Search

Background: The primary goal of studying isospin dynamics via heavy-ion reactions is to explore the isospin dependence of effective interactions within the nuclear equation of state (EOS). Purpose: This work aims to investigate the effects of nuclear incompressibility ($ K_0 $) on neutron-proton equilibration in projectile-like fragments (PLFs). Method: We simulate $^{70}$Zn + $^{70}$Zn collisions at 35 MeV/nucleon using the isospin-dependent quantum molecular dynamics (IQMD) model, coupled with the statistical decay code GEMINI. Results: The IQMD simulations not only reproduce experimental data patterns but also reveal the dynamic mechanisms underlying the binary breakup of PLFs. The rotation of PLFs is influenced by the transformation of angular momentum, which is connected to the isoscalar component of the EOS. This connection explains why shifts in $ K_0 $ affect the description of neutron-proton equilibration as measured by PLF rotation. The simulations demonstrate that a model with a smaller $ K_0 $ paired with a softer symmetry energy, or a larger $ K_0 $ with a slightly stiffer symmetry energy, both offer better indications of neutron-proton equilibration. Conclusion: Considering the uncertainty in $ K_0 $, the slope of the symmetry energy is constrained within the range of $ L = 20 \sim 40 $ MeV, providing valuable insights into the nuclear equation of state.

FlashAttention series has been widely applied in the inference of large language models (LLMs). However, FlashAttention series only supports the high-level GPU architectures, e.g., Ampere and Hopper. At present, FlashAttention series is not easily transferrable to NPUs and low-resource GPUs. Moreover, FlashAttention series is inefficient for multi- NPUs or GPUs inference scenarios. In this work, we propose FastAttention which pioneers the adaptation of FlashAttention series for NPUs and low-resource GPUs to boost LLM inference efficiency. Specifically, we take Ascend NPUs and Volta-based GPUs as representatives for designing our FastAttention. We migrate FlashAttention series to Ascend NPUs by proposing a novel two-level tiling strategy for runtime speedup, tiling-mask strategy for memory saving and the tiling-AllReduce strategy for reducing communication overhead, respectively. Besides, we adapt FlashAttention for Volta-based GPUs by redesigning the operands layout in shared memory and introducing a simple yet effective CPU-GPU cooperative strategy for efficient memory utilization. On Ascend NPUs, our FastAttention can achieve a 10.7$\times$ speedup compared to the standard attention implementation. Llama-7B within FastAttention reaches up to 5.16$\times$ higher throughput than within the standard attention. On Volta architecture GPUs, FastAttention yields 1.43$\times$ speedup compared to its equivalents in \texttt{xformers}. Pangu-38B within FastAttention brings 1.46$\times$ end-to-end speedup using FasterTransformer. Coupled with the propose CPU-GPU cooperative strategy, FastAttention supports a maximal input length of 256K on 8 V100 GPUs. All the codes will be made available soon.

The cold start problem in recommender systems remains a critical challenge. Current solutions often train hybrid models on auxiliary data for both cold and warm users/items, potentially degrading the experience for the latter. This drawback limits their viability in practical scenarios where the satisfaction of existing warm users/items is paramount. Although graph neural networks (GNNs) excel at warm recommendations by effective collaborative signal modeling, they haven't been effectively leveraged for the cold-start issue within a user-item graph, which is largely due to the lack of initial connections for cold user/item entities. Addressing this requires a GNN adept at cold-start recommendations without sacrificing performance for existing ones. To this end, we introduce Graph Neural Patching for Cold-Start Recommendations (GNP), a customized GNN framework with dual functionalities: GWarmer for modeling collaborative signal on existing warm users/items and Patching Networks for simulating and enhancing GWarmer's performance on cold-start recommendations. Extensive experiments on three benchmark datasets confirm GNP's superiority in recommending both warm and cold users/items., Comment: 13 pages, accepted by Australasian Database Conference 2024. arXiv admin note: substantial text overlap with arXiv:2209.12215

Existing works have established multiple benchmarks to highlight the security risks associated with Code GenAI. These risks are primarily reflected in two areas: a model potential to generate insecure code (insecure coding) and its utility in cyberattacks (cyberattack helpfulness). While these benchmarks have made significant strides, there remain opportunities for further improvement. For instance, many current benchmarks tend to focus more on a model ability to provide attack suggestions rather than its capacity to generate executable attacks. Additionally, most benchmarks rely heavily on static evaluation metrics, which may not be as precise as dynamic metrics such as passing test cases. Conversely, expert-verified benchmarks, while offering high-quality data, often operate at a smaller scale. To address these gaps, we develop SecCodePLT, a unified and comprehensive evaluation platform for code GenAIs' risks. For insecure code, we introduce a new methodology for data creation that combines experts with automatic generation. Our methodology ensures the data quality while enabling large-scale generation. We also associate samples with test cases to conduct code-related dynamic evaluation. For cyberattack helpfulness, we set up a real environment and construct samples to prompt a model to generate actual attacks, along with dynamic metrics in our environment. We conduct extensive experiments and show that SecCodePLT outperforms the state-of-the-art (SOTA) benchmark CyberSecEval in security relevance. Furthermore, it better identifies the security risks of SOTA models in insecure coding and cyberattack helpfulness. Finally, we apply SecCodePLT to the SOTA code agent, Cursor, and, for the first time, identify non-trivial security risks in this advanced coding agent.

As AI systems become more advanced, ensuring their alignment with a diverse range of individuals and societal values becomes increasingly critical. But how can we capture fundamental human values and assess the degree to which AI systems align with them? We introduce ValueCompass, a framework of fundamental values, grounded in psychological theory and a systematic review, to identify and evaluate human-AI alignment. We apply ValueCompass to measure the value alignment of humans and language models (LMs) across four real-world vignettes: collaborative writing, education, public sectors, and healthcare. Our findings uncover risky misalignment between humans and LMs, such as LMs agreeing with values like "Choose Own Goals", which are largely disagreed by humans. We also observe values vary across vignettes, underscoring the necessity for context-aware AI alignment strategies. This work provides insights into the design space of human-AI alignment, offering foundations for developing AI that responsibly reflects societal values and ethics.

Human Activity Recognition (HAR) using Inertial Measurement Unit (IMU) sensors is critical for applications in healthcare, safety, and industrial production. However, variations in activity patterns, device types, and sensor placements create distribution gaps across datasets, reducing the performance of HAR models. To address this, we propose LanHAR, a novel system that leverages Large Language Models (LLMs) to generate semantic interpretations of sensor readings and activity labels for cross-dataset HAR. This approach not only mitigates cross-dataset heterogeneity but also enhances the recognition of new activities. LanHAR employs an iterative re-generation method to produce high-quality semantic interpretations with LLMs and a two-stage training framework that bridges the semantic interpretations of sensor readings and activity labels. This ultimately leads to a lightweight sensor encoder suitable for mobile deployment, enabling any sensor reading to be mapped into the semantic interpretation space. Experiments on four public datasets demonstrate that our approach significantly outperforms state-of-the-art methods in both cross-dataset HAR and new activity recognition. The source code will be made publicly available.

LiDAR panoptic segmentation, which jointly performs instance and semantic segmentation for things and stuff classes, plays a fundamental role in LiDAR perception tasks. While most existing methods explicitly separate these two segmentation tasks and utilize different branches (i.e., semantic and instance branches), some recent methods have embraced the query-based paradigm to unify LiDAR panoptic segmentation. However, the distinct spatial distribution and inherent characteristics of objects(things) and their surroundings(stuff) in 3D scenes lead to challenges, including the mutual competition of things/stuff and the ambiguity of classification/segmentation. In this paper, we propose decoupling things/stuff queries according to their intrinsic properties for individual decoding and disentangling classification/segmentation to mitigate ambiguity. To this end, we propose a novel framework dubbed DQFormer to implement semantic and instance segmentation in a unified workflow. Specifically, we design a decoupled query generator to propose informative queries with semantics by localizing things/stuff positions and fusing multi-level BEV embeddings. Moreover, a query-oriented mask decoder is introduced to decode corresponding segmentation masks by performing masked cross-attention between queries and mask embeddings. Finally, the decoded masks are combined with the semantics of the queries to produce panoptic results. Extensive experiments on nuScenes and SemanticKITTI datasets demonstrate the superiority of our DQFormer framework., Comment: 13 pages, 10 figures

World models envision potential future states based on various ego actions. They embed extensive knowledge about the driving environment, facilitating safe and scalable autonomous driving. Most existing methods primarily focus on either data generation or the pretraining paradigms of world models. Unlike the aforementioned prior works, we propose Drive-OccWorld, which adapts a vision-centric 4D forecasting world model to end-to-end planning for autonomous driving. Specifically, we first introduce a semantic and motion-conditional normalization in the memory module, which accumulates semantic and dynamic information from historical BEV embeddings. These BEV features are then conveyed to the world decoder for future occupancy and flow forecasting, considering both geometry and spatiotemporal modeling. Additionally, we propose injecting flexible action conditions, such as velocity, steering angle, trajectory, and commands, into the world model to enable controllable generation and facilitate a broader range of downstream applications. Furthermore, we explore integrating the generative capabilities of the 4D world model with end-to-end planning, enabling continuous forecasting of future states and the selection of optimal trajectories using an occupancy-based cost function. Extensive experiments on the nuScenes dataset demonstrate that our method can generate plausible and controllable 4D occupancy, opening new avenues for driving world generation and end-to-end planning., Comment: 18 pages, 10 figures

Urban traffic is subject to disruptions that cause extended waiting time and safety issues at signalized intersections. While numerous studies have addressed the issue of intelligent traffic systems in the context of various disturbances, traffic signal malfunction, a common real-world occurrence with significant repercussions, has received comparatively limited attention. The primary objective of this research is to mitigate the adverse effects of traffic signal malfunction, such as traffic congestion and collision, by optimizing the control of neighboring functioning signals. To achieve this goal, this paper presents a novel traffic signal control framework (MalLight), which leverages an Influence-aware State Aggregation Module (ISAM) and an Influence-aware Reward Aggregation Module (IRAM) to achieve coordinated control of surrounding traffic signals. To the best of our knowledge, this study pioneers the application of a Reinforcement Learning(RL)-based approach to address the challenges posed by traffic signal malfunction. Empirical investigations conducted on real-world datasets substantiate the superior performance of our proposed methodology over conventional and deep learning-based alternatives in the presence of signal malfunction, with reduction of throughput alleviated by as much as 48.6$\%$., Comment: Paper accepted to CIKM24 Full Research track

Sequential recommendation, where user preference is dynamically inferred from sequential historical behaviors, is a critical task in recommender systems (RSs). To further optimize long-term user engagement, offline reinforcement-learning-based RSs have become a mainstream technique as they provide an additional advantage in avoiding global explorations that may harm online users' experiences. However, previous studies mainly focus on discrete action and policy spaces, which might have difficulties in handling dramatically growing items efficiently. To mitigate this issue, in this paper, we aim to design an algorithmic framework applicable to continuous policies. To facilitate the control in the low-dimensional but dense user preference space, we propose an \underline{\textbf{E}}fficient \underline{\textbf{Co}}ntinuous \underline{\textbf{C}}ontrol framework (ECoC). Based on a statistically tested assumption, we first propose the novel unified action representation abstracted from normalized user and item spaces. Then, we develop the corresponding policy evaluation and policy improvement procedures. During this process, strategic exploration and directional control in terms of unified actions are carefully designed and crucial to final recommendation decisions. Moreover, beneficial from unified actions, the conservatism regularization for policies and value functions are combined and perfectly compatible with the continuous framework. The resulting dual regularization ensures the successful offline training of RL-based recommendation policies. Finally, we conduct extensive experiments to validate the effectiveness of our framework. The results show that compared to the discrete baselines, our ECoC is trained far more efficiently. Meanwhile, the final policies outperform baselines in both capturing the offline data and gaining long-term rewards.

Reinforcement learning (RL) techniques have been increasingly investigated for dynamic HVAC control in buildings. However, most studies focus on exploring solutions in online or off-policy scenarios without discussing in detail the implementation feasibility or effectiveness of dealing with purely offline datasets or trajectories. The lack of these works limits the real-world deployment of RL-based HVAC controllers, especially considering the abundance of historical data. To this end, this paper comprehensively evaluates the strengths and limitations of state-of-the-art offline RL algorithms by conducting analytical and numerical studies. The analysis is conducted from two perspectives: algorithms and dataset characteristics. As a prerequisite, the necessity of applying offline RL algorithms is first confirmed in two building environments. The ability of observation history modeling to reduce violations and enhance performance is subsequently studied. Next, the performance of RL-based controllers under datasets with different qualitative and quantitative conditions is investigated, including constraint satisfaction and power consumption. Finally, the sensitivity of certain hyperparameters is also evaluated. The results indicate that datasets of a certain suboptimality level and relatively small scale can be utilized to effectively train a well-performed RL-based HVAC controller. Specifically, such controllers can reduce at most 28.5% violation ratios of indoor temperatures and achieve at most 12.1% power savings compared to the baseline controller. In summary, this paper presents our well-structured investigations and new findings when applying offline reinforcement learning to building HVAC systems.

Large Multimodal Models (LMMs) have ushered in a new era in artificial intelligence, merging capabilities in both language and vision to form highly capable Visual Foundation Agents. These agents are postulated to excel across a myriad of tasks, potentially approaching general artificial intelligence. However, existing benchmarks fail to sufficiently challenge or showcase the full potential of LMMs in complex, real-world environments. To address this gap, we introduce VisualAgentBench (VAB), a comprehensive and pioneering benchmark specifically designed to train and evaluate LMMs as visual foundation agents across diverse scenarios, including Embodied, Graphical User Interface, and Visual Design, with tasks formulated to probe the depth of LMMs' understanding and interaction capabilities. Through rigorous testing across nine proprietary LMM APIs and eight open models, we demonstrate the considerable yet still developing agent capabilities of these models. Additionally, VAB constructs a trajectory training set constructed through hybrid methods including Program-based Solvers, LMM Agent Bootstrapping, and Human Demonstrations, promoting substantial performance improvements in LMMs through behavior cloning. Our work not only aims to benchmark existing models but also provides a solid foundation for future development into visual foundation agents. Code, train \& test data, and part of fine-tuned open LMMs are available at \url{https://github.com/THUDM/VisualAgentBench}.

Symbolic data analysis (SDA) aggregates large individual-level datasets into a small number of distributional summaries, such as random rectangles or random histograms. Inference is carried out using these summaries in place of the original dataset, resulting in computational gains at the loss of some information. In likelihood-based SDA, the likelihood function is characterised by an integral with a large exponent, which limits the method's utility as for typical models the integral unavailable in closed form. In addition, the likelihood function is known to produce biased parameter estimates in some circumstances. Our article develops a Bayesian framework for SDA methods in these settings that resolves the issues resulting from integral intractability and biased parameter estimation using pseudo-marginal Markov chain Monte Carlo methods. We develop an exact but computationally expensive method based on path sampling and the block-Poisson estimator, and a much faster, but approximate, method based on Taylor expansion. Through simulation and real-data examples we demonstrate the performance of the developed methods, showing large reductions in computation time compared to the full-data analysis, with only a small loss of information.

We explore a large $N$ generalization of the Kitaev model on the honeycomb lattice with a simple nearest-neighbor interacting Hamiltonian. In particular, we focus on the $\mathbb{Z}_4$ case with isotropic couplings, which is characterized by an exact $\mathbb{Z}_4$ one-form symmetry. Guided by symmetry considerations and an analytical study in the single chain limit, on the infinitely long cylinders, we find the model is gapped with an extremely short correlation length. Combined with the $\mathbb{Z}_4$ one-form symmetry, this suggests the model is topologically ordered. To pin down the nature of this phase, we further study the model on both finite and infinitely long strips, where we consistently find a $c=1$ conformal field theory (CFT) description, suggesting the existence of chiral edge modes described by a free boson CFT. Further evidence is found by studying the dimer correlators on infinitely long strips. We find the dimer correlation functions show a power-law decay with the exponent close to 2 on the boundary of the strip, while decay much faster in the bulk. Combined with the topological entanglement entropy extracted from cylinder geometry, we identify the spin liquid is chiral and supports a $\mathrm{U}(1)_{-8}$ chiral topological order. A unified perspective for all $\mathbb{Z}_N$ type Kitaev models is also discussed., Comment: 10 pages, 5 figures

Decision Transformer (DT) has emerged as a promising class of algorithms in offline reinforcement learning (RL) tasks, leveraging pre-collected datasets and Transformer's capability to model long sequences. Recent works have demonstrated that using parts of trajectories from training tasks as prompts in DT enhances its performance on unseen tasks, giving rise to Prompt-DT methods. However, collecting data from specific environments can be both costly and unsafe in many scenarios, leading to suboptimal performance and limited few-shot prompt abilities due to the data-hungry nature of Transformer-based models. Additionally, the limited datasets used in pre-training make it challenging for Prompt-DT type of methods to distinguish between various RL tasks through prompts alone. To address these challenges, we introduce the Language model-initialized Prompt Decision Transformer (LPDT), which leverages pre-trained language models for meta-RL tasks and fine-tunes the model using Low-rank Adaptation (LoRA). We further incorporate prompt regularization to effectively differentiate between tasks based on prompt feature representations. Our approach integrates pre-trained language model and RL tasks seamlessly. Extensive empirical studies demonstrate that initializing with a pre-trained language model significantly enhances the performance of Prompt-DT on unseen tasks compared to baseline methods., Comment: 2 figures, 8 tables. Accepted by the Training Agents with Foundation Models Workshop at RLC 2024

As the recommendation service needs to address increasingly diverse distributions, such as multi-population, multi-scenario, multitarget, and multi-interest, more and more recent works have focused on multi-distribution modeling and achieved great progress. However, most of them only consider modeling in a single multi-distribution manner, ignoring that mixed multi-distributions often coexist and form hierarchical relationships. To address these challenges, we propose a flexible modeling paradigm, named Hierarchical Multi-Distribution Network (HMDN), which efficiently models these hierarchical relationships and can seamlessly integrate with existing multi-distribution methods, such as Mixture of-Experts (MoE) and Dynamic-Weight (DW) models. Specifically, we first design a hierarchical multi-distribution representation refinement module, employing a multi-level residual quantization to obtain fine-grained hierarchical representation. Then, the refined hierarchical representation is integrated into the existing single multi-distribution models, seamlessly expanding them into mixed multi-distribution models. Experimental results on both public and industrial datasets validate the effectiveness and flexibility of HMDN.

The dynamics of N/Z and mass equilibration are investigated in the reactions 112,124Sn + 239Pu by employing the isospin-dependent quantum molecular dynamics model. It is found that N/Z and mass equilibration take place at different collision stages. The N/Z relaxation is observed in the approaching phase (from first contact to deepest contact) with a very short time, whereas interestingly we find for the first time that mass equilibration only takes place in the separation phase (from the deepest contact to re-separation), which are explained by investigating the dynamical asymmetry between the approaching and separation phases. The mass equilibration also could be clarified with a dynamical potential energy surface. Our results provide a new insight into the equilibration dynamics of the quantum systems.

Training with larger mini-batches improves the convergence rate and can yield superior performance. However, training with large mini-batches becomes prohibitive for Large Language Models (LLMs), due to the large GPU memory requirement. To address this problem, an effective approach is finding small mini-batch coresets that closely match the gradient of larger mini-batches. However, this approach becomes infeasible and ineffective for LLMs, due to the highly imbalanced nature of the sources in language data, use of the Adam optimizer, and the very large gradient dimensionality of LLMs. In this work, we address the above challenges by proposing Coresets for Training LLMs (CoLM). First, we show that mini-batch coresets found by gradient matching do not contain representative examples of the small sources w.h.p., and thus including all examples of the small sources in the mini-batch coresets is crucial for optimal performance. Second, we normalize the gradients by their historical exponential to find mini-batch coresets for training with Adam. Finally, we leverage zeroth-order methods to find smooth gradient of the last V -projection matrix and sparsify it to keep the dimensions with the largest normalized gradient magnitude. We apply CoLM to fine-tuning Phi-2, Phi-3, and Zephyr with LoRA on MathInstruct and SuperGLUE benchmark. Remarkably, CoLM reduces the memory requirement of fine-tuning by 2x and even outperforms training with 4x larger mini-batches. Notably, CoLM easily stack with existing memory-efficient training methods, such as LoRA., Comment: 18 pages, 5 figures, 7 tables

The Einstein Probe (EP) achieved its first detection and localization of a bright X-ray flare, EP240219a, on February 19, 2024, during its commissioning phase. Subsequent targeted searches triggered by the EP240219a alert identified a faint, untriggered gamma-ray burst (GRB) in the archived data of Fermi/GBM, Swift/BAT, Insight-HXMT/HE and INTEGRAL/SPI-ACS. The EP/WXT light curve reveals a long duration of approximately 160 seconds with a slow decay, whereas the Fermi/GBM light curve shows a total duration of approximately 70 seconds. The peak in the Fermi/GBM light curve occurs slightly later with respect to the peak seen in the EP/WXT light curve. Our spectral analysis shows that a single cutoff power-law model effectively describes the joint EP/WXT-Fermi/GBM spectra in general, indicating coherent broad emission typical of GRBs. The model yielded a photon index of $\sim -1.70 \pm 0.05$ and a peak energy of $\sim 257 \pm 134$ keV. After detection of GRB 240219A, long-term observations identified several candidates in optical and radio wavelengths, none of which was confirmed as the afterglow counterpart during subsequent optical and near-infrared follow-ups. The analysis of GRB 240219A classifies it as an X-ray rich GRB with a high peak energy, presenting both challenges and opportunities for studying the physical origins of X-ray flashes (XRFs), X-ray rich GRBs (XRRs), and classical GRBs (C-GRBs). Furthermore, linking the cutoff power-law component to non-thermal synchrotron radiation suggests that the burst is driven by a Poynting flux-dominated outflow., Comment: 14 pages, 8 figures, 3 tables

Foundation models (FMs) provide societal benefits but also amplify risks. Governments, companies, and researchers have proposed regulatory frameworks, acceptable use policies, and safety benchmarks in response. However, existing public benchmarks often define safety categories based on previous literature, intuitions, or common sense, leading to disjointed sets of categories for risks specified in recent regulations and policies, which makes it challenging to evaluate and compare FMs across these benchmarks. To bridge this gap, we introduce AIR-Bench 2024, the first AI safety benchmark aligned with emerging government regulations and company policies, following the regulation-based safety categories grounded in our AI risks study, AIR 2024. AIR 2024 decomposes 8 government regulations and 16 company policies into a four-tiered safety taxonomy with 314 granular risk categories in the lowest tier. AIR-Bench 2024 contains 5,694 diverse prompts spanning these categories, with manual curation and human auditing to ensure quality. We evaluate leading language models on AIR-Bench 2024, uncovering insights into their alignment with specified safety concerns. By bridging the gap between public benchmarks and practical AI risks, AIR-Bench 2024 provides a foundation for assessing model safety across jurisdictions, fostering the development of safer and more responsible AI systems.

The multinucleon transfer (MNT) process in low-energy heavy ion collisions can be utilized to produce unknown nuclei far beyond the stability line. However, the reaction products exhibit broad angular and energy distributions, which could lower the experimental detection efficiency. We present a classical approach that employs a parameterized angular distribution to describe the complex issue. By analyzing limited experimental data on angular distribution, we proposed a three-parameter formula to calculate the angular distribution and identified the dependencies of the parameters. We also discuss the sensitivity of these parameters within this method. A comprehensive comparison with microscopic models and experimental data across a wide range of conditions is conducted. The proposed formula offers an efficient and straightforward way to determine the angular distribution of MNT products., Comment: 6 pages, 6 figure

Data curation is commonly considered a "secret-sauce" for LLM training, with higher quality data usually leading to better LLM performance. Given the scale of internet-scraped corpora, data pruning has become a larger and larger focus. Specifically, many have shown that de-duplicating data, or sub-selecting higher quality data, can lead to efficiency or performance improvements. Generally, three types of methods are used to filter internet-scale corpora: embedding-based, heuristic-based, and classifier-based. In this work, we contrast the former two in the domain of finetuning LLMs for code generation. We find that embedding-based methods are often confounded by length, and that a simple heuristic--pruning long files--outperforms other methods in compute-limited regimes. Our method can yield up to a 2x efficiency benefit in training (while matching performance) or a 3.5% absolute performance improvement on HumanEval (while matching compute). However, we find that perplexity on held-out long files can increase, begging the question of whether optimizing data mixtures for common coding benchmarks (HumanEval, MBPP) actually best serves downstream use cases. Overall, we hope our work builds useful intuitions about code data (specifically, the low quality of extremely long code files) provides a compelling heuristic-based method for data pruning, and brings to light questions in how we evaluate code generation models., Comment: 15 pages, 5 figures

The efficiency improvement of hardware accelerators such as single-instruction-multiple-data (SIMD) and coarse-grained reconfigurable architecture (CGRA) empowers the rapid advancement of AI and machine learning applications. These streaming applications consist of numerous vector operations that can be naturally parallelized. Despite the outstanding achievements of today's hardware accelerators, their potential is limited by their instruction set design. Traditional instruction sets, designed for microprocessors and accelerators, focus on computation and pay little attention to instruction composability and instruction-level cooperation. It leads to a rigid instruction set that is difficult to extend and significant control overhead in hardware. This paper presents an instruction set that is composable in both spatial and temporal sense and suitable for streaming applications. The proposed instruction set contains significantly fewer instruction types but can still efficiently implement complex multi-level loop structures, which is essential for accelerating streaming applications. It is also a resource-centric instruction set that can be conveniently extended by adding new hardware resources, thus creating a custom heterogeneous computation machine. Besides presenting the composable instruction set, we propose a simple yet efficient instruction scheduling algorithm. We analyzed the scalability of the scheduling algorithm and compared the efficiency of our compiled programs against RISC-V programs. The results indicate that our scheduling algorithm scales linearly, and our instruction set leads to near-optimal execution latency. The mapped applications on CIS are nearly 10 times faster than the RISC-V version.

We present a comprehensive AI risk taxonomy derived from eight government policies from the European Union, United States, and China and 16 company policies worldwide, making a significant step towards establishing a unified language for generative AI safety evaluation. We identify 314 unique risk categories organized into a four-tiered taxonomy. At the highest level, this taxonomy encompasses System & Operational Risks, Content Safety Risks, Societal Risks, and Legal & Rights Risks. The taxonomy establishes connections between various descriptions and approaches to risk, highlighting the overlaps and discrepancies between public and private sector conceptions of risk. By providing this unified framework, we aim to advance AI safety through information sharing across sectors and the promotion of best practices in risk mitigation for generative AI models and systems.

As large language models (LLMs) appear to behave increasingly human-like in text-based interactions, more and more researchers become interested in investigating personality in LLMs. However, the diversity of psychological personality research and the rapid development of LLMs have led to a broad yet fragmented landscape of studies in this interdisciplinary field. Extensive studies across different research focuses, different personality psychometrics, and different LLMs make it challenging to have a holistic overview and further pose difficulties in applying findings to real-world applications. In this paper, we present a comprehensive review by categorizing current studies into three research problems: self-assessment, exhibition, and recognition, based on the intrinsic characteristics and external manifestations of personality in LLMs. For each problem, we provide a thorough analysis and conduct in-depth comparisons of their corresponding solutions. Besides, we summarize research findings and open challenges from current studies and further discuss their underlying causes. We also collect extensive publicly available resources to facilitate interested researchers and developers. Lastly, we discuss the potential future research directions and application scenarios. Our paper is the first comprehensive survey of up-to-date literature on personality in LLMs. By presenting a clear taxonomy, in-depth analysis, promising future directions, and extensive resource collections, we aim to provide a better understanding and facilitate further advancements in this emerging field.

Recent years have witnessed a growing trend toward employing deep reinforcement learning (Deep-RL) to derive heuristics for combinatorial optimization (CO) problems on graphs. Maximum Coverage Problem (MCP) and its probabilistic variant on social networks, Influence Maximization (IM), have been particularly prominent in this line of research. In this paper, we present a comprehensive benchmark study that thoroughly investigates the effectiveness and efficiency of five recent Deep-RL methods for MCP and IM. These methods were published in top data science venues, namely S2V-DQN, Geometric-QN, GCOMB, RL4IM, and LeNSE. Our findings reveal that, across various scenarios, the Lazy Greedy algorithm consistently outperforms all Deep-RL methods for MCP. In the case of IM, theoretically sound algorithms like IMM and OPIM demonstrate superior performance compared to Deep-RL methods in most scenarios. Notably, we observe an abnormal phenomenon in IM problem where Deep-RL methods slightly outperform IMM and OPIM when the influence spread nearly does not increase as the budget increases. Furthermore, our experimental results highlight common issues when applying Deep-RL methods to MCP and IM in practical settings. Finally, we discuss potential avenues for improving Deep-RL methods. Our benchmark study sheds light on potential challenges in current deep reinforcement learning research for solving combinatorial optimization problems., Comment: This paper has been accepted by VLDB 2024

Enhancing diverse human decision-making processes in an urban environment is a critical issue across various applications, including ride-sharing vehicle dispatching, public transportation management, and autonomous driving. Offline reinforcement learning (RL) is a promising approach to learn and optimize human urban strategies (or policies) from pre-collected human-generated spatial-temporal urban data. However, standard offline RL faces two significant challenges: (1) data scarcity and data heterogeneity, and (2) distributional shift. In this paper, we introduce MODA -- a Multi-Task Offline Reinforcement Learning with Contrastive Data Sharing approach. MODA addresses the challenges of data scarcity and heterogeneity in a multi-task urban setting through Contrastive Data Sharing among tasks. This technique involves extracting latent representations of human behaviors by contrasting positive and negative data pairs. It then shares data presenting similar representations with the target task, facilitating data augmentation for each task. Moreover, MODA develops a novel model-based multi-task offline RL algorithm. This algorithm constructs a robust Markov Decision Process (MDP) by integrating a dynamics model with a Generative Adversarial Network (GAN). Once the robust MDP is established, any online RL or planning algorithm can be applied. Extensive experiments conducted in a real-world multi-task urban setting validate the effectiveness of MODA. The results demonstrate that MODA exhibits significant improvements compared to state-of-the-art baselines, showcasing its capability in advancing urban decision-making processes. We also made our code available to the research community., Comment: KDD 2024

Recent studies successfully learned static graph embeddings that are structurally fair by preventing the effectiveness disparity of high- and low-degree vertex groups in downstream graph mining tasks. However, achieving structure fairness in dynamic graph embedding remains an open problem. Neglecting degree changes in dynamic graphs will significantly impair embedding effectiveness without notably improving structure fairness. This is because the embedding performance of high-degree and low-to-high-degree vertices will significantly drop close to the generally poorer embedding performance of most slightly changed vertices in the long-tail part of the power-law distribution. We first identify biased structural evolutions in a dynamic graph based on the evolving trend of vertex degree and then propose FairDGE, the first structurally Fair Dynamic Graph Embedding algorithm. FairDGE learns biased structural evolutions by jointly embedding the connection changes among vertices and the long-short-term evolutionary trend of vertex degrees. Furthermore, a novel dual debiasing approach is devised to encode fair embeddings contrastively, customizing debiasing strategies for different biased structural evolutions. This innovative debiasing strategy breaks the effectiveness bottleneck of embeddings without notable fairness loss. Extensive experiments demonstrate that FairDGE achieves simultaneous improvement in the effectiveness and fairness of embeddings.

Thompson sampling (TS) is one of the most popular exploration techniques in reinforcement learning (RL). However, most TS algorithms with theoretical guarantees are difficult to implement and not generalizable to Deep RL. While the emerging approximate sampling-based exploration schemes are promising, most existing algorithms are specific to linear Markov Decision Processes (MDP) with suboptimal regret bounds, or only use the most basic samplers such as Langevin Monte Carlo. In this work, we propose an algorithmic framework that incorporates different approximate sampling methods with the recently proposed Feel-Good Thompson Sampling (FGTS) approach (Zhang, 2022; Dann et al., 2021), which was previously known to be computationally intractable in general. When applied to linear MDPs, our regret analysis yields the best known dependency of regret on dimensionality, surpassing existing randomized algorithms. Additionally, we provide explicit sampling complexity for each employed sampler. Empirically, we show that in tasks where deep exploration is necessary, our proposed algorithms that combine FGTS and approximate sampling perform significantly better compared to other strong baselines. On several challenging games from the Atari 57 suite, our algorithms achieve performance that is either better than or on par with other strong baselines from the deep RL literature., Comment: First two authors contributed equally. Accepted to the Reinforcement Learning Conference (RLC) 2024

Strong nonlinear interactions between quantized excitations are an important resource for quantum technologies based on bosonic oscillator modes. However, most electromagnetic and mechanical nonlinearities arising from intrinsic material properties are far too weak compared to dissipation in the system to allow for nonlinear effects to be observed on the single-quantum level. To overcome this limitation, electromagnetic resonators in both the optical and microwave frequency regimes have been coupled to other strongly nonlinear quantum systems such as atoms and superconducting qubits, allowing for the demonstration of effects such as photon blockade and coherent quantum protocols using the Kerr effect. Here, we demonstrate the realization of the single-phonon nonlinear regime in a solid-state mechanical system. The single-phonon anharmonicity in our system exceeds the decoherence rate by a factor of 6.8, allowing us to use the lowest two energy levels of the resonator as a mechanical qubit, for which we show initialization, readout, and a complete set of direct single qubit gates. Our work adds another unique capability to a powerful quantum acoustics platform for quantum simulations, sensing, and information processing., Comment: Comments are welcome

Medical image registration is an essential topic in medical image analysis. In this paper, we propose a method for medical image registration using a pretrained large language model. We find that using the pretrained large language model to encode deep features of the medical images in the registration model can effectively improve image registration accuracy, indicating the great potential of the large language model in medical image registration tasks. We use dual encoders to perform deep feature extraction on image pairs and then input the features into the pretrained large language model. To adapt the large language model to our registration task, the weights of the large language model are frozen in the registration model, and an adapter is utilized to fine-tune the large language model, which aims at (a) mapping the visual tokens to the language space before the large language model computing, (b) project the modeled language tokens output from the large language model to the visual space. Our method combines output features from the fine-tuned large language model with the features output from each encoder layer to gradually generate the deformation fields required for registration in the decoder. To demonstrate the effectiveness of the large prediction model in registration tasks, we conducted experiments on knee and brain MRI and achieved state-of-the-art results.

As one of the most energetic and brightest events, gamma-ray bursts (GRBs) have been used as a standard candle for cosmological probe. Based on the relevant features of GRBs light curves, a plateau phase followed a decay phase, we obtain X-ray samples of 31 GRBs and optical samples of 50 GRBs, which are thought to be caused by the same physical mechanism. We standardize GRBs using the two-dimension fundamental plane relation of the rest-frame luminosity of the plateau emission ($L_{b,z}$) and the end time of plateau ($T_{b,z}$) $L_{b,z}-T_{b,z}$, as well as the three-dimension fundamental plane correlation including the peak energy ($E_{p,i}$) $L_{b,z}-T_{b,z}-E_{p,i}$. For the cosmological probes, we consider the $\omega$CDM model in which the dark energy consists of one component, and mainly focus on the $X_1X_2$CDM model in which the dark energy is made up of two independent components. We obtain the constraints on the related parameters of the cosmological models using the type Ia supernovae (SNe Ia) data and selected X-ray and optical samples. For the $X_1X_2$CDM model, we find that the values of the equations of state parameters of two dark energies, $\omega_1$ and $\omega_2$, are very close. We also conduct the comparison between the models using the Bayesian information criterion, and find that the $\omega$CDM model is favoured., Comment: 13 pages, 8 figures and 3 tables, accepted for publication in Astronomy & Astrophysics

To improve the theoretical prediction power for synthesizing superheavy elements beyond Og, a Bayesian uncertainty quantification method is employed to evaluate the uncertainty of the calculated evaporation residue cross sections (ERCS) for the first time. The key parameters of the dinuclear system (DNS) model, such as the diffusion parameter $\textit{a}$, the damping factor $E_\mathrm{d}$, and the level-density parameter ratio $a_\mathrm{f}/a_\mathrm{n}$ are systematically constrained by the Bayesian analysis of recent ERCS data. One intriguing behavior is shown that the optimal incident energies (OIE) corresponding to the largest ERCS weakly depend on the fission process. We also find that these parameters are strongly correlated and the uncertainty propagation considering the parameters independently is not reasonable. The 2$\sigma$ confidence level of posterior distributions for $a = 0.586_{-0.002}^{+0.002}$ fm, $E_\mathrm{d} = 25.65_{-3.41}^{+3.43}$ MeV, and $a_\mathrm{f}/a_\mathrm{n} = 1.081_{-0.021}^{+0.021}$ are obtained. Furthermore, the confidence levels of the ERCS and OIE for synthesizing Z = 119 via the reactions($^{54}\mathrm{Cr}+^{243}\mathrm{Am}$), (${}^{50}\mathrm{Ti}+{}^{249}\mathrm{Bk}$), and (${}^{51}\mathrm{V}+{}^{248}\mathrm{Cm}$) are predicted. This work sets the stage for future analyses to explore the OIE and reaction systems for the synthesis of superheavy elements., Comment: 7 pages, 4 figures

Traditionally, results given by the direct numerical simulation (DNS) of Navier-Stokes equations are widely regarded as reliable benchmark solutions of turbulence, as long as grid spacing is fine enough (i.e. less than the minimum Kolmogorov scale) and time-step is small enough, say, satisfying the Courant-Friedrichs-Lewy condition. Is this really true? In this paper a two-dimensional sustained turbulent Kolmogorov flow is investigated numerically by the two numerical methods with detailed comparisons: one is the traditional `direct numerical simulation' (DNS), the other is the `clean numerical simulation' (CNS). The results given by DNS are a kind of mixture of the false numerical noise and the true physical solution, which however are mostly at the same order of magnitude due to the butterfly-effect of chaos. On the contrary, the false numerical noise of the results given by CNS is much smaller than the true physical solution of turbulence in a long enough interval of time so that a CNS result is very close to the true physical solution and thus can be used as a benchmark solution. It is found that numerical noise as a kind of artificial tiny disturbances can lead to huge deviations at large scale on the two-dimensional Kolmogorov turbulence, not only quantitatively (even in statistics) but also qualitatively (such as symmetry of flow). Thus, fine enough spatial grid spacing with small enough time-step alone cannot guarantee the validity of the DNS: it is only a necessary condition but not sufficient. This finding might challenge some assumptions in investigation of turbulence. So, DNS results of a few sustained turbulent flows might have huge deviations on both of small and large scales from the true solution of Navier-Stokes equations even in statistics. Hopefully, CNS as a new tool to investigate turbulent flows more accurately than DNS could bring us some new discoveries., Comment: 27 pages, 18 figures

In recent astronomical observations, an almost dark galaxy, designated as Nube, has unveiled an intriguing anomaly in its stellar distribution. Specifically, Nube exhibits an exceptionally low central brightness, with the 2D half-light radius of its stars far exceeding the typical values found in dwarf galaxies, and even surpassing those observed in ultra-diffuse galaxies (UDGs). This phenomenon is difficult to explain within the framework of cold dark matter (CDM). Meanwhile, due to its ultralight particle mass, fuzzy dark matter (FDM) exhibits a de Broglie wavelength on the order of kiloparsecs under the typical velocities of galaxies. The interference between different modes of the FDM wave gives rise to fluctuations in the gravitational field, which can lead to the dynamical heating of stars within galaxies, resulting in an expansion of their spatial distribution. In this paper, we aim to interpret the anomalous stellar distribution observed in Nube as a consequence of the dynamical heating effect induced by FDM. Our findings suggest that a FDM particle mass around $1-2\times 10^{-23}$ eV can effectively account for this anomaly. And we propose that the FDM dynamical heating effect provides a new insight into understanding the formation of field UDGs., Comment: 13 pages, 4 figures

When zooming between dual cameras on a mobile, noticeable jumps in geometric content and image color occur in the preview, inevitably affecting the user's zoom experience. In this work, we introduce a new task, ie, dual-camera smooth zoom (DCSZ) to achieve a smooth zoom preview. The frame interpolation (FI) technique is a potential solution but struggles with ground-truth collection. To address the issue, we suggest a data factory solution where continuous virtual cameras are assembled to generate DCSZ data by rendering reconstructed 3D models of the scene. In particular, we propose a novel dual-camera smooth zoom Gaussian Splatting (ZoomGS), where a camera-specific encoding is introduced to construct a specific 3D model for each virtual camera. With the proposed data factory, we construct a synthetic dataset for DCSZ, and we utilize it to fine-tune FI models. In addition, we collect real-world dual-zoom images without ground-truth for evaluation. Extensive experiments are conducted with multiple FI methods. The results show that the fine-tuned FI models achieve a significant performance improvement over the original ones on DCSZ task. The datasets, codes, and pre-trained models will are available at https://github.com/ZcsrenlongZ/ZoomGS., Comment: 24 pages

Personality Recognition in Conversation (PRC) aims to identify the personality traits of speakers through textual dialogue content. It is essential for providing personalized services in various applications of Human-Computer Interaction (HCI), such as AI-based mental therapy and companion robots for the elderly. Most recent studies analyze the dialog content for personality classification yet overlook two major concerns that hinder their performance. First, crucial implicit factors contained in conversation, such as emotions that reflect the speakers' personalities are ignored. Second, only focusing on the input dialog content disregards the semantic understanding of personality itself, which reduces the interpretability of the results. In this paper, we propose Affective Natural Language Inference (Affective-NLI) for accurate and interpretable PRC. To utilize affectivity within dialog content for accurate personality recognition, we fine-tuned a pre-trained language model specifically for emotion recognition in conversations, facilitating real-time affective annotations for utterances. For interpretability of recognition results, we formulate personality recognition as an NLI problem by determining whether the textual description of personality labels is entailed by the dialog content. Extensive experiments on two daily conversation datasets suggest that Affective-NLI significantly outperforms (by 6%-7%) state-of-the-art approaches. Additionally, our Flow experiment demonstrates that Affective-NLI can accurately recognize the speaker's personality in the early stages of conversations by surpassing state-of-the-art methods with 22%-34%., Comment: Accepted by IEEE PerCom 2024

Despite the effectiveness of data selection for large language models (LLMs) during pretraining and instruction fine-tuning phases, improving data efficiency in supervised fine-tuning (SFT) for specialized domains poses significant challenges due to the complexity of fine-tuning data. To bridge this gap, we introduce an effective and scalable data selection method for SFT, SmallToLarge (S2L), which leverages training trajectories from small models to guide the data selection for larger models. We demonstrate through extensive experiments that S2L significantly improves data efficiency in SFT for mathematical problem-solving, reducing the training data to just 11% of the original MathInstruct dataset (Yue et al., 2023) to match full dataset performance while outperforming state-of-the-art data selection algorithms by an average of 4.7% across 6 in- and out-domain evaluation datasets. Remarkably, selecting only 50K data for SFT, S2L achieves a 32.7% accuracy on the most challenging MATH (Hendrycks et al., 2021) benchmark, improving Phi-2 (Li et al., 2023b) by 16.6%. In clinical text summarization on the MIMIC-III dataset (Johnson et al., 2016), S2L again outperforms training on the full dataset using only 50% of the data. Notably, S2L can perform data selection using a reference model 40x smaller than the target model, proportionally reducing the cost of data selection.

High-dimensional quantum entanglement characterizes the entanglement of quantum systems within a larger Hilbert space, introducing more intricate and complex correlations among the entangled particles' states. The high-dimensional Greenberger-Horne-Zeilinger (GHZ) state, symbolic of this type of entanglement, is of significant importance in various quantum information processing applications. This study proposes integrating a neutral atom platform with quantum reservoir engineering to generate a high-dimensional GHZ state deterministically. Leveraging the advantages of neutral atoms in a modified unconventional Rydberg pumping mechanism, combined with controlled dissipation, we achieve a three-dimensional GHZ state with a fidelity surpassing 99\% through multiple pump and dissipation cycles. This innovative approach paves the way for experimentally feasible, deterministic preparation of high-dimensional GHZ states in Rydberg atom systems, thereby advancing the capabilities of quantum information processing., Comment: Accepted by Applied Physics Letters, 7 pages, 5 figures

Data diversity and abundance are essential for improving the performance and generalization of models in natural language processing and 2D vision. However, 3D vision domain suffers from the lack of 3D data, and simply combining multiple 3D datasets for pretraining a 3D backbone does not yield significant improvement, due to the domain discrepancies among different 3D datasets that impede effective feature learning. In this work, we identify the main sources of the domain discrepancies between 3D indoor scene datasets, and propose Swin3D++, an enhanced architecture based on Swin3D for efficient pretraining on multi-source 3D point clouds. Swin3D++ introduces domain-specific mechanisms to Swin3D's modules to address domain discrepancies and enhance the network capability on multi-source pretraining. Moreover, we devise a simple source-augmentation strategy to increase the pretraining data scale and facilitate supervised pretraining. We validate the effectiveness of our design, and demonstrate that Swin3D++ surpasses the state-of-the-art 3D pretraining methods on typical indoor scene understanding tasks. Our code and models will be released at https://github.com/microsoft/Swin3D, Comment: technical report

Li ion diffusion is fundamentally a thermally activated ion hopping process. Recently, soft lattice, anharmonic phonon and paddlewheel mechanism have been proposed to potentially benefit the ion transport, while the understanding of vibrational couplings of mobile ion and anions is still limited but essential. Herein, we access the ionic conductivity, the stability and the lattice dynamics in LiM(SeO3)2 (M =Al, Ga, In, Sc, Y, and La) with two types of oxygen anions within LiO4 polyhedron, namely edge-shared and corner-shared, the prototype of which, LiGa(SeO3)2, has been experimentally synthesized. We studied in detail the anharmonic and harmonic phonon interactions, as well as couplings between vibrations of edge-bonded or corner-bonded anions in Li polyanions and Li ion diffusion. As M changing from Sc to La, anharmonic phonons increase alongside reduced activation energy for Li diffusion. Phonon modes involving edge-bonded oxygen anions contribute more to Li migration than corner-bonded oxygen anions, owing to greater atomic interactions between Li ions and edge-bonded anions. Thus, rather than the overall lattice softness, attentions shall be paid to reduce the frequency of the critical phonons contributing to Li ion diffusions as well as to increase the anharmonicity, for the design of Li ion superionic conductors for all-solid-state-batteries.

In the era of modern education, addressing cross-school learner diversity is crucial, especially in personalized recommender systems for elective course selection. However, privacy concerns often limit cross-school data sharing, which hinders existing methods' ability to model sparse data and address heterogeneity effectively, ultimately leading to suboptimal recommendations. In response, we propose HFRec, a heterogeneity-aware hybrid federated recommender system designed for cross-school elective course recommendations. The proposed model constructs heterogeneous graphs for each school, incorporating various interactions and historical behaviors between students to integrate context and content information. We design an attention mechanism to capture heterogeneity-aware representations. Moreover, under a federated scheme, we train individual school-based models with adaptive learning settings to recommend tailored electives. Our HFRec model demonstrates its effectiveness in providing personalized elective recommendations while maintaining privacy, as it outperforms state-of-the-art models on both open-source and real-world datasets.