arXiv subjects

Jiaxi Cui

Publications and source records attributed to Jiaxi Cui.

13 recordsLinked to original sources

arXiv2026

Correlation-enhanced metrology from scrambling dynamics in a solid-state spin system

Quantum information scrambling, the dispersal of local information into many-body degrees of freedom, provides a powerful mechanism for generating large-scale correlations and entanglement essential for quantum-enhanced metrology. However, experimentally verifying such quantum-enhanced metrology remains a demanding task. Here, we correlate thousands of spins by engineering chaotic scrambling dynamics in a solid-state nuclear spin system. By leveraging the newly developed scramblon theory, we reveal exponential scaling in both the quantum Fisher information and the signal response to a phase shift. The signal response achieves a correlation-enabled enhancement of $33(2)$ dB over uncorrelated spins. After accounting for signal loss due to imperfect time reversal in the readout stage, we obtain a total metrological gain of 18(1) dB with a phase sensitivity of 40(3) ${\mathrm{\mu rad}}$. Our results bridge quantum chaos with practical quantum metrology, establishing reversible scrambling dynamics as a powerful resource for precision measurements.

quant-ph↗

arXiv2025

Stability of spin dynamics for a spin-orbit coupled boson in a driven non-Hermitian double well

We study the stability of spin dynamics for a spin-orbit (SO) coupled boson held in a driven non-Hermitian double-well potential. Under high-frequency approximation, we analytically derive the Floquet states and complex Floquet quasienergies of the system and reveal a striking parity-dependent stability criterion: when the ratio of the Zeeman field strength to the driving frequency $\Omega/\omega$ is even, stable spin dynamics can be achieved for \emph{arbitrary} SO coupling strength. However, when $\Omega/\omega$ is odd, stability requires the SO coupling strength to be integer or half-integer values. Particularly, we find four types of stability boundary lines for non-zero bias field strength, in sharp contrast to the commonly observed stability regions. These results establish a tunable parity-governed mechanism for stabilizing spin dynamics in non-Hermitian cold atomic systems.

cond-mat.quant-gas↗

arXiv2024

AnyTaskTune: Advanced Domain-Specific Solutions through Task-Fine-Tuning

The pervasive deployment of Large Language Models-LLMs in various sectors often neglects the nuanced requirements of individuals and small organizations, who benefit more from models precisely tailored to their specific business contexts rather than those with broadly superior general capabilities. This work introduces \textbf{AnyTaskTune}, a novel fine-tuning methodology coined as \textbf{Task-Fine-Tune}, specifically developed to elevate model performance on a diverse array of domain-specific tasks. This method involves a meticulous process to identify and define targeted sub-tasks within a domain, followed by the creation of specialized enhancement datasets for fine-tuning, thereby optimizing task-specific model performance. We conducted comprehensive fine-tuning experiments not only in the legal domain for tasks such as keyword extraction and sentence prediction but across over twenty different sub-tasks derived from the domains of finance, healthcare, law, psychology, consumer services, and human resources. To substantiate our approach and facilitate community engagement, we will open-source these bilingual task datasets. Our findings demonstrate that models fine-tuned using the \textbf{Task-Fine-Tune} methodology not only achieve superior performance on these specific tasks but also significantly outperform models with higher general capabilities in their respective domains. Our work is publicly available at \url{https://github.com/PandaVT/DataTager}.

cs.CL↗

arXiv2024

ProLLaMA: A Protein Large Language Model for Multi-Task Protein Language Processing

Recent advances in Protein Language Models (PLMs) have transformed protein engineering, yet unlike their counterparts in Natural Language Processing (NLP), current PLMs exhibit a fundamental limitation: they excel in either Protein Language Understanding (PLU) or Protein Language Generation (PLG), but rarely both. This fragmentation hinders progress in protein engineering. To bridge this gap, we introduce ProLLaMA, a multitask protein language model enhanced by the Evolutionary Protein Generation Framework (EPGF). We construct a comprehensive instruction dataset containing approximately 13 million samples with over 11,000 superfamily annotations to facilitate better modeling of sequence-function landscapes. We leverage a two-stage training approach to develop ProLLaMA, a multitask LLM with protein domain expertise. Our EPGF addresses the mismatch between statistic language modeling and biological constraints through three innovations: a multi-dimensional interpretable scorer, hierarchical efficient decoding, and a probabilistic-biophysical joint selection mechanism. Extensive experiments demonstrate that ProLLaMA excels in both unconditional and controllable protein generation tasks, achieving superior structural quality metrics compared to existing PLMs. Additionally, ProLLaMA demonstrates strong understanding capabilities with a 67.1% exact match rate in superfamily prediction. EPGF significantly enhances the biological viability of generated sequences, as evidenced by improved biophysical scores (+4.3%) and structural metrics (+14.5%). The project is available at https://github.com/PKU-YuanGroup/ProLLaMA.

cs.CE↗

arXiv2023

Aurora:Activating Chinese chat capability for Mixtral-8x7B sparse Mixture-of-Experts through Instruction-Tuning

Existing research has demonstrated that refining large language models (LLMs) through the utilization of machine-generated instruction-following data empowers these models to exhibit impressive zero-shot capabilities for novel tasks, without requiring human-authored instructions. In this paper, we systematically investigate, preprocess, and integrate three Chinese instruction-following datasets with the aim of enhancing the Chinese conversational capabilities of Mixtral-8x7B sparse Mixture-of-Experts model. Through instruction fine-tuning on this carefully processed dataset, we successfully construct the Mixtral-8x7B sparse Mixture-of-Experts model named "Aurora." To assess the performance of Aurora, we utilize three widely recognized benchmark tests: C-Eval, MMLU, and CMMLU. Empirical studies validate the effectiveness of instruction fine-tuning applied to Mixtral-8x7B sparse Mixture-of-Experts model. This work is pioneering in the execution of instruction fine-tuning on a sparse expert-mixed model, marking a significant breakthrough in enhancing the capabilities of this model architecture. Our code, data and model are publicly available at https://github.com/WangRongsheng/Aurora

cs.CL↗

arXiv2023

Machine Mindset: An MBTI Exploration of Large Language Models

We present a novel approach for integrating Myers-Briggs Type Indicator (MBTI) personality traits into large language models (LLMs), addressing the challenges of personality consistency in personalized AI. Our method, "Machine Mindset," involves a two-phase fine-tuning and Direct Preference Optimization (DPO) to embed MBTI traits into LLMs. This approach ensures that models internalize these traits, offering a stable and consistent personality profile. We demonstrate the effectiveness of our models across various domains, showing alignment between model performance and their respective MBTI traits. The paper highlights significant contributions in the development of personality datasets and a new training methodology for personality integration in LLMs, enhancing the potential for personalized AI applications. We also open-sourced our model and part of the data at \url{https://github.com/PKU-YuanGroup/Machine-Mindset}.

cs.CL↗

arXiv2023

Video-Bench: A Comprehensive Benchmark and Toolkit for Evaluating Video-based Large Language Models

Video-based large language models (Video-LLMs) have been recently introduced, targeting both fundamental improvements in perception and comprehension, and a diverse range of user inquiries. In pursuit of the ultimate goal of achieving artificial general intelligence, a truly intelligent Video-LLM model should not only see and understand the surroundings, but also possess human-level commonsense, and make well-informed decisions for the users. To guide the development of such a model, the establishment of a robust and comprehensive evaluation system becomes crucial. To this end, this paper proposes \textit{Video-Bench}, a new comprehensive benchmark along with a toolkit specifically designed for evaluating Video-LLMs. The benchmark comprises 10 meticulously crafted tasks, evaluating the capabilities of Video-LLMs across three distinct levels: Video-exclusive Understanding, Prior Knowledge-based Question-Answering, and Comprehension and Decision-making. In addition, we introduce an automatic toolkit tailored to process model outputs for various tasks, facilitating the calculation of metrics and generating convenient final scores. We evaluate 8 representative Video-LLMs using \textit{Video-Bench}. The findings reveal that current Video-LLMs still fall considerably short of achieving human-like comprehension and analysis of real-world videos, offering valuable insights for future research directions. The benchmark and toolkit are available at: \url{https://github.com/PKU-YuanGroup/Video-Bench}.

cs.CV↗

arXiv2023

Video-LLaVA: Learning United Visual Representation by Alignment Before Projection

The Large Vision-Language Model (LVLM) has enhanced the performance of various downstream tasks in visual-language understanding. Most existing approaches encode images and videos into separate feature spaces, which are then fed as inputs to large language models. However, due to the lack of unified tokenization for images and videos, namely misalignment before projection, it becomes challenging for a Large Language Model (LLM) to learn multi-modal interactions from several poor projection layers. In this work, we unify visual representation into the language feature space to advance the foundational LLM towards a unified LVLM. As a result, we establish a simple but robust LVLM baseline, Video-LLaVA, which learns from a mixed dataset of images and videos, mutually enhancing each other. Video-LLaVA achieves superior performances on a broad range of 9 image benchmarks across 5 image question-answering datasets and 4 image benchmark toolkits. Additionally, our Video-LLaVA also outperforms Video-ChatGPT by 5.8%, 9.9%, 18.6%, and 10.1% on MSRVTT, MSVD, TGIF, and ActivityNet, respectively. Notably, extensive experiments demonstrate that Video-LLaVA mutually benefits images and videos within a unified visual representation, outperforming models designed specifically for images or videos. We aim for this work to provide modest insights into the multi-modal inputs for the LLM. Code address: \href{https://github.com/PKU-YuanGroup/Video-LLaVA}

cs.CV↗

arXiv2023

LanguageBind: Extending Video-Language Pretraining to N-modality by Language-based Semantic Alignment

The video-language (VL) pretraining has achieved remarkable improvement in multiple downstream tasks. However, the current VL pretraining framework is hard to extend to multiple modalities (N modalities, N>=3) beyond vision and language. We thus propose LanguageBind, taking the language as the bind across different modalities because the language modality is well-explored and contains rich semantics. Specifically, we freeze the language encoder acquired by VL pretraining, then train encoders for other modalities with contrastive learning. As a result, all modalities are mapped to a shared feature space, implementing multi-modal semantic alignment. While LanguageBind ensures that we can extend VL modalities to N modalities, we also need a high-quality dataset with alignment data pairs centered on language. We thus propose VIDAL-10M with Video, Infrared, Depth, Audio and their corresponding Language, naming as VIDAL-10M. In our VIDAL-10M, all videos are from short video platforms with complete semantics rather than truncated segments from long videos, and all the video, depth, infrared, and audio modalities are aligned to their textual descriptions. LanguageBind has achieved superior performance on a wide range of 15 benchmarks covering video, audio, depth, and infrared. Moreover, multiple experiments have provided evidence for the effectiveness of LanguageBind in achieving indirect alignment and complementarity among diverse modalities. Code address: https://github.com/PKU-YuanGroup/LanguageBind

cs.CV↗

arXiv2023

Analytical results for a spin-orbit coupled atom held in a non-Hermitian double well under synchronous combined modulation

We propose a simple method of synchronous combined modulations to generate the exact analytic solutions for a spin-orbit (SO) coupled ultracold atom held in a non-Hermitian double-well potential. Based on the obtained analytical solutions, we mainly study the parity-time ($\mathcal{PT}$) symmetry of this system and the system stability for both balanced and unbalanced gain-loss between two wells. Under balanced gain and loss, the effect of the proportional constants between synchronous combined modulations and the SO-coupling strength on the $\mathcal{PT}$-symmetry breaking is revealed analytically. Surprisingly, we find when the Zeeman field is present, the stable spin-flipping tunneling between two wells can not occur in the non-Hermitian SO-coupled ultracold atomic system, but the stable spin-conserving tunneling can be performed. Under unbalanced gain and loss, the unique set of parameter conditions that can cause the system to stabilize is found. The results may provide a possibility for the exact control of $\mathcal{PT}$-symmetry breaking and quantum spin dynamics in a non-Hermitian SO-coupled system.

cond-mat.quant-gas↗

arXiv2023

Chatlaw: A Multi-Agent Legal Assistant based on a Role-Aligned Mixture-of-Experts Architecture

Artificial Intelligence (AI) holds great potential in legal services, yet Large Language Models (LLMs) face two major challenges: limited knowledge of the Chinese legal system and vulnerability to hallucinations. To address these issues, we present Chatlaw, a multi-agent legal assistant. Chatlaw's framework is designed to emulate the Standard Operating Procedures (SOP) of real law firms, where different roles (e.g., assistant, researcher, senior lawyer) collaborate on a case. To computationally mirror this collaborative structure, we developed a novel Role-Aligned Mixture-of-Experts (RA-MoE) architecture. In this system, the internal "experts" are specifically trained to align with the distinct tasks of each agent role (e.g., inquiry, analysis, drafting). These specialized agents (Legal Assistant, Researcher, etc.) then form the collaborative framework. When they interact with users, retrieve legal knowledge, analyze case details, or generate reliable consultations, the RA-MoE architecture intelligently routes their computations to the corresponding dedicated expert, ensuring each step is handled by the most qualified parameters. In evaluations, Chatlaw surpasses general-purpose AI models, including GPT-4, achieving a 7.73% improvement in accuracy on the LawBench benchmark and an 11-point higher score on the Unified Qualification Exam for Legal Professionals. Real-case studies and expert assessments further confirm its robustness. Chatlaw enhances the accessibility and reliability of legal services, advancing the provision of legal support to the public.

cs.CL↗

arXiv2023

ChatFace: Chat-Guided Real Face Editing via Diffusion Latent Space Manipulation

Editing real facial images is a crucial task in computer vision with significant demand in various real-world applications. While GAN-based methods have showed potential in manipulating images especially when combined with CLIP, these methods are limited in their ability to reconstruct real images due to challenging GAN inversion capability. Despite the successful image reconstruction achieved by diffusion-based methods, there are still challenges in effectively manipulating fine-gained facial attributes with textual instructions.To address these issues and facilitate convenient manipulation of real facial images, we propose a novel approach that conduct text-driven image editing in the semantic latent space of diffusion model. By aligning the temporal feature of the diffusion model with the semantic condition at generative process, we introduce a stable manipulation strategy, which perform precise zero-shot manipulation effectively. Furthermore, we develop an interactive system named ChatFace, which combines the zero-shot reasoning ability of large language models to perform efficient manipulations in diffusion semantic latent space. This system enables users to perform complex multi-attribute manipulations through dialogue, opening up new possibilities for interactive image editing. Extensive experiments confirmed that our approach outperforms previous methods and enables precise editing of real facial images, making it a promising candidate for real-world applications. Project page: https://dongxuyue.github.io/chatface/

cs.CV↗

arXiv2019

Printing surface charge as a new paradigm to program droplet transport

Directed, long-range and self-propelled transport of droplets on solid surfaces, especially on water repellent surfaces, is crucial for many applications from water harvesting to bio-analytical devices. One appealing strategy to achieve the preferential transport is to passively control the surface wetting gradients, topological or chemical, to break the asymmetric contact line and overcome the resistance force. Despite extensive progress, the directional droplet transport is limited to small transport velocity and short transport distance due to the fundamental trade-off: rapid transport of droplet demands a large wetting gradient, whereas long-range transport necessitates a relatively small wetting gradient. Here, we report a radically new strategy that resolves the bottleneck through the creation of an unexplored gradient in surface charge density (SCD). By leveraging on a facile droplet printing on superamphiphobic surfaces as well as the fundamental understanding of the mechanisms underpinning the creation of the preferential SCD, we demonstrate the self-propulsion of droplets with a record-high velocity over an ultra-long distance without the need for additional energy input. Such a Leidenfrost-like droplet transport, manifested at ambient condition, is also genetic, which can occur on a variety of substrates such as flexible and vertically placed surfaces. Moreover, distinct from conventional physical and chemical gradients, the new dimension of gradient in SCD can be programmed in a rewritable fashion. We envision that our work enriches and extends our capability in the manipulation of droplet transport and would find numerous potential applications otherwise impossible.

cond-mat.soft↗