Search

Long-sequence causal reasoning seeks to uncover causal relationships within extended time series data but is hindered by complex dependencies and the challenges of validating causal links. To address the limitations of large-scale language models (e.g., GPT-4) in capturing intricate emotional causality within extended dialogues, we propose CauseMotion, a long-sequence emotional causal reasoning framework grounded in Retrieval-Augmented Generation (RAG) and multimodal fusion. Unlike conventional methods relying only on textual information, CauseMotion enriches semantic representations by incorporating audio-derived features-vocal emotion, emotional intensity, and speech rate-into textual modalities. By integrating RAG with a sliding window mechanism, it effectively retrieves and leverages contextually relevant dialogue segments, thus enabling the inference of complex emotional causal chains spanning multiple conversational turns. To evaluate its effectiveness, we constructed the first benchmark dataset dedicated to long-sequence emotional causal reasoning, featuring dialogues with over 70 turns. Experimental results demonstrate that the proposed RAG-based multimodal integrated approach, the efficacy of substantially enhances both the depth of emotional understanding and the causal inference capabilities of large-scale language models. A GLM-4 integrated with CauseMotion achieves an 8.7% improvement in causal accuracy over the original model and surpasses GPT-4o by 1.2%. Additionally, on the publicly available DiaASQ dataset, CauseMotion-GLM-4 achieves state-of-the-art results in accuracy, F1 score, and causal reasoning accuracy., Comment: 7pages

Although sign language recognition aids non-hearing-impaired understanding, many hearing-impaired individuals still rely on sign language alone due to limited literacy, underscoring the need for advanced sign language production and translation (SLP and SLT) systems. In the field of sign language production, the lack of adequate models and datasets restricts practical applications. Existing models face challenges in production accuracy and pose control, making it difficult to provide fluent sign language expressions across diverse scenarios. Additionally, data resources are scarce, particularly high-quality datasets with complete sign vocabulary and pose annotations. To address these issues, we introduce CNText2Sign and CNSign, comprehensive datasets to benchmark SLP and SLT, respectively, with CNText2Sign covering gloss and landmark mappings for SLP, and CNSign providing extensive video-to-text data for SLT. To improve the accuracy and applicability of sign language systems, we propose the AuraLLM and SignMST-C models. AuraLLM, incorporating LoRA and RAG techniques, achieves a BLEU-4 score of 50.41 on the CNText2Sign dataset, enabling precise control over gesture semantics and motion. SignMST-C employs self-supervised rapid motion video pretraining, achieving a BLEU-4 score of 31.03/32.08 on the PHOENIX2014-T benchmark, setting a new state-of-the-art. These models establish robust baselines for the datasets released for their respective tasks.

Modern configurable systems provide tremendous opportunities for engineering future intelligent software systems. A key difficulty thereof is how to effectively self-adapt the configuration of a running system such that its performance (e.g., runtime and throughput) can be optimized under time-varying workloads. This unfortunately remains unaddressed in existing approaches as they either overlook the available past knowledge or rely on static exploitation of past knowledge without reasoning the usefulness of information when planning for self-adaptation. In this paper, we tackle this challenging problem by proposing DLiSA, a framework that self-adapts configurable systems. DLiSA comes with two properties: firstly, it supports lifelong planning, and thereby the planning process runs continuously throughout the lifetime of the system, allowing dynamic exploitation of the accumulated knowledge for rapid adaptation. Secondly, the planning for a newly emerged workload is boosted via distilled knowledge seeding, in which the knowledge is dynamically purified such that only useful past configurations are seeded when necessary, mitigating misleading information. Extensive experiments suggest that the proposed DLiSA significantly outperforms state-of-the-art approaches, demonstrating a performance improvement of up to 229% and a resource acceleration of up to 2.22x on generating promising adaptation configurations. All data and sources can be found at our repository: https://github.com/ideas-labo/dlisa., Comment: Accepted by the 2025 International Conference on Software Engineering (ICSE 2025)

Event cameras are biologically inspired sensors that emit events asynchronously with remarkable temporal resolution, garnering significant attention from both industry and academia. Mainstream methods favor frame and voxel representations, which reach a satisfactory performance while introducing time-consuming transformation, bulky models, and sacrificing fine-grained temporal information. Alternatively, Point Cloud representation demonstrates promise in addressing the mentioned weaknesses, but it ignores the polarity information, and its models have limited proficiency in abstracting long-term events' features. In this paper, we propose a frequency-aware network named FECNet that leverages Event Cloud representations. FECNet fully utilizes 2S-1T-1P Event Cloud by innovating the event-based Group and Sampling module. To accommodate the long sequence events from Event Cloud, FECNet embraces feature extraction in the frequency domain via the Fourier transform. This approach substantially extinguishes the explosion of Multiply Accumulate Operations (MACs) while effectively abstracting spatial-temporal features. We conducted extensive experiments on event-based object classification, action recognition, and human pose estimation tasks, and the results substantiate the effectiveness and efficiency of FECNet., Comment: Under Review

Spatio-temporal Human-Object Interaction (ST-HOI) understanding aims at detecting HOIs from videos, which is crucial for activity understanding. However, existing whole-body-object interaction video benchmarks overlook the truth that open-world objects are diverse, that is, they usually provide limited and predefined object classes. Therefore, we introduce a new open-world benchmark: Grounding Interacted Objects (GIO) including 1,098 interacted objects class and 290K interacted object boxes annotation. Accordingly, an object grounding task is proposed expecting vision systems to discover interacted objects. Even though today's detectors and grounding methods have succeeded greatly, they perform unsatisfactorily in localizing diverse and rare objects in GIO. This profoundly reveals the limitations of current vision systems and poses a great challenge. Thus, we explore leveraging spatio-temporal cues to address object grounding and propose a 4D question-answering framework (4D-QA) to discover interacted objects from diverse videos. Our method demonstrates significant superiority in extensive experiments compared to current baselines. Data and code will be publicly available at https://github.com/DirtyHarryLYL/HAKE-AVA., Comment: To be published in the Proceedings of AAAI 2025. The first three authors contributed equally. Project: https://github.com/DirtyHarryLYL/HAKE-AVA

Intrusion Detection Systems (IDS) are crucial for safeguarding digital infrastructure. In dynamic network environments, both threat landscapes and normal operational behaviors are constantly changing, resulting in concept drift. While continuous learning mitigates the adverse effects of concept drift, insufficient attention to drift patterns and excessive preservation of outdated knowledge can still hinder the IDS's adaptability. In this paper, we propose SSF (Strategic Selection and Forgetting), a novel continual learning method for IDS, providing continuous model updates with a constantly refreshed memory buffer. Our approach features a strategic sample selection algorithm to select representative new samples and a strategic forgetting mechanism to drop outdated samples. The proposed strategic sample selection algorithm prioritizes new samples that cause the `drifted' pattern, enabling the model to better understand the evolving landscape. Additionally, we introduce strategic forgetting upon detecting significant drift by discarding outdated samples to free up memory, allowing the incorporation of more recent data. SSF captures evolving patterns effectively and ensures the model is aligned with the change of data patterns, significantly enhancing the IDS's adaptability to concept drift. The state-of-the-art performance of SSF on NSL-KDD and UNSW-NB15 datasets demonstrates its superior adaptability to concept drift for network intrusion detection., Comment: Accepted by IEEE International Conference on Computer Communications (INFOCOM) 2025

Sidechain techniques improve blockchain scalability and interoperability, providing decentralized exchange and cross-chain collaboration solutions for Internet of Things (IoT) data across various domains. However, current state-of-the-art (SOTA) schemes for IoT multi-domain data exchange are constrained by the need for synchronous networks, hindering efficient cross-chain interactions in discontinuous networks and leading to suboptimal data exchange. In this paper, we propose AsyncSC, a novel asynchronous sidechain construction. It employs a committee to provide Cross-Blockchain as a Service (C-BaaS) for data exchange in multi-domain IoT. To fulfill the need for asynchronous and efficient data exchange, we combine the ideas of aggregate signatures and verifiable delay functions to devise a novel cryptographic primitive called delayed aggregate signature (DAS), which constructs asynchronous cross-chain proofs (ACPs) that ensure the security of cross-chain interactions. To ensure the consistency of asynchronous transactions, we propose a multilevel buffered transaction pool that guarantees the transaction sequencing. We analyze and prove the security of AsyncSC, simulate an asynchronous communication environment, and conduct a comprehensive evaluation. The results show that AsyncSC outperforms SOTA schemes, improving throughput by an average of 1.21 to 3.96 times, reducing transaction latency by 59.76% to 83.61%, and maintaining comparable resource overhead., Comment: Accepted by IEEE INFOCOM 2025

Building on the foundations of language modeling in natural language processing, Next Token Prediction (NTP) has evolved into a versatile training objective for machine learning tasks across various modalities, achieving considerable success. As Large Language Models (LLMs) have advanced to unify understanding and generation tasks within the textual modality, recent research has shown that tasks from different modalities can also be effectively encapsulated within the NTP framework, transforming the multimodal information into tokens and predict the next one given the context. This survey introduces a comprehensive taxonomy that unifies both understanding and generation within multimodal learning through the lens of NTP. The proposed taxonomy covers five key aspects: Multimodal tokenization, MMNTP model architectures, unified task representation, datasets \& evaluation, and open challenges. This new taxonomy aims to aid researchers in their exploration of multimodal intelligence. An associated GitHub repository collecting the latest papers and repos is available at https://github.com/LMM101/Awesome-Multimodal-Next-Token-Prediction, Comment: 69 papes, 18 figures, repo at https://github.com/LMM101/Awesome-Multimodal-Next-Token-Prediction

Conventional frame-based cameras inevitably produce blurry effects due to motion occurring during the exposure time. Event camera, a bio-inspired sensor offering continuous visual information could enhance the deblurring performance. Effectively utilizing the high-temporal-resolution event data is crucial for extracting precise motion information and enhancing deblurring performance. However, existing event-based image deblurring methods usually utilize voxel-based event representations, losing the fine-grained temporal details that are mathematically essential for fast motion deblurring. In this paper, we first introduce point cloud-based event representation into the image deblurring task and propose a Multi-Temporal Granularity Network (MTGNet). It combines the spatially dense but temporally coarse-grained voxel-based event representation and the temporally fine-grained but spatially sparse point cloud-based event. To seamlessly integrate such complementary representations, we design a Fine-grained Point Branch. An Aggregation and Mapping Module (AMM) is proposed to align the low-level point-based features with frame-based features and an Adaptive Feature Diffusion Module (AFDM) is designed to manage the resolution discrepancies between event data and image data by enriching the sparse point feature. Extensive subjective and objective evaluations demonstrate that our method outperforms current state-of-the-art approaches on both synthetic and real-world datasets., Comment: 12 pages, 8 figures

Secure outsourced computation (SOC) provides secure computing services by taking advantage of the computation power of cloud computing and the technology of privacy computing (e.g., homomorphic encryption). Expanding computational operations on encrypted data (e.g., enabling complex calculations directly over ciphertexts) and broadening the applicability of SOC across diverse use cases remain critical yet challenging research topics in the field. Nevertheless, previous SOC solutions frequently lack the computational efficiency and adaptability required to fully meet evolving demands. To this end, in this paper, we propose a toolkit for TEE-assisted (Trusted Execution Environment) SOC over integers, named TRUST. In terms of system architecture, TRUST falls in a single TEE-equipped cloud server only through seamlessly integrating the computation of REE (Rich Execution Environment) and TEE. In consideration of TEE being difficult to permanently store data and being vulnerable to attacks, we introduce a (2, 2)-threshold homomorphic cryptosystem to fit the hybrid computation between REE and TEE. Additionally, we carefully design a suite of SOC protocols supporting unary, binary and ternary operations. To achieve applications, we present \texttt{SEAT}, secure data trading based on TRUST. Security analysis demonstrates that TRUST enables SOC, avoids collusion attacks among multiple cloud servers, and mitigates potential secret leakage risks within TEE (e.g., from side-channel attacks). Experimental evaluations indicate that TRUST outperforms the state-of-the-art and requires no alignment of data as well as any network communications. Furthermore, \texttt{SEAT} is as effective as the \texttt{Baseline} without any data protection.

In light of the inherently complex and dynamic nature of real-world environments, incorporating risk measures is crucial for the robustness evaluation of deep learning models. In this work, we propose a Risk-Averse Certification framework for Bayesian neural networks called RAC-BNN. Our method leverages sampling and optimisation to compute a sound approximation of the output set of a BNN, represented using a set of template polytopes. To enhance robustness evaluation, we integrate a coherent distortion risk measure--Conditional Value at Risk (CVaR)--into the certification framework, providing probabilistic guarantees based on empirical distributions obtained through sampling. We validate RAC-BNN on a range of regression and classification benchmarks and compare its performance with a state-of-the-art method. The results show that RAC-BNN effectively quantifies robustness under worst-performing risky scenarios, and achieves tighter certified bounds and higher efficiency in complex tasks.

Decomposing physically-based materials from images into their constituent properties remains challenging, particularly when maintaining both computational efficiency and physical consistency. While recent diffusion-based approaches have shown promise, they face substantial computational overhead due to multiple denoising steps and separate models for different material properties. We present SuperMat, a single-step framework that achieves high-quality material decomposition with one-step inference. This enables end-to-end training with perceptual and re-render losses while decomposing albedo, metallic, and roughness maps at millisecond-scale speeds. We further extend our framework to 3D objects through a UV refinement network, enabling consistent material estimation across viewpoints while maintaining efficiency. Experiments demonstrate that SuperMat achieves state-of-the-art PBR material decomposition quality while reducing inference time from seconds to milliseconds per image, and completes PBR material estimation for 3D objects in approximately 3 seconds. The project page is at https://hyj542682306.github.io/SuperMat/., Comment: https://hyj542682306.github.io/SuperMat/

Facial personalization represents a crucial downstream task in the domain of text-to-image generation. To preserve identity fidelity while ensuring alignment with user-defined prompts, current mainstream frameworks for facial personalization predominantly employ identity embedding mechanisms to associate identity information with textual embeddings. However, our experiments show that identity embeddings compromise the effectiveness of other tokens within the prompt, thereby hindering high prompt consistency, particularly when prompts involve multiple facial attributes. Moreover, previous works overlook the fact that their corresponding foundation models hold great potential to generate faces aligning to prompts well and can be easily leveraged to cure these ill-aligned attributes in personalized models. Building upon these insights, we propose FreeCure, a training-free framework that harnesses the intrinsic knowledge from the foundation models themselves to improve the prompt consistency of personalization models. First, by extracting cross-attention and semantic maps from the denoising process of foundation models, we identify easily localized attributes (e.g., hair, accessories, etc). Second, we enhance multiple attributes in the outputs of personalization models through a novel noise-blending strategy coupled with an inversion-based process. Our approach offers several advantages: it eliminates the need for training; it effectively facilitates the enhancement for a wide array of facial attributes in a non-intrusive manner; and it can be seamlessly integrated into existing popular personalization models. FreeCure has demonstrated significant improvements in prompt consistency across a diverse set of state-of-the-art facial personalization models while maintaining the integrity of original identity fidelity.

In plasma wakefield accelerators, the structure of the blowout sheath is vital for the blowout radius and the electromagnetic field distribution inside the blowout. Previous theories assume artificial distribution functions for the sheath, which are either inaccurate or require prior knowledge of parameters. In this study, we develop an adiabatic sheath model based on force balancing, which leads to a self-consistent form of the sheath distribution. This model gives a better estimate of the blowout channel balancing radius than previous models., Comment: 5 pages, 3 figures

This study presents a comprehensive evaluation of GPT-4's translation capabilities compared to human translators of varying expertise levels. Through systematic human evaluation using the MQM schema, we assess translations across three language pairs (Chinese$\longleftrightarrow$English, Russian$\longleftrightarrow$English, and Chinese$\longleftrightarrow$Hindi) and three domains (News, Technology, and Biomedical). Our findings reveal that GPT-4 achieves performance comparable to junior-level translators in terms of total errors, while still lagging behind senior translators. Unlike traditional Neural Machine Translation systems, which show significant performance degradation in resource-poor language directions, GPT-4 maintains consistent translation quality across all evaluated language pairs. Through qualitative analysis, we identify distinctive patterns in translation approaches: GPT-4 tends toward overly literal translations and exhibits lexical inconsistency, while human translators sometimes over-interpret context and introduce hallucinations. This study represents the first systematic comparison between LLM and human translators across different proficiency levels, providing valuable insights into the current capabilities and limitations of LLM-based translation systems., Comment: Work in progress

To train generalist robot policies, machine learning methods often require a substantial amount of expert human teleoperation data. An ideal robot for humans collecting data is one that closely mimics them: bimanual arms and dexterous hands. However, creating such a bimanual teleoperation system with over 50 DoF is a significant challenge. To address this, we introduce Bidex, an extremely dexterous, low-cost, low-latency and portable bimanual dexterous teleoperation system which relies on motion capture gloves and teacher arms. We compare Bidex to a Vision Pro teleoperation system and a SteamVR system and find Bidex to produce better quality data for more complex tasks at a faster rate. Additionally, we show Bidex operating a mobile bimanual robot for in the wild tasks. The robot hands (5k USD) and teleoperation system (7k USD) is readily reproducible and can be used on many robot arms including two xArms (16k USD). Website at https://bidex-teleop.github.io/, Comment: In CoRL 2024. Website at https://bidex-teleop.github.io/

The problem of interfacing quantum mechanics and gravity has long been an unresolved issue in physics. Recent advances in precision measurement technology suggest that detecting gravitational effects in massive quantum systems, particularly gravity-induced entanglement (GIE) in the oscillator system, could provide crucial empirical evidence for revealing the quantum nature of the gravitational field. However, thermal decoherence imposes strict constraints on system parameters. For entanglement to occur, the inequality $2 \gamma_m k_B T < \hbar G \Lambda \rho$ must be satisfied, linking mechanical dissipation $\gamma_m$, effective temperature $T$, oscillator density $\rho$ and form factor $\Lambda$ determined by the geometry and spatial arrangement of the oscillators. This inequality, based on the inherent property of the noise model of GIE, is considered universally across experimental systems and cannot be improved by quantum control. Given the challenges in further optimizing $\gamma_m$, $\rho$, and $T$ near their limits, optimizing the form factor $\Lambda$ may reduce demands on other parameters. In this work, we prove that the form factor $\Lambda$ has a supremum of $2\pi$, revealing a fundamental limit of the oscillator system. We propose design schemes that enable the form factor to approach this supremum, which is nearly an order of magnitude higher than typical spherical oscillators. This optimization may ease experimental constraints, bringing GIE-based validation of quantum gravity closer to realization.

This paper is concerned with effective approximations and wall laws of viscous laminar flows in 3D pipes with randomly rough boundaries. The random roughness is characterized by the boundary oscillation scale $\varepsilon \ll 1 $ and a probability space with ergodicity quantified by functional inequalities. The results in this paper generalize the previous work for 2D channel flows with random Lipschitz boundaries to 3D pipe flows with random boundaries of John type. Moreover, we establish the optimal convergence rates and substantially improve the stochastic integrability obtained in the previous studies. Additionally, we prove a refined version of the Poiseuille's law in 3D pipes with random boundaries, which seems unaddressed in the literature. Our systematic approach combines several classical and recent ideas (particularly from homogenization theory), including the Saint-Venant's principle for pipe flows, the large-scale regularity theory over rough boundaries, and applications of functional inequalities.

Code Language Models (CLMs) have achieved tremendous progress in source code understanding and generation, leading to a significant increase in research interests focused on applying CLMs to real-world software engineering tasks in recent years. However, in realistic scenarios, CLMs are exposed to potential malicious adversaries, bringing risks to the confidentiality, integrity, and availability of CLM systems. Despite these risks, a comprehensive analysis of the security vulnerabilities of CLMs in the extremely adversarial environment has been lacking. To close this research gap, we categorize existing attack techniques into three types based on the CIA triad: poisoning attacks (integrity \& availability infringement), evasion attacks (integrity infringement), and privacy attacks (confidentiality infringement). We have collected so far the most comprehensive (79) papers related to adversarial machine learning for CLM from the research fields of artificial intelligence, computer security, and software engineering. Our analysis covers each type of risk, examining threat model categorization, attack techniques, and countermeasures, while also introducing novel perspectives on eXplainable AI (XAI) and exploring the interconnections between different risks. Finally, we identify current challenges and future research opportunities. This study aims to provide a comprehensive roadmap for both researchers and practitioners and pave the way towards more reliable CLMs for practical applications., Comment: Under a reviewing process since Sep. 3, 2024

The Automated Verification of Textual Claims (AVeriTeC) shared task asks participants to retrieve evidence and predict veracity for real-world claims checked by fact-checkers. Evidence can be found either via a search engine, or via a knowledge store provided by the organisers. Submissions are evaluated using AVeriTeC score, which considers a claim to be accurately verified if and only if both the verdict is correct and retrieved evidence is considered to meet a certain quality threshold. The shared task received 21 submissions, 18 of which surpassed our baseline. The winning team was TUDA_MAI with an AVeriTeC score of 63%. In this paper we describe the shared task, present the full results, and highlight key takeaways from the shared task.

Low-Rank Adaptation (LoRA) is a parameter-efficient technique for rapidly fine-tuning foundation models. In standard LoRA training dynamics, models tend to quickly converge to a local optimum near the initialization. However, this local optimum may not be ideal for out-of-distribution data or tasks such as merging and pruning. In this work, we propose a novel progressive training strategy for LoRA with random layer dropping. This strategy also optimizes the Shapley value of LoRA parameters in each layer, treating each layer as a player in a cooperative game. We refer to this method as Cooperative LoRA (CopRA). Our experimental results demonstrate that parameters trained with CopRA exhibit linear mode connectivity, which enables efficient model merging. This also paves the way for federated learning and multi-task learning via LoRA merging. Additionally, by optimizing the Shapley value, CopRA shows superior performance in pruning tasks., Comment: Published in UniReps Workshop (Extended Abstract Track), NeurIPS 2024

Fine-tuning large language models (LLMs) has become essential for adapting pretrained models to specific downstream tasks. In this paper, we propose Linear Chain Transformation (LinChain), a novel approach that introduces a sequence of linear transformations during fine-tuning to enrich optimization dynamics. By incorporating multiple linear transformations into the parameter update process, LinChain expands the effective rank of updates and enhances the model's ability to learn complex task-specific representations. We demonstrate that this method significantly improves the performance of LLM fine-tuning over state-of-the-art methods by providing more flexible optimization paths during training, while maintaining the inference efficiency of the resulting model. Our experiments on various benchmark tasks show that LinChain leads to better generalization, fewer learnable parameters, and improved task adaptation, making it a compelling strategy for LLM fine-tuning., Comment: 9 pages, 2 figures, 4 tables

We report the discovery of a peculiar X-ray transient, EP240408a, by Einstein Probe (EP) and follow-up studies made with EP, Swift, NICER, GROND, ATCA and other ground-based multi-wavelength telescopes. The new transient was first detected with Wide-field X-ray Telescope (WXT) on board EP on April 8th, 2024, manifested in an intense yet brief X-ray flare lasting for 12 seconds. The flare reached a peak flux of 3.9x10^(-9) erg/cm2/s in 0.5-4 keV, about 300 times brighter than the underlying X-ray emission detected throughout the observation. Rapid and more precise follow-up observations by EP/FXT, Swift and NICER confirmed the finding of this new transient. Its X-ray spectrum is non-thermal in 0.5-10 keV, with a power-law photon index varying within 1.8-2.5. The X-ray light curve shows a plateau lasting for about 4 days, followed by a steep decay till becoming undetectable about 10 days after the initial detection. Based on its temporal property and constraints from previous EP observations, an unusual timescale in the range of 7-23 days is found for EP240408a, which is intermediate between the commonly found fast and long-term transients. No counterparts have been found in optical and near-infrared, with the earliest observation at 17 hours after the initial X-ray detection, suggestive of intrinsically weak emission in these bands. We demonstrate that the remarkable properties of EP240408a are inconsistent with any of the transient types known so far, by comparison with, in particular, jetted tidal disruption events, gamma-ray bursts, X-ray binaries and fast blue optical transients. The nature of EP240408a thus remains an enigma. We suggest that EP240408a may represent a new type of transients with intermediate timescales of the order of about 10 days. The detection and follow-ups of more of such objects are essential for revealing their origin., Comment: 25 pages, 11 figures

In this paper, we develop an uncertainty-aware decision-making and motion-planning method for an autonomous ego vehicle in forced merging scenarios, considering the motion uncertainty of surrounding vehicles. The method dynamically captures the uncertainty of surrounding vehicles by online estimation of their acceleration bounds, enabling a reactive but rapid understanding of the uncertainty characteristics of the surrounding vehicles. By leveraging these estimated bounds, a non-conservative forward occupancy of surrounding vehicles is predicted over a horizon, which is incorporated in both the decision-making process and the motion-planning strategy, to enhance the resilience and safety of the planned reference trajectory. The method successfully fulfills the tasks in challenging forced merging scenarios, and the properties are illustrated by comparison with several alternative approaches., Comment: Accepted by the 63rd IEEE Conference on Decision and Control, 2024

Vision-Language Models (VLMs) have recently made significant progress, but the limited scale and quality of open-source instruction data hinder their performance compared to closed-source models. In this work, we address this limitation by introducing Infinity-MM, a large-scale multimodal instruction dataset with 40 million samples, enhanced through rigorous quality filtering and deduplication. We also propose a synthetic instruction generation method based on open-source VLMs, using detailed image annotations and diverse question generation. Using this data, we trained a 2-billion-parameter VLM, Aquila-VL-2B, achieving state-of-the-art (SOTA) performance for models of similar scale. This demonstrates that expanding instruction data and generating synthetic data can significantly improve the performance of open-source models.

Advanced Persistent Threats (APTs) represent sophisticated cyberattacks characterized by their ability to remain undetected within the victim system for extended periods, aiming to exfiltrate sensitive data or disrupt operations. Existing detection approaches often struggle to effectively identify these complex threats, construct the attack chain for defense facilitation, or resist adversarial attacks. To overcome these challenges, we propose Slot, an advanced APT detection approach based on provenance graphs and graph reinforcement learning. Slot excels in uncovering multi-level hidden relationships, such as causal, contextual, and indirect connections, among system behaviors through provenance graph mining. By pioneering the integration of graph reinforcement learning, Slot dynamically adapts to new user activities and evolving attack strategies, enhancing its resilience against adversarial attacks. Additionally, Slot automatically constructs the attack chain according to detected attacks with clustering algorithms, providing precise identification of attack paths and facilitating the development of defense strategies. Evaluations with real-world datasets demonstrate Slot's outstanding accuracy, efficiency, adaptability, and robustness in APT detection, with most metrics surpassing state-of-the-art methods. Additionally, case studies conducted to assess Slot's effectiveness in supporting APT defense further establish it as a practical and reliable tool for cybersecurity protection.

Recent works show that sensitive user data can be reconstructed from gradient updates, breaking the key privacy promise of federated learning. While success was demonstrated primarily on image data, these methods do not directly transfer to other domains, such as spatiotemporal data. To understand privacy risks in spatiotemporal federated learning, we first propose Spatiotemporal Gradient Inversion Attack (ST-GIA), a gradient attack algorithm tailored to spatiotemporal data that successfully reconstructs the original location from gradients. Furthermore, the absence of priors in attacks on spatiotemporal data has hindered the accurate reconstruction of real client data. To address this limitation, we propose ST-GIA+, which utilizes an auxiliary language model to guide the search for potential locations, thereby successfully reconstructing the original data from gradients. In addition, we design an adaptive defense strategy to mitigate gradient inversion attacks in spatiotemporal federated learning. By dynamically adjusting the perturbation levels, we can offer tailored protection for varying rounds of training data, thereby achieving a better trade-off between privacy and utility than current state-of-the-art methods. Through intensive experimental analysis on three real-world datasets, we reveal that the proposed defense strategy can well preserve the utility of spatiotemporal federated learning with effective security protection., Comment: arXiv admin note: substantial text overlap with arXiv:2407.08529

Current flight procedure design methods heavily rely on human-led design process, which is not only low auto-mation but also suffer from complex algorithm modelling and poor generalization. To address these challenges, this paper proposes an agent-driven flight procedure design method based on large language model, named Au-toFPDesigner, which utilizes multi-agent collaboration to complete procedure design. The method enables end-to-end automated design of performance-based navigation (PBN) procedures. In this process, the user input the design requirements in natural language, AutoFPDesigner models the flight procedure design by loading the design speci-fications and utilizing tool libraries complete the design. AutoFPDesigner allows users to oversee and seamlessly participate in the design process. Experimental results show that AutoFPDesigner ensures nearly 100% safety in the designed flight procedures and achieves 75% task completion rate, with good adaptability across different design tasks. AutoFPDesigner introduces a new paradigm for flight procedure design and represents a key step towards the automation of this process. Keywords: Flight Procedure Design; Large Language Model; Performance-Based Navigation (PBN); Multi Agent, Comment: 21 pages, 18 figures, 5 tables

Lenticular galaxies (S0s) are formed mainly from the gas stripping of spirals in the cluster. But how S0s form and evolve in the field is still untangled. Based on spatially resolved observations from the optical Hispanic Astronomical Center in Andalusia 3.5-m telescope with the PPAK Integral Field Spectroscopy instrument and NOrthern Extended Millimeter Array, we study a dwarf (M*<10^9 Msun) S0, PGC 44685, with triple star-forming regions in the central region, namely A, B, and C, respectively. In northwest region C, we clearly detect the spectral features of Wolf-Rayet (WR) stars and quantify the WR population by stacking spectra with high WR significance. Most of the molecular gas is concentrated in the region C(WR), and there is diffuse gas around regions A and B. The WR region possesses the strongest intensities of Ha, CO(1-0), and 3mm continuum, indicating its ongoing violent star formation (gas depletion timescale $\lesssim$25 Myr) with tentative hundreds (<500) km/s stellar winds accompanied by the WR phase. Most (~96%) of three star-forming regions show relatively low metallicity distributions, suggesting possible (minor) accretions of metal-poor gas that trigger the subsequent complex star formation in a field S0 galaxy. We speculate that PGC 44685 will become quiescent in less than 30 Myr if there is no new molecular gas to provide raw materials for star formation. The existence of this dwarf star-forming S0 presents an example of star formation in the low-mass/metallicity S0 galaxy., Comment: 19 pages, 12 figures, 3 tables, ApJ accepted

An important paradigm in 3D object detection is the use of multiple modalities to enhance accuracy in both normal and challenging conditions, particularly for long-tail scenarios. To address this, recent studies have explored two directions of adaptive approaches: MoE-based adaptive fusion, which struggles with uncertainties arising from distinct object configurations, and late fusion for output-level adaptive fusion, which relies on separate detection pipelines and limits comprehensive understanding. In this work, we introduce Cocoon, an object- and feature-level uncertainty-aware fusion framework. The key innovation lies in uncertainty quantification for heterogeneous representations, enabling fair comparison across modalities through the introduction of a feature aligner and a learnable surrogate ground truth, termed feature impression. We also define a training objective to ensure that their relationship provides a valid metric for uncertainty quantification. Cocoon consistently outperforms existing static and adaptive methods in both normal and challenging conditions, including those with natural and artificial corruptions. Furthermore, we show the validity and efficacy of our uncertainty metric across diverse datasets., Comment: 23 pages

In this paper, we introduce the big.LITTLE Vision Transformer, an innovative architecture aimed at achieving efficient visual recognition. This dual-transformer system is composed of two distinct blocks: the big performance block, characterized by its high capacity and substantial computational demands, and the LITTLE efficiency block, designed for speed with lower capacity. The key innovation of our approach lies in its dynamic inference mechanism. When processing an image, our system determines the importance of each token and allocates them accordingly: essential tokens are processed by the high-performance big model, while less critical tokens are handled by the more efficient little model. This selective processing significantly reduces computational load without sacrificing the overall performance of the model, as it ensures that detailed analysis is reserved for the most important information. To validate the effectiveness of our big.LITTLE Vision Transformer, we conducted comprehensive experiments on image classification and segment anything task. Our results demonstrate that the big.LITTLE architecture not only maintains high accuracy but also achieves substantial computational savings. Specifically, our approach enables the efficient handling of large-scale visual recognition tasks by dynamically balancing the trade-offs between performance and efficiency. The success of our method underscores the potential of hybrid models in optimizing both computation and performance in visual recognition tasks, paving the way for more practical and scalable deployment of advanced neural networks in real-world applications.

Relaxation to equilibrium in Bose-Hubbard rings is numerically investigated in the time domain. We show that for small mean site populations in the Mott insulator regime, characterized by strong onsite interaction as compared to hopping, equipartition starting from nonequilibrium initial states is significantly faster in these bosonic quantum many-body systems than in their classical counterparts, given by discrete Gross-Pitaevskii lattices. We attribute this phenomenon of thermalization enhancement to a genuine quantum transport mechanism akin to tunneling, allowing for pathways that are more efficient than chaotic diffusion across the ergodic domain of the classical systems., Comment: 12 pages, 6 figures

Autonomous driving necessitates the ability to reason about future interactions between traffic agents and to make informed evaluations for planning. This paper introduces the \textit{Gen-Drive} framework, which shifts from the traditional prediction and deterministic planning framework to a generation-then-evaluation planning paradigm. The framework employs a behavior diffusion model as a scene generator to produce diverse possible future scenarios, thereby enhancing the capability for joint interaction reasoning. To facilitate decision-making, we propose a scene evaluator (reward) model, trained with pairwise preference data collected through VLM assistance, thereby reducing human workload and enhancing scalability. Furthermore, we utilize an RL fine-tuning framework to improve the generation quality of the diffusion model, rendering it more effective for planning tasks. We conduct training and closed-loop planning tests on the nuPlan dataset, and the results demonstrate that employing such a generation-then-evaluation strategy outperforms other learning-based approaches. Additionally, the fine-tuned generative driving policy shows significant enhancements in planning performance. We further demonstrate that utilizing our learned reward model for evaluation or RL fine-tuning leads to better planning performance compared to relying on human-designed rewards. Project website: https://mczhi.github.io/GenDrive.

Recently, there is growing demand for effective and efficient long sequence modeling, with State Space Models (SSMs) proving to be effective for long sequence tasks. To further reduce energy consumption, SSMs can be adapted to Spiking Neural Networks (SNNs) using spiking functions. However, current spiking-formalized SSMs approaches still rely on float-point matrix-vector multiplication during inference, undermining SNNs' energy advantage. In this work, we address the efficiency and performance challenges of long sequence learning in SNNs simultaneously. First, we propose a decoupled reset method for parallel spiking neuron training, reducing the typical Leaky Integrate-and-Fire (LIF) model's training time from $O(L^2)$ to $O(L\log L)$, effectively speeding up the training by $6.57 \times$ to $16.50 \times$ on sequence lengths $1,024$ to $32,768$. To our best knowledge, this is the first time that parallel computation with a reset mechanism is implemented achieving equivalence to its sequential counterpart. Secondly, to capture long-range dependencies, we propose a Parallel Resonate and Fire (PRF) neuron, which leverages an oscillating membrane potential driven by a resonate mechanism from a differentiable reset function in the complex domain. The PRF enables efficient long sequence learning while maintaining parallel training. Finally, we demonstrate that the proposed spike-driven architecture using PRF achieves performance comparable to Structured SSMs (S4), with two orders of magnitude reduction in energy consumption, outperforming Transformer on Long Range Arena tasks., Comment: arXiv admin note: text overlap with arXiv:2208.04933 by other authors

We consider a family of variable time-stepping Dahlquist-Liniger-Nevanlinna (DLN) schemes, which is unconditional non-linear stable and second order accurate, for the Allen-Cahn equation. The finite element methods are used for the spatial discretization. For the non-linear term, we combine the DLN scheme with two efficient temporal algorithms: partially implicit modified algorithm and scalar auxiliary variable algorithm. For both approaches, we prove the unconditional, long-term stability of the model energy under any arbitrary time step sequence. Moreover, we provide rigorous error analysis for the partially implicit modified algorithm with variable time-stepping. Efficient time adaptive algorithms based on these schemes are also proposed. Several one- and two-dimensional numerical tests are presented to verify the properties of the proposed time adaptive DLN methods.

While next-token prediction is considered a promising path towards artificial general intelligence, it has struggled to excel in multimodal tasks, which are still dominated by diffusion models (e.g., Stable Diffusion) and compositional approaches (e.g., CLIP combined with LLMs). In this paper, we introduce Emu3, a new suite of state-of-the-art multimodal models trained solely with next-token prediction. By tokenizing images, text, and videos into a discrete space, we train a single transformer from scratch on a mixture of multimodal sequences. Emu3 outperforms several well-established task-specific models in both generation and perception tasks, surpassing flagship models such as SDXL and LLaVA-1.6, while eliminating the need for diffusion or compositional architectures. Emu3 is also capable of generating high-fidelity video via predicting the next token in a video sequence. We simplify complex multimodal model designs by converging on a singular focus: tokens, unlocking great potential for scaling both during training and inference. Our results demonstrate that next-token prediction is a promising path towards building general multimodal intelligence beyond language. We open-source key techniques and models to support further research in this direction., Comment: Project Page: https://emu.baai.ac.cn

Semantic communication technology emerges as a pivotal bridge connecting AI with classical communication. The current semantic communication systems are generally modeled as an Auto-Encoder (AE). AE lacks a deep integration of AI principles with communication strategies due to its inability to effectively capture channel dynamics. This gap makes it difficult to justify the need for joint source-channel coding (JSCC) and to explain why performance improves. This paper begins by exploring lossless and lossy communication, highlighting that the inclusion of data distortion distinguishes semantic communication from classical communication. It breaks the conditions for the separation theorem to hold and explains why the amount of data transferred by semantic communication is less. Therefore, employing JSCC becomes imperative for achieving optimal semantic communication. Moreover, a Variational Source-Channel Coding (VSCC) method is proposed for constructing semantic communication systems based on data distortion theory, integrating variational inference and channel characteristics. Using a deep learning network, we develop a semantic communication system employing the VSCC method and demonstrate its capability for semantic transmission. We also establish semantic communication systems of equivalent complexity employing the AE method and the VAE method. Experimental results reveal that the VSCC model offers superior interpretability compared to AE model, as it clearly captures the semantic features of the transmitted data, represented as the variance of latent variables in our experiments. In addition, VSCC model exhibits superior semantic transmission capabilities compared to VAE model. At the same level of data distortion evaluated by PSNR, VSCC model exhibits stronger human interpretability, which can be partially assessed by SSIM.

Foundation models for natural language processing, powered by the transformer architecture, exhibit remarkable in-context learning (ICL) capabilities, allowing pre-trained models to adapt to downstream tasks using few-shot prompts without updating their weights. Recently, transformer-based foundation models have also emerged as versatile tools for solving scientific problems, particularly in the realm of partial differential equations (PDEs). However, the theoretical foundations of the ICL capabilities in these scientific models remain largely unexplored. This work develops a rigorous error analysis for transformer-based ICL applied to solution operators associated with a family of linear elliptic PDEs. We first demonstrate that a linear transformer, defined by a linear self-attention layer, can provably learn in-context to invert linear systems arising from the spatial discretization of PDEs. This is achieved by deriving theoretical scaling laws for the prediction risk of the proposed linear transformers in terms of spatial discretization size, the number of training tasks, and the lengths of prompts used during training and inference. These scaling laws also enable us to establish quantitative error bounds for learning PDE solutions. Furthermore, we quantify the adaptability of the pre-trained transformer on downstream PDE tasks that experience distribution shifts in both tasks (represented by PDE coefficients) and input covariates (represented by the source term). To analyze task distribution shifts, we introduce a novel concept of task diversity and characterize the transformer's prediction error in terms of the magnitude of task shift, assuming sufficient diversity in the pre-training tasks. We also establish sufficient conditions to ensure task diversity. Finally, we validate the ICL-capabilities of transformers through extensive numerical experiments., Comment: Code available at https://github.com/ LuGroupUMN/ICL-EllipticPDEs

Simulation stands as a cornerstone for safe and efficient autonomous driving development. At its core a simulation system ought to produce realistic, reactive, and controllable traffic patterns. In this paper, we propose ProSim, a multimodal promptable closed-loop traffic simulation framework. ProSim allows the user to give a complex set of numerical, categorical or textual prompts to instruct each agent's behavior and intention. ProSim then rolls out a traffic scenario in a closed-loop manner, modeling each agent's interaction with other traffic participants. Our experiments show that ProSim achieves high prompt controllability given different user prompts, while reaching competitive performance on the Waymo Sim Agents Challenge when no prompt is given. To support research on promptable traffic simulation, we create ProSim-Instruct-520k, a multimodal prompt-scenario paired driving dataset with over 10M text prompts for over 520k real-world driving scenarios. We will release code of ProSim as well as data and labeling tools of ProSim-Instruct-520k at https://ariostgx.github.io/ProSim., Comment: Accepted to CoRL 2024. Website available at https://ariostgx.github.io/ProSim

Multi-robot swarms utilize swarm intelligence to collaborate on tasks and play an increasingly significant role in a variety of practical scenarios. However, due to the complex design, multi-robot swarm systems often have vulnerabilities caused by logical errors, which can severely disrupt the normal operations of multi-robot swarms. Despite the significant security threats that logical vulnerabilities pose to multi-robot swarms, there are still considerable challenges in testing and identifying these vulnerabilities, and related research still faces two major challenges: 1) the explosion of input space for testing, 2) the lack of effective test-guidance strategies. Therefore, in this paper, we overcome the two major challenges mentioned above, and propose a formal verification method to discover logical flaws in multi-robot swarms. Specifically, we abstract linear temporal logic constraints of the swarm and compute swarm robustness based on these constraints thus guiding fuzzing, we call this approach LiTelFuzz (Fuzzing based on Linear Temporal Logic Constraints). The core idea of LiTelFuzz is to design a metric based on behavioral constraints to assess the state of the multi-robot swarm at different moments, and guide fuzz testing based on the assessment results. Based on this idea, we overcome the two challenges of excessive test case input space and the lack of fuzzing guidance. Consequently, we implement a single attack drone fuzzing scheme and a multiple attack drones scheme based on LiTelFuzz. These are named SA-Fuzzing and MA-Fuzzing, respectively. Finally, we tested three popular swarm algorithms using LiTelFuzz with an average success rate of 87.35% for SA-Fuzzing and 91.73% for MA-Fuzzing to find vulnerabilities. The success rate and efficiency are better than the existing state-of-the-art fuzzer SWARMFLAWFINDER., Comment: Submitted to USENIX security'25

We introduce a novel method for estimating and conducting inference about extreme quantile treatment effects (QTEs) in the presence of endogeneity. Our approach is applicable to a broad range of empirical research designs, including instrumental variables design and regression discontinuity design, among others. By leveraging regular variation and subsampling, the method ensures robust performance even in extreme tails, where data may be sparse or entirely absent. Simulation studies confirm the theoretical robustness of our approach. Applying our method to assess the impact of job training provided by the Job Training Partnership Act (JTPA), we find significantly negative QTEs for the lowest quantiles (i.e., the most disadvantaged individuals), contrasting with previous literature that emphasizes positive QTEs for intermediate quantiles.

Recent advancements in text-to-image diffusion models have significantly transformed visual content generation, yet their application in specialized fields such as interior design remains underexplored. In this paper, we present RoomDiffusion, a pioneering diffusion model meticulously tailored for the interior design industry. To begin with, we build from scratch a whole data pipeline to update and evaluate data for iterative model optimization. Subsequently, techniques such as multiaspect training, multi-stage fine-tune and model fusion are applied to enhance both the visual appeal and precision of the generated results. Lastly, leveraging the latent consistency Distillation method, we distill and expedite the model for optimal efficiency. Unlike existing models optimized for general scenarios, RoomDiffusion addresses specific challenges in interior design, such as lack of fashion, high furniture duplication rate, and inaccurate style. Through our holistic human evaluation protocol with more than 20 professional human evaluators, RoomDiffusion demonstrates industry-leading performance in terms of aesthetics, accuracy, and efficiency, surpassing all existing open source models such as stable diffusion and SDXL.

The cold interstellar medium (ISM) as the raw material for star formation is critical to understanding galaxy evolution. It is generally understood that galaxies stop making stars when, in one way or another, they run out of gas. However, here we provide evidence that central spiral galaxies remain rich in atomic gas even if their star formation rate and molecular gas fraction have dropped significantly compared to "normal" star-forming galaxies of the same mass. Since HI is sensitive to external processes, here we investigate central spiral galaxies using a combined sample from SDSS, ALFALFA, and xGASS surveys. After proper incompleteness corrections, we find that the key HI scaling relations for central spirals show significant but regular systematic dependence on stellar mass. At any given stellar mass, the HI gas mass fraction is about constant with changing specific star formation rate (sSFR), which suggests that HI reservoir is ubiquitous in central spirals with any star formation status down to M* ~ 10^9 Msun. Together with the tight correlation between the molecular gas mass fraction and sSFR for galaxies across a wide range of different properties, it suggests that the decline of SFR of all central spirals in the local universe is due to the halt of H2 supply, though there is plenty of HI gas around. These hence provide critical observations of the dramatically different behavior of the cold multi-phase ISM, and a key to understand the star formation process and quenching mechanism., Comment: 18 pages, 7 figures; Accepted for publication in the ApJL; This is the fourth paper in the "From Haloes to Galaxies" series

This paper investigates a wireless network consisting of an unmanned aerial vehicle (UAV) base station (BS), a fully-connected reconfigurable intelligent surface (FC-RIS), and multiple users, where the downlink signal can simultaneously be captured by an aerial eavesdropper at a random location. To improve the physical-layer security (PLS) of the considered downlink multiuser communications, we propose the fully-connected reconfigurable intelligent surface aided round-robin scheduling (FCR-RS) and the FC-RIS and ground channel state information (CSI) aided proportional fair scheduling (FCR-GCSI-PFS) schemes. Thereafter, we derive closed-form expressions of the zero secrecy rate probability (ZSRP). Numerical results not only validate the closed-form ZSRP analysis, but also verify that the proposed GCSI-PFS scheme obtains the same performance gain as the full-CSI-aided PFS in FC-RIS-aided communications. Furthermore, optimizing the hovering altitude remarkably enhances the PLS of the FC-RIS and UAV empowered multiuser communications., Comment: submitted to IEEE Wireless Communications letters

Legal Judgment Prediction (LJP) aims to form legal judgments based on the criminal fact description. However, researchers struggle to classify confusing criminal cases, such as robbery and theft, which requires LJP models to distinguish the nuances between similar crimes. Existing methods usually design handcrafted features to pick up necessary semantic legal clues to make more accurate legal judgment predictions. In this paper, we propose a Semantic-Aware Dual Encoder Model (SEMDR), which designs a novel legal clue tracing mechanism to conduct fine-grained semantic reasoning between criminal facts and instruments. Our legal clue tracing mechanism is built from three reasoning levels: 1) Lexicon-Tracing, which aims to extract criminal facts from criminal descriptions; 2) Sentence Representation Learning, which contrastively trains language models to better represent confusing criminal facts; 3) Multi-Fact Reasoning, which builds a reasons graph to propagate semantic clues among fact nodes to capture the subtle difference among criminal facts. Our legal clue tracing mechanism helps SEMDR achieve state-of-the-art on the CAIL2018 dataset and shows its advance in few-shot scenarios. Our experiments show that SEMDR has a strong ability to learn more uniform and distinguished representations for criminal facts, which helps to make more accurate predictions on confusing criminal cases and reduces the model uncertainty during making judgments. All codes will be released via GitHub.

The impressive performance of Large Language Models (LLMs) has consistently surpassed numerous human-designed benchmarks, presenting new challenges in assessing the shortcomings of LLMs. Designing tasks and finding LLMs' limitations are becoming increasingly important. In this paper, we investigate the question of whether an LLM can discover its own limitations from the errors it makes. To this end, we propose a Self-Challenge evaluation framework with human-in-the-loop. Starting from seed instances that GPT-4 fails to answer, we prompt GPT-4 to summarize error patterns that can be used to generate new instances and incorporate human feedback on them to refine these patterns for generating more challenging data, iteratively. We end up with 8 diverse patterns, such as text manipulation and questions with assumptions. We then build a benchmark, SC-G4, consisting of 1,835 instances generated by GPT-4 using these patterns, with human-annotated gold responses. The SC-G4 serves as a challenging benchmark that allows for a detailed assessment of LLMs' abilities. Our results show that only 44.96\% of instances in SC-G4 can be answered correctly by GPT-4. Interestingly, our pilot study indicates that these error patterns also challenge other LLMs, such as Claude-3 and Llama-3, and cannot be fully resolved through fine-tuning. Our work takes the first step to demonstrate that LLMs can autonomously identify their inherent flaws and provide insights for future dynamic and automatic evaluation., Comment: COLM 2024

In this paper, we introduce a major advancement in Quantum Singular Value Transformation (QSVT) through the development of Feedforward QSVT (FQSVT), a framework that significantly enhances the efficiency and robustness of quantum algorithm design. By leveraging intermediate measurements and feedforward operations, FQSVTs reclaim quantum information typically discarded in conventional QSVT, enabling more efficient transformations. Our results show that FQSVTs can exponentially accelerate the projection of quantum states onto energy subspaces, outperforming probabilistic projection and adiabatic algorithms with superior efficiency and a drastic reduction in query complexity. In the context of superconducting qubits, FQSVTs offer a powerful tool for managing energy subspaces, improving efficiency for state preparation and leakage detection.

This paper introduces the Aquila2 series, which comprises a wide range of bilingual models with parameter sizes of 7, 34, and 70 billion. These models are trained based on an innovative framework named HeuriMentor (HM), which offers real-time insights into model convergence and enhances the training process and data management. The HM System, comprising the Adaptive Training Engine (ATE), Training State Monitor (TSM), and Data Management Unit (DMU), allows for precise monitoring of the model's training progress and enables efficient optimization of data distribution, thereby enhancing training effectiveness. Extensive evaluations show that the Aquila2 model series performs comparably well on both English and Chinese benchmarks. Specifically, Aquila2-34B demonstrates only a slight decrease in performance when quantized to Int4. Furthermore, we have made our training code (https://github.com/FlagOpen/FlagScale) and model weights (https://github.com/FlagAI-Open/Aquila2) publicly available to support ongoing research and the development of applications.

Despite significant advancements in active learning and adversarial attacks, the intersection of these two fields remains underexplored, particularly in developing robust active learning frameworks against dynamic adversarial threats. The challenge of developing robust active learning frameworks under dynamic adversarial attacks is critical, as these attacks can lead to catastrophic forgetting within the active learning cycle. This paper introduces Robust Active Learning (RoAL), a novel approach designed to address this issue by integrating Elastic Weight Consolidation (EWC) into the active learning process. Our contributions are threefold: First, we propose a new dynamic adversarial attack that poses significant threats to active learning frameworks. Second, we introduce a novel method that combines EWC with active learning to mitigate catastrophic forgetting caused by dynamic adversarial attacks. Finally, we conduct extensive experimental evaluations to demonstrate the efficacy of our approach. The results show that RoAL not only effectively counters dynamic adversarial threats but also significantly reduces the impact of catastrophic forgetting, thereby enhancing the robustness and performance of active learning systems in adversarial environments.

Large Language Models (LLMs) have shown promising potentials in program generation and no-code automation. However, LLMs are prone to generate hallucinations, i.e., they generate text which sounds plausible but is incorrect. Although there has been a recent surge in research on LLM hallucinations for text generation, similar hallucination phenomenon can happen in code generation. Sometimes the generated code can have syntactical or logical errors as well as more advanced issues like security vulnerabilities, memory leaks, etc. Given the wide adaptation of LLMs to enhance efficiency in code generation and development in general, it becomes imperative to investigate hallucinations in code generation. To the best of our knowledge, this is the first attempt at studying hallucinations in the code generated by LLMs. We start by introducing the code hallucination definition and a comprehensive taxonomy of code hallucination types. We propose the first benchmark CodeMirage dataset for code hallucinations. The benchmark contains 1,137 GPT-3.5 generated hallucinated code snippets for Python programming problems from two base datasets - HumanEval and MBPP. We then propose the methodology for code hallucination detection and experiment with open source LLMs such as CodeLLaMA as well as OpenAI's GPT-3.5 and GPT-4 models using one-shot prompt. We find that GPT-4 performs the best on HumanEval dataset and gives comparable results to the fine-tuned CodeBERT baseline on MBPP dataset. Towards the end, we discuss various mitigation strategies for code hallucinations and conclude our work., Comment: Accepted at AutoMates @ IJCAI 2024