Search

The two-dimensional (2D) Cu(CrSe$_2$)$_2$ monolayer stands out for its combined ferromagnetic (FM), ferroelectric (FE), and metallic properties, marking itself as a prominent 2D multiferroic metal. This work studies those properties and the relevant physics, using density functional calculations, Monte Carlo simulations, and $ab$ $initio$ molecular dynamics. Our results show that Cu(CrSe$_2$)$_2$ monolayer is in the Cr$^{3+}$ $t_{2g}^3$ state with $S$ = 3/2 and Cu$^{1+}$ $3d^{10}$ with $S$ = 0. A ligand hole in the Se 4$p$ orbitals gives rise to metallic behavior and enhances the FM coupling between the local Cr$^{3+}$ $S$ = 3/2 spins. The observed in-plane magnetic anisotropy primarily arises from exchange anisotropy, which is associated with the Cr-Se-Cr itinerant ferromagnetism. In contrast, both single-ion anisotropy and shape magnetic anisotropy contribute negligibly. The Dzyaloshinskii-Moriya interaction is also quite weak, only about 3\% of the intralayer exchange parameters. Our Monte Carlo simulations show a FM Curie temperature ($T_{\rm C}$) of 190 K. Moreover, the monolayer exhibits a vertical FE polarization of 1.79 pC/m and a FE polarization switching barrier of 182 meV/f.u., and the FE state remains stable above 800 K as shown by $ab$ $initio$ molecular dynamics simulations. Furthermore, a magnetoelectric coupling is partially manifested by a magnetization rotation from in-plane to out-of-plane associated with a FE-to-paraelectric transition. The magnetization rotation can also be induced by either hole or electron doping, and the hole doping increases the $T_{\rm C}$ up to 238 K. In addition, tensile strain reduces the FE polarization but enhances $T_{\rm C}$ to 290 K, while a compressive strain gives an opposite effect. Therefore, the multiferroic metallic Cu(CrSe$_2$)$_2$ monolayer may be explored for advanced multifunctional electronic devices., Comment: 14 pages, 7 figures

Large Language Models (LLMs) have ignited an innovative surge of AI applications, marking a new era of exciting possibilities equipped with extended context windows. However, hosting these models is cost-prohibitive mainly due to the extensive memory consumption of KV Cache involving long-context modeling. Despite several works proposing to evict unnecessary tokens from the KV Cache, most of them rely on the biased local statistics of accumulated attention scores and report performance using unconvincing metric like perplexity on inadequate short-text evaluation. In this paper, we propose NACL, a general framework for long-context KV cache eviction that achieves more optimal and efficient eviction in a single operation during the encoding phase. Due to NACL's efficiency, we combine more accurate attention score statistics in PROXY TOKENS EVICTION with the diversified random eviction strategy of RANDOM EVICTION, aiming to alleviate the issue of attention bias and enhance the robustness in maintaining pivotal tokens for long-context modeling tasks. Notably, our method significantly improves the performance on short- and long-text tasks by 80% and 76% respectively, reducing KV Cache by up to 50% with over 95% performance maintenance. The code is available at https://github.com/PaddlePaddle/Research/tree/master/NLP/ACL2024-NACL., Comment: Accepted by ACL 2024 (main conference, long paper)

This study investigates the spin-orbital states of FePS3 and FePSe3 monolayers and the origin of their Ising zigzag AFM, using DFT, crystal field level diagrams, superexchange analyses, and parallel tempering MC simulations. Our calculations show that under the trigonal elongation of the FeS6 (FeSe6) octahedra, the $e_g^\pi$ doublet of the Fe 3d crystal field levels lies lower than the $a_{1g}$ singlet by about 108 meV (123 meV), which is much larger than the strength of Fe 3d SOC. Then, the half-filled minority-spin $e_g^\pi$ doublet of the high-spin Fe$^{2+}$ ions ($d^{5\uparrow,1\downarrow}$) splits by the SOC into the lower $L_{z+}$ and higher $L_{z-}$ states. The spin-orbital ground state $d^{5\uparrow}$$L_{z+}^{1\downarrow}$ formally with $S_z$ = 2 and $L_z$ = 1 gives the large z-axis spin/orbital moments of 3.51/0.76 $\mu_{B}$ (3.41/0.67 $\mu_{B}$) for FePS$_3$ (FePSe$_3$) monolayer, and both the moments are reduced by the strong (stronger) Fe 3d hybridizations with S 3p (Se 4p) states. As a result, FePS3 (FePSe3) monolayer has a huge perpendicular single-ion anisotropy energy of 19.4 meV (14.9 meV), giving an Ising-type magnetism. Moreover, via the maximally localized Wannier functions, we find that the first nearest neighboring (1NN) Fe-Fe pair has large hopping parameters in between some specific orbitals, and so does the 3NN Fe-Fe pair. In contrast, the 2NN Fe-Fe pair has much smaller hopping parameters and the 4NN Fe-Fe pair has negligibly small ones. Then, a combination of those hopping parameters and the superexchange picture can readily explain the computed strong 1NN ferromagnetic coupling and the strong 3NN antiferromagnetic one but the relatively much smaller 2NN antiferromagnetic coupling. Furthermore, our PTMC simulations give TN of 119 K for FePS3 monolayer and also predict for FePSe3 monolayer the same magnetic structure with a close or even higher TN., Comment: 14 pages, 9 figures, 3 tables

Deep learning-based models are widely deployed in autonomous driving areas, especially the increasingly noticed end-to-end solutions. However, the black-box property of these models raises concerns about their trustworthiness and safety for autonomous driving, and how to debug the causality has become a pressing concern. Despite some existing research on the explainability of autonomous driving, there is currently no systematic solution to help researchers debug and identify the key factors that lead to the final predicted action of end-to-end autonomous driving. In this work, we propose a comprehensive approach to explore and analyze the causality of end-to-end autonomous driving. First, we validate the essential information that the final planning depends on by using controlled variables and counterfactual interventions for qualitative analysis. Then, we quantitatively assess the factors influencing model decisions by visualizing and statistically analyzing the response of key model inputs. Finally, based on the comprehensive study of the multi-factorial end-to-end autonomous driving system, we have developed a strong baseline and a tool for exploring causality in the close-loop simulator CARLA. It leverages the essential input sources to obtain a well-designed model, resulting in highly competitive capabilities. As far as we know, our work is the first to unveil the mystery of end-to-end autonomous driving and turn the black box into a white one. Thorough close-loop experiments demonstrate that our method can be applied to end-to-end autonomous driving solutions for causality debugging. Code will be available at https://github.com/bdvisl/DriveInsight.

World models are receiving increasing attention in autonomous driving for their ability to predict potential future scenarios. In this paper, we present BEVWorld, a novel approach that tokenizes multimodal sensor inputs into a unified and compact Bird's Eye View (BEV) latent space for environment modeling. The world model consists of two parts: the multi-modal tokenizer and the latent BEV sequence diffusion model. The multi-modal tokenizer first encodes multi-modality information and the decoder is able to reconstruct the latent BEV tokens into LiDAR and image observations by ray-casting rendering in a self-supervised manner. Then the latent BEV sequence diffusion model predicts future scenarios given action tokens as conditions. Experiments demonstrate the effectiveness of BEVWorld in autonomous driving tasks, showcasing its capability in generating future scenes and benefiting downstream tasks such as perception and motion prediction. Code will be available at https://github.com/zympsyche/BevWorld., Comment: 10 pages

Large language models (LLMs) with one or more fine-tuning phases have become a necessary step to unlock various capabilities, enabling LLMs to follow natural language instructions or align with human preferences. However, it carries the risk of catastrophic forgetting during sequential training, the parametric knowledge or the ability learned in previous stages may be overwhelmed by incoming training data. In this paper, we find that by regularly resetting partial parameters, LLMs can restore some of the original knowledge. Inspired by this, we introduce Half Fine-Tuning (HFT) for LLMs, as a substitute for full fine-tuning (FFT), to mitigate the forgetting issues, where half of the parameters are selected to learn new tasks while the other half are frozen to remain previous knowledge. We provide a feasibility analysis from the perspective of optimization and interpret the parameter selection operation as a regularization term. Without changing the model architecture, HFT could be seamlessly integrated into existing fine-tuning frameworks. Extensive experiments and analysis on supervised fine-tuning, direct preference optimization, and continual learning consistently demonstrate the effectiveness, robustness, and efficiency of HFT. Compared with FFT, HFT not only significantly alleviates the forgetting problem, but also achieves the best performance in a series of downstream benchmarks, with an approximately 30% reduction in training time., Comment: Work in progress

Harnessing visual texts represents a burgeoning frontier in the evolution of language modeling. In this paper, we introduce a novel pre-training framework for a suite of pixel-based autoregressive language models, pre-training on a corpus of over 400 million documents rendered as RGB images. Our approach is characterized by a dual-modality training regimen, engaging both visual data through next patch prediction with a regression head and textual data via next token prediction with a classification head. This study is particularly focused on investigating the synergistic interplay between visual and textual modalities of language. Our comprehensive evaluation across a diverse array of benchmarks reveals that the confluence of visual and textual data substantially augments the efficacy of pixel-based language models. Notably, our findings show that a unidirectional pixel-based model, devoid of textual data during training, can match the performance levels of advanced bidirectional pixel-based models on various language understanding benchmarks. This work highlights the considerable untapped potential of integrating visual and textual information for language modeling purposes. We will release our code, data, and checkpoints to inspire further research advancement.

Amidst the rapid advancements in generative language models, the investigation of how training data shapes the performance of GPT models is still emerging. This paper presents GPTfluence, a novel approach that leverages a featurized simulation to assess the impact of training examples on the training dynamics of GPT models. Our approach not only traces the influence of individual training instances on performance trajectories, such as loss and other key metrics, on targeted test points but also enables a comprehensive comparison with existing methods across various training scenarios in GPT models, ranging from 14 million to 2.8 billion parameters, across a range of downstream tasks. Contrary to earlier methods that struggle with generalization to new data, GPTfluence introduces a parameterized simulation of training dynamics, demonstrating robust generalization capabilities to unseen training data. This adaptability is evident across both fine-tuning and instruction-tuning scenarios, spanning tasks in natural language understanding and generation. We will make our code and data publicly available.

Quantum spin Hall insulators hosting edge spin currents hold great potential for low-power spintronic devices. In this paper, we present a universal approach to achieve a high and near-quantized spin Hall conductance plateau within a sizable bulk gap. Using a nonmagnetic four-band model Hamiltonian, we demonstrate that an even-spin Chern (ESC) insulator can be accessed by tuning the sign of spin-orbit coupling (SOC) within a crystal symmetry-enforced orbital doublet. With the assistance of a high spin Chern number of $C_{S}=-2$ and spin $U$(1) quasisymmetry, this orbital-doublet-driven ESC phase is endowed with the near-double-quantized spin Hall conductance. We identify 12 crystallographic point groups supporting such a sign-tunable SOC. Furthermore, we apply our theory to realistic examples, and show the phase transition from a trivial insulator governed by positive SOC in the RuI$_{3}$ monolayer to an ESC insulator dominated by negative SOC in the RuBr$_{3}$ monolayer. This orbital-doublet-driven ESC insulator, RuBr$_{3}$, showcases nontrivial characteristics including helical edge states, near-double-quantized spin Hall conductance, and robust corner states. Our work provides different pathways in the pursuit of the long-sought quantum spin Hall insulators., Comment: 8 pages, 7 figures

Due to the excellent capacities of large language models (LLMs), it becomes feasible to develop LLM-based agents for reliable user simulation. Considering the scarcity and limit (e.g., privacy issues) of real user data, in this paper, we conduct large-scale user simulation for web search, to improve the analysis and modeling of user search behavior. Specially, we propose BASES, a novel user simulation framework with LLM-based agents, designed to facilitate comprehensive simulations of web search user behaviors. Our simulation framework can generate unique user profiles at scale, which subsequently leads to diverse search behaviors. To demonstrate the effectiveness of BASES, we conduct evaluation experiments based on two human benchmarks in both Chinese and English, demonstrating that BASES can effectively simulate large-scale human-like search behaviors. To further accommodate the research on web search, we develop WARRIORS, a new large-scale dataset encompassing web search user behaviors, including both Chinese and English versions, which can greatly bolster research in the field of information retrieval. Our code and data will be publicly released soon.

Quantum spin Hall (QSH) effect, where electrons with opposite spin channels are deflected to opposite sides of a two-dimensional system with a quantized conductance, was believed to be characterized by a nontrivial topological index $Z_{2}$. However, spin mixing effects in realistic materials often lead to deviation of the spin Hall conductance from exact quantization. In this Letter, we present a universal symmetry indicator for diagnosing QSH effect in realistic materials, termed spin $U$(1) quasi-symmetry. Such a symmetry eliminates the first-order spin-mixing perturbation and thus protects the near-quantization of SHC, applicable to time-reversal-preserved cases with either $Z_{2}=1$ or $Z_{2}=0$, as well as time-reversal-broken scenarios. We propose that spin $U$(1) quasi-symmetry is hidden in the subspace spanned by the doublets with unquenched orbital momentum and emerges when SOC is present, which can be realized in 19 crystallographic point groups. Particularly, we identify a previous overlooked even spin Chern phase with a trivial $Z_{2}$ index as an ideal platform for achieving a near-double-quantized SHC, as exemplified by twisted bilayer transition metal dichalcogenides and monolayer RuBr$_{3}$. Our work offers a new perspective for understanding QSH effect and significantly expands the material pool for the screening of exemplary material candidates., Comment: 6 pages, 4 figures

Two-dimensional van der Waals magnetic materials are of great current interest for their promising applications in spintronics. In this work, using density functional theory calculations in combination with the maximally localized Wannier functions method and the magnetic anisotropy analyses, we study the electronic and magnetic properties of MnPSe$_3$ monolayer. Our results show that it is a charge transfer antiferromagnetic (AF) insulator. For this Mn$^{2+}$ $3d^5$ system, although it seems straightforward to explain the AF ground state using the direct exchange, we find that the near 90$^\circ$ Mn-Se-Mn charge transfer type superexchange plays a dominant role in stabilizing the AF ground state. Moreover, our results indicate that although the shape anisotropy favors an out-of-plane spin orientation, the spin-orbit coupling (SOC) leads to the experimentally observed in-plane spin orientation. We prove that the actual dominant contribution to the magnetic anisotropy comes from the second-order perturbation of the SOC, by analyzing its distribution over the reciprocal space. Using the AF exchange and anisotropy parameters obtained from our calculations, our Monte Carlo simulations give the N\'eel temperature $T_{\rm N}=47$ K for MnPSe$_3$ monolayer, which agrees with the experimental 40 K. Furthermore, our calculations show that under a uniaxial tensile (compressive) strain, N\'eel vector would be parallel (perpendicular) to the strain direction, which well reproduces the recent experiments. We also predict that $T_{\rm N}$ would be increased by a compressive strain., Comment: 8 pages, 9 figures

Graph Neural Networks (GNNs) have shown great power for learning and mining on graphs, and Graph Structure Learning (GSL) plays an important role in boosting GNNs with a refined graph. In the literature, most GSL solutions either primarily focus on structure refinement with task-specific supervision (i.e., node classification), or overlook the inherent weakness of GNNs themselves (e.g., over-squashing), resulting in suboptimal performance despite sophisticated designs. In light of these limitations, we propose to study self-supervised graph structure-feature co-refinement for effectively alleviating the issue of over-squashing in typical GNNs. In this paper, we take a fundamentally different perspective of the Ricci curvature in Riemannian geometry, in which we encounter the challenges of modeling, utilizing and computing Ricci curvature. To tackle these challenges, we present a self-supervised Riemannian model, DeepRicci. Specifically, we introduce a latent Riemannian space of heterogeneous curvatures to model various Ricci curvatures, and propose a gyrovector feature mapping to utilize Ricci curvature for typical GNNs. Thereafter, we refine node features by geometric contrastive learning among different geometric views, and simultaneously refine graph structure by backward Ricci flow based on a novel formulation of differentiable Ricci curvature. Finally, extensive experiments on public datasets show the superiority of DeepRicci, and the connection between backward Ricci flow and over-squashing. Codes of our work are given in https://github.com/RiemanGraph/., Comment: Accepted by IEEE ICDM 2023, Full paper, 10 pages

Two-dimensional (2D) magnetic materials are of current great interest for their promising applications in spintronics. Strong magnetic coupling and anisotropy are both highly desirable for the achievement of a high temperature magnetic order. Here we propose the unusual high valent FeS$_2$ hexagonal monolayer as such a candidate for a strong Ising 2D ferromagnet (FM), by spin-orbital state analyses, first-principles calculations, and the renormalized spin-wave theory (RSWT). We find that very importantly, the high valent Fe$^{4+}$ ion is in the low-spin state ($t_{2g}^{4}$, $S$=1) with degenerate $t_{2g}$ orbitals rather than the high-spin state ($t_{2g}^{3}e_g^{1}$, $S$=2). It is the low-spin state that allows to carry a large perpendicular orbital moment and then produces a huge single ion anisotropy (SIA) of 25 meV/Fe. Moreover, the negative charge transfer character associated with the unusual high valence, strong Fe $3d$-S $3p$ hybridization, wide bands, and a small band gap all help to establish a strong superexchange. Indeed, our first-principles calculations confirm the strong FM superexchange and the huge perpendicular SIA, both of which are further enhanced by a compressive strain. Then, our RSWT calculations predict that the FM $T_{\rm C}$ is 261 K for the pristine FeS$_2$ monolayer and could be increased to 409 K under the compressive --5\% strain. The high $T_{\rm C}$ is also reproduced by our Monte Carlo (MC) simulations. Therefore, it is worth exploring the high-$T_{\rm C}$ Ising FMs in the high valent 2D magnetic materials with degenerate orbitals., Comment: 13 pages, 5 figures

The training paradigm for machine translation has gradually shifted, from learning neural machine translation (NMT) models with extensive parallel corpora to instruction finetuning on multilingual large language models (LLMs) with high-quality translation pairs. In this paper, we focus on boosting many-to-many multilingual translation of LLMs with an emphasis on zero-shot translation directions. We demonstrate that prompt strategies adopted during finetuning are crucial to zero-shot translation and introduce a cross-lingual consistency regularization, XConST, to bridge the representation gap among different languages and improve zero-shot translation performance. XConST is not a new method, but a version of CrossConST (Gao et al., 2023a) adapted for translation instruction finetuning with LLMs. Experimental results on ALMA (Xu et al., 2023), Tower (Team, 2024), and LLaMA-2 (Touvron et al., 2023) show that our approach consistently improves translation performance. Our implementations are available at https://github.com/gpengzhi/CrossConST-LLM.

As commonly-used methods for debiasing natural language understanding (NLU) models, dataset refinement approaches heavily rely on manual data analysis, and thus maybe unable to cover all the potential biased features. In this paper, we propose IBADR, an Iterative Bias-Aware Dataset Refinement framework, which debiases NLU models without predefining biased features. We maintain an iteratively expanded sample pool. Specifically, at each iteration, we first train a shallow model to quantify the bias degree of samples in the pool. Then, we pair each sample with a bias indicator representing its bias degree, and use these extended samples to train a sample generator. In this way, this generator can effectively learn the correspondence relationship between bias indicators and samples. Furthermore, we employ the generator to produce pseudo samples with fewer biased features by feeding specific bias indicators. Finally, we incorporate the generated pseudo samples into the pool. Experimental results and in-depth analyses on two NLU tasks show that IBADR not only significantly outperforms existing dataset refinement approaches, achieving SOTA, but also is compatible with model-centric methods., Comment: EMNLP2023 main conference

We investigate the electronic and magnetic properties of the newly synthesized double perovskites Y$_{2}$NiIrO$_{6}$ and La$_{2}$NiIrO$_{6}$, using density functional calculations, crystal field theory, superexchange pictures, and Monte Carlo simulations. We find that both systems are antiferromagnetic (AFM) Mott insulators, with the high-spin Ni$^{2+}$ $t_{2g}$$^{6}e_{g}$$^{2}$ ($S=1$) and the low-spin Ir$^{4+}$ $t_{2g}$$^{5}$ ($S=1/2$) configurations. We address that their lattice distortion induces $t_{2g}$-$e_{g}$ orbital mixing and thus enables the normal Ni$^{+}$-Ir$^{5+}$ charge excitation with the electron hopping from the Ir `$t_{2g}$' to Ni `$e_g$' orbitals, which promotes the AFM Ni$^{2+}$-Ir$^{4+}$ coupling. Therefore, the increasing $t_{2g}$-$e_{g}$ mixing accounts for the enhanced $T_{\rm N}$ from the less distorted La$_{2}$NiIrO$_{6}$ to the more distorted Y$_{2}$NiIrO$_{6}$. Moreover, our test calculations find that in the otherwise ideally cubic Y$_{2}$NiIrO$_{6}$, the Ni$^{+}$-Ir$^{5+}$ charge excitation is forbidden, and only the abnormal Ni$^{3+}$-Ir$^{3+}$ excitation gives a weakly ferromagnetic (FM) behavior. Furthermore, we find that owing to the crystal field splitting, Hund exchange, and broad band formation in the highly coordinated fcc sublattice, Ir$^{4+}$ ions are not in the $j_{\rm eff}=1/2$ state but in the $S=1/2$ state carrying a finite orbital moment by spin-orbit coupling (SOC). This work clarifies the varying magnetism in Y$_{2}$NiIrO$_{6}$ and La$_{2}$NiIrO$_{6}$ associated with the lattice distortions., Comment: 7 pages, 7 figures, 1 table

Reward modeling (a.k.a., preference modeling) is instrumental for aligning large language models with human preferences, particularly within the context of reinforcement learning from human feedback (RLHF). While conventional reward models (RMs) have exhibited remarkable scalability, they oft struggle with fundamental functionality such as arithmetic computation, code execution, and factual lookup. In this paper, we propose a tool-augmented preference modeling approach, named Themis, to address these limitations by empowering RMs with access to external environments, including calculators and search engines. This approach not only fosters synergy between tool utilization and reward grading but also enhances interpretive capacity and scoring reliability. Our study delves into the integration of external tools into RMs, enabling them to interact with diverse external sources and construct task-specific tool engagement and reasoning traces in an autoregressive manner. We validate our approach across a wide range of domains, incorporating seven distinct external tools. Our experimental results demonstrate a noteworthy overall improvement of 17.7% across eight tasks in preference ranking. Furthermore, our approach outperforms Gopher 280B by 7.3% on TruthfulQA task in zero-shot evaluation. In human evaluations, RLHF trained with Themis attains an average win rate of 32% when compared to baselines across four distinct tasks. Additionally, we provide a comprehensive collection of tool-related RM datasets, incorporating data from seven distinct tool APIs, totaling 15,000 instances. We have made the code, data, and model checkpoints publicly available to facilitate and inspire further research advancements\footnote{\url{https://github.com/ernie-research/Tool-Augmented-Reward-Model}}., Comment: ICLR 2024 Spotlight

Pre-trained models have achieved success in Chinese Short Text Matching (STM) tasks, but they often rely on superficial clues, leading to a lack of robust predictions. To address this issue, it is crucial to analyze and mitigate the influence of superficial clues on STM models. Our study aims to investigate their over-reliance on the edit distance feature, commonly used to measure the semantic similarity of Chinese text pairs, which can be considered a superficial clue. To mitigate STM models' over-reliance on superficial clues, we propose a novel resampling training strategy called Gradually Learn Samples Containing Superficial Clue (GLS-CSC). Through comprehensive evaluations of In-Domain (I.D.), Robustness (Rob.), and Out-Of-Domain (O.O.D.) test sets, we demonstrate that GLS-CSC outperforms existing methods in terms of enhancing the robustness and generalization of Chinese STM models. Moreover, we conduct a detailed analysis of existing methods and reveal their commonality.

Consistency regularization methods, such as R-Drop (Liang et al., 2021) and CrossConST (Gao et al., 2023), have achieved impressive supervised and zero-shot performance in the neural machine translation (NMT) field. Can we also boost end-to-end (E2E) speech-to-text translation (ST) by leveraging consistency regularization? In this paper, we conduct empirical studies on intra-modal and cross-modal consistency and propose two training strategies, SimRegCR and SimZeroCR, for E2E ST in regular and zero-shot scenarios. Experiments on the MuST-C benchmark show that our approaches achieve state-of-the-art (SOTA) performance in most translation directions. The analyses prove that regularization brought by the intra-modal consistency, instead of modality gap, is crucial for the regular E2E ST, and the cross-modal consistency could close the modality gap and boost the zero-shot E2E ST performance.

Knowledge-intensive tasks (e.g., open-domain question answering (QA)) require a substantial amount of factual knowledge and often rely on external information for assistance. Recently, large language models (LLMs) (e.g., ChatGPT), have demonstrated impressive prowess in solving a wide range of tasks with world knowledge, including knowledge-intensive tasks. However, it remains unclear how well LLMs are able to perceive their factual knowledge boundaries, particularly how they behave when incorporating retrieval augmentation. In this study, we present an initial analysis of the factual knowledge boundaries of LLMs and how retrieval augmentation affects LLMs on open-domain QA. Specially, we focus on three primary research questions and analyze them by examining QA performance, priori judgement and posteriori judgement of LLMs. We show evidence that LLMs possess unwavering confidence in their capabilities to respond to questions and the accuracy of their responses. Furthermore, retrieval augmentation proves to be an effective approach in enhancing LLMs' awareness of knowledge boundaries, thereby improving their judgemental abilities. Additionally, we also find that LLMs have a propensity to rely on the provided retrieval results when formulating answers, while the quality of these results significantly impacts their reliance. The code to reproduce this work is available at https://github.com/RUCAIBox/LLM-Knowledge-Boundary.

Multilingual sentence representations are the foundation for similarity-based bitext mining, which is crucial for scaling multilingual neural machine translation (NMT) system to more languages. In this paper, we introduce MuSR: a one-for-all Multilingual Sentence Representation model that supports more than 220 languages. Leveraging billions of English-centric parallel corpora, we train a multilingual Transformer encoder, coupled with an auxiliary Transformer decoder, by adopting a multilingual NMT framework with CrossConST, a cross-lingual consistency regularization technique proposed in Gao et al. (2023). Experimental results on multilingual similarity search and bitext mining tasks show the effectiveness of our approach. Specifically, MuSR achieves superior performance over LASER3 (Heffernan et al., 2022) which consists of 148 independent multilingual sentence encoders.

Current Transformer-based natural language understanding (NLU) models heavily rely on dataset biases, while failing to handle real-world out-of-distribution (OOD) instances. Many methods have been proposed to deal with this issue, but they ignore the fact that the features learned in different layers of Transformer-based NLU models are different. In this paper, we first conduct preliminary studies to obtain two conclusions: 1) both low- and high-layer sentence representations encode common biased features during training; 2) the low-layer sentence representations encode fewer unbiased features than the highlayer ones. Based on these conclusions, we propose a simple yet effective self-debiasing framework for Transformer-based NLU models. Concretely, we first stack a classifier on a selected low layer. Then, we introduce a residual connection that feeds the low-layer sentence representation to the top-layer classifier. In this way, the top-layer sentence representation will be trained to ignore the common biased features encoded by the low-layer sentence representation and focus on task-relevant unbiased features. During inference, we remove the residual connection and directly use the top-layer sentence representation to make predictions. Extensive experiments and indepth analyses on NLU tasks show that our framework performs better than several competitive baselines, achieving a new SOTA on all OOD test sets.

Recently, model-based retrieval has emerged as a new paradigm in text retrieval that discards the index in the traditional retrieval model and instead memorizes the candidate corpora using model parameters. This design employs a sequence-to-sequence paradigm to generate document identifiers, which enables the complete capture of the relevance between queries and documents and simplifies the classic indexretrieval-rerank pipeline. Despite its attractive qualities, there remain several major challenges in model-based retrieval, including the discrepancy between pre-training and fine-tuning, and the discrepancy between training and inference. To deal with the above challenges, we propose a novel two-stage model-based retrieval approach called TOME, which makes two major technical contributions, including the utilization of tokenized URLs as identifiers and the design of a two-stage generation architecture. We also propose a number of training strategies to deal with the training difficulty as the corpus size increases. Extensive experiments and analysis on MS MARCO and Natural Questions demonstrate the effectiveness of our proposed approach, and we investigate the scaling laws of TOME by examining various influencing factors., Comment: ACL 2023

Named entity recognition in real-world applications suffers from the diversity of entity types, the emergence of new entity types, and the lack of high-quality annotations. To address the above problems, this paper proposes an in-context learning-based NER approach, which can effectively inject in-context NER ability into PLMs and recognize entities of novel types on-the-fly using only a few demonstrative instances. Specifically, we model PLMs as a meta-function $\mathcal{ \lambda_ {\text{instruction, demonstrations, text}}. M}$, and a new entity extractor can be implicitly constructed by applying new instruction and demonstrations to PLMs, i.e., $\mathcal{ (\lambda . M) }$(instruction, demonstrations) $\to$ $\mathcal{F}$ where $\mathcal{F}$ will be a new entity extractor, i.e., $\mathcal{F}$: text $\to$ entities. To inject the above in-context NER ability into PLMs, we propose a meta-function pre-training algorithm, which pre-trains PLMs by comparing the (instruction, demonstration)-initialized extractor with a surrogate golden extractor. Experimental results on 4 few-shot NER datasets show that our method can effectively inject in-context NER ability into PLMs and significantly outperforms the PLMs+fine-tuning counterparts., Comment: Accepted to ACL 2023 Main Conference

The multilingual neural machine translation (NMT) model has a promising capability of zero-shot translation, where it could directly translate between language pairs unseen during training. For good transfer performance from supervised directions to zero-shot directions, the multilingual NMT model is expected to learn universal representations across different languages. This paper introduces a cross-lingual consistency regularization, CrossConST, to bridge the representation gap among different languages and boost zero-shot translation performance. The theoretical analysis shows that CrossConST implicitly maximizes the probability distribution for zero-shot translation, and the experimental results on both low-resource and high-resource benchmarks show that CrossConST consistently improves the translation performance. The experimental analysis also proves that CrossConST could close the sentence representation gap and better align the representation space. Given the universality and simplicity of CrossConST, we believe it can serve as a strong baseline for future multilingual NMT research., Comment: Accepted to Findings of ACL 2023

The spin-orbital entangled states are of great interest as they hold exotic phases and intriguing properties. Here we use first-principles calculations to investigate the electronic and magnetic properties of RuI$_{3}$ and RuCl$_{3}$ in both bulk and monolayer cases. Our results show that RuI$_{3}$ bulk is a paramagnetic metal, which is in agreement with recent experiments. We find that the Ru$^{3+}$ ion of RuI$_{3}$ is in the spin-orbital entangled $j_{\rm eff}=\frac{1}{2}$ state. More interestingly, a metal-insulator transition occurs from RuI$_{3}$ bulk to monolayer, and this is mainly due to the band narrowing with the decreasing lattice dimensionality and to the Ru-I hybridization altered by the I $5p$ spin-orbit coupling. In contrast, RuCl$_{3}$ bulk and monolayer both show Mott-insulating behavior, the Ru$^{3+}$ ion is in the formal $S=\frac{1}{2}$ and $L=1$ state with a large in-plane orbital moment, and this result well explains the experimental large effective magnetic moment of RuCl$_{3}$ and the strong in-plane magnetization. The present work demonstrates the contrasting spin-orbital states and the varying properties of RuI$_{3}$ and RuCl$_{3}$., Comment: 12 pages, 12 figures