arXiv subjects

Bo Zhao

Publications and source records attributed to Bo Zhao.

At least 127 records · Page 7Linked to original sources

arXiv2025

EARL: Efficient Agentic Reinforcement Learning Systems for Large Language Models

Reinforcement learning (RL) has become a pivotal component of large language model (LLM) post-training, and agentic RL extends this paradigm to operate as agents through multi-turn interaction and tool use. Scaling such systems exposes two practical bottlenecks: (1) context length grows rapidly during training, inflating memory usage and latency, and triggering out-of-memory (OOM) failures; and (2) intermediate tensors accumulate with context length, making cross-device data movement a major system bottleneck. We present EARL, a scalable system for efficient agentic RL. EARL designs a parallelism selector that dynamically adapts model and training parallelism across RL stages based on sequence length and system load, and a data dispatcher that performs layout-aware, decentralized exchange of intermediate data batches. Together, these components increase throughput, reduce long-context failures, and enable stable large-scale training of agentic LLMs without relying on hard limits or penalties of context length.

cs.DC↗

arXiv2025

QUASAR: Quantum Assembly Code Generation Using Tool-Augmented LLMs via Agentic RL

Designing and optimizing task-specific quantum circuits are crucial to leverage the advantage of quantum computing. Recent large language model (LLM)-based quantum circuit generation has emerged as a promising automatic solution. However, the fundamental challenges remain unaddressed: (i) parameterized quantum gates require precise numerical values for optimal performance, which also depend on multiple aspects, including the number of quantum gates, their parameters, and the layout/depth of the circuits. (ii) LLMs often generate low-quality or incorrect quantum circuits due to the lack of quantum domain-specific knowledge. We propose QUASAR, an agentic reinforcement learning (RL) framework for quantum circuits generation and optimization based on tool-augmented LLMs. To align the LLM with quantum-specific knowledge and improve the generated quantum circuits, QUASAR designs (i) a quantum circuit verification approach with external quantum simulators and (ii) a sophisticated hierarchical reward mechanism in RL training. Extensive evaluation shows improvements in both syntax and semantic performance of the generated quantum circuits. When augmenting a 4B LLM, QUASAR has achieved the validity of 99.31% in Pass@1 and 100% in Pass@10, outperforming industrial LLMs of GPT-4o, GPT-5 and DeepSeek-V3 and several supervised-fine-tuning (SFT)-only and RL-only baselines.

cs.AI↗

arXiv2025

Zero-field Anomalous Hall Effect in Bulk Single Crystal Mn3Ir

The L1_2-phase non-collinear antiferromagnet (AFM) Mn_3Ir has emerged as a pioneering platform for realizing the zero-field anomalous Hall effect (AHE), thereby catalyzing rapid advances in antiferromagnetic spintronics. Despite its significant potential, experimental investigations of the intrinsic magnetic and electronic properties of Mn_3Ir have been greatly hindered by the formidable challenges in growing bulk single crystals. Here, we report the growth of stoichiometric Mn_3Ir bulk single crystals and their characterization in terms of magnetization and the AHE. Using a high-throughput flux method, we obtained (111)-oriented hexagonal Mn_3Ir single crystals. A small AHE signal was detected, which we attribute to the coexistence of A- and B-type antiferromagnetic domains that mutually cancel the net AHE response. Our results reveal key aspects of the intrinsic magnetic properties and AHE in bulk Mn_3Ir, providing a critical material platform for the development of advanced spintronic devices.

cond-mat.mtrl-sci↗

arXiv2025

DD-Ranking: Rethinking the Evaluation of Dataset Distillation

In recent years, dataset distillation has provided a reliable solution for data compression, where models trained on the resulting smaller synthetic datasets achieve performance comparable to those trained on the original datasets. To further improve the performance of synthetic datasets, various training pipelines and optimization objectives have been proposed, greatly advancing the field of dataset distillation. Recent decoupled dataset distillation methods introduce soft labels and stronger data augmentation during the post-evaluation phase and scale dataset distillation up to larger datasets (e.g., ImageNet-1K). However, this raises a question: Is accuracy still a reliable metric to fairly evaluate dataset distillation methods? Our empirical findings suggest that the performance improvements of these methods often stem from additional techniques rather than the inherent quality of the images themselves, with even randomly sampled images achieving superior results. Such misaligned evaluation settings severely hinder the development of DD. Therefore, we propose DD-Ranking, a unified evaluation framework, along with new general evaluation metrics to uncover the true performance improvements achieved by different methods. By refocusing on the actual information enhancement of distilled datasets, DD-Ranking provides a more comprehensive and fair evaluation standard for future research advancements.

cs.CV↗

arXiv2025

BDPM: A Machine Learning-Based Feature Extractor for Parkinson's Disease Classification via Gut Microbiota Analysis

Background: Parkinson's disease remains a major neurodegenerative disorder with high misdiagnosis rates, primarily due to reliance on clinical rating scales. Recent studies have demonstrated a strong association between gut microbiota and Parkinson's disease, suggesting that microbial composition may serve as a promising biomarker. Although deep learning models based ongut microbiota show potential for early prediction, most approaches rely on single classifiers and often overlook inter-strain correlations or temporal dynamics. Therefore, there is an urgent need for more robust feature extraction methods tailored to microbiome data. Methods: We proposed BDPM (A Machine Learning-Based Feature Extractor for Parkinson's Disease Classification via Gut Microbiota Analysis). First, we collected gut microbiota profiles from 39 Parkinson's patients and their healthy spouses to identify differentially abundant taxa. Second, we developed an innovative feature selection framework named RFRE (Random Forest combined with Recursive Feature Elimination), integrating ecological knowledge to enhance biological interpretability. Finally, we designed a hybrid classification model to capture temporal and spatial patterns in microbiome data.

cs.AI↗

arXiv2025

Agent-Q: Fine-Tuning Large Language Models for Quantum Circuit Generation and Optimization

Large language models (LLMs) have achieved remarkable outcomes in complex problems, including math, coding, and analyzing large amounts of scientific reports. Yet, few works have explored the potential of LLMs in quantum computing. The most challenging problem is to leverage LLMs to automatically generate quantum circuits at a large scale. Fundamentally, the existing pre-trained LLMs lack the knowledge of quantum circuits. In this paper, we address this challenge by fine-tuning LLMs and injecting the domain-specific knowledge of quantum computing. We describe Agent-Q, an LLM fine-tuning system to generate and optimize quantum circuits. In particular, Agent-Q implements the mechanisms to generate training data sets and constructs an end-to-end pipeline to fine-tune pre-trained LLMs to generate parameterized quantum circuits for various optimization problems. Agent-Q provides 14,000 quantum circuits covering a large spectrum of the quantum optimization landscape: 12 optimization problem instances and their optimized QAOA, VQE, and adaptive VQE circuits. Based thereon, Agent-Q fine-tunes LLMs and constructs syntactically correct parametrized quantum circuits in OpenQASM 3.0. We have evaluated the quality of the LLM-generated circuits and parameters by comparing them to the optimized expectation values and distributions. Experimental results show superior performance of Agent-Q, compared to several state-of-the-art LLMs and better parameters than random. Agent-Q can be integrated into an agentic workflow, and the generated parametrized circuits with initial parameters can be used as a starting point for further optimization, e.g., as templates in quantum machine learning and as benchmarks for compilers and hardware.

quant-ph↗

arXiv2025

Higher-Order Portfolio Optimization with Quantum Approximate Optimization Algorithm

Portfolio optimization is one of the most studied optimization problems at the intersection of quantum computing and finance. In this work, we develop the first quantum formulation for a portfolio optimization problem with higher-order moments, skewness and kurtosis. Including higher-order moments leads to more detailed modeling of portfolio return distributions. Portfolio optimization with higher-order moments has been studied in classical portfolio optimization approaches but with limited exploration within quantum formulations. In the context of quantum optimization, higher-order moments generate higher-order terms in the cost Hamiltonian. Thus, instead of obtaining a quadratic unconstrained binary optimization problem, we obtain a higher-order unconstrained binary optimization (HUBO) problem, which has a natural formulation as a parametrized circuit. Additionally, we employ realistic integer variable encoding and a capital-based budget constraint. We consider the classical continuous variable solution with integer programming-based discretization to be the computationally efficient classical baseline for the problem. Our extensive experimental evaluation of 100 portfolio optimization problems shows that the solutions to the HUBO formulation often correspond to better portfolio allocations than the classical baseline. This is a promising result for those who want to perform computationally challenging portfolio optimization on quantum hardware, as portfolio optimization with higher moments is classically complex. Moreover, the experimental evaluation studies QAOA's performance with higher-order terms in this practically relevant problem.

quant-ph↗

arXiv2025

Unbiased Reasoning for Knowledge-Intensive Tasks in Large Language Models via Conditional Front-Door Adjustment

Large Language Models (LLMs) have shown impressive capabilities in natural language processing but still struggle to perform well on knowledge-intensive tasks that require deep reasoning and the integration of external knowledge. Although methods such as Retrieval-Augmented Generation (RAG) and Chain-of-Thought (CoT) have been proposed to enhance LLMs with external knowledge, they still suffer from internal bias in LLMs, which often leads to incorrect answers. In this paper, we propose a novel causal prompting framework, Conditional Front-Door Prompting (CFD-Prompting), which enables the unbiased estimation of the causal effect between the query and the answer, conditional on external knowledge, while mitigating internal bias. By constructing counterfactual external knowledge, our framework simulates how the query behaves under varying contexts, addressing the challenge that the query is fixed and is not amenable to direct causal intervention. Compared to the standard front-door adjustment, the conditional variant operates under weaker assumptions, enhancing both robustness and generalisability of the reasoning process. Extensive experiments across multiple LLMs and benchmark datasets demonstrate that CFD-Prompting significantly outperforms existing baselines in both accuracy and robustness.

cs.CL↗

arXiv2025

Understanding Data Influence with Differential Approximation

Data plays a pivotal role in the groundbreaking advancements in artificial intelligence. The quantitative analysis of data significantly contributes to model training, enhancing both the efficiency and quality of data utilization. However, existing data analysis tools often lag in accuracy. For instance, many of these tools even assume that the loss function of neural networks is convex. These limitations make it challenging to implement current methods effectively. In this paper, we introduce a new formulation to approximate a sample's influence by accumulating the differences in influence between consecutive learning steps, which we term Diff-In. Specifically, we formulate the sample-wise influence as the cumulative sum of its changes/differences across successive training iterations. By employing second-order approximations, we approximate these difference terms with high accuracy while eliminating the need for model convexity required by existing methods. Despite being a second-order method, Diff-In maintains computational complexity comparable to that of first-order methods and remains scalable. This efficiency is achieved by computing the product of the Hessian and gradient, which can be efficiently approximated using finite differences of first-order gradients. We assess the approximation accuracy of Diff-In both theoretically and empirically. Our theoretical analysis demonstrates that Diff-In achieves significantly lower approximation error compared to existing influence estimators. Extensive experiments further confirm its superior performance across multiple benchmark datasets in three data-centric tasks: data cleaning, data deletion, and coreset selection. Notably, our experiments on data pruning for large-scale vision-language pre-training show that Diff-In can scale to millions of data points and outperforms strong baselines.

cs.LG↗

arXiv2025

EarthSynth: Generating Informative Earth Observation with Diffusion Models

Remote sensing image (RSI) interpretation typically faces challenges due to the scarcity of labeled data, which limits the performance of RSI interpretation tasks. To tackle this challenge, we propose EarthSynth, a diffusion-based generative foundation model that enables synthesizing multi-category, cross-satellite labeled Earth observation for downstream RSI interpretation tasks. To the best of our knowledge, EarthSynth is the first to explore multi-task generation for remote sensing, tackling the challenge of limited generalization in task-oriented synthesis for RSI interpretation. EarthSynth, trained on the EarthSynth-180K dataset, employs the Counterfactual Composition training strategy with a three-dimensional batch-sample selection mechanism to improve training data diversity and enhance category control. Furthermore, a rule-based method of R-Filter is proposed to filter more informative synthetic data for downstream tasks. We evaluate our EarthSynth on scene classification, object detection, and semantic segmentation in open-world scenarios. There are significant improvements in open-vocabulary understanding tasks, offering a practical solution for advancing RSI interpretation.

cs.CV↗

arXiv2025

Is Green Enough? A Remote Sensing Assessment of Environmental Impacts and Green Commitments at Beijing Daxing International Airport

Beijing Daxing International Airport has been promoted as a model of green infrastructure under China's ecological modernization agenda. Featuring energy-efficient design, renewable energy systems, and smart environmental controls, the airport embodies multiple green commitments. This study evaluates its environmental outcomes using multi-source remote sensing data -- including NDVI, NDBI, Land Surface Temperature (LST), VIIRS night-time lights, and PM2.5 -- from 2014 to 2019. Through spatial and temporal comparisons, we assess landscape-level changes during and after construction. Findings indicate partial gains from green initiatives but also reveal substantial vegetation loss, increased built-up surfaces, and intensified surface temperatures. The results suggest a gap between sustainable design and ecological impact. We propose a remote-sensing-based framework for evaluating future infrastructure projects, emphasizing the need for spatially explicit, independent monitoring to ensure environmental accountability.

physics.soc-ph↗

arXiv2025

BOOD: Boundary-based Out-Of-Distribution Data Generation

Harnessing the power of diffusion models to synthesize auxiliary training data based on latent space features has proven effective in enhancing out-of-distribution (OOD) detection performance. However, extracting effective features outside the in-distribution (ID) boundary in latent space remains challenging due to the difficulty of identifying decision boundaries between classes. This paper proposes a novel framework called Boundary-based Out-Of-Distribution data generation (BOOD), which synthesizes high-quality OOD features and generates human-compatible outlier images using diffusion models. BOOD first learns a text-conditioned latent feature space from the ID dataset, selects ID features closest to the decision boundary, and perturbs them to cross the decision boundary to form OOD features. These synthetic OOD features are then decoded into images in pixel space by a diffusion model. Compared to previous works, BOOD provides a more training efficient strategy for synthesizing informative OOD features, facilitating clearer distinctions between ID and OOD data. Extensive experimental results on common benchmarks demonstrate that BOOD surpasses the state-of-the-art method significantly, achieving a 29.64% decrease in average FPR95 (40.31% vs. 10.67%) and a 7.27% improvement in average AUROC (90.15% vs. 97.42%) on the CIFAR-100 dataset.

cs.LG↗

arXiv2025

Optimizing $α''$-Fe$_{16}$N$_2$ as permanent magnet via alloying

Based on systematic first-principles calculations, we investigate the effects of 27 alloying elements on the intrinsic magnetic properties of Fe$_{16}$N$_2$, in order to further optimize its properties for permanent magnet applications. Analysis on the thermodynamic stabilities based on formation energy and distance to the convex hull reveals that 20 elements can be substituted into Fe$_{16}$N$_2$, where there is no strong site-preference upon doping. It is observed that all alloying elements can essentially reduce the saturation magnetization, whereas the magnetic anisotropy can be significantly modified. In terms of the Boltzmann-average intrinsic properties, we identify 8 elements as interesting candidates, with Co, Mo, and W as the most promising cases for further experimental validations.

cond-mat.mtrl-sci↗

arXiv2025

AU-LLM: Micro-Expression Action Unit Detection via Enhanced LLM-Based Feature Fusion

The detection of micro-expression Action Units (AUs) is a formidable challenge in affective computing, pivotal for decoding subtle, involuntary human emotions. While Large Language Models (LLMs) demonstrate profound reasoning abilities, their application to the fine-grained, low-intensity domain of micro-expression AU detection remains unexplored. This paper pioneers this direction by introducing \textbf{AU-LLM}, a novel framework that for the first time uses LLM to detect AUs in micro-expression datasets with subtle intensities and the scarcity of data. We specifically address the critical vision-language semantic gap, the \textbf{Enhanced Fusion Projector (EFP)}. The EFP employs a Multi-Layer Perceptron (MLP) to intelligently fuse mid-level (local texture) and high-level (global semantics) visual features from a specialized 3D-CNN backbone into a single, information-dense token. This compact representation effectively empowers the LLM to perform nuanced reasoning over subtle facial muscle movements.Through extensive evaluations on the benchmark CASME II and SAMM datasets, including stringent Leave-One-Subject-Out (LOSO) and cross-domain protocols, AU-LLM establishes a new state-of-the-art, validating the significant potential and robustness of LLM-based reasoning for micro-expression analysis. The codes are available at https://github.com/ZS-liu-JLU/AU-LLMs.

cs.CV↗

arXiv2025

Megrez2 Technical Report

We present Megrez2, a novel lightweight and high-performance language model architecture optimized for device native deployment. Megrez2 introduces a novel cross-layer expert sharing mechanism, which significantly reduces total parameter count by reusing expert modules across adjacent transformer layers while maintaining most of the model's capacity. It also incorporates pre-gated routing, enabling memory-efficient expert loading and faster inference. As the first instantiation of the Megrez2 architecture, we introduce the Megrez2-Preview model, which is pre-trained on a 5-trillion-token corpus and further enhanced through supervised fine-tuning and reinforcement learning with verifiable rewards. With only 3B activated and 7.5B stored parameters, Megrez2-Preview demonstrates competitive or superior performance compared to larger models on a wide range of tasks, including language understanding, instruction following, mathematical reasoning, and code generation. These results highlight the effectiveness of the Megrez2 architecture to achieve a balance between accuracy, efficiency, and deployability, making it a strong candidate for real-world, resource-constrained applications.

cs.CL↗

arXiv2025

STI-Bench: Are MLLMs Ready for Precise Spatial-Temporal World Understanding?

The use of Multimodal Large Language Models (MLLMs) as an end-to-end solution for Embodied AI and Autonomous Driving has become a prevailing trend. While MLLMs have been extensively studied for visual semantic understanding tasks, their ability to perform precise and quantitative spatial-temporal understanding in real-world applications remains largely unexamined, leading to uncertain prospects. To evaluate models' Spatial-Temporal Intelligence, we introduce STI-Bench, a benchmark designed to evaluate MLLMs' spatial-temporal understanding through challenging tasks such as estimating and predicting the appearance, pose, displacement, and motion of objects. Our benchmark encompasses a wide range of robot and vehicle operations across desktop, indoor, and outdoor scenarios. The extensive experiments reveals that the state-of-the-art MLLMs still struggle in real-world spatial-temporal understanding, especially in tasks requiring precise distance estimation and motion analysis.

cs.CV↗

arXiv2025

MMCR: Benchmarking Cross-Source Reasoning in Scientific Papers

Fully comprehending scientific papers by machines reflects a high level of Artificial General Intelligence, requiring the ability to reason across fragmented and heterogeneous sources of information, presenting a complex and practically significant challenge. While Vision-Language Models (VLMs) have made remarkable strides in various tasks, particularly those involving reasoning with evidence source from single image or text page, their ability to use cross-source information for reasoning remains an open problem. This work presents MMCR, a high-difficulty benchmark designed to evaluate VLMs' capacity for reasoning with cross-source information from scientific papers. The benchmark comprises 276 high-quality questions, meticulously annotated by humans across 7 subjects and 10 task types. Experiments with 18 VLMs demonstrate that cross-source reasoning presents a substantial challenge for existing models. Notably, even the top-performing model, GPT-4o, achieved only 48.55% overall accuracy, with only 20% accuracy in multi-table comprehension tasks, while the second-best model, Qwen2.5-VL-72B, reached 39.86% overall accuracy. Furthermore, we investigated the impact of the Chain-of-Thought (CoT) technique on cross-source reasoning and observed a detrimental effect on small models, whereas larger models demonstrated substantially enhanced performance. These results highlight the pressing need to develop VLMs capable of effectively utilizing cross-source information for reasoning.

cs.CL↗

arXiv2025

Task-Aware KV Compression For Cost-Effective Long Video Understanding

Long-video understanding (LVU) remains a severe challenge for existing multimodal large language models (MLLMs), primarily due to the prohibitive computational cost. Recent approaches have explored KV compression to mitigate this issue, but they often suffer from significant information loss at high compression ratios. In this paper, we introduce Video-X^2L, which flexibly preserves critical video information for each LVU task. Video-X^2L involves two key operations. The first one is called bi-level KV compression. During the MLLM's pre-filling stage, Video-X^2L generates two types of compressed KVs: low-compression KVs (L-KVs) to capture fine-grained video details and high-compression KVs (H-KVs) to offer compact video representations. The second one is called selective KV re-loading. During the MLLM's decoding stage, Video-X^2L selectively re-loads L-KVs for the most critical video chunks while using H-KVs for other less important ones. This allows the MLLM to fully utilize task-specific information while maintaining the overall compactness. Video-X^2L is simple yet effective: it is free from additional training and directly compatible with existing KV-compressible MLLMs. We evaluate Video-X^2L with a variety of popular LVU benchmarks, including VideoMME, MLVU, LongVideoBench, and VNBench. Our experiment result shows that Video-X^2L outperforms existing KV-compression methods by a huge advantage while substantially saving the computation cost.

cs.CV↗