arXiv subjects

Xiaojing Li

Publications and source records attributed to Xiaojing Li.

7 recordsLinked to original sources

arXiv2025

FAMMA: A Benchmark for Financial Domain Multilingual Multimodal Question Answering

In this paper, we introduce FAMMA, an open-source benchmark for \underline{f}in\underline{a}ncial \underline{m}ultilingual \underline{m}ultimodal question \underline{a}nswering (QA). Our benchmark aims to evaluate the abilities of large language models (LLMs) in answering complex reasoning questions that require advanced financial knowledge. The benchmark has two versions: FAMMA-Basic consists of 1,945 questions extracted from university textbooks and exams, along with human-annotated answers and rationales; FAMMA-LivePro consists of 103 novel questions created by human domain experts, with answers and rationales held out from the public for a contamination-free evaluation. These questions cover advanced knowledge of 8 major subfields in finance (e.g., corporate finance, derivatives, and portfolio management). Some are in Chinese or French, while a majority of them are in English. Each question has some non-text data such as charts, diagrams, or tables. Our experiments reveal that FAMMA poses a significant challenge on LLMs, including reasoning models such as GPT-o1 and DeepSeek-R1. Additionally, we curated 1,270 reasoning trajectories of DeepSeek-R1 on the FAMMA-Basic data, and fine-tuned a series of open-source Qwen models using this reasoning data. We found that training a model on these reasoning trajectories can significantly improve its performance on FAMMA-LivePro. We released our leaderboard, data, code, and trained models at https://famma-bench.github.io/famma/.

cs.CL↗

arXiv2024

PEER: Expertizing Domain-Specific Tasks with a Multi-Agent Framework and Tuning Methods

In domain-specific applications, GPT-4, augmented with precise prompts or Retrieval-Augmented Generation (RAG), shows notable potential but faces the critical tri-lemma of performance, cost, and data privacy. High performance requires sophisticated processing techniques, yet managing multiple agents within a complex workflow often proves costly and challenging. To address this, we introduce the PEER (Plan, Execute, Express, Review) multi-agent framework. This systematizes domain-specific tasks by integrating precise question decomposition, advanced information retrieval, comprehensive summarization, and rigorous self-assessment. Given the concerns of cost and data privacy, enterprises are shifting from proprietary models like GPT-4 to custom models, striking a balance between cost, security, and performance. We developed industrial practices leveraging online data and user feedback for efficient model tuning. This study provides best practice guidelines for applying multi-agent systems in domain-specific problem-solving and implementing effective agent tuning strategies. Our empirical studies, particularly in the financial question-answering domain, demonstrate that our approach achieves 95.0% of GPT-4's performance, while effectively managing costs and ensuring data privacy.

cs.AI↗

arXiv2024

P/D-Serve: Serving Disaggregated Large Language Model at Scale

Serving disaggregated large language models (LLMs) over tens of thousands of xPU devices (GPUs or NPUs) with reliable performance faces multiple challenges. 1) Ignoring the diversity (various prefixes and tidal requests), treating all the prompts in a mixed pool is inadequate. To facilitate the similarity per scenario and minimize the inner mismatch on P/D (prefill and decoding) processing, fine-grained organization is required, dynamically adjusting P/D ratios for better performance. 2) Due to inaccurate estimation on workload (queue status or maintained connections), the global scheduler easily incurs unnecessary timeouts in prefill. 3) Block-fixed device-to-device (D2D) KVCache transfer over cluster-level RDMA (remote direct memory access) fails to achieve desired D2D utilization as expected. To overcome previous problems, this paper proposes an end-to-end system P/D-Serve, complying with the paradigm of MLOps (machine learning operations), which models end-to-end (E2E) P/D performance and enables: 1) fine-grained P/D organization, mapping the service with RoCE (RDMA over converged ethernet) as needed, to facilitate similar processing and dynamic adjustments on P/D ratios; 2) on-demand forwarding upon rejections for idle prefill, decoupling the scheduler from regular inaccurate reports and local queues, to avoid timeouts in prefill; and 3) efficient KVCache transfer via optimized D2D access. P/D-Serve is implemented upon Ascend and MindSpore, has been deployed over tens of thousands of NPUs for more than eight months in commercial use, and further achieves 60\%, 42\% and 46\% improvements on E2E throughput, time-to-first-token (TTFT) SLO (service level objective) and D2D transfer time. As the E2E system with optimizations, P/D-Serve achieves 6.7x increase on throughput, compared with aggregated LLMs.

cs.DC↗

arXiv2019

Tunable anisotropic behaviors in phosphorene under periodic potentials in arbitrary directions

We investigate theoretically the anisotropic electronic and optical behaviors of a monolayer black phosphorus (phosphorene) modulated by periodic potential superlattices in arbitrary directions. We demonstrate that different strength and orientation of the phosphorene potential superlattice can give rise to distinct energy spectra, i.e., tuning the intrinsic electronic anisotropy. Accordingly, the anisotropic effective mass, and optical absorption modulated by superlattice strength and orientation are addressed systematically. This feature enables tuning capability more than one order of magnitude in the optical absorption spectrum. Our findings should be useful in building phosphorene optical and (opto)electronic devices by applying external potential superlattice.

cond-mat.mes-hall↗

arXiv2018

Electric field tuning spin splitting in topological insulator quantum dots doped with a single magnetic ion

We investigate theoretically the electron spin states in disk-shaped HgTe topological insulator quantum dots (TIQDs) containing a single magnetic $Mn^{2+}$ ion. We show that the energy spectrum and the electron density distribution of the topological edge states in HgTe TIQD can be modulated significantly by the position of the magnetic $Mn^{2+}$ ion. The numerical results further demonstrate that the electric fields not only tune the spin splittings of edge states via the $\emph{sp-d}$ exchange interaction between the electron (hole) and the magnetic $Mn^{2+}$ ion but also give rise to the bright-to-dark transitions and anti-crossing behaviors in the photoluminescence (PL) spectra. Such spin properties of HgTe TIQDs with single $Mn^{2+}$ ion as illustrated in this work could offer a new platform for topological electro-optical devices.

cond-mat.mes-hall↗

arXiv2017

Aharonov-Bohm effect in monolayer black phosphorus (phosphorene) nanorings

This work presents theoretical demonstration of Aharonov-Bohm (AB) effect in monolayer phosphorene nanorings (PNR). Atomistic quantum transport simulations of PNR are employed to investigate the impact of multiple modulation sources on the sample conductance. In presence of a perpendicular magnetic field, we find that the conductance of both armchair and zigzag PNR oscillate periodically in a low-energy window as a manifestation of the AB effect. Our numerical results have revealed a giant magnetoresistance (MR) in zigzag PNR (with a maximum magnitude approaching two thousand percent). It is attributed to the AB effect induced destructive interference phase in a wide energy range below the bottom of the second subband. We also demonstrate that PNR conductance is highly anisotropic, offering an additional way to modulate MR. The giant MR in PNR is maintained at room temperature in the presence of thermal broadening effect.

cond-mat.mes-hall↗

arXiv2008

Electric-field switching of exciton spin splitting in coupled quantum dots

We investigate theoretically the spin splitting of the exciton states in semiconductor coupled quantum dots (CQDs) containing a single magnetic ion. We find that the spin splitting can be switched on/off in the CQDs \textit{via} the \textit{sp-d}exchange interaction using the electric field. An interesting bright-to-dark exciton transition can be found and significantly affects the photoluminescence spectrum. This phenonmenon is induced by the transition of the ground exciton state, arising from the hole mixing effect, between the bonding and antibonding states.

cond-mat.mtrl-sci↗