arXiv subjects

Yiming Qiu

Publications and source records attributed to Yiming Qiu.

At least 37 records · Page 2Linked to original sources

arXiv2025

Curie: Toward Rigorous and Automated Scientific Experimentation with AI Agents

Scientific experimentation, a cornerstone of human progress, demands rigor in reliability, methodical control, and interpretability to yield meaningful results. Despite the growing capabilities of large language models (LLMs) in automating different aspects of the scientific process, automating rigorous experimentation remains a significant challenge. To address this gap, we propose Curie, an AI agent framework designed to embed rigor into the experimentation process through three key components: an intra-agent rigor module to enhance reliability, an inter-agent rigor module to maintain methodical control, and an experiment knowledge module to enhance interpretability. To evaluate Curie, we design a novel experimental benchmark composed of 46 questions across four computer science domains, derived from influential research papers, and widely adopted open-source projects. Compared to the strongest baseline tested, we achieve a 3.4$\times$ improvement in correctly answering experimental questions. Curie is open-sourced at https://github.com/Just-Curieous/Curie.

cs.AI↗

arXiv2024

Disaggregating Embedding Recommendation Systems with FlexEMR

Efficiently serving embedding-based recommendation (EMR) models remains a significant challenge due to their increasingly large memory requirements. Today's practice splits the model across many monolithic servers, where a mix of GPUs, CPUs, and DRAM is provisioned in fixed proportions. This approach leads to suboptimal resource utilization and increased costs. Disaggregating embedding operations from neural network inference is a promising solution but raises novel networking challenges. In this paper, we discuss the design of FlexEMR for optimized EMR disaggregation. FlexEMR proposes two sets of techniques to tackle the networking challenges: Leveraging the temporal and spatial locality of embedding lookups to reduce data movement over the network, and designing an optimized multi-threaded RDMA engine for concurrent lookup subrequests. We outline the design space for each technique and present initial results from our early prototype.

cs.IR↗

arXiv2024

Generative Retrieval with Preference Optimization for E-commerce Search

Generative retrieval introduces a groundbreaking paradigm to document retrieval by directly generating the identifier of a pertinent document in response to a specific query. This paradigm has demonstrated considerable benefits and potential, particularly in representation and generalization capabilities, within the context of large language models. However, it faces significant challenges in E-commerce search scenarios, including the complexity of generating detailed item titles from brief queries, the presence of noise in item titles with weak language order, issues with long-tail queries, and the interpretability of results. To address these challenges, we have developed an innovative framework for E-commerce search, called generative retrieval with preference optimization. This framework is designed to effectively learn and align an autoregressive model with target data, subsequently generating the final item through constraint-based beam search. By employing multi-span identifiers to represent raw item titles and transforming the task of generating titles from queries into the task of generating multi-span identifiers from queries, we aim to simplify the generation process. The framework further aligns with human preferences using click data and employs a constrained search method to identify key spans for retrieving the final item, thereby enhancing result interpretability. Our extensive experiments show that this framework achieves competitive performance on a real-world dataset, and online A/B tests demonstrate the superiority and effectiveness in improving conversion gains.

cs.IR↗

arXiv2024

Optimizing E-commerce Search: Toward a Generalizable and Rank-Consistent Pre-Ranking Model

In large e-commerce platforms, search systems are typically composed of a series of modules, including recall, pre-ranking, and ranking phases. The pre-ranking phase, serving as a lightweight module, is crucial for filtering out the bulk of products in advance for the downstream ranking module. Industrial efforts on optimizing the pre-ranking model have predominantly focused on enhancing ranking consistency, model structure, and generalization towards long-tail items. Beyond these optimizations, meeting the system performance requirements presents a significant challenge. Contrasting with existing industry works, we propose a novel method: a Generalizable and RAnk-ConsistEnt Pre-Ranking Model (GRACE), which achieves: 1) Ranking consistency by introducing multiple binary classification tasks that predict whether a product is within the top-k results as estimated by the ranking model, which facilitates the addition of learning objectives on common point-wise ranking models; 2) Generalizability through contrastive learning of representation for all products by pre-training on a subset of ranking product embeddings; 3) Ease of implementation in feature construction and online deployment. Our extensive experiments demonstrate significant improvements in both offline metrics and online A/B test: a 0.75% increase in AUC and a 1.28% increase in CVR.

cs.IR↗

arXiv2023

Differentiable Retrieval Augmentation via Generative Language Modeling for E-commerce Query Intent Classification

Retrieval augmentation, which enhances downstream models by a knowledge retriever and an external corpus instead of by merely increasing the number of model parameters, has been successfully applied to many natural language processing (NLP) tasks such as text classification, question answering and so on. However, existing methods that separately or asynchronously train the retriever and downstream model mainly due to the non-differentiability between the two parts, usually lead to degraded performance compared to end-to-end joint training. In this paper, we propose Differentiable Retrieval Augmentation via Generative lANguage modeling(Dragan), to address this problem by a novel differentiable reformulation. We demonstrate the effectiveness of our proposed method on a challenging NLP task in e-commerce search, namely query intent classification. Both the experimental results and ablation study show that the proposed method significantly and reasonably improves the state-of-the-art baselines on both offline evaluation and online A/B test.

cs.IR↗

arXiv2023

Disorder-induced excitation continuum in a spin-1/2 cobaltate on a triangular lattice

A spin-1/2 triangular-lattice antiferromagnet is a prototypical frustrated quantum magnet, which exhibits remarkable quantum many-body effects that arise from the synergy between geometric spin frustration and quantum fluctuations. It can host quantum frustrated magnetic topological phenomena like quantum spin liquid (QSL) states, highlighted by the presence of fractionalized quasiparticles within a continuum of magnetic excitations. In this work, we use neutron scattering to study CoZnMo$_3$O$_8$, which has a triangular lattice of Jeff = 1/2 Co2+ ions with octahedral coordination. We found a wave-vector-dependent excitation continuum at low energy that disappears with increasing temperature. Although these excitations are reminiscent of a spin excitation continuum in a QSL state, their presence in CoZnMo$_3$O$_8$ originates from magnetic intersite disorder-induced dynamic spin states with peculiar excitations. Our results, therefore, give direct experimental evidence for the presence of a disorder-induced spin excitation continuum.

cond-mat.str-el↗

arXiv2023

A Multi-Granularity Matching Attention Network for Query Intent Classification in E-commerce Retrieval

Query intent classification, which aims at assisting customers to find desired products, has become an essential component of the e-commerce search. Existing query intent classification models either design more exquisite models to enhance the representation learning of queries or explore label-graph and multi-task to facilitate models to learn external information. However, these models cannot capture multi-granularity matching features from queries and categories, which makes them hard to mitigate the gap in the expression between informal queries and categories. This paper proposes a Multi-granularity Matching Attention Network (MMAN), which contains three modules: a self-matching module, a char-level matching module, and a semantic-level matching module to comprehensively extract features from the query and a query-category interaction matrix. In this way, the model can eliminate the difference in expression between queries and categories for query intent classification. We conduct extensive offline and online A/B experiments, and the results show that the MMAN significantly outperforms the strong baselines, which shows the superiority and effectiveness of MMAN. MMAN has been deployed in production and brings great commercial value for our company.

cs.IR↗

arXiv2022

Magnetic ground state of the Kitaev Na$_2$Co$_2$TeO$_6$ spin liquid candidate

As a candidate Kitaev material, Na$_2$Co$_2$TeO$_6$ exhibits intriguing magnetism on a honeycomb lattice that is believed to be $C_3$-symmetric. Here we report a neutron diffraction study of high quality single crystals under $a$-axis magnetic fields. Our data support the less common notion of a magnetic ground state that corresponds to a triple-$\mathbf{q}$ magnetic structure with $C_3$ symmetry, rather than the multi-domain zigzag structure typically assumed in prototype Kitaev spin liquid candidates. In particular, we find that the field is unable to repopulate the supposed zigzag domains, where the only alternative explanation is that the domains are strongly pinned by hitherto unidentified structural reasons. If the triple-$\mathbf{q}$ structure is correct then this requires reevaluation of many candidate Kitaev materials. We also find that fields beyond about 10 Tesla suppress the long range antiferromagnetic order, allowing new magnetic behavior to emerge different from that expected for a spin liquid.

cond-mat.str-el↗

arXiv2022

Beyond Single Tetrahedron Physics of Breathing Pyrochlore Compound Ba3Yb2Zn5O11

Recently a new class of quantum magnets, the so-called breathing pyrochlore spin systems, have attracted much attention due to their potential to host exotic emergent phenomena. Here, we present magnetometry, heat capacity, thermal conductivity, Muon-spin relaxation, and polarized inelastic neutron scattering measurements performed on high-quality single-crystal samples of breathing pyrochlore compound Ba3Yb2Zn5O11. We interpret these results using a simplified toy model and provide a new insight into the low-energy physics of this system beyond the single-tetrahedron physics proposed previously.

cond-mat.str-el↗

arXiv2022

Pre-training Tasks for User Intent Detection and Embedding Retrieval in E-commerce Search

BERT-style models pre-trained on the general corpus (e.g., Wikipedia) and fine-tuned on specific task corpus, have recently emerged as breakthrough techniques in many NLP tasks: question answering, text classification, sequence labeling and so on. However, this technique may not always work, especially for two scenarios: a corpus that contains very different text from the general corpus Wikipedia, or a task that learns embedding spacial distribution for a specific purpose (e.g., approximate nearest neighbor search). In this paper, to tackle the above two scenarios that we have encountered in an industrial e-commerce search system, we propose customized and novel pre-training tasks for two critical modules: user intent detection and semantic embedding retrieval. The customized pre-trained models after fine-tuning, being less than 10% of BERT-base's size in order to be feasible for cost-efficient CPU serving, significantly improve the other baseline models: 1) no pre-training model and 2) fine-tuned model from the official pre-trained BERT using general corpus, on both offline datasets and online system. We have open sourced our datasets for the sake of reproducibility and future works.

cs.IR↗

arXiv2022

Bootstrapped Dimensional Crossover of a Spin Density Wave

Quantum materials display rich and myriad types of magnetic, electronic, and structural ordering, often with these ordering modes either competing with one another or 'intertwining', that is, reinforcing one another. Low dimensional quantum materials, influenced strongly by competing interactions and/or geometric frustration, are particularly susceptible to such ordering phenomena and thus offer fertile ground for understanding the consequent emergent collective quantum phenomena. Such is the case of the quasi-2D materials R4Ni3O10(R=La, Pr), in which intertwined charge-and spin-density waves (CDW and SDW) on the Ni sublattice have been identified and characterized. Not unexpectedly, these density waves are largely 2D as a result of weak coupling between planes, compounded with magnetic frustration. In the case of R=Pr, however, we show here that exchange coupling between the transition metal and rare earth sublattices upon cooling overcomes both obstacles, leading to a dimensional crossover into a fully 3D ordered and coupled SDW state on both sublattices, as an induced moment on notionally nonmagnetic Pr3+ opens exchange pathways in the third dimension. In the process, the structure of the SDW on the Ni sublattice is irreversibly altered, an effect that survives reheating of the material until the underlying CDW melts. This 'bootstrapping' mechanism linking incommensurate SDWs on the two sublattices illustrates a new member of the multitude of quantum states that low-dimensional magnets can express, driven by coupled orders and modulated by frustrated exchange pathways.

cond-mat.str-el↗

arXiv2022

Evidence for pressure induced unconventional quantum criticality in the coupled spin ladder antiferromagnet C$_9$H$_{18}$N$_2$CuBr$_4$

Quantum phase transitions in quantum matter occur at zero temperature between distinct ground states by tuning a nonthermal control parameter. Often, they can be accurately described within the Landau theory of phase transitions, similarly to conventional thermal phase transitions. However, this picture can break down under certain circumstances. Here, we present a comprehensive study of the effect of hydrostatic pressure on the magnetic structure and spin dynamics of the spin-1/2 ladder compound C$_9$H$_{18}$N$_2$CuBr$_4$. Single-crystal heat capacity and neutron diffraction measurements reveal that the N$\rm \acute{e}$el-ordered phase breaks down beyond a critical pressure of $P_{\rm c}$$\sim$1.0 GPa through a continuous quantum phase transition. Estimates of the critical exponents suggest that this transition may fall outside the traditional Landau paradigm. The inelastic neutron scattering spectra at 1.3 GPa are characterized by two well-separated gapped modes, including one continuum-like and another resolution-limited excitation in distinct scattering channels, which further indicates an exotic quantum-disordered phase above $P_{\rm c}$.

cond-mat.str-el↗

arXiv2022

Givens Coordinate Descent Methods for Rotation Matrix Learning in Trainable Embedding Indexes

Product quantization (PQ) coupled with a space rotation, is widely used in modern approximate nearest neighbor (ANN) search systems to significantly compress the disk storage for embeddings and speed up the inner product computation. Existing rotation learning methods, however, minimize quantization distortion for fixed embeddings, which are not applicable to an end-to-end training scenario where embeddings are updated constantly. In this paper, based on geometric intuitions from Lie group theory, in particular the special orthogonal group $SO(n)$, we propose a family of block Givens coordinate descent algorithms to learn rotation matrix that are provably convergent on any convex objectives. Compared to the state-of-the-art SVD method, the Givens algorithms are much more parallelizable, reducing runtime by orders of magnitude on modern GPUs, and converge more stably according to experimental studies. They further improve upon vanilla product quantization significantly in an end-to-end training scenario.

cs.IR↗

arXiv2021

Highly tunable magnetic phases in transition metal dichalcogenide Fe$_{1/3+δ}$NbS$_2$

Layered transition metal dichalcogenides (TMDCs) host a plethora of interesting physical phenomena ranging from charge order to superconductivity. By introducing magnetic ions into 2H-NbS$_2$, the material forms a family of magnetic intercalated TMDCs T$_x$NbS$_2$ (T = 3d transition metal). Recently, Fe$_{1/3+δ}$NbS$_2$ has been found to possess intriguing resistance switching and magnetic memory effects coupled to the Néel temperature of T$_N \sim 45$ K [1,2]. We present comprehensive single crystal neutron diffraction measurements on under-intercalated ($δ\sim -0.01$), stoichiometric, and over-intercalated ($δ\sim 0.01$) samples. Magnetic defects are usually considered to suppress magnetic correlations and, concomitantly, transition temperatures. Instead, we observe highly tunable magnetic long-ranged states as the Fe concentration is varied from under-intercalated to over-intercalated, that is from Fe vacancies to Fe interstitials. The under- and over- intercalated samples reveal distinct antiferromagnetic stripe and zig-zag orders, associated with wave vectors $k_1$ = (0.5, 0, 0) and $k_2$ = (0.25, 0.5, 0), respectively. The stoichiometric sample shows two successive magnetic phase transitions for these two wave vectors with an unusual rise-and-fall feature in the intensities connected to $k_1$. We ascribe this sensitive tunability to the competing next nearest neighbor exchange interactions and the oscillatory nature of the Ruderman-Kittel-Kasuya-Yosida (RKKY) mechanism. We discuss experimental observations that relate to the observed intriguing switching resistance behaviors. Our discovery of a magnetic defect tuning of the magnetic structure in bulk crystals Fe$_{1/3+δ}$NbS$_2$ provides a possible new avenue to implement controllable antiferromagnetic spintronic devices.

cond-mat.mtrl-sci↗

arXiv2021

Query Rewriting via Cycle-Consistent Translation for E-Commerce Search

Nowadays e-commerce search has become an integral part of many people's shopping routines. One critical challenge in today's e-commerce search is the semantic matching problem where the relevant items may not contain the exact terms in the user query. In this paper, we propose a novel deep neural network based approach to query rewriting, in order to tackle this problem. Specifically, we formulate query rewriting into a cyclic machine translation problem to leverage abundant click log data. Then we introduce a novel cyclic consistent training algorithm in conjunction with state-of-the-art machine translation models to achieve the optimal performance in terms of query rewriting accuracy. In order to make it practical in industrial scenarios, we optimize the syntax tree construction to reduce computational cost and online serving latency. Offline experiments show that the proposed method is able to rewrite hard user queries into more standard queries that are more appropriate for the inverted index to retrieve. Comparing with human curated rule-based method, the proposed model significantly improves query rewriting diversity while maintaining good relevancy. Online A/B experiments show that it improves core e-commerce business metrics significantly. Since the summer of 2020, the proposed model has been launched into our search engine production, serving hundreds of millions of users.

cs.IR↗

arXiv2021

Joint Learning of Deep Retrieval Model and Product Quantization based Embedding Index

Embedding index that enables fast approximate nearest neighbor(ANN) search, serves as an indispensable component for state-of-the-art deep retrieval systems. Traditional approaches, often separating the two steps of embedding learning and index building, incur additional indexing time and decayed retrieval accuracy. In this paper, we propose a novel method called Poeem, which stands for product quantization based embedding index jointly trained with deep retrieval model, to unify the two separate steps within an end-to-end training, by utilizing a few techniques including the gradient straight-through estimator, warm start strategy, optimal space decomposition and Givens rotation. Extensive experimental results show that the proposed method not only improves retrieval accuracy significantly but also reduces the indexing time to almost none. We have open sourced our approach for the sake of comparison and reproducibility.

cs.IR↗

arXiv2021

High-energy magnetic excitations from heavy quasiparticles in CeCu$_2$Si$_2$

Magnetic fluctuations is the leading candidate for pairing in cuprate, iron-based and heavy fermion superconductors. This view is challenged by the recent discovery of nodeless superconductivity in CeCu$_2$Si$_2$, and calls for a detailed understanding of the corresponding magnetic fluctuations. Here, we mapped out the magnetic excitations in \ys{superconducting (S-type)} CeCu$_2$Si$_2$ using inelastic neutron scattering, finding a strongly asymmetric dispersion for $E\lesssim1.5$~meV, which at higher energies evolve into broad columnar magnetic excitations that extend to $E\gtrsim 5$ meV. While low-energy magnetic excitations exhibit marked three-dimensional characteristics, the high-energy magnetic excitations in CeCu$_2$Si$_2$ are almost two-dimensional, reminiscent of paramagnons found in cuprate and iron-based superconductors. By comparing our experimental findings with calculations in the random-phase approximation,we find that the magnetic excitations in CeCu$_2$Si$_2$ arise from quasiparticles associated with its heavy electron band, which are also responsible for superconductivity. Our results provide a basis for understanding magnetism and superconductivity in CeCu$_2$Si$_2$, and demonstrate the utility of neutron scattering in probing band renormalization in heavy fermion metals.

cond-mat.str-el↗

arXiv2021

Beyond magnons in Nd2ScNbO7: An Ising pyrochlore antiferromagnet with all in all out order and random fields

We report the low temperature magnetic properties of Nd$^{3+}$ pyrochlore $\rm Nd_2ScNbO_7$. Susceptibility and magnetization show an easy-axis moment, and heat capacity reveals a phase transition to long range order at $T_N=371(2)$ mK with a fully recovered $ΔS = R \ln(2)$, 53\% of it recovered for $T>T_N$. Elastic neutron scattering shows a long-range all-in-all-out magnetic order with low-$Q$ diffuse elastic scattering. Inelastic neutron scattering shows a low-energy flat-band, indicating a magnetic Hamiltonian similar to $\rm Nd_2Zr_2O_7$. Nuclear hyperfine excitations measured by ultra-high-resolution neutron backscattering indicates a distribution of static electronic moments below $T_N$, which may be due to B-site disorder influencing Nd crystal electric fields. Analysis of heat capacity data shows an unexpected $T$-linear or $T^{3/2}$ term which is inconsistent with conventional magnon quasiparticles, but is consistent with fractionalized spinons or gapless local spin excitations. We use legacy data to show similar behavior in $\rm Nd_2Zr_2O_7$. Comparing local static moments also reveals a suppression of the nuclear Schottky anomaly in temperature, evidencing a fraction of Nd sites with nearly zero static moment, consistent with exchange-disorder-induced random singlet formation. Taken together, these measurements suggest an unusual fluctuating magnetic ground state which mimics a spin-liquid -- but may not actually be one.

cond-mat.str-el↗