arXiv subjects

Qian Liu

Publications and source records attributed to Qian Liu.

At least 181 records · Page 10Linked to original sources

arXiv2024

Purifying Large Language Models by Ensembling a Small Language Model

The emerging success of large language models (LLMs) heavily relies on collecting abundant training data from external (untrusted) sources. Despite substantial efforts devoted to data cleaning and curation, well-constructed LLMs have been reported to suffer from copyright infringement, data poisoning, and/or privacy violations, which would impede practical deployment of LLMs. In this study, we propose a simple and easily implementable method for purifying LLMs from the negative effects caused by uncurated data, namely, through ensembling LLMs with benign and small language models (SLMs). Aside from theoretical guarantees, we perform comprehensive experiments to empirically confirm the efficacy of ensembling LLMs with SLMs, which can effectively preserve the performance of LLMs while mitigating issues such as copyright infringement, data poisoning, and privacy violations.

cs.CL↗

arXiv2024

OctoPack: Instruction Tuning Code Large Language Models

Finetuning large language models (LLMs) on instructions leads to vast performance improvements on natural language tasks. We apply instruction tuning using code, leveraging the natural structure of Git commits, which pair code changes with human instructions. We compile CommitPack: 4 terabytes of Git commits across 350 programming languages. We benchmark CommitPack against other natural and synthetic code instructions (xP3x, Self-Instruct, OASST) on the 16B parameter StarCoder model, and achieve state-of-the-art performance among models not trained on OpenAI outputs, on the HumanEval Python benchmark (46.2% pass@1). We further introduce HumanEvalPack, expanding the HumanEval benchmark to a total of 3 coding tasks (Code Repair, Code Explanation, Code Synthesis) across 6 languages (Python, JavaScript, Java, Go, C++, Rust). Our models, OctoCoder and OctoGeeX, achieve the best performance across HumanEvalPack among all permissive models, demonstrating CommitPack's benefits in generalizing to a wider set of languages and natural coding tasks. Code, models and data are freely available at https://github.com/bigcode-project/octopack.

cs.CL↗

arXiv2024

Test-Time Backdoor Attacks on Multimodal Large Language Models

Backdoor attacks are commonly executed by contaminating training data, such that a trigger can activate predetermined harmful effects during the test phase. In this work, we present AnyDoor, a test-time backdoor attack against multimodal large language models (MLLMs), which involves injecting the backdoor into the textual modality using adversarial test images (sharing the same universal perturbation), without requiring access to or modification of the training data. AnyDoor employs similar techniques used in universal adversarial attacks, but distinguishes itself by its ability to decouple the timing of setup and activation of harmful effects. In our experiments, we validate the effectiveness of AnyDoor against popular MLLMs such as LLaVA-1.5, MiniGPT-4, InstructBLIP, and BLIP-2, as well as provide comprehensive ablation studies. Notably, because the backdoor is injected by a universal perturbation, AnyDoor can dynamically change its backdoor trigger prompts/harmful effects, exposing a new challenge for defending against backdoor attacks. Our project page is available at https://sail-sg.github.io/AnyDoor/.

cs.CL↗

arXiv2024

UniVision: A Unified Framework for Vision-Centric 3D Perception

The past few years have witnessed the rapid development of vision-centric 3D perception in autonomous driving. Although the 3D perception models share many structural and conceptual similarities, there still exist gaps in their feature representations, data formats, and objectives, posing challenges for unified and efficient 3D perception framework design. In this paper, we present UniVision, a simple and efficient framework that unifies two major tasks in vision-centric 3D perception, \ie, occupancy prediction and object detection. Specifically, we propose an explicit-implicit view transform module for complementary 2D-3D feature transformation. We propose a local-global feature extraction and fusion module for efficient and adaptive voxel and BEV feature extraction, enhancement, and interaction. Further, we propose a joint occupancy-detection data augmentation strategy and a progressive loss weight adjustment strategy which enables the efficiency and stability of the multi-task framework training. We conduct extensive experiments for different perception tasks on four public benchmarks, including nuScenes LiDAR segmentation, nuScenes detection, OpenOccupancy, and Occ3D. UniVision achieves state-of-the-art results with +1.5 mIoU, +1.8 NDS, +1.5 mIoU, and +1.8 mIoU gains on each benchmark, respectively. We believe that the UniVision framework can serve as a high-performance baseline for the unified vision-centric 3D perception task. The code will be available at \url{https://github.com/Cc-Hy/UniVision}.

cs.CV↗

arXiv2024

End-to-End Learning for SLP-Based ISAC Systems

Integrated sensing and communication (ISAC) is an encouraging wireless technology which can simultaneously perform both radar and communication functionalities by sharing the same transmit waveform, spectral resource, and hardware platform. Recently emerged symbol-level precoding (SLP) technique exhibits advancement in ISAC systems by leveraging the waveform design degrees of freedom (DoFs) in both temporal and spatial domains. However, traditional SLP-based ISAC systems are designed in a modular paradigm, which potentially limits the overall performance of communication and radar sensing. The high complexity of existing SLP design algorithms is another issue that hurdles the practical deployment. To break through the bottleneck of these approaches, in this paper we propose an end-to-end approach to jointly design the SLP-based dual-functional transmitter and receivers of communication and radar sensing. In particular, we aim to utilize deep learning-based methods to minimize the symbol error rate (SER) of communication users, maximize the detection probability, and minimize the root mean square error (RMSE) of the target angle estimation. Multi-layer perceptron (MLP) networks and a long short term memory (LSTM) network are respectively applied to the transmitter, communication users and radar receiver. Simulation results verify the feasibility of the proposed deep-learning-based end-to-end optimization for ISAC systems and reveal the effectiveness of the proposed neural networks for the end-to-end design.

eess.SP↗

arXiv2024

Comparing discriminating abilities of evaluation metrics in link prediction

Link prediction aims to predict the potential existence of links between two unconnected nodes within a network based on the known topological characteristics. Evaluation metrics are used to assess the effectiveness of algorithms in link prediction. The discriminating ability of these evaluation metrics is vitally important for accurately evaluating link prediction algorithms. In this study, we propose an artificial network model, based on which one can adjust a single parameter to monotonically and continuously turn the prediction accuracy of the specifically designed link prediction algorithm. Building upon this foundation, we show a framework to depict the effectiveness of evaluating metrics by focusing on their discriminating ability. Specifically, a quantitative comparison in the abilities of correctly discerning varying prediction accuracies was conducted encompassing nine evaluation metrics: Precision, Recall, F1-Measure, Matthews Correlation Coefficient (MCC), Balanced Precision (BP), the Area Under the receiver operating characteristic Curve (AUC), the Area Under the Precision-Recall curve (AUPR), Normalized Discounted Cumulative Gain (NDCG), and the Area Under the magnified ROC (AUC-mROC). The results indicate that the discriminating abilities of the three metrics, AUC, AUPR, and NDCG, are significantly higher than those of other metrics.

cs.SI↗

arXiv2024

A Practical Beamforming Design for Active RIS-assisted MU-MISO Systems

Reconfigurable Intelligent Surfaces (RIS) have been proposed as a revolutionary technology with the potential to address several critical requirements of 6G communication systems. Despite its powerful ability for radio environment reconfiguration, the ``double fading'' effect constricts the practical system performance enhancements due to the significant path loss. A new active RIS architecture has been recently proposed to overcome this challenge. However, existing active RIS studies rely on an ideal amplification model without considering the practical hardware limitation of amplifiers, which may cause performance degradation using such inaccurate active RIS modeling. Motivated by this fact, in this paper we first investigate the amplification principle of typical active RIS and propose a more accurate amplification model based on amplifier hardware characteristics. Then, based on the new amplification model, we propose a novel joint transmit beamforming and RIS reflection beamforming design considering the incident signal power on practical active RIS for multiuser multi-input single-output (MU-MISO) communication system. Fractional programming (FP), majorization minimization (MM) and block coordinate descent (BCD) methods are used to solve for the complex problem. Simulation results indicate the importance of the consideration of practical amplifier hardware characteristics in the joint beamforming designs and demonstrate the effectiveness of the proposed algorithm compared to other benchmarks.

eess.SP↗

arXiv2024

Cooperative Cell-Free ISAC Networks: Joint BS Mode Selection and Beamforming Design

Owing to the promising ability of saving hardware cost and spectrum resources, integrated sensing and communication (ISAC) is regarded as a revolutionary technology for future sixth-generation (6G) networks. The mono-static ISAC systems considered in most of existing works can only achieve limited sensing performance due to the single observation angle and easily blocked transmission links, which motivates researchers to investigate cooperative ISAC networks. In order to further improve the degrees of freedom (DoFs) of cooperative ISAC networks, the transmitter-receiver selection, i.e., base station (BS) mode selection problem, is meaningful to be studied. However, to our best knowledge, this crucial problem has not been extensively studied in existing works. In this paper, we consider the joint BS mode selection, transmit beamforming, and receive filter designs for cooperative cell-free ISAC networks, where multi-BSs cooperatively serve communication users and detect targets. An efficient joint beamforming design algorithm and three different heuristic BS mode selection methods are proposed to solve the non-convex NP-hard problem. Simulation results demonstrates the advantages of cooperative ISAC networks, the importance of BS mode selection, and the effectiveness of proposed algorithms.

cs.IT↗

arXiv2024

Astraios: Parameter-Efficient Instruction Tuning Code Large Language Models

The high cost of full-parameter fine-tuning (FFT) of Large Language Models (LLMs) has led to a series of parameter-efficient fine-tuning (PEFT) methods. However, it remains unclear which methods provide the best cost-performance trade-off at different model scales. We introduce Astraios, a suite of 28 instruction-tuned OctoCoder models using 7 tuning methods and 4 model sizes up to 16 billion parameters. Through investigations across 5 tasks and 8 different datasets encompassing both code comprehension and code generation tasks, we find that FFT generally leads to the best downstream performance across all scales, and PEFT methods differ significantly in their efficacy based on the model scale. LoRA usually offers the most favorable trade-off between cost and performance. Further investigation into the effects of these methods on both model robustness and code security reveals that larger models tend to demonstrate reduced robustness and less security. At last, we explore the relationships among updated parameters, cross-entropy loss, and task performance. We find that the tuning effectiveness observed in small models generalizes well to larger models, and the validation loss in instruction tuning can be a reliable indicator of overall downstream performance.

cs.CL↗

arXiv2023

Toward Spatial Temporal Consistency of Joint Visual Tactile Perception in VR Applications

With the development of VR technology, especially the emergence of the metaverse concept, the integration of visual and tactile perception has become an expected experience in human-machine interaction. Therefore, achieving spatial-temporal consistency of visual and tactile information in VR applications has become a necessary factor for realizing this experience. The state-of-the-art vibrotactile datasets generally contain temporal-level vibrotactile information collected by randomly sliding on the surface of an object, along with the corresponding image of the material/texture. However, they lack the position/spatial information that corresponds to the signal acquisition, making it difficult to achieve spatiotemporal alignment of visual-tactile data. Therefore, we develop a new data acquisition system in this paper which can collect visual and vibrotactile signals of different textures/materials with spatial and temporal consistency. In addition, we develop a VR-based application call "V-Touching" by leveraging the dataset generated by the new acquisition system, which can provide pixel-to-taxel joint visual-tactile perception when sliding over the surface of objects in the virtual environment with distinct vibrotactile feedback of different textures/materials.

cs.RO↗

arXiv2023

Sparsity Exploitation via Joint Receive Processing and Transmit Beamforming Design for MIMO-OFDM ISAC Systems

Integrated sensing and communication (ISAC) is widely recognized as a pivotal enabling technique for the advancement of future wireless networks. This paper aims to efficiently exploit the inherent sparsity of echo signals for the multi-input-multi-output (MIMO) orthogonal frequency division multiplexing (OFDM) based ISAC system. A novel joint receive echo processing and transmit beamforming design is presented to achieve this goal. Specifically, we first propose a compressive sensing (CS)-assisted estimation approach to facilitate ISAC receive echo processing, which can not only enable accurate recovery of target information, but also allow substantial reduction in the number of sensing subcarriers to be sampled and processed. Then, based on the proposed CS-assisted processing method, the associated transmit beamforming design is formulated with the objective of maximizing the sum-rate of multiuser communications while satisfying the transmit power budget and ensuring the received signal-to-noise ratio (SNR) for the designated sensing subcarriers. In order to address the formulated non-convex problem involving high-dimensional variables, an effective iterative algorithm employing majorization minimization (MM), fractional programming (FP), and the nonlinear equality alternative direction method of multipliers (neADMM) with closed-form solutions has been developed. Finally, extensive numerical simulations are conducted to verify the effectiveness of the proposed algorithm and the superior performance of the introduced sparsity exploitation strategy.

cs.IT↗

arXiv2023

StarCoder: may the source be with you!

The BigCode community, an open-scientific collaboration working on the responsible development of Large Language Models for Code (Code LLMs), introduces StarCoder and StarCoderBase: 15.5B parameter models with 8K context length, infilling capabilities and fast large-batch inference enabled by multi-query attention. StarCoderBase is trained on 1 trillion tokens sourced from The Stack, a large collection of permissively licensed GitHub repositories with inspection tools and an opt-out process. We fine-tuned StarCoderBase on 35B Python tokens, resulting in the creation of StarCoder. We perform the most comprehensive evaluation of Code LLMs to date and show that StarCoderBase outperforms every open Code LLM that supports multiple programming languages and matches or outperforms the OpenAI code-cushman-001 model. Furthermore, StarCoder outperforms every model that is fine-tuned on Python, can be prompted to achieve 40\% pass@1 on HumanEval, and still retains its performance on other programming languages. We take several important steps towards a safe open-access model release, including an improved PII redaction pipeline and a novel attribution tracing tool, and make the StarCoder models publicly available under a more commercially viable version of the Open Responsible AI Model license.

cs.CL↗

arXiv2023

Joint Sensing and Communication Optimization in Target-Mounted STARS-Assisted Vehicular Networks: A MADRL Approach

The utilization of integrated sensing and communication (ISAC) technology has the potential to enhance the communication performance of road side units (RSUs) through the active sensing of target vehicles. Furthermore, installing a simultaneous transmitting and reflecting surface (STARS) on the target vehicle can provide an extra boost to the reflection of the echo signal, thereby improving the communication quality for in-vehicle users. However, the design of this target-mounted STARS system exhibits significant challenges, such as limited information sharing and distributed STARS control. In this paper, we propose an end-to-end multi-agent deep reinforcement learning (MADRL) framework to tackle the challenges of joint sensing and communication optimization in the considered target-mounted STARS assisted vehicle networks. By deploying agents on both RSU and vehicle, the MADRL framework enables RSU and vehicle to perform beam prediction and STARS pre-configuration using their respective local information. To ensure efficient and stable learning for continuous decision-making, we employ the multi-agent soft actor critic (MASAC) algorithm and the multi-agent proximal policy optimization (MAPPO) algorithm on the proposed MADRL framework. Extensive experimental results confirm the effectiveness of our proposed MADRL framework in improving both sensing and communication performance through the utilization of target-mounted STARS. Finally, we conduct a comparative analysis and comparison of the two proposed algorithms under various environmental conditions.

eess.SP↗

arXiv2023

SCITAB: A Challenging Benchmark for Compositional Reasoning and Claim Verification on Scientific Tables

Current scientific fact-checking benchmarks exhibit several shortcomings, such as biases arising from crowd-sourced claims and an over-reliance on text-based evidence. We present SCITAB, a challenging evaluation dataset consisting of 1.2K expert-verified scientific claims that 1) originate from authentic scientific publications and 2) require compositional reasoning for verification. The claims are paired with evidence-containing scientific tables annotated with labels. Through extensive evaluations, we demonstrate that SCITAB poses a significant challenge to state-of-the-art models, including table-based pretraining models and large language models. All models except GPT-4 achieved performance barely above random guessing. Popular prompting techniques, such as Chain-of-Thought, do not achieve much performance gains on SCITAB. Our analysis uncovers several unique challenges posed by SCITAB, including table grounding, claim ambiguity, and compositional reasoning. Our codes and data are publicly available at https://github.com/XinyuanLu00/SciTab.

cs.CL↗

arXiv2023

Low-Range-Sidelobe Waveform Design for MIMO-OFDM ISAC Systems

Integrated sensing and communication (ISAC) is a promising technology in future wireless systems owing to its efficient hardware and spectrum utilization. In this paper, we consider a multi-input multi-output (MIMO) orthogonal frequency division multiplexing (OFDM) ISAC system and propose a novel waveform design to provide better radar ranging performance by taking range sidelobe suppression into consideration. In specific, we aim to design MIMO-OFDM dual-function waveform to minimize its integrated sidelobe level (ISL) while satisfying the quality of service (QoS) requirements of multi-user communications and the transmit power constraint. To achieve a lower ISL, the symbol-level precoding (SLP) technique is employed to fully exploit the degrees of freedom (DoFs) of the waveform design in both temporal and spatial domains. An efficient algorithm utilizing majorization-minimization (MM) framework is developed to solve the non-convex waveform design problem. Simulation results reveal radar ranging performance improvement and demonstrate the benefits of the proposed SLP-based low-range-sidelobe waveform design in ISAC systems.

eess.SP↗

arXiv2023

Active Retrieval Augmented Generation

Despite the remarkable ability of large language models (LMs) to comprehend and generate language, they have a tendency to hallucinate and create factually inaccurate output. Augmenting LMs by retrieving information from external knowledge resources is one promising solution. Most existing retrieval augmented LMs employ a retrieve-and-generate setup that only retrieves information once based on the input. This is limiting, however, in more general scenarios involving generation of long texts, where continually gathering information throughout generation is essential. In this work, we provide a generalized view of active retrieval augmented generation, methods that actively decide when and what to retrieve across the course of the generation. We propose Forward-Looking Active REtrieval augmented generation (FLARE), a generic method which iteratively uses a prediction of the upcoming sentence to anticipate future content, which is then utilized as a query to retrieve relevant documents to regenerate the sentence if it contains low-confidence tokens. We test FLARE along with baselines comprehensively over 4 long-form knowledge-intensive generation tasks/datasets. FLARE achieves superior or competitive performance on all tasks, demonstrating the effectiveness of our method. Code and datasets are available at https://github.com/jzbjyb/FLARE.

cs.CL↗

arXiv2023

OpenAgents: An Open Platform for Language Agents in the Wild

Language agents show potential in being capable of utilizing natural language for varied and intricate tasks in diverse environments, particularly when built upon large language models (LLMs). Current language agent frameworks aim to facilitate the construction of proof-of-concept language agents while neglecting the non-expert user access to agents and paying little attention to application-level designs. We present OpenAgents, an open platform for using and hosting language agents in the wild of everyday life. OpenAgents includes three agents: (1) Data Agent for data analysis with Python/SQL and data tools; (2) Plugins Agent with 200+ daily API tools; (3) Web Agent for autonomous web browsing. OpenAgents enables general users to interact with agent functionalities through a web user interface optimized for swift responses and common failures while offering developers and researchers a seamless deployment experience on local setups, providing a foundation for crafting innovative language agents and facilitating real-world evaluations. We elucidate the challenges and opportunities, aspiring to set a foundation for future research and development of real-world language agents.

cs.CL↗

arXiv2023

Approximation algorithms for $k$-submodular maximization subject to a knapsack constraint

In this paper, we study the problem of maximizing $k$-submodular functions subject to a knapsack constraint. For monotone objective functions, we present a $\frac{1}{2}(1-e^{-2})\approx 0.432$ greedy approximation algorithm. For the non-monotone case, we are the first to consider the knapsack problem and provide a greedy-type combinatorial algorithm with approximation ratio $\frac{1}{3}(1-e^{-3})\approx 0.317$.

cs.DS↗