning from LLMs with efficient communication and federating techniques. Our approach enables federated clients to assign soft labels to unlabeled data using gradually learned knowledge from the global model. Through iterative soft-label assigning, we continually expand the labeled set during the FL process. Additionally, to reduce computation and communication costs, LP-FL utilizes the Low-Rank Adaptation (LoRA) technique for compact learnable parameter construction, efficient local model fine-tuning, and affordable global model federation. LP-FL consistently outperforms Full-Parameter Federated Learning (FP-FL) in sentiment analysis tasks across various FL settings. Its resistance to overfitting allows LP-FL to equal or surpass centralized training in few-shot scenarios.

cs.DC↗

arXiv2023

OmniForce: On Human-Centered, Large Model Empowered and Cloud-Edge Collaborative AutoML System

Automated machine learning (AutoML) seeks to build ML models with minimal human effort. While considerable research has been conducted in the area of AutoML in general, aiming to take humans out of the loop when building artificial intelligence (AI) applications, scant literature has focused on how AutoML works well in open-environment scenarios such as the process of training and updating large models, industrial supply chains or the industrial metaverse, where people often face open-loop problems during the search process: they must continuously collect data, update data and models, satisfy the requirements of the development and deployment environment, support massive devices, modify evaluation metrics, etc. Addressing the open-environment issue with pure data-driven approaches requires considerable data, computing resources, and effort from dedicated data engineers, making current AutoML systems and platforms inefficient and computationally intractable. Human-computer interaction is a practical and feasible way to tackle the problem of open-environment AI. In this paper, we introduce OmniForce, a human-centered AutoML (HAML) system that yields both human-assisted ML and ML-assisted human techniques, to put an AutoML system into practice and build adaptive AI in open-environment scenarios. Specifically, we present OmniForce in terms of ML version management; pipeline-driven development and deployment collaborations; a flexible search strategy framework; and widely provisioned and crowdsourced application algorithms, including large models. Furthermore, the (large) models constructed by OmniForce can be automatically turned into remote services in a few minutes; this process is dubbed model as a service (MaaS). Experimental results obtained in multiple search spaces and real-world use cases demonstrate the efficacy and efficiency of OmniForce.

cs.LG↗